W E AT H E R : A C O N C I S E I N T R O D U C T I O N

From a world-renowned team at the Department of Atmospheric Sciences at

the University of Washington, Seattle, Weather: A Concise Introduction is an
accessible and beautifully illustrated text covering the foundations of
meteorology in a concise, clear, and engaging manner. Designed to provide
students with a strong foundation in the physical, dynamical, and chemical
processes taking place in the atmosphere, this introductory textbook will
appeal to students with a wide range of mathematical and scientific
backgrounds.
This textbook provides a practical approach to the study of meteorology. It
features: a single case study of a midlatitude cyclone which is referred to
throughout the whole book to illustrate the basic principles driving
atmospheric dynamics and phenomena; boxes on more advanced topics;
appendices for additional coverage; chapter summaries listing the “take-
home” points discussed; and color figures and charts clearly illustrating the
fundamental concepts. Key terms are evident throughout, and a glossary
explains the terms that students will need to understand and become familiar
with.

Gregory J. Hakim has undergraduate degrees in Mathematics and

Atmospheric Science and a PhD in Atmospheric Science from the University
at Albany, State University of New York. He joined the Department of
Atmospheric Sciences at the University of Washington in 1999, where he
served as Department Chair from 2012 to 2017 and is currently a Professor.
He is also a leading scientist in the areas of weather analysis, predictability,
and dynamics, and his research interests include weather and climate
prediction, hurricanes, past climates, and polar circulation patterns.

He has served on the advisory panel for the Directorate of Geosciences at the
National Science Foundation, as Chair of the advisory panel for the
Mesoscale and Microscale Meteorology Laboratory at the National Center for
Atmospheric Research (NCAR), as a member of the NCAR Advisory Panel,
as a member of the NCAR Strategic Planning Council, and as Chair of the
University Corporation for Atmospheric Research’s President’s Advisory
Committee on University Relations.

Jérôme Patoux earned a Master in Environmental Engineering from the

University of Texas at Austin and a PhD in Atmospheric Science from the
University of Washington. He has been funded by the National Science
Foundation (NSF), the National Aeronautics and Space Administration
(NASA), the Office of Naval Research (ONR), and the National Oceanic and
Atmospheric Administration (NOAA). He has taught undergraduate
introductory meteorology for many years, and has been funded by the NSF to
develop weather and climate curriculum. He is a former faculty member from
the Department of Atmospheric Sciences at the University of Washington,
and currently teaches meteorology at the University of Nantes in France.
 WEATHER
A Concise Introduction
GREGORY HAKIM
University of Washington

JÉRÔME PATOUX
University of Washington
 University Printing House, Cambridge CB2 8BS, United Kingdom

One Liberty Plaza, 20th Floor, New York, NY 10006, USA

477 Williamstown Road, Port Melbourne, VIC 3207, Australia

314–321, 3rd Floor, Plot 3, Splendor Forum, Jasola District Centre, New Delhi –
110025, India

79 Anson Road, #06–04/06, Singapore 079906

Cambridge University Press is part of the University of Cambridge.

It furthers the University’s mission by disseminating knowledge in the pursuit of

education, learning, and research at the highest international levels of excellence.

[Link]

Information on this title: [Link]/9781108417167

DOI: 10.1017/9781108264983

© Gregory Hakim and Jérôme Patoux 2018

This publication is in copyright. Subject to statutory exception and to the

provisions of relevant collective licensing agreements, no reproduction of any
part may take place without the written permission of Cambridge University
Press.

First published 2018

Printed in the United States of America by Sheridan Books, Inc.

 A catalogue record for this publication is available from the British Library.

ISBN 978-1-108-41716-7 Hardback

ISBN 978-1-108-40465-5 Paperback

Cambridge University Press has no responsibility for the persistence or accuracy of

URLs for external or third-party internet websites referred to in this publication and
does not guarantee that any content on such websites is, or will remain, accurate or
appropriate.
Contents
Preface
Introduction

CHAPTER 1 Weather Variables

1.1 Temperature
1.1.1 Heat and Temperature
1.1.2 Thermometers
1.1.3 Temperature Measurements
1.1.4 Temperature Scales
1.1.5 Radiosonde Profiles

1.2 Pressure
1.2.1 Force and Pressure
1.2.2 Atmospheric Pressure
1.2.3 Vertical Distribution of Pressure
1.2.4 Barometers
1.2.5 Pressure Units
1.2.6 Some Useful Numbers

1.3 Wind
1.3.1 Measuring Wind
1.3.2 Reporting Wind
1.3.3 Additional Sources of Wind Information

1.4 Precipitation
1.5 Weather Stations
Summary

CHAPTER 2 Spatial Representations of Weather Data

2.1 The Station Model

2.2 Surface Maps

2.2.1 Isotherms and Temperature Maps
2.2.2 Temperature Fronts
2.2.3 Isobars and Pressure Maps
2.2.4 Highs, Lows, Ridges, and Troughs

2.3 Upper-Level Maps

2.4 Radar

2.5 Satellites
2.5.1 Visible Satellite Images
2.5.2 Infrared Satellite Images
2.5.3 Water Vapor Images
2.5.4 Geostationary Satellites
2.5.5 Polar-Orbiting Satellites
Summary

Appendix 2.1 Important Satellite Cloud Signatures

Appendix 2.2 Contiguous USA Reference Map

CHAPTER 3 Our Atmosphere: Origin, Composition, and Structure

3.1 Aspect

3.2 Composition

3.3 Origin and Evolution

3.4 Future Evolution

3.5 Vertical Structure

Summary

Appendix 3.1 Dynamic Equilibrium

CHAPTER 4 Heat and Energy Transfer

4.1 Conduction

4.2 Convection

4.3 Radiation
4.3.1 The Nature of Electromagnetic Radiation
4.3.2 Temperature and Radiation

4.4 Radiative Interactions

4.4.1 Absorption
4.4.2 Reflection
4.4.3 Scattering
4.4.4 Radiative Equilibrium
4.4.5 Selective Absorbers
4.4.6 A Window to the Sky
4.4.7 The Greenhouse Effect
4.5 Radiation and Weather
4.5.1 Heat Imbalance
4.5.2 Seasonal Variations
4.5.3 Diurnal Variations
4.5.4 The Influence of Clouds
4.5.5 Land–Ocean Contrasts
Summary

CHAPTER 5 Water

5.1 The Water Cycle

5.2 Saturation

5.3 Humidity

5.4 Relative Humidity

5.5 Humidity and Temperature

5.5.1 Relative vs. Absolute Humidity
5.5.2 Condensation

5.6 Dew Point Temperature

5.7 Applications of the Dew point Temperature

5.7.1 Surface Weather Maps
5.7.2 Meteograms
5.7.3 Radiosonde Profiles
5.7.4 Back to Relative Humidity
5.7.5 How to Saturate
 Summary

CHAPTER 6 Cloud Formation

6.1 Adiabatic Processes

6.2 Adiabatic Processes in the Atmosphere

6.3 Dry Adiabatic Lapse Rate

6.4 Relative Humidity

6.5 Moist Adiabatic Lapse Rate

6.6 Orographic Lifting

6.7 Lifting by Convergence

6.8 Frontal Lifting

6.9 Convection
6.9.1 Stable Air
6.9.2 Unstable Air and Thermals
6.9.3 Stable vs. Unstable
6.9.4 Fair-Weather Cumulus Clouds
6.9.5 Conditional Instability and Cumulonimbus
Summary

Appendix 6.1 A Cloud Family Album

CHAPTER 7 Precipitation
7.1 Warm vs. Cold Clouds

7.2 Collision and Coalescence

7.3 Ice-Crystal Growth

7.4 Precipitation Types

Summary

Appendix 7.1 Some Optical Phenomena

CHAPTER 8 Wind

8.1 Force and Acceleration

8.2 Pressure Gradient Force

8.3 Sea Breeze and Land Breeze

8.4 Coriolis Force

8.5 Geostrophic Wind

8.6 Gradient Wind

8.7 Surface Winds

8.8 Friction

8.9 Topography
8.9.1 Mountain Breeze and Valley Breeze
8.9.2 Katabatic Winds
 Summary

CHAPTER 9 Global Wind Systems

9.1 The Averaged Atmosphere

9.1.1 Surface Temperature
9.1.2 Upper-Level Heights
9.1.3 Surface Pressure
9.1.4 Precipitation

9.2 The Single-Cell Model

9.3 The Three-Cell Model

9.4 Some Large-Scale Circulations

9.4.1 West Coast vs. East Coast
9.4.2 Antarctica
9.4.3 The Sahel
9.4.4 The Indian Monsoon
9.4.5 El Niño
Summary

CHAPTER 10 Air Masses, Fronts, and Midlatitude Cyclones

10.1 Air Masses

10.2 Fronts
10.2.1 Stationary Fronts
10.2.2 Cold Fronts
10.2.3 Warm Fronts
 10.2.4 Occluded Fronts
10.2.5 Large-Scale Influences on Cyclone Structure, and the T-
bone Model

10.3 Midlatitude Cyclone Development

10.3.1 The Life Cycle of a Midlatitude Cyclone
10.3.2 Vertical Structure of Cyclones
10.3.3 The February 2014 Cyclone
10.3.4 Where do Cyclones Form?
Summary

Appendix 10.1 Southern Hemisphere Midlatitude Cyclones

Appendix 10.2 The Bergen School of Meteorology

CHAPTER 11 Thunderstorms and Tornadoes

11.1 Ordinary Thunderstorm

11.2 Severe Thunderstorm

11.3 Lightning and Thunder

11.4 Supercells

11.5 Tornadoes
11.5.1 Description
11.5.2 Tornado Development
11.5.3 Tornado Alley
Summary
CHAPTER 12 Tropical Cyclones

12.1 Facts and Figures

12.2 Tropical Cyclone Structure

12.3 Tropical Cyclone Development

12.3.1 Tropical Easterly Wave
12.3.2 Tropical Depression
12.3.3 Tropical Storm
12.3.4 Tropical Cyclone (Hurricane)
12.3.5 Tropical Cyclone Decay

12.4 Conditions for Tropical Cyclone Development

Summary

CHAPTER 13 Weather Forecasting

13.1 Weather Forecasts and Uncertainty

13.2 Prognostic Equations

13.3 Ensemble Forecasting

13.4 Chaos and Weather Prediction

13.5 From Forecast Grids to Reliable Forecast Values

13.6 Making a Forecast

13.6.1 Medium to Long-Range Forecasting
13.6.2 Seasonal Outlook
 Summary

CHAPTER 14 Air Pollution

14.1 Pollutants
14.1.1 Gases and Compounds
14.1.2 Particulates
14.1.3 Photochemical Smog

14.2 Wind and Stability

14.3 Large-Scale Patterns

14.4 Topography
Summary

CHAPTER 15 Climate Change and Weather

15.1 Past and Future

15.2 Changing Composition

15.3 A Warmer World

15.4 An Altered Water Cycle

15.5 Changing Global Wind Systems

15.6 Midlatitude and Tropical Cyclones in a Warmer World

15.7 Beyond Weather

15.8 The Forecast

 Summary

Glossary

References

Credits

Index
Preface
Having taught introductory classes on weather many times, we came to see
the need for a textbook on the subject that covers the foundations of
meteorology in a concise, clear, and engaging manner. We set out to create an
informative, cost-effective text that meets the needs of students who may not
have any background in mathematics and science. The result – Weather: A
Concise Introduction – is an introductory meteorology textbook designed
from scratch to provide students with a strong foundation in the physical,
dynamical, and chemical processes taking place in the atmosphere.
This textbook is unique in that it:

► provides a concise and practical approach to understanding the

atmosphere;
► introduces the basic physical laws early on and then ties them together
with a single case study spanning the book;
► presents weather analysis tools early in the book to allow instructors to
engage in discussions of current weather in tandem with the basic
concepts, thus attracting and retaining student interest; and
► facilitates students’ learning and understanding of the fundamental
aspects of weather analysis and forecasting, as well as practical skills,
through a careful description of the forecasting process. Modern
methods, such as ensemble forecasting, are central to the approach.
 Features

Case Study: February 2014 Cyclone

The main concepts of the book are illustrated in Chapters 2–13 by a single
case study: a midlatitude cyclone that swept through the eastern half of the
USA between February 19 and 22, 2014. This rich case study serves as a
common thread throughout the book, allowing students to study it from
multiple perspectives. Viewing the storm in the context of different topics
provides a familiar setting for mastering new subjects and for developing an
holistic understanding of midlatitude cyclones.
 Boxes on More Advanced Topics
Instructors have the option of including more advanced coverage through use
of boxes that provide insights on various topics. For example, in Chapter 1,
Weather Variables, boxes include an in-depth description of the four laws of
physics that are central to the study of the atmosphere. The book contains 25
boxes, affording instructors the opportunity to tailor the level of the material
that they present to students in their course.
 Appendixes for Additional Coverage
Appendixes at the ends of Chapters 2, 3, 6, 7, and 10 include additional
material on important cloud signatures found in satellite imagery, the concept
of dynamic equilibrium, the cloud classification, some optical phenomena,
southern hemisphere midlatitude cyclones, and the Bergen School of
meteorology.
 Summary
A summary of key points has been included at the end of each chapter so that
students can, at a glance, confirm that they have understood the significant
take-away facts and ideas.
 Figures, Charts, and Maps
Figures have been designed to convey the key concepts in a simple and self-
explanatory way, keeping in mind that clean representations of information
are more helpful to students than complex drawings. Graphs and maps have
been created with real data as much as possible, obtained from NOAA,
NASA, ECMWF, and similar research-quality sources referenced in the text.
 Key Terms and Glossary
The main text contains terms (in bold) that students need to understand and
become familiar with. Many of these terms are listed in the Glossary at the
back of the book. The Glossary allows the reader to look up terms easily
whenever needed and can also be used to review important topics and key
facts.
 SI Units
We have consistently used SI units throughout the book, while providing
alternative units whenever possible or relevant.
 Organization
The first two chapters provide a general overview of key variables and
weather maps used by meteorologists, which facilitates daily weather map
discussions early in the course. We have found that motivating lecture topics
with real-time examples using weather map discussions is a very effective
way to engage students in the lecture material, and it allows instructors to
introduce aspects of weather forecasting at their discretion well in advance of
discussing the material more completely in Chapter 13. As a result, students
are more invested in adding to their knowledge, which builds systematically
toward understanding and predicting weather systems.
Chapters 3–8 provide foundational material on the composition and
structure of the atmosphere, along with the application of the laws of classical
physics to emphasize and explain the role of energy, water, and wind in
weather systems.
Chapters 9–12 apply the foundational material to understanding the
general circulation of the atmosphere (Chapter 9), midlatitude cyclones and
fronts (Chapter 10), thunderstorms (Chapter 11), and tropical cyclones
(Chapter 12).
Chapters 13–15 build further on the first twelve chapters by applying the
concepts developed to explain processes that affect how weather forecasts are
made (Chapter 13), air pollution (Chapter 14), and climate change (Chapter
15).
 Instructor Resources
A companion website at [Link]/weather contains PowerPoint
slides of the figures in the text as well as a testbank of questions.
 Acknowledgments
We thank: NOAA, NASA, and ECMWF for providing access to data and
images; Reto Knutti, Jan Sedlacek, and Urs Beyerle for providing access to
IPCC data; Rick Kohrs from the University of Wisconsin-Madison for
providing global composite satellite imagery; and Paul Sirvatka from the
College of DuPage for providing radar imagery.
We also thank Ángel Adames, Becky Alexander, Ileana Blade, Peter
Blossey, Michael Diamond, Ralph Foster, Dargan Frierson, Qiang Fu, Dennis
Hartmann, Lynn McMurdie, Paul Markowski, Cliff Mass, Max Menchaca,
Yumin Moon, Scott Powell, Virginia Rux, David Schultz, Justin Sharp, Brian
Smoliak, Mike Warner, Steve Warren, Rachel White, Darren Wilton, Matt
Wyant, and Qi Zhong, as well as 13 anonymous reviewers, for their help in
the preparation of this book.
This project would not have come to life without the support, help,
influence, and constructive criticism from many fellow professors, teaching
assistants, and students. We cannot acknowledge them all here by name, but
we thank them nevertheless for the important role they have played in
shaping the development of this book.
Introduction
Why should we study our atmosphere? Why should we learn about the causes
and mechanisms of our weather? Weather affects our daily life: the clothes
we wear (rain coat, shorts, hat, should we take an umbrella or sunglasses...?),
the means of transportation we choose (walk, take a bus, ride our bike...?),
our activities (ski, sail, water our plants, read a book in a coffee shop...?), and
probably more. But beyond our daily concerns, weather affects society at
large. Schools close when snow impedes traffic. Visitors to ski resorts might
be more impatient for snow, while the ski instructors will be keeping an eye
on the possibility of avalanches. Rangers are concerned with fog,
thunderstorms, and flash floods. Fire patrols look for weather patterns that are
conducive to forest fires (dryness, wind). Electricity providers are concerned
by wind storms that can damage the infrastructure of the electrical grid and,
on larger timescales, also need to plan how weather will affect upcoming
energy needs (minimum temperatures impact heating, while maximum
temperatures impact air-conditioning). Weather averages, such as prevailing
winds, the typical temperature range, and mean precipitation determine how
we build our homes and what locations are sensitive to extreme events, such
as droughts, floods, hurricanes, tornadoes, etc. On longer timescales, we can
ask how humans are changing the atmosphere, and what those changes imply
for the weather and climate of the future.
To start answering those questions, we need to understand how the
atmosphere works. We need to identify the basic processes that drive the
atmosphere, and the laws that govern atmospheric processes. By doing so, we
will be able to explain the weather phenomena we experience around the year
and throughout the world. Furthermore, we will also be able to apply these
laws to the current state of the atmosphere, and predict how it will evolve in
the future.

© Caroline Planque
There is a lot of value in becoming a knowledgeable observer of the
atmosphere. After reading this book, you will look at the sky differently, you
will gain an understanding of weather and climate that will make you more
attentive to the world around you. You will have a basic understanding of
weather phenomena, of cyclones, thunderstorms, and hurricanes, and you will
understand the basic aspects of weather forecasting. You will see beyond the
weather forecast you get on your phone, radio, TV, or the internet, and you
will be able to make your own forecast in many situations.
 Weather and Climate
Before we continue, let us clarify an important distinction between weather
and climate. Weather is the condition of the atmosphere at a particular time
and location. Weather varies on timescales of minutes to days. Climate, by
contrast, is an average of the weather. It varies on timescales of decades to
centuries and beyond. In this textbook, we will be mostly concerned with
weather – even though many of the concepts have direct application to
climate.
 Getting Started
Our exploration of weather will start with a quick overview of important
weather elements that we can observe or measure, and analyze. The choice of
variables to observe is influenced by the laws of physics that govern the
atmosphere. As we will see shortly, the atmosphere is made of matter (air and
water etc.), it contains energy (heat), and it is in motion (wind, convection).
Our understanding of weather is based on the fundamental notion that matter,
energy, and motion obey conservation laws. To apply these conservation
laws to the atmosphere requires observations of temperature (conservation of
energy), pressure (conservation of mass), wind (conservation of momentum),
along with humidity, precipitation, and clouds. One step at a time, and one
building block over another, we will then investigate the physical processes
that underlie the atmosphere at work. Finally, we will articulate these
processes together to build a picture of weather systems such as midlatitude
cyclones, thunderstorms, and hurricanes. In doing so, we will follow the
precepts of René Descartes, who advocated, as early as 1637, that every
difficult problem should be divided into small parts, and that one should
always proceed from the more simple to the more complex. This cornerstone
of the scientific method, still in favor today, will be an important aspect of
our exploration as we elaborate a thorough understanding of the atmosphere
from its most fundamental constituents at the molecular scale to its most
complex inner workings as a system for moving heat at the global scale.
CHAPTER 1
Weather Variables
CONTENTS
1.1 Temperature
1.2 Pressure
1.3 Wind
1.4 Precipitation
1.5 Weather Stations
Summary

Where should we start with our study of the atmosphere? How should we
first approach the weather? Like many scientists, meteorologists first make
observations. Then they raise questions, and try to answer them. In this first
chapter, we will quickly describe four of the elements, also called variables,
of weather that meteorologists regularly observe, measure, and chart on
weather maps, before we return to each of them for a more thorough
exploration in subsequent chapters. Three of these elements are fairly
intuitive: when concerned with the weather, we like to know how warm or
cold it will be (temperature), whether it will be windy or not (wind), and
whether it will rain or not (precipitation). The fourth variable, atmospheric
pressure, is less intuitive, but it may be the most important to a meteorologist,
as we will soon discover.
Weather results from atmospheric changes. These changes obey certain rules,
dictated by the laws of physics. In meteorology, three laws are of particular
importance: the law of conservation of energy, the law of conservation of
mass, and the law of conservation of momentum. Each describes a particular
aspect of the atmosphere, and each requires that we measure certain
variables of the atmosphere. The object of this first chapter will be to
provide an overview of these variables, a starting point for our exploration of
atmospheric changes. We will then return to each variable in subsequent
chapters for a more thorough description and analysis.
 1.1 Temperature
Of primary interest to us is the law of conservation of energy (see Box 1.1). It
states that energy is never created or destroyed, but only transferred between
locations or transformed between different types of energy. In the atmosphere
one form of energy is heat, and weather is largely the result of heat contrasts
and heat transfers. Therefore we need to design ways of describing the
amount and fluxes of heat throughout the atmosphere, which is accomplished
by measuring temperature.

Box 1.1. The Law of Conservation of Energy

The law of conservation of energy states that the total energy of a

system remains constant, if we account for the gains and losses of
energy from and to the outside. Energy can be transferred between
different parts of the system, or transformed into different types of
energy inside the system (e.g., from potential energy to kinetic
energy, or chemical energy, or radiative energy), but energy cannot
be created (out of nothing) or destroyed.
If we think of the atmosphere as a system, we can apply the law
of conservation of energy to describe how energy, and in particular
heat, is transferred or transformed in the atmosphere. And since heat
transfers are related to temperature differences, we need to measure
temperature and map these differences. That is one reason why
temperature measurements are an integral part of weather
observations.
 We will return to the conservation of energy in Chapter 6, in the
form of the first law of thermodynamics.
 1.1.1 Heat and Temperature
We all have an intuitive feel for temperature, for whether things are hot or
cold. But that feeling is very subjective – warm water feels cold after
stepping out of a sauna. Science requires objectivity, which is sought by
measuring and quantifying variables and processes. But how do we quantify
that feeling of warm and cold? What is temperature, really? First, we need to
return to the fundamental definition of heat as it is transferred to our body
and the environment by contact and interactions at the molecular level.
We can think of the atmosphere as a mixture of gases made up of
molecules in motion (Figure 1.1). And we can think of heat as the energy
associated with this molecular motion. In warm air, molecules are moving
more rapidly than in cold air, and therefore have more energy of motion. We
call this energy of motion, kinetic energy. Temperature is an indirect
measurement of the average kinetic energy of the molecules constituting the
air. Here it is useful to think of the air around us as made of individual
volumes of air of uniform characteristics, called air parcels. [Although the
size of such volumes of air is somewhat arbitrary, the concept of an air parcel
will be useful later to explain various processes at work in the atmosphere.]
While individual molecules in the parcel might all have different speeds, their
average state is indicative of a higher or lower energy level, hence a higher or
lower temperature of the air parcel.
 Figure 1.1. Random molecular motion in a volume of air.

When fast-moving air molecules (at a higher temperature) are in contact

with slow-moving air molecules (at a lower temperature), they impart some
of their energy to the slow molecules, by contact, through collisions. In doing
so, they lose energy, and therefore cool down, while the slower molecules
gain energy, and therefore warm up. Kinetic energy, and therefore heat, has
been transferred from the faster air molecules to the slower molecules, from
the warmer to the colder air. We will see in Chapter 4 that this is the basic
mechanism for transferring energy at the molecular level through
conduction, but it is not a primary heat transfer mechanism for most of the
atmosphere.
 1.1.2 Thermometers
How do we measure temperature? We make indirect measurements of
average kinetic energy using thermometers. Liquid-in-glass thermometers,
for example, work on the basis that substances expand or contract when
temperature increases or decreases. A glass tube is filled with a liquid such as
alcohol or mercury. When placed in warmer air, the energy of motion of the
air molecules is transferred to the liquid in the reservoir by conduction
through the glass. The molecules constituting the liquid now have more
energy than before. Being more active, they push each other apart, which
makes the liquid expand. In particular, it makes the liquid rise in the tube.
Exposed to the same temperature, the liquid will always rise to the same
level. Thus, if we attach a scale and units to the glass tube, and calibrate the
instrument against known temperatures (such as the freezing point and the
boiling point of water), we obtain an instrument that can measure any
temperature (Figure 1.2).
 Figure 1.2. Thermometer in a standard weather shelter.

Many other types of thermometer exist, and all make use of a property of
matter to determine temperature indirectly. A bimetallic strip, for example, is
made of two thin pieces of different types of metal attached to each other.
Because the metals expand and contract at different rates, the combined strip
bends when the temperature changes. If the tip of the metal strip is made to
bend toward a temperature scale, the device can be calibrated and turned into
an instrument.
Some electronic thermometers use a material whose electrical resistance
depends on temperature (a thermistor). Others, called radiometers, measure
the radiation emitted by bodies. Because the emitted radiation is a function of
the temperature of these bodies, we can once again indirectly deduce their
temperature, as we will describe in Chapter 4.
 1.1.3 Temperature Measurements
As we will discuss at the end of this chapter, temperature measurements, like
other weather measurements, are standardized, so that they can be compared
and mapped. Weather stations always measure temperature in the shade, to
avoid contamination by sunlight – recall that we are interested in measuring
the kinetic energy of the air molecules surrounding the thermometer, and not
the radiative energy contained in sunlight that might be absorbed by the
thermometer. For this reason, temperatures are measured in a shelter, which
is elevated at 2 meters (6½ feet) height above a vegetated surface, to avoid
contamination by surface effects. Indeed, the air temperature can change
rapidly near the ground, even within 2 meters (see Chapter 4 and Figure 4.14
in particular), and it is important that we measure the temperature of the air,
not that of the surface. The shelter is painted white, and ventilated, to limit
the absorption of sunlight and the concentration of heat inside the box, which
could otherwise produce an artificially high temperature.
 1.1.4 Temperature Scales
You are probably familiar with the Fahrenheit temperature scale, named after
Gabriel Fahrenheit, a German scientist who constructed the first mercury
thermometer in the eighteenth century and calibrated it against three fixed
points: 0, as defined by a mixture of ice, water, and sea salt; 32 in water and
ice; and 96 “in the mouth or armpit of a healthy man.” Degrees Fahrenheit
are commonly used in the United States. Scientists, and most other countries
of the world, prefer to use the Celsius (or Centigrade) temperature scale,
named after Anders Celsius, a Swedish astronomer who also lived in the
eighteenth century and proposed the temperature of melting ice and the
temperature of boiling water as fixed points (0 and 100, respectively). To
convert from one scale to another, use the following formulas:

∘C=5/9×(∘F−32°)∘F=9/5×(∘C)+32°

(Note that, in the first formula, 32 is subtracted before multiplying by 5/9,

while, in the second formula, 32 is added after multiplying by 9/5 – a
common source of mistake.)
Scientists also use the Kelvin temperature scale, named after Lord Kelvin,
a nineteenth-century scientist whose original name was William Thomson.
The Kelvin scale is merely an offset version of the Celsius scale, translated in
such a way that temperature measurements are always positive numbers.

K= ∘C+273.15 °∘C=K−273.15
The smallest possible temperature, zero kelvin (0 K), is called “absolute
zero,” and corresponds to the theoretical state in which all molecular motion
stops, in which case molecules have zero kinetic energy, and therefore zero
temperature.
(Note that the symbol for kelvin (K) does not have the degree symbol –
the little circle – in contrast with °F and °C.)
Here, in keeping with the International System of Units, we use degrees
Celsius on weather maps and temperature profiles. If you are more
accustomed to degrees Fahrenheit, however, it will be useful to be able to
convert from one scale to another, using the formulas above, or Table 1.1 for
quick reference. If you hear or read temperatures in degrees Celsius and want
to convert them quickly to degrees Fahrenheit, you can take advantage of the
fact that 9/5 is approximately equal to 2, and 32 is approximately equal to 30.
Thus, you can use the following approximation:

∘F≈2× ∘C+30

This is a very quick calculation that you can easily do in your head: multiply
by 2 and add 30. For example, 20 °C multiplied by 2 is equal to 40. Add 30 to
obtain 70 °F, which is close enough to the exact conversion, 68 °F.

Table 1.1. Common temperature values in degrees Celsius and Fahrenheit

°C °F °C °F

−40 −40 0 32

−35 −31 5 41

−30 −22 10 50
−25 −13 15 59

−20 −4 20 68

−15 5 25 77

−10 14 30 86

−5 23 35 95

0 32 40 104
 1.1.5 Radiosonde Profiles
As we will see later in the book, a large part of our weather is dictated by
what takes place aloft, as weather systems often extend up into the
atmosphere and surface phenomena are often driven by upper air currents.
Therefore it is useful to obtain information about the vertical structure of the
atmosphere (Figure 1.3). We do so by launching helium-filled balloons
carrying instruments that record weather variables up to 35 km altitude as the
balloons ascend – helium is used because it is a very light gas (Figure 1.4).
These balloons are called radiosondes, since the measurements are radioed
back to a receiving station at the surface.
 Figure 1.3. Temperature profile obtained by radiosonde at Amarillo,
Texas, on February 18, 2014, at 00:00 UTC. The red curve indicates
temperature (T ) and the green curve indicates the dew point temperature
(Td ). Pressure levels (given in hectopascals, hPa) are shown in blue and
will be described in the text.

Figure 1.4. Radiosonde being launched in Hawaii.

Figure 1.3 shows the lower section of a temperature profile, in red,

obtained by launching a radiosonde from Amarillo, Texas, on February 18,
2014, at 00:00 UTC. [Except where specified, we use Universal Time
Coordinates (UTC) in the rest of the book, i.e., time referenced at the
meridian of Greenwich, England.] The green curve indicates the dew point
temperature, an important measure of the amount of water vapor in the air,
which we will discuss in Chapter 5. Figure 1.3 also gives us an opportunity to
describe a particular type of graph that we will encounter many times in this
book. In such a graph, altitude is displayed on the y-axis (i.e., the vertical
axis), while the variable of interest (temperature, pressure, wind speed) is
displayed on the x-axis. Thus, the red curve in Figure 1.3 tells us that the
temperature decreases from about 18 °C at 1 km altitude to −60 °C at 12 km
altitude, which is in fact fairly typical. For reference, pressure levels are
indicated by dashed blue lines, labeled in hectopascals (see Section 1.2.5).
We can see that the temperature decreases up to about 200 hPa. We will soon
learn that this level of the atmosphere (which changes with time and location)
is called the tropopause and defines the upper limit of the troposphere – the
layer of the atmosphere where essentially all weather takes place.
 1.2 Pressure
Atmospheric changes, and therefore weather, involve the movement of air,
and the redistribution of air mass in the atmosphere. Such redistribution
obeys the law of conservation of mass, which, in meteorology, is expressed in
terms of pressure (see Box 1.2). It is therefore important that we measure
pressure to analyze and predict the weather. As we did for temperature,
however, it is useful that we first understand the nature of pressure in the
atmosphere before we describe instruments to measure it.

Box 1.2. The Law of Conservation of Mass

Much in the same way as energy is conserved, matter, and therefore

mass, is conserved in a system. It cannot be created out of nothing
and cannot be destroyed: it can only enter or leave the system, or be
transformed inside the system (by chemical reaction, for example, or
by phase changes, as we will discuss in Chapter 5). This is described
by the law of conservation of mass.
In meteorology, the distribution of mass in the atmosphere is
described by the distribution of density, which is defined as mass per
volume. However, because density is difficult to measure, we instead
rely on measurements of pressure and temperature to infer density
(see Box 1.3). As a result, we can consider pressure measurements as
informing us about the distribution of mass in the atmosphere.
Notions of mass conservation and pressure will be important for
understanding wind and weather systems in Chapter 8 and 10.
Box 1.3. The Ideal Gas Law

To very good approximation, the atmosphere behaves like an ideal

gas, in which molecular collisions result only in a transfer of kinetic
energy. As a result, pressure, temperature, and density do not vary
independently of each other, but are related by the ideal gas law,

pV=nR∗T

where p stands for pressure, V denotes volume, n is the number of

molecules in volume V, and T is temperature. R* is a universal
constant that applies to any gas.
For the atmosphere, the ideal gas law is more conveniently
written:

p=ρRT

where ρ is the density (mass per volume) and R is a gas constant

specific to the atmosphere.
For example, if we heat a fixed volume and mass of air (i.e., the
density remains constant), the kinetic energy of the molecules
increases (i.e., the temperature increases), and the molecules bump
into each other with more impetus (i.e., the pressure also increases).
 1.2.1 Force and Pressure
As we discussed for temperature, air is a gas made up of molecules in
motion, and heat can be equated with the kinetic energy of the molecules.
However, we also recognized that these molecules constantly bump into each
other. Loosely speaking, the amount of bumping is what we call pressure.
(We will come to a more exact definition shortly.) If the molecules bump into
each other more frequently, or if the collisions themselves are stronger, there
is more pressure. Thus, in a closed container, we could increase the pressure
by either decreasing the size of the container at fixed temperature, which
would bring the molecules closer to each other and would increase the
number of the collisions, or increasing the temperature by heating the air,
which would provide more kinetic energy to the gas molecules, and would
result in more intense collisions. We can see here that pressure, temperature,
and the volume of the container are indeed related, as expressed by the ideal
gas law (see box 1.3). However, this is not to say that pressure and
temperature are the same thing. Temperature, as a measure of heat, relates to
the average kinetic energy of the molecules, independently of the collisions.
Conversely, pressure relates to the collisions between molecules, which is not
uniquely determined by the speed of the molecules, i.e., it is also a function
of the number of molecules in the volume, and therefore the density of the
air.
In the atmosphere, things are slightly more complicated, as the air is not
enclosed in a container, but we will address that issue shortly. In the
meantime, we can think of pressure as the amount of bumping between
molecules. And since the molecules are moving in all directions, we can see
that pressure is applied in all directions as well.
In practice, we often need to know what happens when air pushes on
particular surfaces, either real, like the ground, or imaginary. For example,
we sometimes think of the atmosphere as being made of air columns standing
next to each other, and we are interested in knowing how much the air in two
adjacent columns is pushing against the “wall” in between the columns. At
other times, as described earlier, we will think of the atmosphere as being
made of air parcels, i.e., volumes of air delimited by an imaginary envelope.
Then we will be interested in knowing how much the air inside and outside
the parcel pushes on the envelope. In all cases, pressure is applied to a
surface, which results in a force pushing on that surface. As the area on
which pressure is applied increases (i.e., more air molecules bump into that
surface), the force increases proportionately. In symbolic notation:

F=p ×A

where p stands for pressure applied to a surface of area A, and F is the

resulting force. Equivalently, pressure can be thought of as the force applied
to a surface divided by the area of the surface:

p=F/A

which is, in reality, how pressure is defined in physics. In the above formula,
if the force is distributed over a large area, the resulting pressure is small. But
if the force is concentrated on a small area, the pressure is high.
A simple application comes to mind. If you are wearing regular shoes,
your weight, which provides the force, is distributed over the entire sole of
your shoes, which results in a small pressure being applied to the floor. With
high heels, however, weight is concentrated on a very small area, which
results in a much greater pressure, and could dent a soft floor surface.
 1.2.2 Atmospheric Pressure
By analogy, we now consider the weight of the entire atmosphere applied
onto Earth’s surface at sea level. Air might seem weightless, as philosophers
of antiquity, including Aristotle, used to think, but gravity pulls air
downward, keeping the atmosphere, like everything else, around Earth. In a
column of air extending from Earth’s surface all the way up to the top of the
atmosphere, the accumulated weight of all the air pushes down on the
surface. Recall that weight is a force, so when we divide the weight of the air
column by the area of the base of the atmospheric column, we obtain the
pressure of the atmosphere, or atmospheric pressure.
Note that, in this explanation, we are thinking about atmospheric pressure
as pushing essentially downward, but recall that pressure at a point acts
equally in all directions.
 1.2.3 Vertical Distribution of Pressure
We do not need to be at sea level for the concept of atmospheric pressure to
make sense. We can repeat the same exercise at some altitude above sea
level, say, 3000 m. Air pressure at that altitude is determined by how much
the air is compressed due to the accumulated weight of the overlying layers
of air. (Note that the amount of air below 3000 m is irrelevant in calculating
pressure at 3000 m.) Since there is necessarily less air above 3000 m than
there is above sea level, air pressure will be less than at sea level (Figure 1.5).

Figure 1.5. Atmospheric pressure is tantamount to the weight of the

atmosphere above us.

To make a familiar connection with this decrease in pressure with

altitude, notice the feeling you experience when you swim to the bottom of a
pool – you experience an increase in pressure in your ears and nose. The
atmosphere is like a swimming pool, and we live at the bottom of the pool – a
pool of air. In the same way that a diver will experience higher pressure at
depth in the ocean, atmospheric pressure is highest at ground level, and
decreases upward (Figure 1.6 – pressure units are defined in Section 1.2.5).
 Figure 1.6. Pressure decreases with height.

In summary, it is useful to think of atmospheric pressure as the weight of

air above a point, or above a unit area. If we are at sea level, we are
experiencing the weight of the total atmospheric column, and therefore higher
pressure. If we are higher up in the atmosphere, the column above us contains
less air than at sea level, and as a result the pressure is lower.
 1.2.4 Barometers
As the weather changes, air moves around, resulting in atmospheric pressure
changes at a given location. We monitor these changes with an instrument
that measures atmospheric pressure, which we call a barometer.
In the seventeenth century, Evangelista Torricelli, an Italian scientist who
was greatly influenced by the writings of Galileo, invented the principle of
the mercury barometer. He filled a glass tube with mercury, inverted it in a
cistern, and observed that the level of the liquid in the tube would not
completely drop, but would stabilize at a certain height above the cistern
(about 760 mm). He concluded that the air pressure applied downward onto
the mercury in the cistern was forcing the liquid up into the tube to that level.
He correctly speculated that the weight of the column of mercury in the tube
was balanced by the weight of the atmosphere pressing down on the mercury
outside the tube; i.e., the weight of the mercury column is balanced by
atmospheric pressure (Figure 1.7).
 Figure 1.7. Concept of a mercury barometer.

Torricelli’s design remained the basic principle behind most barometers

for more than 300 years. If atmospheric pressure increases, the mercury level
rises in the tube. Conversely, when a storm approaches and atmospheric
pressure drops, a shorter column of mercury is required to balance the weight
of the atmosphere, and the mercury level drops. That is why atmospheric
pressure is often reported as the height of the mercury column, namely about
76 cm, 760 mm, or 30 inches Hg (“Hg” being the chemical symbol for
mercury).
Note that, in reality, Torricelli first experimented with water. However,
about 10 meters of water are required to balance the weight of the
atmosphere, i.e., about 33 feet, which made the experiment quite
cumbersome! Since mercury is much denser, and thus heavier than water, a
much shorter column of mercury is required to achieve the same weight.
 If your barometer at home is relatively small and round, you must be
wondering how it can possibly contain 30 inches of mercury in a tube. And it
does not, of course. You probably own an aneroid barometer, which is
made of a small empty chamber (evacuated of some air to create a partial
vacuum) that can expand and contract. When atmospheric pressure increases,
it squeezes the chamber and forces it to contract. When atmospheric pressure
decreases, the chamber expands. By connecting the chamber to a moving
needle, the expanding chamber can be turned into a precise instrument
(Figure 1.8).

Figure 1.8. Aneroid barometer.

 1.2.5 Pressure units
In the International System of Units, pressure is expressed in pascal (Pa),
where 1 Pa = 1 N/m2 (one newton per square meter), or alternatively in
kilopascal (1 kPa = 1000 Pa). Note that a newton is a unit of force and a
square meter is a unit of surface area, which is consistent with our earlier
definition of pressure as force divided by area. In meteorology, however,
since we think of atmospheric pressure as the weight of the atmosphere
exerted on a surface, it can be more intuitive to express it in pounds per
square inch (psi), or kilograms per square centimeter (kg/cm2) in the metric
system. In that context, we can picture a column of air extending from
Earth’s surface up to the top of the atmosphere. If the base of the column has
a surface area of one square inch, then the weight of that air column is on
average 14.6 pounds (at sea level); thus, atmospheric pressure is, on average,
14.6 psi.
Meteorologists, however, go one step further. Since we are interested in
the pressure exerted by the atmosphere, we may express pressure... in
atmospheres! The average atmospheric pressure would then be close to “one
atmosphere.” For historical reasons, one atmosphere is called “one bar,” after
the Greek “baros,” meaning weight. Since daily variations in pressure are on
the order of a thousandth of an atmosphere (i.e., a thousandth of a bar),
meteorologists have historically used millibars (mb), where 1 bar = 1000 mb.
The exact value of a bar has been chosen as 100 kPa for practical purposes,
which is equal to 1000 hectopascal (hPa, where hecto- is the prefix for 100),
so millibar and hectopascal can be used interchangeably. Putting it all
together, the average pressure at sea level is approximately:
 1013hPa−1013mb−29.92°Hg−14.6psi

To be consistent with the International System of Units, we will use

hectopascals (hPa) in the rest of this book.
 1.2.6 Some Useful Numbers
If you observe the variations in atmospheric pressure at a given location over
time using a barometer, you will find that the typical range of pressure at sea
level is about 980 to 1030 hPa. The pressure will only drop below 980 hPa
during intense storms. Similarly, it will only rise above 1030 hPa during
exceptionally high pressure events.
If you carry your barometer upward, you will find that pressure drops
very quickly with height, as we saw in Figure 1.6. As a rough rule of thumb, -
pressure near the surface drops about 8 hPa for every 60 m of altitude – this
is easily noticeable, and measurable, in elevator trips in tall buildings. Recall
the pool analogy: pressure is greatest at the bottom of the pool, or ocean of
air, and decreases as we rise in the pool. Since we will be interested in what
happens higher up in the atmosphere, it is useful to have some standard levels
of pressure in mind and know their approximate altitude.
Figure 1.9 shows some useful numbers to which we will refer often. We
saw that the average pressure at sea level is 1013 hPa. We will often look at
maps of weather variables at 850 hPa, which is around 1.5 km altitude (about
5000 ft), and near the top of what we call the “boundary layer” (i.e., near the
top of the layer where the effects of the daily variations in temperature near
Earth’s surface are felt). At 500 hPa, about half of the atmosphere is above us
and the other half below (in terms of mass and weight), and this happens at
about 5.5 km, or 18,000 ft. Aircraft cruising altitude is at about 250 hPa
(roughly 11 km, or 34,000 ft), which is also the top of the troposphere. Thus,
the troposphere, i.e., the first layer of the atmosphere, where most of our
weather takes place, contains about 75% of the mass of the atmosphere. Note
that 250 hPa also corresponds to the altitude where we find the jet stream, as
will be discussed in Chapter 9. Above the troposphere, we enter the
stratosphere, where the ozone layer is found, as we will describe in more
detail in Chapter 3.

Figure 1.9. Altitude and pressure in the atmosphere: some useful numbers
to remember.

Since water is 1000 times more dense than air, the increase in pressure
with depth happens much faster in the ocean than in the atmosphere. Recall
that the weight of the atmosphere is balanced by a 10 m column of water in a
water barometer. In other words, a 10 m column of water has the same
weight as a column of air extending from sea level to the top of the
atmosphere. Therefore, we will add the equivalent of another atmosphere of
pressure (i.e., one additional bar) if we dive 10 meters down into the ocean –
another good number to remember. And for each additional 10 m layer of
ocean, we will add another bar of pressure. That is why it is important for
divers to pause regularly, especially on their way up to the surface, to let their
blood and other internal fluids adjust to the change in external pressure.
 1.3 Wind
As mentioned before, weather results from atmospheric motions, and air
motion is also constrained by the laws of physics. As for energy and mass, air
motion is constrained by a conservation law: the law of conservation of
momentum (see Box 1.4). Momentum is the product of mass and velocity.
The density of air is estimated indirectly from measurements of pressure and
temperature, using the ideal gas law (see Box 1.3). The wind velocity,
however, needs to be measured. It has two components, direction and
magnitude, which need to be measured separately.

Box 1.4. The Law of Conservation of Momentum

Weather is a result of the atmosphere in motion to redistribute heat. In

meteorology, it is useful to describe air motion in the form of
momentum, a product of the mass, or density, of the air and its speed,
or velocity. Momentum tells us both how much air is in motion, and
how fast it moves.
As it turns out, much like energy and mass obey conservation
laws, momentum is governed by the law of conservation of
momentum, which is described by Newton’s laws of motion. In
particular, Newton’s second law, which relates forces to changes in
momentum, describes how the wind evolves. That is why we measure
wind as part of our routine weather observations.
Wind and Newton’s second law of motion will be the subject of
Chapter 8.
 1.3.1 Measuring wind
Wind direction is measured using a wind vane. The vane points toward the
direction that the wind is coming from (Figure 1.10), which is why
meteorologists report wind direction as the direction that the wind is coming
from. An easterly wind, for example, is a wind blowing from the east. A
northerly wind blows from the north, etc. Figure 1.11(a) shows wind in terms
of compass direction.

Figure 1.10. A wind vane points toward the direction that the wind is
coming from.
 Figure 1.11. Wind direction convention in meteorology. (a) Compass
direction. (b) Degrees from true north.

Meteorologists also use degrees from true north to report wind direction
(Figure 1.11(b)). Degrees from true north are more convenient for
computation and graphing, for example, since they are numerical quantities.
Wind speed is measured with an anemometer. The most common type
is the cup anemometer (Figure 1.12(a)). As the cups catch the wind, the
spindle rotates, and the rate of rotation of the spindle is proportional to the
wind speed. It is similar to propeller anemometers, for which the wind turns
the blades in proportion to the wind speed (Figure 1.12(b)).

Figure 1.12. Two different instruments to measure wind speed. (a) Cup
anemometer. (b) Propeller anemometer.

The units for wind speed are miles per hour (mph) in the English system,
and kilometers per hour or meters per second in the metric system.
(Remember that speed is distance divided by time.) Unfortunately, the most
common unit used by meteorologists is knots, for historical reasons: mariners
were affected by weather before airplanes were. Sailors used to measure the
speed of their ship by letting a knotted rope run into the sea, and counting the
number of knots per time as a measure of speed. One knot is one nautical
mile per hour, and since one nautical mile is the distance of one minute of
latitude on Earth (1.85 km, or 1.15 mi):

1 knot−1.85 km/h−1.15 mph−0.5 m/s.

Mariners sometimes also use the Beaufort wind scale, which was first
designed to relate wind speed to the sea state in the absence of anemometers;
the rougher the sea surface, the higher the wind speed. It is now also
occasionally used on land and describes typical damage observed during
strong wind events (Table 1.2).

Table 1.2. Beaufort wind scale

Force Description Wind speed Wind

(knots) speed
(km/h)

0 Calm 0 0

1 Light air 1–3 1–5

2 Light breeze 4–6 6–11

3 Gentle breeze 7–10 12–19

4 Moderate breeze 11–16 20–28

5 Fresh breeze 17–21 29–38

6 Strong breeze 22–27 39–49

7 High wind 28–33 50–61

8 Gale 34–40 62–74

9 Strong gale 41–47 75–88

10 Storm 48–55 89–

102
11 Violent storm 56–63 103–
117

12 Hurricane force >63 >117

 1.3.2 Reporting Wind
As you know from experience, the wind does not blow constantly and
regularly. It stops, resumes, increases, and decreases in strength over short
time intervals, producing wind gusts. Figure 1.13, for example, shows a lot
of “gusty” variability over the course of 2 minutes. Wind gusts are often due
to the presence of swirls of air called “turbulent eddies” that mix the air up
and down on a variety of scales (i.e., eddies of different sizes). Since wind
speed typically increases with altitude and decreases as it approaches the
surface, where there is more friction, turbulent eddies tend to bring higher
wind speeds down to the surface intermittently (Figure 1.14). The resulting
wind gusts can be twice as strong as the average wind and can damage trees,
for example, or affect sailing. In the example shown in Figure 1.13, wind
gusts reach speeds up to 20 knots.

Figure 1.13. Wind variability, wind gusts, and sustained wind (one-minute
average wind).
 Figure 1.14. Turbulent eddies bring higher wind speeds down to the
surface.

To reduce the effect of turbulent eddies on wind measurements, we

measure the wind away from the surface, at a standard height of 10 m (30 ft).
Even at this distance above the surface, the wind speed fluctuates
significantly, and an instantaneous wind measurement is not very
representative. Therefore, we average the wind speed over a period of time
and report the wind as both sustained wind and gusts, where the gusts denote
the higher values of wind speed that only last for short periods of time. The
standard time average for sustained surface winds is one minute. In the
example shown in Figure 1.13, the sustained wind is about 12 knots, and the
wind would be reported as a 12 knot wind with 20 knot gusts.
Note that the effect of friction on the wind varies with the underlying
surface. For example, water surfaces are relatively smooth compared to land
surfaces, where trees and buildings act as a drag on the air and slow it down.
As a result, under similar weather conditions, sustained winds at the surface
are typically 30–50% higher over water than over land (Figure 1.15).
Figure 1.15. The ocean is a smooth surface that offers less resistance to the
wind than land surfaces, resulting in higher wind speeds over the ocean.
 1.3.3 Additional Sources of Wind Information
Our understanding of atmospheric motions requires that we measure not only
surface winds, but also upper-level winds. Meteorologists use radiosonde
measurements to learn about the wind at various altitudes. Wind speed and
direction are inferred from the balloon drift, which give us a vertical wind
profile (see Figure 5.9 for an example). Measurements are also made
routinely on commercial aircraft, which inform us about the weather along
the main carrier routes. Finally, wind measurements can also be obtained
from satellites, either by tracking cloud movement (speed and direction) or
by using radar to measure the small ripples caused by the wind blowing on
the surface of the ocean, which is an effect you can also observe on a lake:
stronger winds make the surface “rougher.”
 1.4 Precipitation
The last element of weather that is of particular interest to us is precipitation,
a generic term that encompasses rain, snow, hail, sleet, etc. Generally, we can
define precipitation as solid or liquid water falling out of the atmosphere. We
typically measure precipitation with a rain gauge, of which there are two
primary types.
In a tipping bucket, water is collected by a funnel and falls onto a tipping
plate that tips when enough water has accumulated in the bucket (Figure
1.16). One tip usually corresponds to 0.25 mm (0.01 inches) of rain. The
number of tips indicates the amount of rain that has fallen over time.

Figure 1.16. Concept of the “tipping bucket” rain gauge.

In a simpler type of rain gauge, a funnel also collects raindrops and

magnifies the rain accumulation into a thinner cylinder so that the total depth
may be accurately read on the scale (Figure 1.17). However, only the total
depth of accumulated rain is read on the scale. In other words, we have no
indication of the changes in precipitation rate between two readings. Such a
rain gauge usually measures up to 25 mm, and the overflow is caught in the
outside cylinder, for later measurements.

Figure 1.17. Concept of the “funnel” rain gauge.

Precipitation of less than 1 mm is reported as a trace. Frozen precipitation

like snow is melted and its liquid depth is reported. (The frozen depth is also
reported.)
On weather maps, precipitation is indicated using specific symbols (see
Figure 2.1 in Chapter 2). The most useful symbols are those for rain and
snow. However, we will discuss very important weather conditions
associated with sleet (frozen raindrops in the form of ice pellets), freezing
rain (liquid raindrops that freeze when they hit objects at the ground),
and fog (which is not precipitation).
 1.5 Weather Stations
All the instruments described above are set up in a standardized fashion at
thousands of weather stations around the world. Most are now automated, as
is the case for the Automated Weather Observing Systems (AWOS) of the
Federal Aviation Administration and the Automated Surface Observing
Systems (ASOS) of the US National Weather Service (Figure 1.18). At
standard sites, thermometers are enclosed in a shelter to prevent direct
exposure to the sun, wind, and precipitation. Nevertheless, the shelter needs
to be well ventilated so that the measurements are representative of the
surrounding air. That is why weather instrument shelters are painted white (to
reflect sunlight) and have open slats to promote air circulation (Figure 1.19).
Temperature is measured at a standard height of 2 m.
Figure 1.18. Example of an automated weather station.
 Figure 1.19. Standard weather station shelter.

Anemometers are set at the top of “masts” where they can measure wind
speed and direction at the standard height of 10 m.
Over the ocean, weather instruments are attached to masts on buoys,
which are themselves anchored to the sea floor (Figure 1.20). Due to the
difficulty and expense of installing and maintaining instruments at sea, there
are far fewer buoys than there are land stations, as shown in Figure 1.21.
When comparing the number of weather buoys at sea to the number of
weather stations on land, it is immediately apparent that weather is much
better directly monitored over land, and especially in urban areas, than over
the ocean. (Note that some regions have chosen to set up denser networks of
weather stations, one of which is apparent in Alberta, Canada.) This disparity
has important implications, as weather observations are crucial for
constraining forecasting models (see Chapter 13). However, the advent of
satellite observations in the last few decades has very much reduced the
impact of this disparity, as we will see in Chapter 2.

Figure 1.20. Example of a weather buoy.

Figure 1.21. Map showing the distribution of land-based stations and
weather buoys in and around North America.
 Summary
Atmospheric changes obey the laws of physics for the conservation of
energy, mass, and momentum. We observe the atmosphere by measuring a
number of variables constrained by these laws and relevant to explaining and
predicting weather patterns.

► We measure temperature using thermometers and use three different

temperature scales: Celsius, Fahrenheit, and Kelvin.
► We measure atmospheric pressure with barometers, and express it
using hectopascals.
► We measure wind direction with wind vanes, and wind speed with
anemometers. We report wind speed as sustained winds and wind
gusts, using knots, miles per hour, meters per second, and kilometers
per hour.
► We measure precipitation depth with “funnel” rain gauges with units
of millimeters or inches, and precipitation rate with “tipping bucket”
rain gauges.

Finally, we assemble weather instruments at dedicated weather stations

on land using shelters and masts, and buoys at sea. Weather observations are
more dense over land than over water.
We can use these measurements to create time series of the variables
(i.e., variations in time) and weather maps (i.e., variations in space at a given
time, as we will see in Chapter 2).
Instruments are also attached to a weather balloon, a radiosonde, which
records the corresponding variables as the balloon ascends, building a
vertical profile of the atmosphere above a given location (i.e., variations
with height).
CHAPTER 2
Spatial Representations of Weather Data
CONTENTS
2.1 The Station Model
2.2 Surface Maps
2.3 Upper-Level Maps
2.4 Radar
2.5 Satellites
Summary

Local weather is largely the result of large weather systems in motion. Thus,
meteorologists gain insight into the manifestations of weather by studying
maps and images of the atmosphere on a number of scales, from global, to
regional, down to local scales. In this chapter, we will learn how weather
information is represented on weather maps and images to reveal the two-
dimensional dynamics of weather systems.
In Chapter 1 we introduced some foundational concepts to initiate a study of
the atmosphere guided by scientific principles and the laws of physics and, in
particular, conservation laws that constrain the behavior of particular
variables: temperature, pressure, wind, and water. Understanding weather,
and therefore the atmosphere in motion, now requires that we display these
variables on maps, to reveal their spatial distribution. Furthermore, because
the analysis of raw observations is often quite challenging, we need to design
tools that will enable us to analyze these variables in a way that reveals
salient features of the weather.
The spatial representation and analysis of data will be the object of this
chapter, before we return to each element of weather individually for a more
thorough exploration in later chapters.
 2.1 The Station Model
The variables described in Chapter 1 and measured at weather stations or
buoys are typically displayed on weather maps using the station model partly
described in Box 2.1. Pressure, temperature, and dew point temperature are
displayed as numerical values, while we use symbols for precipitation. The
central circle indicates cloud cover. Wind is displayed with wind barbs using
the meteorological convention. Wind direction is represented with a line
extending from the weather station toward the direction the wind is coming
from, while wind speed is indicated with barbs and pennants.

Box 2.1. A (Simplified) Weather Station Model

Weather observations are reported on surface maps using the weather

station model shown in Figure 2.1.1.*

Figure 2.1.1. Simplified weather station model.

Cloud cover is indicated by shading the circle, as shown in

Figure 2.1.2.
Figure 2.1.2. Some cloud cover symbols.

Significant weather by type and intensity is indicated on

weather maps using symbols such as those shown in Figure 2.1.3.

Figure 2.1.3. Some weather symbols.

Wind direction is represented by a line extending from the

weather station toward the direction the wind is coming from, while
wind speed is indicated with barbs and pennants, as shown in Figure
2.1.4.
Figure 2.1.4. Wind convention.

Strictly speaking, the barbs and pennants indicate a range of wind

speeds around the central value. For example, a half barb indicates 3
to 7 knots. Hence the values shown in Figure 2.1.5.

Figure 2.1.5. Reporting wind speed with barbs and pennants.

 * See
[Link] for a
complete description.

An example of a surface weather map showing measurements at land and

buoy stations on February 20, 2014, is shown in Figure 2.1 (only
temperature, dew point temperature, pressure, and wind are shown, for
clarity). This is our first look at a weather event that will follow us through
the rest of this book: an extratropical cyclone generated east of the Rocky
Mountains and sweeping through the eastern half of the United States
between February 19 and 22, 2014. Extratropical cyclones are also called
midlatitude storms, or frontal cyclones, as cyclones form fronts, but fronts
can also form cyclones.

Figure 2.1. Example of surface station map on February 20, 2014, at 18:00
UTC. To facilitate reading, temperature is indicated in red, dew point
 temperature in blue, and sea-level pressure in black, in addition to
following the weather station model layout described in Box 2.1.

At this stage of its development, the storm (let’s call it “our February
2014 cyclone”) appears in our weather map as a counterclockwise circulation
occurring over the eastern half of the United States. By following the
direction of the wind barbs, we can get a sense of its size and locate its center
somewhere along the border between Iowa and Missouri (see Appendix 2.2
for geographical reference). By contrast, the western United States is
relatively quiet with low wind speeds and no organized circulation.
 2.2 Surface Maps
As evident in Figure 2.1, it is quite difficult to appreciate the spatial
distribution of the different variables at a glance. That is why we analyze and
complement the surface maps with additional elements.
 2.2.1 Isotherms and Temperature Maps
We first color the temperature field using an intuitive color scale, typically
going from blue (low temperatures) to red (high temperatures), as shown in
Figure 2.2. We further contour the temperature observations with contour
lines at regular intervals. These contours are called isotherms, from “iso,”
meaning “equal,” and “therm”, meaning “heat.” Isotherms are lines of
constant temperature. Indeed, if we were to walk along an isotherm with a
thermometer, the temperature would remain constant. If we decided to depart
from the isotherm, the temperature would either increase or decrease.

Figure 2.2. Example of a surface temperature map on February 20, 2014 at

18:00 UTC.
 Coloring and contouring the temperature map allows us to quickly
identify the air masses, where the temperature is relatively uniform. In
Figure 2.2, for example, a warm air mass occupies the southeastern sector of
the United States, while a cold air mass occupies the northwestern half. In
between these air masses are areas of strong temperature change, where the
temperature increases from cold to warm over relatively short distances.
These regions of sharp temperature change are called temperature fronts –
see Section 2.2.2.
The rate of change of a variable over a certain distance is called a
gradient – a notion we will use often in the rest of this book. In Figure 2.2,
for example, we can see that the temperature gradient is relatively small in
the cold and warm air masses, i.e., if we drove across an air mass with a
thermometer, for example from Austin, Texas, to Atlanta, Georgia, a 1500
km drive, the temperature would not vary much. Frontal regions, however,
are, by definition, regions of strong temperature gradient, i.e., if we drove
across a temperature front, our thermometer would record a strong
temperature increase or decrease over a short distance. Going from Kansas to
Arkansas, for example, the temperature increases by 30 °F over only 200
miles.
temperature increases by
 2.2.2 Temperature Fronts
Weather in the midlatitudes is largely the result of air masses in motion. In
fact, by matching the air masses visible in Figure 2.2 with the corresponding
wind directions in Figure 2.1, you can see that the warm air mass over the
southeastern United States is advancing northward, toward colder air, while
the cold air mass over the Great Plains is advancing to the southeast, toward
the warm air mass. The temperature fronts visible in between are largely the
result of these air masses in motion. As we will see in Chapter 10, clouds and
precipitation tend to be concentrated along the temperature fronts, i.e., along
the boundaries between the air masses. Therefore, we indicate the leading
edge of air masses on weather maps, using a blue line and blue triangles for
cold fronts, and a red line and red semi-circles for warm fronts, as shown in
Figure 2.3. Note that, in each case, the symbols are pointing in the direction
of motion. In Figure 2.3(a), the cold air mass is replacing the warm air to the
southeast, and therefore the blue triangles point to the southeast. Similarly in
Figure 2.3(b), the warm air replaces the cold air to the northeast, and
therefore the red semi-circles point to the northeast. [For future reference,
know that fronts are drawn on the warm side of the largest magnitude in the
temperature gradient.]

Figure 2.3. Conventional representation of (a) a cold front and (b) a warm
 front on weather maps.

We will learn more about air masses and fronts in Chapter 10. In the
meantime, we will soon add cold and warm fronts to our surface map, but
first we need to introduce pressure contours, or isobars, which will help
identify the fronts.
 2.2.3 Isobars and Pressure Maps
As for temperature, when studying isolated point measurements of pressure
on a surface weather map, it is quite challenging to capture visually at once
the entire structure of the pressure field, with its minimum and maximum
values, its areas of rapid horizontal change as opposed to slow change, etc.
That is why we also contour the pressure observations, to obtain a two-
dimensional picture of the pressure field (Figure 2.4). Note that pressure
measurements are adjusted to mean sea level before contouring (Box 2.2).
Contours of pressure are called isobars, and each isobar represents a
particular value of pressure – typically at a 4 hPa interval. We can talk, for
example, about the 1000 hPa isobar. By definition, the pressure will be 1000
hPa everywhere along that contour.

Figure 2.4. Example of surface pressure map on February 20, 2014, at

18:00 UTC.

Box 2.2. Adjustment of Pressure to Mean Sea Level

Pressure maps are an important tool of the weather analyst, as

pressure is intimately related to wind and weather systems. We will
learn in Chapter 8 that wind is caused by horizontal pressure
variations. Pressure, however, decreases far more rapidly upward than
it changes horizontally. A good rule of thumb is that pressure
decreases by about 8 hPa with each 60 m gain in altitude (at low
elevation). Thus, the pressure at the top of the Empire State Building
(381 m) is about 50 hPa lower than at street level. The pressure at the
top of the Rocky Mountains could be as low as 600 hPa. However,
the horizontal pressure variations causing wind are much more subtle
than this. Recall that the typical range of pressure at sea level is about
980 to 1030 hPa. Therefore, if we reported actual pressure
measurements on a surface weather map, as they are measured at
weather stations on the ground, we would observe very low pressure
values wherever there are mountains. We would suggest the presence
of artificially low pressure areas where the mountains stand, with a
drop in pressure of 300 or 400 hPa in places.
To avoid such artefacts, we calculate what the pressure would be
if there were no mountains and we were at sea level. We replace the
mountains with air, so to speak, and add the corresponding pressure
to the value measured at the weather station. In other words, we
adjust the values of pressure to mean sea level. Figure 2.2.1(b) shows
the unadjusted pressure field, and a comparison with the topography
(Figure 2.2.1(c)) confirms that the low pressure areas correspond to
the Rocky Mountains and the Appalachians, at the expense of our
February 2014 extratropical cyclone over the Midwest. After
adjusting the pressure measurements to mean sea level (Figure
2.2.1(a)), the low pressure center and the troughs corresponding to the
cold and warm fronts appear clearly. [Note that we have used a green
color scale for the purpose of this exercise, although pressure fields
are usually displayed with isobars alone.]
Figure 2.2.1. Pressure measurements adjusted to mean sea level (a),
with the unadjusted pressure field (b) and topography (c) for
comparison.
 A good analogy for isobars is the lines of constant height on a
topographic map. You know that if you are hiking along such a line, you stay
at the same height. If you depart from that contour, however, you either climb
up toward higher elevations or down toward lower elevations. Similarly on a
pressure map, pressure necessarily increases on one side of the isobar and
decreases on the other side. Furthermore, if topographic contours are tightly
packed together, you know that you are facing a steep slope. Similarly, tight
isobars indicate a steep pressure change, i.e., a sharp pressure gradient, and
we will learn in Chapter 8 that strong pressure gradients induce strong winds.
 2.2.4 Highs, Lows, Ridges, and Troughs
The surface pressure field often contains regions of minimum and maximum
pressure. Figure 2.5(a) shows an example of a high pressure region where the
pressure increases toward a maximum value, which we indicate with a big H
(for “high”). We will see that high pressure regions, or “highs,” correspond
to anticyclones and are often associated with fair weather (i.e., weak wind
and clear skies). In Figure 2.4 the western half of the United States is under a
broad area of high pressure.

Figure 2.5. Example of (a) high pressure region, or “high,” and (b) low
pressure center, or “low,” as would appear on a surface pressure map.

Figure 2.5(b) shows an example of a low pressure region in which the

pressure decreases to a central minimum, indicated with a big L (for “low”).
Note that there is no 988 hPa contour. Therefore, we know that the pressure
is above 988 hPa everywhere inside the interior contour. The big L, however,
tells us that there is a minimum value less than 992 hPa. The exact value of
the minimum pressure is often indicated on standard weather maps. We will
see that such low pressure centers correspond to cyclonic weather systems,
such as midlatitude storms and hurricanes. They are often associated with
active weather, such as clouds, precipitation, and strong wind. We will refer
to such minima in pressure as “low pressure areas”, “low pressure centers,”
or simply as “lows.” Our February 2014 cyclone, for example, as depicted in
Figure 2.4, corresponds to a low pressure center located over northern
Missouri.
To continue with the topography analogy, it is sometimes useful to think
of a high pressure region as a mountain of high pressure, and to think of a
low as a basin. For practice, try to convince yourself that you could hike up
to the top of the mountain shown in Figure 2.5(a) and down into the bowl
shown in Figure 2.5(b).
Similarly, if the contours are not closed, we can still find useful features
to identify in the pressure field. For example, if higher pressure values extend
in one direction, this is like a mountain ridge, as indicated in Figure 2.6(a).
Even though there is no isolated high pressure center, the crest of the ridge is
still a line of relative maximum pressure, compared to the lower values of
pressure down the slopes on each side of the crest. We will refer to this
feature as a “ridge of high pressure,” and to the line of relative maximum
pressure as the “axis” of the ridge.

Figure 2.6. Example of (a) high pressure ridge, or “ridge,” and (b) low
pressure trough, or “trough,” as would appear on a surface pressure map.
 We will see that ridges are especially apparent on upper-level maps, and
are associated with fewer clouds and lighter winds. Conversely, when the
pressure field is relatively lower along a line, we have the equivalent of a
valley. Although we do not call it a “valley of pressure,” we use a similar
analogy and call it a “trough of low pressure.” Even though there is not a
closed low pressure center, the axis of the trough is still a line of relative
minimum pressure (Figure 2.6(b)). We will see that troughs are also
frequently observed on upper-level maps and are often associated with
regions of clouds and precipitation.
Note that we rarely color the pressure field as we did for temperature,
because it is more informative to overlay the isobars on top of the (colored) -
temperature field, as shown in Figure 2.7. We will see that this is a powerful
tool in meteorology, as it allows us to analyze temperature and pressure
jointly. After adding the highs (H), the lows (L), and the temperature fronts,
we obtain a compelling picture of the state of the atmosphere at any given
time. To convince yourself fully, return to Figure 2.1 and compare it to
Figure 2.7 to appreciate how much more readable our surface map has
become. At a glance, we have what we call a synoptic view of the
atmosphere, or a view of the “synoptic situation” over the United States
(from the Greek “syn” meaning “together” and “opsis” meaning “view”).
 Figure 2.7. Example of surface weather map showing temperature (colors
and isotherms), pressure (isobars), highs (H) and lows (L), as well as
temperature fronts on February 20, 2014, at 18:00 UTC.

Here we can see that the cold and warm air masses, the cold and warm
fronts, and the low pressure center all coincide to form our February 2014
extratropical cyclone. We will see in Chapter 10 that the air masses and fronts
circle around the low, and the warm sector, i.e., the sector of warm air
delimited by the two fronts, decreases in size as the cyclone evolves, while
clouds and rain form along the fronts and at the center of the cyclone. By
looking carefully at the pressure field, you will also notice that the fronts
happen to be located where the isobars “kink” and change directions. If you
think of these two depressed areas as pressure troughs, then the fronts follow
the axis of the troughs. This is indeed a useful tip for placing temperature
fronts on a surface map, the reason for which we will discover when we
know a bit more about wind.
 2.3 Upper-Level Maps
We tend to think of weather as happening close to Earth’s surface, where we
live and where we can observe it. But most of the action actually takes place
aloft. Surface weather in the midlatitudes is largely determined by the
position and structure of the jet stream, a fast-flowing air current blowing at
the top of the troposphere. Disturbances in the jet stream create certain
conditions that are more or less conducive to the formation, or demise, of
cyclones. That is why meteorologists spend a lot of time studying upper-level
maps.
We could create such maps by contouring pressure observations at
specified heights, as we did for our surface weather map. For example, we
could collect pressure measurements at an altitude of 5000 m and build a
5000 meter pressure map. We would then find highs and lows, troughs and
ridges, as we did at the surface. In practice, however, it is more convenient to
do the opposite: we choose a particular pressure, for example, 500 hPa, and
we indicate how high in the atmosphere this 500 hPa value is located. If we
now contour these height measurements, we have a 500 hPa “height” map.
We can think of it as a two-dimensional surface where the pressure is
everywhere 500 hPa, roughly at 5.5 km altitude, but with dips and bumps that
make it sag lower or rise higher in places.
This seems complicated, but in practice it yields about the same result,
and it is more convenient from a mathematical point of view, because
working at constant pressure allows us to simplify our equations.
[Specifically, the air density disappears from the equations. Recall that
density is not measured, but estimated indirectly from other measurements.
Therefore, disposing of density is very convenient.]
In the end, we can treat this “height” map exactly as we would a pressure
map, as illustrated in Figure 2.8. If you picture a series of pressure surfaces at
different altitudes, they look like the blue lines shown in Figure 2.8(a) on a
cross-section. Recall that pressure decreases with height, and therefore the
850 hPa surface is necessarily above the 900 hPa surface and below the 800
hPa surface. If you follow the 850 hPa surface from west to east, you will see
that it is higher to the west (blue H) and lower to the east (blue L). If you now
follow the 1.5 km height line (dashed) from west to east and read the
corresponding pressure (on the blue lines), you will see that the pressure is
higher to the west (about 900 hPa, indicated by a black H) and lower to the
east (about 800 hPa, indicated by a black L). In other words, the height
surface is high (H) when the pressure is high (H), and it is low (L) when the
pressure is low (L). On the actual 850 hPa map, the high and the low might
look like Figure 2.8(b), which tells us that the 850 hPa surface is sloping
from higher to lower altitudes from H to L. The high (at the constant pressure
of 850 hPa) also corresponds to higher pressures (at the constant height of 1.5
km), while the low corresponds to lower pressures. Therefore, for our
purposes, we can think about the features on the upper-level height maps as if
they were on a pressure map.
 Figure 2.8. Example of a high and a low as would appear on (a) a cross-
section, and (b) a 850 hPa map. The blue lines in panel (a) represent
pressure surfaces, while the black lines in panel (b) represent isobars. The
blue H represents a region where the 850 hPa surface is high, while the
black H represents the corresponding region where the pressure is high at
1.5 km altitude (and similarly for low pressures).

Meteorologists frequently use the 850, 500, and 300 hPa pressure
surfaces for upper-level analyses. In addition to height contours, we usually
plot station data on these maps, as shown in Figure 2.9 for a 500 hPa map,
which helps us to visualize the speed of air currents flowing along the
contours (see Chapter 8). Finally, recall from the ideal gas law that, for a
fixed mass of air, warmer air has a greater volume than colder air. Therefore,
for a fixed area, a warmer column of air is “thicker” and associated with
higher upper-level heights on pressure surfaces. Thus, meteorologists can
also analyze height fields in terms of thickness and heat content of the
underlying air columns.
 Figure 2.9. Example of station data on a 500 hPa upper-level map. Height
given in meters.

An example of a simplified upper-level map, corresponding to our

February 2014 cyclone, is shown in Figure 2.10, where wind speed is also
indicated with colors in the background. Closed highs and lows are not so
common at 500 hPa, but troughs and ridges are, and often correspond to
specific weather features at the surface. Upper-level ridges, for example,
often indicate fewer clouds, while upper-level troughs create surface
conditions that are conducive to the formation of cyclones (see Chapter 10).
Here, the Great Lakes are under a ridge of high pressure, but the Midwest is
under a trough of low pressure, with strong winds blowing around the trough,
and particularly high winds on the southeastern flank of the trough, over
Missouri. The alert reader will notice that this feature of the upper-level jet
coincides with the location of the surface low in Figure 2.7. Indeed, we will
learn in Chapter 10 that this configuration is conducive to the deepening of
extratropical cyclones.
Figure 2.10. Example of a 500 hPa map on February 20, 2014, at 18:00
UTC.
 2.4 Radar
Radars (originally RAdio Detection And Ranging when developed for
military purposes) are very useful for measuring and tracking precipitation.
They emit radio waves nearly horizontally, since the troposphere is so thin.
Some of these waves are transmitted through the atmosphere and are lost if
they do not encounter any obstacle. If precipitation is present on their path,
however, the raindrops will reflect some of the waves back to the radar,
where the returned energy can be measured and analyzed (Figure 2.11).

Figure 2.11. Concept of radar as used to measure precipitation. The black

lines show the signal emitted by the radar. The red lines show the part of
the signal that returns to the radar after being reflected by raindrops.

The return signal is an indirect measurement of precipitation. It contains

two pieces of information. First, more precipitation causes more reflected
signal. Therefore, the radar can be used to estimate the rate of precipitation.
Second, by calculating the time it took for the signal to reach the raindrops
and come back, we can estimate the distance between the radar and the
precipitation, and we can indeed map the structure of the precipitation around
the radar. Therefore, a radar provides information about both the location and
the intensity of precipitation.
By tilting the radar up and down, we can also learn about the vertical
structure of the precipitation. By rotating the radar horizontally, we can build
a 360º picture of precipitation all around the radar, out to a distance of about
250 km. By aggregating measurements from a network of radars, we can
build a composite picture for large areas, as shown in Figure 2.12 for our
February 2014 cyclone at a later stage of its development. Recall that
precipitation tends to concentrate along the temperature fronts and in the low
pressure center. On February 21, the cyclone center had moved over
Wisconsin, and the bands of precipitation can be seen to delineate the cold
front (from East Texas to Illinois), the warm front (from Michigan to New
York), and what we will call the occluded front (arching back over
Wisconsin).
 Figure 2.12. Example of a radar image of precipitation over the United
States on February 21, 2014, at 00:00 UTC.

A more advanced type of radar, called Doppler radar, can also detect the
shift in frequency of the return signal, as compared to the emitted signal,
caused by the horizontal motion of the raindrops. This shift in frequency can
be translated into a speed, which informs us about the wind – specifically,
whether the raindrops are moving toward or away from the radar.
Finally, the latest radars have the ability to utilize polarized radar signals,
i.e., radio waves that are oriented in the vertical and horizontal plane, to
determine the shape of precipitation particles.
 2.5 Satellites
Since the first weather satellite was placed in space in 1960 – TIROS
(Television Infrared Observation Satellite) – satellites have revolutionized
our view and understanding of the atmosphere and weather. They provide
information about cloud coverage (cloud location and height), as well as
information about the structure of weather systems. Importantly, they provide
information both over land and over the ocean, where very few direct
measurements are available. (Recall that the ocean covers two thirds of the
planet.)
Weather satellites carry sensors that measure different kinds of radiation.
These measurements can then be translated into variables of interest for
weather and climate (temperature, cloud coverage, cloud top height, cloud
motion, wind, etc.). For our purposes, we will be interested in three important
types of satellite sensor: visible, infrared, and water vapor.
 2.5.1 Visible Satellite Images
A satellite can be equipped with a visible light sensor. Since Earth does not
emit visible radiation, the only light that will be measured by such a sensor is
visible light from the sun reflected by bright features on Earth, such as
clouds, ice, snow, and to some extent bright land areas. Figure 2.13 shows a
visible image of our February 2014 cyclone captured by GOES-East, an
American geostationary satellite (see Section 2.5.4). Later on February 21,
the cyclone was starting to decay, as we will explain in Chapter 10. At this
fully developed stage, however, the visible image reveals a beautiful display
of clouds spanning thousands of miles. The cold front has now moved further
east and extends from the Gulf of Mexico, along the East Coast, to New
England. The extensive cloud cover over the low pressure area suggests -
precipitation over Canada, and it was indeed snowing heavily over the Great
Lakes. As for the warm front, it extends over the Atlantic Ocean and connects
with another cyclone in its decaying phase over Europe.
 Figure 2.13. Example of a visible satellite image on February 21, 2014, at
17:45 UTC. For orientation, the United States and Mexico appear in light
gray in the western half of the image.

Visible images are useful for identifying and tracking cloud features
when they are illuminated by the sun. As Earth and the satellite rotate away
from the sun to the night side, however, no more sunlight is reflected and no
observation is possible (Figure 2.14). Therefore, we resort to a second type of
sensor to obtain information about Earth independently of the position of the
sun.

Figure 2.14. Geostationary satellites can only take visible images during
the day. A few hours later, both Earth and the satellite have spun away to
the night side.
 2.5.2 Infrared Satellite Images
As we will see in Chapter 4, all objects emit some level of radiation, and the
warmer the object, the more radiation it emits and at a higher frequency. The
sun, for example, emits a large amount of radiation, in the visible part of the
spectrum. Most objects on Earth are not as hot, however, and therefore emit a
lesser amount of radiation, in the infrared (IR) part of the spectrum –
radiation of lower frequency than visible light. Now, the infrared spectrum
encompasses a broad range of frequencies and slight variations in the
temperature of Earth features will lead to slight variations in the amount and
frequency of the infrared radiation they emit. An infrared sensor on a satellite
can be tuned to detect such variations in radiation, and therefore variations in
temperature. Thus, the full picture taken by an infrared sensor, i.e., an
infrared satellite image, is really a “temperature image” of Earth.
It is relatively straightforward to display a visible satellite image, since it
shows in shades of gray what our eyes would see if we were sitting on that
satellite looking down at Earth. A visible satellite image is really a black and
white photograph, as can be seen in Figure 2.14 – or, lately, a color
photograph, with the latest full-color satellite imagery. But how do we
display an infrared satellite image? Our eyes have evolved to see visible
light, but not infrared radiation. Therefore, there is no such thing as “infrared
colors.” So we need to create a “fake” color scale to display the temperature
information contained in the image (Figure 2.15). We display warm features
with darker shades of gray and colder features with lighter shades of gray
(Figure 2.16). Of course, the scale was not chosen randomly: it reflects the
fact that cloud tops, being higher in the atmosphere, are colder than the
surface. Recall that temperature decreases with height. Since it is more
intuitive to represent cloud tops in white, we chose our scale to show cold as
white, and warm as dark gray.

Figure 2.15. Example of an infrared satellite image on February 21, 2014,

at 17:45 UTC, corresponding to Figure 2.13.

Figure 2.16. Color scale for displaying satellite infrared images.

At first glance, the white features in Figure 2.15 indeed seem to match the
clouds of Figure 2.13. After closer inspection, however, we notice that some
of the clouds present in the visible image are absent from the infrared image.
Indeed, clouds are not all found at the same altitude, and therefore do not all
have the same temperature. Clouds with a light gray shading in an infrared
satellite image are colder and therefore higher in the atmosphere, while
clouds with a dark gray shading are warmer and therefore lower in the
atmosphere (Figure 2.17). [Note that, strictly speaking, what we are seeing
are the cloud tops. Therefore, a cloud with a light gray shading can be either a
high cloud or a tall cloud.]

Figure 2.17. The amount of infrared radiation emitted by cloud tops tells
us their temperature, which in turn tells us their altitude. (Note that “light
gray” and “dark gray” here refer to the appearance of cloud tops on an
infrared image, not to the color of the clouds themselves.)

Figure 2.15 tells us that, at this later stage of its development, our
February 2014 extratropical cyclone contains different types of clouds, in
contrast to what seems to be fairly homogeneous cloud cover in the visible
image. This is very instructive to the weather analyst, as it reveals different
aspects of the circulation pattern. The cloud band along the cold front, for
example, is bright white and reaches high in the atmosphere. In contrast, the
cloud spiral that extends over Canada and arches back toward the Great
Lakes, producing heavy snow, is darker in color, and therefore lower in the
atmosphere.
In the end, we should never forget that an infrared image shows us
temperature, rather than the brightness of the clouds. This explains, for
example, why some clouds over the Atlantic Ocean appear in the visible
image shown in Figure 2.13, but not in the infrared image shown in Figure
2.15. They are clouds at very low altitude. They are made of liquid cloud
droplets and highly reflective; therefore, they clearly appear on a visible
image. However, being low in the atmosphere, and therefore at temperatures
close to the temperature of the ocean underneath, they are hardly
distinguishable from the ocean itself. If we were not aware of the true nature
of an infrared image, we could simply overlook the existence of these clouds.
 2.5.3 Water Vapor Images
Our satellite sensor can also be tuned to detect radiation emitted by specific
molecules in the atmosphere. As we will see in Chapter 4, gas molecules
absorb and emit radiation at specific wavelengths due to their molecular
structure. A gas of particular interest to us is water vapor, as it transports
energy through phase changes and plays a major role in the development of
weather systems (see Chapter 5). Therefore, we also equip our satellite with a
water vapor sensor to produce images such as shown in Figure 2.18.

Figure 2.18. Example of a water vapor satellite image on February 21,

2014, at 17:45 UTC, corresponding to Figure 2.13.

In this image, darker colors indicate a drier troposphere, whereas lighter

colors indicate a higher water vapor content. As for infrared imagery, the
altitude of the atmospheric water vapor is a factor in the amount of radiation
recorded by the satellite. In particular, dark and light colors are more
indicative of the presence or absence of water vapor in the mid to upper
troposphere. In other words, a dark region on a water vapor satellite image
might still contain a fair amount of water vapor in the lower troposphere, and
even low clouds. Note, for example, the presence of numerous low cumulus
clouds over the tropical Atlantic in Figure 2.13, even though it is depicted as
“dry” in Figure 2.18.
Water vapor satellite imagery is particularly useful to weather analysts
when it is animated, as it shows the fluxes of water vapor in and out of
weather systems, as well as the intrusions of dry air. Note, for example, the
tongue of dry upper-tropospheric air wrapping around and into the cyclone in
Figure 2.18 (west of the cold front). This highlights another important
attribute of the water vapor image: it shows us structure in the atmosphere
even where there are no clouds.
 2.5.4 Geostationary Satellites
Meteorologists rely primarily on two types of satellite: geostationary and
polar-orbiting. As their name indicates, geostationary satellites are stationary
with respect to Earth (“geo-”). More precisely, they are stationary with
respect to a specific location on Earth, so that they constantly observe that
same location over time. Since Earth rotates, the satellite needs to move at the
exact same speed, otherwise it will start leading or lagging behind the
location it is supposed to observe. More exactly, it needs to orbit Earth at the
same angular speed, i.e., it needs to describe the same angle over time. At
that speed, the satellite will experience a centrifugal force that will tend to
pull it out of its orbit. This is similar to the outward pull you experience when
you turn at high speed with your car. For the satellite to stay in orbit, the
gravitational pull of Earth needs to balance the centrifugal force exactly. This
perfect combination happens in the equatorial plane at about 36 000 km from
Earth’s center (about 22 000 miles), which defines the geostationary orbit
(see Figure 2.19). For reference, Earth’s radius is about 6400 km. So, a
geostationary satellite is at about five to six times that distance from Earth,
above the equator. At this location, it can hover above a point on Earth,
without falling toward Earth.

Figure 2.19. Geostationary satellites orbit Earth at the same angular speed
as Earth.
 Since a geostationary satellite always remains above the same location, it
is useful for tracking weather changes over a specific country or region. In
the United States, for example, one satellite called GOES-West
(Geostationary Operational Environmental Satellite-West, currently GOES-
15) covers the western USA and part of the Pacific Ocean, while a second
satellite called GOES-East (currently GOES-16) covers the eastern USA and
part of the Atlantic Ocean. The full-disk GOES-East visible, infrared, and
water vapor images corresponding to Figures 2.13, 2.15, and 2.18 are shown
in Figure 2.20.
 Figure 2.20. Example of full-disk visible (a), infrared (b), and water vapor
(c) images taken by GOES-East on February 21, 2014, at 17:45 UTC.

To obtain a full view of Earth, we need several satellites orbiting Earth in

concert, each taking snapshots from a different vantage point. We can do so
with a constellation of satellites, as shown in Figure 2.21 with a view of Earth
above the North Pole. Europe and Africa are currently covered by two
Meteosat satellites, India and Central Asia by INSAT (Indian National
Satellite System) and a number of Russian, European, and Chinese satellites,
while Japan, Eastern Asia, and Australia are covered by MTSAT
(Multifunctional Transport Satellite). Since Earth and the satellites rotate at
the same speed, you can turn this book to model their motion over time. If
you turn it counterclockwise by 90º, you will obtain their new position, as
well as that of Earth, in six hours.
 Figure 2.21. Polar view of Earth (looking down at the North Pole)
showing the constellation of geostationary satellites observing Earth from
all directions.

The images obtained by the different satellites can be composited to

produce a global image such as the one shown in Figure 2.22. These global
images are very instructive, as they reveal many aspects of the general
circulation of the atmosphere, to which we will return in Chapter 9. In
particular, it appears that some weather patterns are characteristic of specific
latitudinal bands. At the equator, for example, the patchy clouds correspond
to convective systems due to the convergence of surface winds (see Chapter
6). North and south of the equator, we can see two bands that are relatively
free of clouds, at the latitude of the Sahara desert in the northern hemisphere
and at the latitude of Australia in the southern hemisphere. These regions are
called the subtropics and are characterized by subsidence (i.e., downward
motion), which prevents the formation of clouds (see Chapter 9). Farther
north and south, at the latitude of the United States and the Southern Ocean,
we find the midlatitudes, i.e., the “middle” latitudes, halfway between the
equator and the poles. That is where we find our February 2014 midlatitude
cyclone, along with four other cyclones in the northern hemisphere, while
eight to ten midlatitude cyclones can be observed over the Southern Ocean at
different stages of development (see Chapter 10). They can be recognized by
their comma cloud signature. Northern and southern midlatitude cyclones are
mirror images of each other, for reasons we will explain in Chapter 8.
Because the midlatitudes are “outside” of the tropics, i.e., north and south of
the tropics, we sometimes refer to them as the extratropics, and to
midlatitude cyclones as extratropical cyclones.
Figure 2.22. Global satellite image obtained by compositing several
images from different satellites on February 20, 2014, at 18:00 UTC.
 2.5.5 Polar-Orbiting Satellites
Geostationary satellites are convenient to monitor the weather at a specific
location. However, they need to do so from quite far away (36 000 km) and
their sensors cannot detect small clouds or small features of Earth’s surface.
Much as a digital camera can only zoom up to a point, and produces pixelated
images beyond that point, satellite sensors can only detect features down to a
certain size. Their current resolution is 0.5 km for visible images and 2 km
for infrared images, which means that any detail inside a 0.5 km box will be
averaged into a single pixel in a visible image. The small cumulus clouds
shown schematically in Figure 2.23, for example, are not detectable and will
blend into a white or gray pixel in the visible image.

Figure 2.23. Features smaller than 0.5 km blend into a unique pixel of
homogeneous color in a visible satellite image.

To gain in resolution, we need to bring our satellite closer to Earth. But

this has a cost: the gravitational pull will be stronger, and the satellite will
need to go faster to generate enough centrifugal force to balance gravity and
remain in orbit. That is the choice we make with polar-orbiting satellites.
They orbit Earth at about 800 km (one eighth of Earth’s radius), which allows
them to resolve details of Earth’s surface down to about 250 m, but they
complete an orbit in about 100 minutes, which prevents them from staying
above the same location. They orbit Earth in a pole-to-pole direction,
however, and provide valuable information about the polar regions, which are
not captured by geostationary satellites. The main disadvantage of the polar-
orbiting satellite is that, being so close to Earth, it cannot get a full view of
the planet, but only a narrow swath along its path. While it completes one
orbit, however, Earth rotates eastward, which means that successive swaths
sample different regions of Earth (Figure 2.24). By compositing successive
swaths, we can obtain a full picture of Earth in about 24 hours – with some
gaps (Figure 2.25). When combining both ascending and descending swaths,
we can obtain a picture of Earth in 12 to 14 hours.

Figure 2.24. Polar-orbiting satellites orbit Earth in about 100 min at an

elevation of about 800 km. As Earth rotates underneath, the satellite
samples a new region of Earth in each orbit.
 Figure 2.25. A polar-orbiting satellite covers Earth in about 24 hours, with
minor gaps between the successive swaths, as illustrated with this image
from the Moderate-resolution Imaging Spectroradiometer (MODIS) aboard
NASA’s Terra satellite on April 19, 2000.

Figure 2.25 shows a number of comma-shaped cloud structures typical of

extratropical cyclones, spiraling in opposite directions, such as we saw in
Figure 2.22. It also features a number of isolated clouds at the equator – the
signature of convective systems. In the northern hemisphere, the cloud-free
subtropical band extending from Mexico to North Africa, the Arab Peninsula,
and India, is particularly well delineated. We will explain the existence of all
these features in time, as we learn more about the general circulation of the
atmosphere and individual weather systems. In the meantime, Appendix 2.1
shows some examples of useful cloud shapes to memorize.
 Summary
We depict weather patterns by creating a variety of maps and images.

► Surface station maps contain measurements taken at individual

weather stations and buoys using a conventional station model.
► Temperature maps show the structure of the temperature field using
colors and contours called isotherms. They are useful for identifying
air masses and fronts. Temperature fronts appear as regions of larger
temperature gradients, characterized by a concentration of isotherms
over short distances.
► Pressure maps reveal the structure of the atmospheric pressure field
using contours called isobars. They are useful for identifying regions
of high and low pressure, either closed (i.e., highs and lows), or open
(ridges and troughs). High pressure often corresponds to fair weather,
while lower pressures often correspond to clouds and precipitation.
► Troughs and ridges also appear on upper-level maps, where we
display the height of a given pressure surface, rather than the pressure
at a given height (for the same overall interpretation).
► Radar images show us the position and intensity of precipitating
particles. Doppler radar images show us the speed of the precipitation
particles relative to the radar.
► Visible satellite images show us the visible light from the sun that is
reflected, more or less efficiently, by the various elements of the Earth
system, most importantly by clouds.
 ► Infrared satellite images show us the temperature of the various
elements of the Earth system, and in particular the temperature of
cloud tops. Since temperature typically decreases with height over the
depth of the troposphere, this information can be used to infer the
altitude of cloud tops, and therefore the type of clouds and the
structure of weather systems.
► Water vapor satellite images show us the water vapor content of the
middle and upper troposphere, and are useful for identifying the fluxes
of water vapor and dry air.
► Geostationary satellites orbit Earth at about 36 000 km above the
equator, where their angular velocity is the same as that of Earth. They
can observe the same location over time, but with less resolution than
polar-orbiting satellites, which orbit Earth at about 800 km. These
satellites, however, observe the planet along successive swaths that
cover different areas of Earth at different times.

Let’s get started

Even though our meteorologist toolkit is not complete, it is sufficient to start
investigating the atmosphere and begin learning about the mechanisms of
weather. With our basic tools in hand, let’s get started on our journey, and lay
down the first building blocks of our understanding.

Appendix 2.1 Important Satellite Cloud Signatures

Satellite images allow us to track the cloud signature of various weather

systems, such as fronts, cyclones, hurricanes, and thunderstorms. A
developing wave, for example, in the first stage of cyclone development, will
appear with two levels of clouds (Figure 2.26).
 Figure 2.26. Developing wave cyclone, with two levels of cloud.

As the cyclone matures, the fronts become more substantial. Cold fronts
are usually associated with long, narrow bands of clouds, whereas warm
fronts are associated with more extensive cloud decks (Figure 2.27).
 Figure 2.27. Cloud signature of cold and warm fronts.

Low pressure systems are often associated with a cloud swirl spinning
counterclockwise (in the northern hemisphere) around the low (Figure 2.28).
 Figure 2.28. Cloud structure around a midlatitude low pressure system.

Cold fronts can spawn strong convection in the form of individual,

growing convective cells (Figure 2.29).
 Figure 2.29. Convective cells in satellite imagery.

Appendix 2.2 Contiguous USA Reference Map

Figure 2.30. Contiguous states of the USA and surrounding geography.
CHAPTER 3
Our Atmosphere: Origin, Composition, and Structure
CONTENTS
3.1 Aspect
3.2 Composition
3.3 Origin and Evolution
3.4 Future Evolution
3.5 Vertical Structure
Summary

The atmosphere is a thin layer of gas and particles with a well-defined

vertical structure. Weather systems continually stir the lowest layer of the
atmosphere, and we experience the clouds, precipitation, and wind that they
produce. Before we investigate the formation and evolution of these weather
systems, let us look deeper into the nature and composition of the
atmosphere, as well as into its origin and future evolution. We will also
examine the vertical structure and properties of different layers of the
atmosphere.
 3.1 Aspect
We can think of the atmosphere as a thin layer of gas surrounding Earth. It is
attracted to Earth by gravity and is extremely thin: 99% of the atmosphere is
found below 30 km while, by comparison, Earth’s radius is about 6380 km. A
common analogy compares the thickness of the atmosphere around Earth to
the thickness of the skin around an apple. This very thin aspect of the
atmosphere has important implications: since large weather systems such as
midlatitude cyclones and storms can be several thousand miles wide but less
than 10 miles high, we need to think of them as essentially “flat” with a large
horizontal extent. This is not to say that vertical air motion is unimportant in
storms and hurricanes – in fact, it is precisely vertical air motion that
determines the formation of clouds and precipitation. But it is useful to
remember that air masses are mostly flat and sliding around near the surface
of Earth.
 3.2 Composition
Besides some ions and larger solid and liquid particles, the atmosphere is
made primarily of gas molecules. The atmosphere mostly contains molecular
nitrogen, N2 (78%) and molecular oxygen, O2 (21%), while argon and other
trace gases make up the remaining one percent (see Table 3.1). Because the
proportion of the most abundant gases does not change much with time and
altitude (up to about 80 km), we call them permanent gases. These
permanent gases, however, coexist with many variable constituents that, in
spite of their small concentration, can greatly impact weather and climate.

Table 3.1. Composition of the atmosphere

Permanent gases (% of volume of Variable gases (% of volume of

dry air) moist air)

Nitrogen 78 Water vapor 0–4

Oxygen 21 Carbon dioxide 0.04

Argon 0.9 Methane 0.0002

Neon 0.002 Nitrous oxide 0.00003

Helium 0.0005 Ozone 0.000004

Water vapor, for example, represents anywhere from almost 0 to 4% of

the atmosphere, and can condense to form liquid water (dew, cloud droplets,
and rain) or solid water (frost, snow, and ice). The concentration of water
vapor can change greatly in time and space. For example, it is typically
greater over the ocean and less over deserts, and it fluctuates at a given
location as relatively drier and moister air masses move around with weather
systems. Also, the water vapor concentration is typically much greater in the
lower atmosphere, closer to the surface, since the ocean is the primary source
of water for the atmosphere and, as we will see, warmer air near the surface
can contain more water vapor than cold air.
Carbon dioxide (CO2) represents about 0.04% of the atmosphere, and
has been increasing since the Industrial Revolution (a topic we will explore
further in Chapter 15). Its concentration also varies in time and space. It
decreases annually in northern hemisphere spring and summer, when
vegetation grows and takes up CO2 during photosynthesis on land areas of
the northern hemisphere, and increases in fall and winter in the northern
hemisphere, when vegetation dies, decays, and releases the CO2 back into the
atmosphere. It also tends to be higher around cities and industrial regions,
where the combustion of fossil fuels in factories, power plants, and cars, for
example, releases CO2.
Methane (CH4) has even smaller concentrations, but can be found in
elevated concentration near sources such as natural gas reserves, swamps,
rice paddies, garbage dumps, and livestock.
Ozone (O3) is found in very small concentrations in the lower
atmosphere, which is a good thing for us, as ozone is a dangerous pollutant
that can cause severe respiratory problems. Ozone, however, plays an
important role higher in the atmosphere, at about 20 to 50 km altitude, in a
region called the stratosphere (see Section 3.5).
The atmosphere also contains many other gases, such as nitrous oxide
(N2O), released mostly by bacteria in nature and by fertilized soils and
various industrial processes. It also contains small solid particles called
aerosols, mostly blown from the surface into the air, such as soil particles,
dust, pollen, smoke from forest fires, and pollution particles from factories
and vehicle exhausts. Salt particles contained in water droplets blown by the
wind from the surface of the ocean as sea spray can also turn into aerosols, as
the airborne salty water droplets evaporate, leaving behind tiny salt particles.
We will see in Chapter 6 that aerosols play a crucial role in the formation of
cloud droplets and, ultimately, rain.
 3.3 Origin and Evolution
When was the atmosphere created? Has it always had the same composition?
Should we expect to see it change further? To answer these questions, we
need to step back and gain some perspective on the formation and evolution
of Earth.
Strictly speaking, our short history of the atmosphere starts 13.7 billion
years ago, when the universe appeared, and with it the first elements,
hydrogen (H) and helium (He). About 4 to 5 billion years ago the solar
system formed when the solar nebula flattened into a spinning disk and dust
grains started colliding and sticking together to form clumps of matter – a
process known as accretion. These clumps of matter further grew over 100
million years to become the planets of our solar system, traveling around the
sun in almost circular orbits.
It is reasonable to think that, when Earth was formed, its atmosphere
must have consisted of the prevalent gases in the solar system, in particular
such light gases as hydrogen, helium, methane, ammonia, and water vapor.
However, a large fraction of these light gases probably escaped to space.
Very early in Earth’s development, gravity was much lower than today, and
the temperature much higher, so that it was easier for light molecules like
hydrogen and helium to reach their escape velocity and overcome the weak
pull of gravity. The collision of Earth with other large bodies traveling the
solar system might also have helped eject the lighter gases out of the early
atmosphere of Earth.
 The scientific consensus is that a secondary atmosphere formed from
gases released from Earth’s interior by volcanic eruptions. This outgassing
process enriched the atmosphere in water vapor, methane, and carbon
dioxide, as well as nitrogen and ammonia. Molecular nitrogen is a stable,
nearly inert gas and the non-reactive property of nitrogen explains how it
simply accumulated over time and why it now represents a large fraction of
the air we breathe.
It is also very likely that a large number of comets containing water in
the form of ice bombarded Earth and enriched our atmosphere with water
vapor. This led to an early atmosphere containing mostly water vapor (about
80%), carbon dioxide (10%), 1 to 5% of nitrogen, and about 5% other gases.
Our present atmosphere, however, contains mainly nitrogen (78%) and
oxygen (21%), with little water vapor, but a vast ocean underneath. What is
the origin of the oceans and the water that makes clouds, rain, snow, and the
rivers and aquifers that are so crucial to life on Earth?
It was possible for water to exist in the vapor phase because the early
atmosphere was also very hot. As Earth slowly cooled over millions of years,
water vapor condensed into liquid water (as we will see in Chapter 5 and 6)
and precipitated to form rivers, lakes, and oceans. In the process of creating
reservoirs of liquid water at Earth’s surface, carbon dioxide was also removed
from the atmosphere. In the same way that carbon dioxide can be dissolved in
a soft drink, early atmospheric carbon dioxide dissolved and accumulated in
the oceans. Most of the carbon dioxide did not remain in the ocean, however,
but combined with other elements to form carbonates that eventually settled
as sediments in the sea floor. On very long timescales, these sediments move
under the action of plate tectonics and subduct under continental margins.
The sediments melt in Earth’s crust and volcanic eruptions that tap into these
regions restore carbon dioxide to the atmosphere.
Another important influence on atmospheric composition comes from
chemical weathering of rocks in contact with air and water. For example,
during oxidation, iron in rocks reacts with oxygen to form rust, which
reduces atmospheric oxygen. Another example is carbonation, when carbon
dioxide dissolved in rainwater reacts with limestone, which reduces
atmospheric carbon dioxide. There are other chemical, biological, and
physical controls on atmospheric composition, but the processes described
here are thought to be the most important on long timescales.
Our story so far accounts for the small concentration of lighter gases
such as hydrogen and helium, for the dominance of nitrogen, the existence of
water oceans, and for the presence of carbon dioxide in sediments and in our
atmosphere. But it does not account for the presence of oxygen, without
which animal and human life would be impossible. About 2.3 billion years
ago, primitive, anaerobic bacteria developed and started removing carbon
dioxide from the atmosphere while replacing it with oxygen. The increase in
oxygen then allowed the evolution of aerobic organisms and subsequent
forms of multicellular life (see Box 3.1).

Box 3.1. Photosynthesis and the Carbon Cycle

Carbon dioxide can be removed from the atmosphere by

photosynthesis, by which CO2 combines with water (H2O) to form
organic matter (CH2O) and release oxygen (O2):

CO2+H2O →lightCH2O+O2
 Conversely, through combustion or decay (i.e., oxidation), organic
matter turns into water vapor and carbon dioxide:

CH2O+O2→ H2O + CO2

The two chemical reactions form a short cycle through which carbon
is exchanged between plant matter and the atmosphere. This short
cycle essentially follows the Northern Hemisphere midlatitude
seasons and therefore its timescale is on the order of a year. By
contrast, the removal of carbon by absorption of carbon dioxide in the
ocean, formation of carbonates (sea shells) and sedimentation in the
sea floor occurs on timescales of hundreds to millions of years,
similarly to its counterpart, the release of carbon dioxide during deep
sea volcanic eruptions and the collision of tectonic plates. Thus, it is
important to note that a disruption of the short cycle (for example,
increasing the concentration of carbon dioxide by increasing the
burning of fossil fuels) cannot be fully accommodated by the long
cycle (removal by absorption in the ocean), because the latter takes
too long.

Our story is still missing one aspect. The incoming solar radiation was
too destructive for life to survive in direct contact with sunlight. As we will
discuss in Chapter 4, sunlight is composed of ultraviolet (UV) radiation, a
high energy form of light that damages DNA and protein, causing, for
example, sun burns and skin cancer in humans. Land areas on Earth only
became habitable when the atmosphere contained enough oxygen that
significant amounts of stratospheric ozone (O3) formed, which is highly
efficient at absorbing ultraviolet radiation (see Box 3.2). This speaks to the
importance of the ozone layer – see Section 3.4 – and demonstrates how the
existence of life on Earth is the result of a subtle balance of atmospheric
composition, radiation, and heat, reached over the course of billions of years.

Box 3.2. Chapman Reactions

Ozone (O3) is naturally created in the stratosphere by the following

reactions:

O2→UVO+O (1)

O+O2+M→ O3+M (2)

In reaction (1), an oxygen molecule is photolyzed by UV radiation to

form two oxygen radicals. In reaction (2), one of the radicals
combines with a new oxygen molecule to form an ozone molecule.
This second reaction releases heat, which is absorbed by the third
molecule, represented by M, which can be anything, but most likely
nitrogen.
Ozone is also naturally destroyed by either of the following
reactions:

O3→UVO+O2 (3)

O3+O→ O2+O2 (4)

In reaction (3), an ozone molecule is destroyed by UV radiation to
form an oxygen radical and an oxygen molecule. In reaction (4), an
ozone molecule combines with a radical to form two oxygen
molecules. These four reactions are referred to as “Chapman
reactions” after Sydney Chapman, who first proposed them in 1930 to
explain the presence of the ozone layer in the stratosphere. They are
simultaneously taking place, with ozone molecules constantly created
and destroyed, keeping the system in dynamic equilibrium (see
Appendix 3.1).
 3.4 Future Evolution
Is the composition of our atmosphere still changing? And if so, how fast? We
have been monitoring the concentration of atmospheric gases for more than
50 years now, and climate scientists have been reconstructing past
concentrations from a variety of sources, such as gas bubbles trapped in
Antarctic ice. We can now ascertain that, while the concentration of
permanent gases such as oxygen and nitrogen has not changed much recently,
the concentrations of certain variable gases, such as carbon dioxide and
methane, have changed drastically over the past 200 years. The concentration
of carbon dioxide, for example, has increased from about 280 ppm (parts per
million, or 0.00028) to 370 ppm (see Figure 3.1) between about 1750 and
2000 – and to over 400 ppm since then. We know that this increase is largely
due to the burning of fossil fuels since the Industrial Revolution, and that this
influx of carbon dioxide is taking place at a much faster rate than the removal
of carbon dioxide from the atmosphere by the ocean. Since carbon dioxide,
among other gases whose concentrations are also changing, absorbs infrared
radiation and heats Earth (through the greenhouse effect, to be discussed in
Chapter 4), the increase in the concentration of these greenhouse gases is also
altering the heat budget of Earth and is increasing its temperature, leading to
global warming – a subject to which we will return in Chapter 15.
 Figure 3.1. Carbondioxide concentration during the last millennium.

In the twentieth century, human activities have also altered the

concentration of ozone in the atmosphere. Recall that ozone is found in very
small concentrations in the lower atmosphere, but plays a significant role in
the stratosphere, at about 20 to 50 km altitude. There it absorbs UV radiation
from the sun (see Box 3.2) and prevents much of it from reaching the surface,
where it would damage life (plants and animals) and cause severe problems
for us, including skin cancer and eye cataracts. While the ozone concentration
remains low in the troposphere, it increases in the stratosphere in proportion
to the other gases. We refer to that region as the ozone layer. The existence
of that layer was crucial for the appearance and development of life on Earth
(especially on land).
The concentration of stratospheric ozone was reduced globally in the
late twentieth century, especially over Antarctica in late winter and early
spring, creating the so-called ozone hole (Figure 3.2), and to a lesser extent
over the Arctic. The reduction in ozone concentration was found to be due to
the emission of chlorofluorocarbons (CFCs) into the atmosphere. CFCs are
synthesized gases containing chlorine that were used as refrigerants,
propellants in spray cans, solvents, and as blowing gas to make foam. CFCs
are very stable, which is attractive for industrial use, but also means that,
under the right conditions, a very small amount of CFCs can destroy many
ozone molecules. Through the Montreal Protocol, a 1987 international
agreement, the production of CFCs has been largely eliminated in the last few
decades and there are signs that the ozone layer is beginning to recover.
Unfortunately, the replacement for CFCs, hydrofluorocarbons (HFCs), turn
out to be very powerful greenhouse gases, contributing to global warming.

Figure 3.2. Ozone concentration in the atmosphere in October 2015.

Dobson units measure the total amount of ozone in an atmospheric
column. The Antarctic ozone hole corresponds to the region of reduced
concentration in blue and purple.
 3.5 Vertical Structure
Now that we have a better understanding of the composition of the
atmosphere, we can have a closer look at the distribution of the constituent
gases and the resulting structure of the atmosphere, in particular with respect
to density, pressure, and temperature.
We already know that pressure decreases rapidly with height (Figure
1.6), or, conversely, pressure increases as we approach the surface. If air were
not compressible (like water, to a large extent), air density would remain the
same throughout the atmosphere. But air, being a gas, can be compressed,
and indeed is very much compressed close to the surface, due to the weight of
the overlying air pressing down on it. Thus, as we approach the surface, not
only does pressure increase, but also the air molecules are compressed closer
together, which means that the air density also increases (i.e., there are more
air molecules in a given volume). A cubic meter typically contains about 1 kg
of air at sea level, whereas it contains only about 0.7 kg of air at 500 hPa
(about 5.5 km altitude).
One might expect temperature also to decrease with height, since the
atmosphere is essentially heated from below, through conduction of heat
from the surface (as well as convection and radiation, as we will see in
Chapter 4). If you think of the surface as a radiator, you can easily picture
how the atmosphere should be warmer closer to the radiator (closer to the
surface) and colder as we move away from the radiator (up in the
atmosphere). And this is indeed what we observe in the first 10 km or so of
the atmosphere – a layer called the troposphere (see Figure 3.3).
 Figure 3.3. Average vertical temperature profile of the atmosphere.

Recall that ozone absorbs UV radiation from the sun, which heats the
atmosphere where the radiation is absorbed. Since ozone concentration peaks
around 20 to 50 km altitude, temperature also reaches a maximum there,
although it is only a local maximum, relative to regions above and below (the
temperature reaches only about 0 °C at the maximum). The presence of ozone
creates an interesting temperature profile, in which the temperature first
decreases with height in the troposphere, then increases in the stratosphere,
before decreasing again in the mesosphere. Still higher up, in the
thermosphere, the air density is extremely low, and the very small amount of
air molecules there absorb the most energetic rays from the sun, which
provides them with extremely high kinetic energy. This high kinetic energy
of the molecules equates to a high temperature, but, due to the low air
density, there is almost no heat there (summed over the number of air
molecules). With this overall structure in mind, let us lay out some facts and
numbers that will help us in our later description of weather phenomena.
The troposphere is about 10 km deep and extends from the surface
(1000 hPa) to about 200 hPa. The temperature drops at an average rate of
about 6.5 °C/km (or 3.6 °F per thousand feet). We call this rate of decrease
with altitude the average environmental lapse rate. Most of the water vapor
in the atmosphere is found in the troposphere, and the troposphere is also the
layer of the atmosphere where most weather takes place. For reasons that we
will explore later, weather systems cannot extend easily into the stratosphere,
and nearly all clouds are found below the tropopause, the transition between
the troposphere and the stratosphere. The tropopause is found at lower
altitudes at high latitudes (about 7 or 8 km) and can be found as high as 16
km near the equator. As we will see, air motion near the tropopause is very
important for understanding the evolution of weather patterns, as that is
where the midlatitude jet streams are found, the “rivers” of high winds that
meander and steer midlatitude cyclones along their course.
The stratosphere extends from about 10 to 50 km (200 to 1 hPa). Due
to the presence of ozone and the absorption of ultraviolet radiation, the
temperature increases with height. Since temperature normally decreases with
height, the situation in the stratosphere is the inverse of what it should be, and
we refer to such a temperature change with height as a temperature
inversion, or just an inversion. We sometimes talk about the “stratospheric
temperature inversion.” As for the troposphere and the tropopause, the
stratosphere has an upper limit called the stratopause.
 The mesosphere extends from about 50 to 90 km (1 to 0.01 hPa). The
ozone concentration decreases with height, along with temperature, up to the
mesopause. Finally, we find the thermosphere above 90 km, with pressures
less than 0.01 hPa and high kinetic energy molecules at extremely low
densities. In some applications, it is useful to define a region of the
atmosphere above 60 km where atoms and molecules tend to gain or lose
electrons, becoming ions. We refer to this region as the ionosphere, which is
important for the reflection of radio waves, for example.
In the rest of this book, we will be mostly concerned with the
troposphere, where most weather takes place, and with the presence of the
temperature inversion above the tropopause, which creates stable conditions
in the stratosphere and has important implications for cloud formation and
storm development.
 Summary
The atmosphere is relatively thin compared to its horizontal extent: 99%
of its mass is found below an altitude of 30 km.
The atmosphere is composed of:

► Permanent gases, whose proportion does not change much over

time and below about 80 km: nitrogen (78%), oxygen (21%), and
trace gases (1%) such as argon.

► Variable gases, whose concentration can vary greatly over time

and space, including: water vapor, carbon dioxide, methane,
ozone, and nitrous oxide.

► Small airborne solid particles called aerosols, including: soil

particles, dust, pollen, smoke from forest fires, exhaust from
power plants, factories and vehicles, and salt particles left behind
after sea spray evaporates.
The composition of the atmosphere has varied over the course of Earth’s
4.5 billion year history, because of various processes such as volcanic
eruptions, photosynthesis, and chemical weathering of rocks.
The concentration of certain gases has rapidly increased in recent
years, in particular since the Industrial Revolution, due to human
activities. For example:

► The concentration of carbon dioxide has increased due to the

burning of fossil fuels, rising from 280 ppm to more than 400
ppm in about 200 years.
 ► The concentration of stratospheric ozone over the South Pole in
early Spring has been greatly reduced since the late twentieth
century, due to the emission of chlorofluorocarbons (CFCs) into
the atmosphere, synthesized gases containing chlorine that can
destroy ozone molecules.

Pressure and density both decrease with height in the atmosphere.

Temperature decreases with height on average in the troposphere (from the
surface (1000 hPa) to roughly 200 hPa, at about 10 km altitude), at an
average environmental lapse rate of about 6.5 °C/km. Temperature then
increases with height in the stratosphere (from about 200 to 1 hPa, or 10 to 50
km), due to the presence of ozone and the absorption of ultraviolet radiation,
creating a temperature inversion.
The region separating the troposphere from the stratosphere is called the
tropopause, and is found at about 200 or 250 hPa, between 7 and 8 km (at
the poles) and as high as 16 km (at the equator).

Appendix 3.1 Dynamic Equilibrium

As we explore the composition and evolution of the atmosphere, an important

concept emerges that we will use extensively in the rest of this book: the
notion that a system can be subject to change through inputs and outputs of
matter and energy, but nevertheless remain relatively constant through these
changes.
A useful analogy is that of a bucket that we are filling with water. As
water flows into the bucket (the input), the bucket fills up; this is not an
equilibrium situation. But if the bucket has a sufficiently large hole, some
water flows out at the same time (the output), so that, at some point, the
amount of water that flows out is exactly equal to the amount of water that
flows in (output = input). Beyond this point, the amount of water in the
bucket does not change: the bucket “system” is in dynamic equilibrium
(Figure 3.4). Change is taking place in the bucket, but the amount of water in
the bucket remains constant.

Figure 3.4. Bucket analogy for dynamic equilibrum.

To use another analogy, the bucket is similar to a bank account in which

you deposit funds, while withdrawing money at the same time. If you spend
exactly the same amount as you deposit, your account balance does not
change.
The bucket analogy illustrates, for example, the dynamic equilibrium
described in Box 3.2: the input of ozone molecules created by Chapman
reactions (1) and (2) is balanced by the destruction of ozone molecules
described by equations (3) and (4). The input is balanced by the output and,
under normal conditions, the concentration of ozone molecules in the
stratosphere remains relatively constant.
Now, if more water flows out of the bucket through the hole than water
flows in from the tap, i.e., if the output is greater than the input, the amount
of water in the bucket decreases. This is a good analogy for the destruction of
the ozone layer that took place in the late twentieth century. As more ozone
was destroyed by the chlorofluorocarbons than was created by the Chapman
reactions (in particular in the southern hemisphere at high latitude), the ozone
layer over Antarctica became depleted over time (in particular in southern
spring).
Conversely, if more water flows in from the tap than water flows out
through the hole, i.e., if the input is greater than the output, then the bucket
fills up. This is a good analogy for the increase in atmospheric carbon dioxide
that has taken place since the Industrial Revolution. Since more carbon
dioxide has been injected into the atmosphere (via the burning of fossil fuels,
for example) than has been removed from the atmosphere (via absorption by
the ocean, for example), carbon dioxide has been accumulating in the
atmosphere, as evidenced by the increasing concentrations shown in Figure
3.1.
This analogy applies to many other phenomena in the atmosphere, and
the concept of dynamic equilibrium will be useful in subsequent chapters
when discussing the flow of matter, energy, and momentum through parts of
the atmosphere.
CHAPTER 4
Heat and Energy Transfer
CONTENTS
4.1 Conduction
4.2 Convection
4.3 Radiation
4.4 Radiative Interactions
4.5 Radiation and Weather
Summary

Energy enters the Earth system in the form of solar radiation, preferentially
heating the ground and the tropical latitudes. The resulting temperature
contrasts set the atmosphere in motion, as heat is transferred upward and
poleward. Weather is largely the result of this transfer of heat by atmospheric
motions. In this chapter, we will explore the nature of heat and radiation, the
origin of temperature contrasts on Earth, the mechanisms by which energy is
transferred, and the implications for weather.
Although we might tend to think about weather in terms of wind and rain,
weather is in fact largely the result of the redistribution of heat in the
atmosphere. A heat imbalance, i.e., a disequilibrium in the distribution of
heat between different regions of Earth, causes a transfer and redistribution of
heat in the atmosphere. The resulting movement of air masses creates wind,
storms, clouds, and rain. Therefore, if we are to understand the development
of weather systems, we need first to understand the source of the heat
imbalance and the nature of heat transfer in the atmosphere.
It is accepted in thermodynamics (the field of science that is concerned
with the energy of systems) that heat does not normally flow from cold to
warm objects (unless some other process is at work): it necessarily flows
from warm to cold, i.e., in a direction that will bring the system to a state of
equilibrium – technically, a state of lower potential energy. In the
atmosphere, this transfer of heat can be achieved by three processes,
conduction, convection, and radiation, which we will now explore in more
detail.
 4.1 Conduction
Earlier we defined heat as the kinetic energy of atoms and molecules, or their
energy of motion. In solid matter, motion is reduced to the vibrations of the
atoms and molecules. In a liquid or a gas, molecules can move around and
their kinetic energy describes both their vibrations and their displacement.
When very energetic molecules (i.e., a hot substance, such as hot tea) are in
contact with slower molecules (e.g., a cold spoon placed in the tea), the more
energetic molecules necessarily impart some of their energy to the slower
molecules when they bump into them (Figure 4.1). In the process, the slower
molecules gain energy (i.e., the spoon warms up) while the faster molecules
lose energy (i.e., the tea cools down). As a result, energy is redistributed
more uniformly throughout the system. Following the same reasoning, the
atoms in the bottom part of the spoon, being now more energetic than those
in the top part of the spoon, will impart kinetic energy to the slower atoms by
collision, in the form of faster vibrations, and heat will rise up the spoon.
Note that the atoms in the spoon do not need to be displaced to transfer their
heat up the spoon: they merely need to bump into adjacent atoms and share
their extra energy of motion by contact. We refer to this transfer of heat by
contact, collision, and transfer of molecular vibrations as conduction. By its
very nature, conduction always transfers heat from warm to cold regions.
Thus, when you touch a cold object, even though you might be under the
impression that “coldness” is conducted into your hand, it is really your own
heat that is conducted into the cold object.
 Figure 4.1. Conduction of heat through a solid.

While conduction is very efficient in solids (in particular in metals), that

is not the case for fluids such as liquids and gases. In particular, air is a very
poor conductor, which is one reason why conduction is not the main process
by which heat is transferred in the atmosphere. Indeed, conduction plays a
significant role only at the surface, where heat is exchanged by contact
between solid material and the very first layer of air in contact with it, at the
centimeter scale. That is the case, for example, when heat is transferred from
a hot surface to the air on a hot summer afternoon, making the lowest few
centimeters of the atmosphere very warm, and when heat is transferred from
the air to a cold surface in winter, making the lowest few centimeters of the
atmosphere very cold. If conduction were the only process at play, however,
the temperature of the troposphere would be fairly uniform, while large
temperature changes would be observed in the first centimeters of the
atmosphere close to the surface. When air near the surface warms by
conduction, however, other processes come into play.
 4.2 Convection
When an air parcel warms, its temperature, pressure, and density all change
according to the ideal gas law (recall Box 1.3). The pressure inside the air
parcel, however, always instantaneously adjusts to the pressure of the
environment. If the air parcel is heated by conduction from a hot surface, for
example, and the environmental pressure is 1013 hPa, the pressure of the air
parcel remains equal to 1013 hPa (as long as it stays at sea level). Therefore,
as described by the ideal gas law (pV = nR*T), the volume and the
temperature of the air parcel must change in proportion to each other; i.e.,
since the pressure is fixed, if the temperature increases, then the volume must
be increasing as well. Indeed, since the mass of the parcel is unchanging (n is
fixed), the heated parcel expands: it is subject to thermal expansion.
Using the other form of the ideal gas law (p = ρRT), we can also say
that, since the pressure is fixed, the density and the temperature of the air
parcel are inversely proportional, i.e., if the temperature increases, then the
density must decrease. Indeed, when an air parcel is heated and expands, the
same mass of air is contained in a larger volume, and the density decreases.
Being less dense than its surroundings, by Archimedes’ principle, the air
parcel is buoyant and rises. Hence the commonly used expression, “warm air
rises”. (We will discuss Archimedes’ principle and buoyancy in more detail
in Chapter 6, Box 6.4.)
We refer to this vertical fluid motion (gas or liquid) due to buoyancy as
convection (not to be confused with “advection,” which refers to horizontal
motion). It is important to note that, as the warm fluid rises, not only the fluid
molecules, but also the heat they carry, are displaced upward. Therefore,
convection is primarily a heat transfer process. Moreover, because entire
blobs are displaced, the corresponding heat is transferred much more
efficiently than it would be by conduction (in which the molecules vibrate but
are not displaced).
You do not need to study the atmosphere to observe convection. In fact,
convection is taking place every day in your home as you use radiators,
stoves, and other heating devices. When you heat a pot of water, heat is first
conducted from the flames to the water through the metallic bottom of the
pan, but as soon as the first millimeters of water at the bottom of the pan
become warmer than the rest of the pan, warm buoyant blobs of water start
rising while colder water sinks to replace them (Figure 4.2). In the process,
heat is transferred and redistributed throughout the pan. The convective
process is very quick and efficient. In fact, if you were to place two
thermometers in the water, one at the top and one at the bottom, you would
observe similar temperatures, as convective overturning is almost
instantaneous. Through convective motion, the system is responding to the
heat imbalance and adjusts to a new equilibrium. It is possible because fluids
move freely and can become organized as currents moving heat around; this
cannot happen in solids.
 Figure 4.2. Convection and transfer of heat in a pan of water heated on a
stove.

The same process is constantly at work in the atmosphere. On a sunny

day, the sun heats the surface, which warms the lower atmosphere by
conduction (Figure 4.3(a)). Warm air rises, while colder air sinks to replace
it, resulting in a net upward transfer of heat (Figure 4.3(b)). We refer to these
rising pockets of warm air as thermals. In Chapter 6, we will learn that, as
warm air rises, it also expands and cools in response to lower pressure, which
can lead to condensation and the formation of clouds. But before we get
distracted by clouds, let us explore the third way by which heat is transferred
in the atmosphere.

Figure 4.3. Formation of a thermal by absorption of radiation and

conduction (a) and convection (b).
 4.3 Radiation
All objects in the universe contain some level of heat, even very cold objects.
The vibration of molecules creates an electromagnetic field and the emission
of electromagnetic radiation. While this is quite difficult to conceptualize
using classic imagery, we can think of radiation as waves of energy that
propagate through space at a speed of 300 000 kilometers per second (i.e.,
186 000 miles per second). These waves correspond to the joint oscillation of
the electrical and magnetic fields, and the properties of the waves depend on
the temperature of the object creating them. Colder objects generate low-
energy waves, while warmer objects generate high-energy waves. In fact, you
know at least one family of electromagnetic waves: the sun, at a temperature
of 6000 K, generates waves that peak in visible light, which our eyes have
evolved to detect. The sun also generates higher energy radiation, such as
ultraviolet, and lower energy radiation, such as infrared. Although our eyes
cannot detect ultraviolet and infrared radiation, our skin is sensitive to both.
Since we have a temperature and also contain heat, our bodies also emit
radiation at all times – infrared radiation. Although our eyes cannot see it, it
can be measured, and can be observed to vary with changes in our
temperature.
 4.3.1 The Nature of Electromagnetic Radiation
Among the properties of electromagnetic radiation, two important measures
are the wave frequency and wavelength, which are related by the wave
speed. These waves can travel through space, and in particular they can travel
through a complete vacuum, where they travel at a speed of 300 000 km/s,
i.e., the speed of light. We can think of these waves as undulating in every
direction of space around the source, the troughs and crests of the waves
being the signature of the pulsations at a certain frequency. For example,
Figure 4.4 represents a wave with a crest-to-crest distance of about 2.5 cm.

Figure 4.4. Schematic representation of two electromagnetic waves of

different wavelength (λ). (a) λ = 2.5 cm; (b) λ = 1.25 cm.

This crest-to-crest distance is called the wavelength and is commonly

represented by the Greek letter lambda (λ). Low energy radiation has
wavelengths on the order of a meter or more, such as radio and TV waves.
High energy radiation, however, pulsates much faster, which means that the
crests and troughs happen more frequently (hence, “higher frequency”) and
the wavelength is much shorter. In Figure 4.4(b), for example, the
wavelength is about 1.25 cm, half the wavelength of the wave in Figure
4.4(a). Since the blue wave travels at the same speed as the red one (i.e., at
the speed of light), the frequency is twice as large.
Since high energy radiation has wavelengths on the order of a thousandth,
millionth, or billionth of a meter, we resort to submultiple units:

1 meter=1000 millimeters (mm)=1 000 000 micrometers (μm)=1

000 000 000 nanometers (nm)

In this unit system, visible light has a wavelength on the order of 0.5 µm, or
500 nm, while infrared radiation has a wavelength on the order of 10 µm (see
Table 4.1).

Table 4.1. Spectrum of electromagnetic radiation

Radiation band Wavelength

Gamma rays <0.0001 μm

X-rays 0.0001 to 0.01 μm

Ultraviolet 0.01 to 0.4 μm

Visible 0.4 to 0.7 μm

Near infrared 0.7 to 4.0 μm

Thermal infrared 4 to 100 μm

Microwave 100 μm to 1 m

Radio waves >1 m

 4.3.2 Temperature and radiation
The radiation emitted by an object is a function of its temperature. As a rule
of thumb, warmer objects emit more electromagnetic radiation (i.e., a greater
quantity of energy), and they do so at higher frequency. This behavior is
captured by Planck’s law, named after the famous German physicist, Max
Planck (1858–1947). Figure 4.5 shows an example of Planck’s law applied to
the sun. It shows the amount of energy emitted by an object with the
temperature of the sun (6000 K), as a function of wavelength. The sun emits
a large quantity of visible light. It also emits ultraviolet radiation, and a fair
amount of infrared radiation (as it turns out, about 50% of the radiation
emitted by the sun is actually in the infrared).

Figure 4.5. Comparison of the emission spectra of the sun and Earth.
 Although the amplitude of the emission spectrum varies with the
temperature of the object, its shape is similar at all temperatures: it has a peak
of maximum emission (at 0.5 µm in the case of the sun) and two tails
showing the decreasing amounts of radiation emitted at lower and higher
frequencies. (See Box 4.1 for a description of other formulas derived from
Planck’s law and describing the emission spectrum.) Thus, Earth’s emission
spectrum is very similar in shape to that of the sun, but it is very different
both in magnitude and spread (see insert in Figure 4.5). Because the average
temperature of Earth is 288 K, the peak of maximum emission is shifted to 10
µm (in the infrared), and the intensity of the emission is several orders of
magnitude smaller than that of the sun. Thus, if we were to superimpose the
emission spectrum of Earth on that of the sun in Figure 4.5, it would be
stretched out to the right (i.e., toward longer wavelengths), in the infrared,
and would be so flat (i.e., so small) that it would blend with the abscissa. This
is consistent with the fact that the sun is a star, and therefore releases large
quantities of energy through nuclear fusion, whereas Earth is a planet at much
lower temperature, which mostly receives energy from the sun and radiates it
back to space in the form of infrared radiation.
Let us now explore how this visible and infrared radiation interacts with
the atmosphere.
 4.4 Radiative Interactions
When radiation encounters an object, it can interact with it in different ways.
It can be transmitted through the object, as it would be through a window,
for example, or through space. Radiation can also be absorbed, reflected,
and scattered.
 4.4.1 Absorption
When radiation is absorbed by an object, it interacts with the atoms and
molecules of that object in such a way that it is converted to kinetic energy.
Recall the law of conservation of energy (Box 1.1), by which energy cannot
be created or destroyed, but only transferred or, in our present case,
transformed (from radiative energy to energy of motion). As a result, the
atoms and molecules either move or vibrate faster. In other words, when
absorbed, radiation is converted to heat. We certainly know that from
standing in the sun: our body warms because it absorbs the incoming
sunlight.

Box 4.1. Radiation Laws

In the second half of the nineteenth century, a series of experimental

and theoretical findings revolutionized our understanding of radiation
and were critical for the advancement of physics. Around 1860,
Gustav Kirchhoff introduced the concept of a “black body” – a body
that, simply put, absorbs all radiation – and stated that, if we know
how much this theoretical black body emits at each wavelength at a
given temperature, then we know how much any body emits, in
proportion to how much it absorbs. Thus, for example, if a body
absorbs half as much as a black body, it also emits half as much as a
black body. Kirchhoff’s law is better known in the form,“good
absorbers are good emitters; poor absorbers are poor emitters.”
Figure 4.1.1. Gustav Kirchoff.

A black body at a given temperature, however, does not emit the

same amount of radiation at all wavelengths. The exact emission
function would later be determined by Max Planck – see below. In
the meantime, a series of scientists laid the ground for him by
establishing several important relationships. At the end of the
nineteenth century, Jožef Stefan (1835–1893) and Ludwig Boltzmann
(1844–1906) determined that the total emission of radiation (E) by a
black body at a given temperature T is proportional to the fourth
power of temperature:

E∝T4

We refer to this relationship as the Stefan–Boltzmann law. It tells us

that, even with a small increase in temperature (expressed in kelvin),
the total amount of energy emitted by an object (in W/m2) increases
with the fourth power of temperature. For example, if the temperature
is multiplied by 2, the total amount of energy is not simply multiplied
by 2, but by 24, which is equal to 16.
Figure 4.1.2. Jožef Stefan.

Figure 4.1.3. Ludwig Boltzmann.

This relationship has many applications in physics. In certain

situations, we prefer to use the inverse relationship by which the
temperature of an object changes as the fourth root of the radiative
energy it emits:

TαE4

Thus, if we can somehow measure or estimate the total amount of

energy emitted by an object, such as with a satellite in space, we can
estimate the object’s temperature.
In 1893, Wilhelm Wien (1864–1928) derived another important
relationship which tells us that the wavelength of peak emission (λm)
of a black body is inversely proportional to the temperature T of that
black body:

λm≈2897/T

where λm is expressed in micrometers and T in kelvin. In this

relationship of inverse proportionality, known as Wien’s
displacement law, when T increases, λm decreases, which
corresponds to a shorter wavelength, and therefore a higher
frequency. The sun, for example, with a temperature of 6000 K, emits
mostly around a wavelength of 2897/6000 ≈ 0.5 µm (i.e., in the
visible part of the electromagnetic spectrum). By contrast, Earth, with
an average temperature of 288 K, emits mostly around a wavelength
of 2897/288 ≈ 10 µm (i.e., in the infrared part of the spectrum).
In a different context, we sometimes prefer to use the inverse
relationship:

T≈2897/λm
Figure 4.1.4. Wilhelm Wien.

Now, if we can somehow measure the wavelength of peak emission

of a body (for example, a distant star), we can estimate its
temperature. These inverse relationships are used to design
instruments that estimate the temperature of objects based on their
radiative properties. Since they estimate temperature, these
instruments are really thermometers, although they use a different
technology from that of the mercury or alcohol thermometers you
might have at home. The most common such instrument is the so-
called “infrared thermometer,” which is specifically designed to
measure radiation in the infrared band of the electromagnetic
spectrum, i.e., typical radiation emitted by objects on Earth.
These radiation laws are really derived forms of a more exact and
complete relationship established by Max Planck (1858–1947) around
1900, and shown graphically in Figure 4.5. Planck’s law describes
the amount of radiation emitted by a black body in thermal
equilibrium at a given temperature as a function of wavelength. It
 helps us understand how a body can be a good absorber and a good
emitter (Kirchhoff’s law) and yet not emit the same amount of
radiation as it absorbs at a given wavelength, because the emitted
amount depends only on temperature (see Box 4.3). Planck’s law can
be integrated to produce an exact form of the Stefan–Boltzmann law
and differentiated to produce Wien’s law.

Figure 4.1.5. Max Planck.

Planck’s findings were revolutionary in many respects and opened

the door to the development of quantum field theory and modern
physics.

From standing in the sun, we also know that our clothing determines
how much we will warm. When we wear white clothes, for example, we stay
cooler than when we wear dark clothes. This tells us that materials do not
absorb radiation in the same way. Indeed, some materials only absorb part of
the incoming radiation, and the rest is reflected.
 4.4.2 Reflection
Whatever is not transmitted or absorbed by a material is reflected, as it
would be by a mirror. To continue with the clothing example, we know that
white clothes reflect most of the visible radiation that strikes them, which
keeps us relatively cool. Dark clothes, however, absorb much of the incoming
visible radiation, and we quickly become warm.
Thus, absorption and reflection are intimately related. With solid,
opaque surfaces, whatever is not absorbed is reflected, and vice versa. A red
apple, for example, absorbs most visible radiation except red light, which is
reflected back in all directions, including into our eyes: that is why we see the
apple as red.
On Earth, some of the sun’s radiation is reflected by bright surfaces such
as clouds, snow, and ice. This is most evident in a visible satellite image such
as that shown in Figure 2.13. Note that land areas are also relatively bright,
compared to the ocean. Land can reflect a significant fraction of the incoming
sunlight, whereas light tends to penetrate into ocean waters where it is first
transmitted and then eventually absorbed.
Since features on Earth reflect various amounts of visible light, it is
useful to define a new quantity, the albedo, as the fraction (or percentage) of
incoming sunlight that is reflected by an object or surface. Table 4.2 shows
typical values of albedo for common surface types. Note how forests, for
example, have a low albedo (10 to 20%). Indeed, they often look very dark
from a distance, or from a plane, because most of the incoming sunlight
penetrates through the branches and foliage and never comes back out. Note
also how Venus has a high albedo (78%): it is surrounded by clouds that
reflect a large fraction of the incoming sunlight. In contrast, the average
albedo of Earth is relatively low (~30%), which means that most of the
incoming sunlight is not reflected back out to space, but is absorbed by Earth.

Table 4.2. Albedo of typical surface types

Surface type Albedo(%)

Fresh snow 80–90

Thin cloud 30

Thick cloud 90

Ice 30–60

Grass 10–30

Water 10

Forest 10–20

Venus 78

Earth as a whole ~30

 4.4.3 Scattering
When radiation encounters individual atoms, molecules, aerosols, and cloud
droplets, it is not exactly reflected as it would be by a uniform, flat surface,
but is deflected in various directions, or scattered. Scattering is a result of
complex interactions between the small particles and the electromagnetic
waves. Radiation can be scattered forward, backward, and to the sides,
depending on the wavelength of the radiation and the size of the particle.
Cloud droplets are large enough to scatter effectively all wavelengths equally,
which is why clouds appear white. Molecules and aerosols, however,
preferentially scatter certain wavelengths, which explains why the sky is blue
and sunsets are red (see Box 4.2).

Box 4.2. Applications of Scattering

When radiation interacts with individual atoms, molecules, aerosols,

or cloud droplets, it can be scattered in various directions, depending
on the wavelength of the radiation and the size of the particle.
Nitrogen and oxygen molecules, for example, are very efficient at
scattering visible light from the sun at shorter wavelengths around
0.45 µm, which corresponds to blue light. That is why the sky is blue:
when you look up at the atmosphere, away from the sun, you see sun
rays that have been deflected by nitrogen or oxygen molecules, either
once or multiple times (Figure 4.2.1).
Figure 4.2.1. Scattering of short wavelengths by oxygen and
nitrogen molecules.

When the sun is low on the horizon, at sunset or sunrise, sunlight

must travel a longer distance through the atmosphere before reaching
your eyes, and most of the blue light has been scattered away, leaving
only longer wavelengths to reach you, such as yellows and reds
(Figure 4.2.2).
Figure 4.2.2. Depletion of short wavelengths at sunrise and sunset.
 4.4.4 Radiative Equilibrium
Transmission, absorption, reflection, and scattering also relate to emission of
radiation and the balance between incoming and outgoing radiation. Recall
that all objects emit some form of radiation. By inverse reasoning, since
absorption results in a gain of heat, you should not be surprised that emission
results in a loss of heat. Since objects emit radiation all the time, they are
constantly losing energy. You certainly know this from sleeping without a
blanket. In addition to losing heat by conduction, you also lose heat by
emission of infrared radiation, and you can get cold very fast – which is why
you use a blanket to keep warm. The same reasoning applies to Earth. It is
constantly losing energy to space in the form of infrared radiation. Why is
Earth not much colder than it is? Because half of Earth is also exposed to the
sun at all times, and the planet is constantly absorbing sunlight. Thus, there is
a balance between the amount of radiation absorbed by Earth on the day side,
and the amount of infrared radiation emitted by the whole planet. Earth is in
radiative equilibrium – another example of dynamic equilibrium (see
Appendix 3.1). If Earth emitted more radiative energy than it absorbed, it
would cool – and it would warm if it emitted less than it absorbed.
The fact that Earth emits nearly the same amount of energy as it absorbs
is no coincidence: nature seeks equilibrium, and Earth is, by nature, in
radiative balance. Our red apple is also in radiative equilibrium in the kitchen.
It constantly emits infrared radiation, but it also constantly absorbs infrared
radiation from the surrounding kitchen objects and walls. If we place it in full
sunlight, it absorbs more energy than it emits, and warms. But as it warms, it
emits more radiative energy (by the Stefan–Boltzmann law – see Box 4.1),
and reaches a new equilibrium at a higher temperature.
We will see shortly that, when some regions of Earth absorb more
energy than they emit, heat accumulates, and if other areas emit more than
they absorb, heat is lost. This imbalance causes a transfer of heat from
surplus areas to deficit areas, which in the atmosphere is achieved through
motion, and gives birth to weather systems.
But before we turn to weather systems, let us finish our radiative picture
of the atmosphere, and mention one important climate application.
 4.4.5 Selective Absorbers
So far we have assumed that Earth absorbs and emits radiation at maximum
efficiency. An object that has such characteristics is called a black body, or
ideal radiator (see Box 4.1). While this is a useful concept in theory, many
objects are in fact not fully efficient (see Box 4.3). In particular, the
atmosphere does not behave like a black body, because it is composed of
different gases that interact in very specific ways with electromagnetic
radiation (see Box 4.4). Ozone, for example, mostly absorbs radiation in the
ultraviolet, at wavelengths less than 0.3 µm. Recall that this absorption was
crucial for the appearance of life on Earth, because it prevents most
ultraviolet radiation from reaching the surface, where it is detrimental to plant
and animal life. Ozone also absorbs some infrared radiation in a very narrow
band at about 9 µm, which makes it a greenhouse gas. But it is largely
transparent to other types of radiation, such as visible light, for example, and
most of the infrared spectrum. Water vapor, in contrast, absorbs a significant
fraction of infrared radiation, but is mostly transparent to visible and
ultraviolet radiation.

Box 4.3. Radiation and Temperature

The radiation laws described in Box 4.1 are useful in understanding

many radiative processes happening on Earth. A good application of
Kirchhoff’s law, for example, is the radiative properties of a space
blanket (often referred to as a Mylar® blanket). As you might know,
such a blanket is designed to keep you warm, in particular by limiting
the loss of infrared radiation from your body. Indeed, the blanket is
made of a material that is a good reflector, and therefore a poor
absorber. According to Kirchhoff’s law, a poor absorber is also a poor
emitter. Therefore, the blanket emits very little infrared radiation,
while your own infrared radiation is reflected back to your body.
Kirchhoff’s law can be confusing, however, when dealing with
bodies at different temperatures. In this case, a body can be a good
absorber and therefore a good emitter, and yet not emit the same
amount of radiation as the amount it absorbed at a given wavelength.
That is because, according to Planck’s law, the amount of emitted
radiation depends on the temperature of the body. Let us illustrate this
subtlety with an example of great importance in climate science.
Let us assume that Earth is a black body with zero albedo. One
could ask, since Earth absorbs mostly visible radiation from the sun,
why Earth does not also emit a lot of visible radiation. According to
Kirchhoff’s law, being a good absorber of visible radiation, Earth
should also be a good emitter of visible radiation. And indeed it is,
but it does not have the right temperature to emit a lot. Planck’s law
tells us that, to emit a significant amount of visible radiation, a body
needs to have a temperature approaching that of the sun, i.e., 6000 K.
The temperature of Earth, however, is closer to 288 K, much less than
that of the sun. Therefore, it emits mostly infrared radiation. Note
that, strictly speaking, Earth does emit visible radiation at maximum
efficiency, or has the potential to do so, but that maximum is so small
that Earth emits virtually none. To illustrate this point graphically, we
could return to the insert of Figure 4.5 (repeated in Figure 4.3.1 for
convenience), showing the emission curve (i.e., Planck’s law) for
Earth. The temperature of Earth is such that its peak emission is in the
infrared (around 10 µm). Note that visible radiation (around 0.5 µm)
is in the leftmost tail of the curve, and is virtually zero.

Figure 4.3.1. Emission spectrum of Earth.

Finally, note that most objects are not black bodies, i.e., they do
not emit at maximum efficiency at all wavelengths, as a black body
would. In other words, there exist specific wavelengths at which they
absorb and emit well, and others at which they don’t. The
atmosphere, being made of different gases, is a good example, which
creates the possibility of visible and infrared windows, as well as the
greenhouse effect.

Box 4.4. Selective Absorption and Emission of Radiation

There exist two mechanisms by which gases interact with

electromagnetic waves. The first mechanism involves changes in the
energy state of electrons at the atomic level and explains, for
example, the absorption of ultraviolet radiation by ozone. The second
mechanism involves changes in the vibration and rotation of the
molecules and explains the absorption of infrared radiation by
greenhouse gases.
Electronic Transition
You might know that atoms are made of a nucleus and a specific
number of electrons orbiting the nucleus at a particular distance from
it. We can picture them as existing on a sphere around the nucleus,
that sphere having a specific radius. We refer to such a sphere as a
“shell.” Moreover, electrons can exist on separate shells, each shell
being successively farther away from the nucleus. We find two
electrons on the first shell, then eight on the second shell, and so on.
Once a shell is full, electrons populate the next one. The last shell
might be full or not. The total number of electrons determines the
type of atom we are dealing with, i.e., the element, hydrogen, helium,
carbon, oxygen, etc.
To interact with an electromagnetic wave by electronic transition,
an atom or molecule must contain an electron that can use the energy
carried by the wave to leap from its current level (i.e., its current
shell) to a higher level – to another shell that is more distant from the
nucleus (see Figure 4.4.1). Such a leap requires energy, and that
energy is captured from the electromagnetic wave. However, because
the successive shells occur at very specific distances from the
nucleus, these leaps require a very specific amount of energy (we say
that they are quantized). These leaps can only happen if the
electromagnetic wave carries exactly the appropriate amount of
energy. In other words, that electromagnetic wave must have a very
specific wavelength. In the case of ozone, for example, the electrons
will only be able to perform the leap if the ozone molecules are hit by
ultraviolet radiation.

Figure 4.4.1. Electronic transition during the absorption and

emission of radiation.

Vibration and Rotation

Certain molecules can also absorb electromagnetic radiation to
activate specific modes of vibration and rotation called “internal
modes.” The atoms constituting a water vapor molecule, for example,
represented in Figure 4.4.2 schematically by two smaller white circles
(hydrogen atoms) and one larger red circle (oxygen atom), can vibrate
with respect to each other in different ways.

Figure 4.4.2. Internal modes of vibration and rotation of the water

vapor molecule. (a) Atoms bending up and down. (b) Atoms vibrating
symmetrically. (c) Atoms vibrating asymmetrically. White circles
represent hydrogen atoms and red circles represent oxygen atoms.
Numbers 1 and 2 indicate the two steps of each vibration mode.

The molecule can also spin on itself, and all these different modes
of vibration and rotation require a specific amount of energy. Here
again, the water vapor molecule will only absorb electromagnetic
waves that carry the exact amount of energy required to excite these
 internal modes, i.e., electromagnetic radiation with the adequate
wavelength. Water vapor and the other greenhouse gases absorb
primarily in the infrared, which explains their role in the greenhouse
effect and global warming.
The wavelengths and intensity of absorption of individual gases
can all be determined precisely in a laboratory with an instrument
called a spectrophotometer. Because all these gases absorb radiation
selectively at certain wavelengths but not others, we call them
selective absorbers.

Figure 4.6 illustrates the impact of the absorption of radiation by

atmospheric gases on solar radiation. It shows how much radiation is
transmitted through the atmosphere at various wavelengths. A low
transmittance at a certain wavelength is due to high absorption by a specific
gas. The main selective absorbers of interest to us are indicated at the top of
the figure for reference. Water vapor absorbs in a broad band around 6–7 µm,
in addition to several other isolated peaks. Carbon dioxide also efficiently
absorbs infrared radiation in several narrow bands of the spectrum, and in a
large band around 15 µm. The absorption by ozone is more confined to about
5 and 10 µm, but ozone strongly absorbs in the ultraviolet (leftmost part of
Figure 4.6).
Figure 4.6. Infrared transmission spectrum of the atmosphere. Some of the
wavelengths at which the main selective absorbers (water vapor, carbon
dioxide, and ozone) absorb radiation are indicated at the top.
 4.4.6 A Window to the Sky
Since visible light is mostly transmitted through the atmosphere, we can use
the analogy of a window and say that there is a “visible window” in the
absorption spectrum of the atmosphere, around 0.5 µm. Similarly, because
some infrared radiation is transmitted through the atmosphere in a broad band
around 10 µm, without much absorption, we can say that there exists an
infrared window. The presence of these two windows is very important, as
the first allows visible light to reach the surface and warm the planet (that is
also why our eyes have evolved to detect visible light) and the second allows
some infrared radiation to escape to space (recall from Chapter 2 that this is
useful for infrared satellite observations). If this infrared window did not
exist, the radiative energy received from the sun would be reemitted by Earth
and absorbed by the atmosphere, and the planet would be warmer than it is.
These two windows allow Earth to be in radiative balance in a temperature
range that permits water to exist in all three phases, which has been
conducive to the evolution of life.
Although some infrared radiation escapes to space, a significant fraction
is absorbed by selectively absorbing gases, which gives rise to the
greenhouse effect.
 4.4.7 The Greenhouse Effect
Let us first imagine a planet Earth without an atmosphere. If there were no
gases either to absorb or emit radiation, Earth would simply absorb radiation
received directly from the sun and would emit infrared radiation back out to
space (Figure. 4.7(a)). We can calculate the amount of solar radiation
intercepted by Earth, and, considering the planet is in radiative equilibrium,
we may use the Stefan–Boltzmann law to estimate the corresponding
temperature (see Box 4.1). This is a classic calculation in climate science and
yields an equilibrium temperature of about −18 ºC – extremely cold
compared to the 15 ºC in reality.

Figure 4.7. Comparison of radiative equilibrium with and without an

atmosphere containing water vapor.

Let us now add an atmosphere to our planet Earth, made of nitrogen,

oxygen, and water vapor. Nitrogen and oxygen do not change the picture
much, as they do not absorb radiation in the visible and infrared. The fact that
water vapor is a good absorber of infrared radiation, however, has important
implications. Since Earth’s surface (land, ocean, vegetation, lakes, ice, snow,
etc.) constantly emits infrared radiation, some of this radiation is absorbed by
water vapor in the atmosphere. By absorbing infrared radiation, water vapor
heats the atmosphere (the heat is distributed by conduction, convection, and
radiation). Thus, the presence of water vapor implies a warmer atmosphere
compared to one lacking water vapor. Our story does not stop there, though.
Good absorbers are also good emitters, so water vapor also strongly emits
infrared radiation. Some of this radiation eventually makes its way to space.
But some of it is emitted downward, where it is reabsorbed by the surface,
and to the sides, in every direction, where it can be reabsorbed by other water
vapor molecules. Since the surface absorbs infrared radiation, it warms, and
therefore emits more infrared radiation, which is then reabsorbed by water
vapor, reemitted, and so on, until a new equilibrium is reached (Figure.
4.7(b)). Under this new equilibrium, the lower atmosphere and the surface are
much warmer than they would be without water vapor.
Because there is an apparent similarity with a greenhouse, in which
sunlight enters but heat does not leave easily, resulting in a higher
temperature, the radiative process we just described is called the greenhouse
effect. Know that this is somewhat of a misnomer, however, as the reason
why a greenhouse is very warm is not because infrared radiation is trapped in
the greenhouse: it is primarily because the warm air itself is trapped inside,
preventing convection and the mixing of warm and cold air that would
naturally take place outside in the free atmosphere. We sometimes refer to
this effect as “suppressed convection.” Therefore, it is a convective effect,
before being a radiative effect. The phenomenon that takes place in the
atmosphere is primarily a radiative effect, and some have proposed to call it
the “atmosphere effect.” Nevertheless, the term “greenhouse effect” has
caught on and is now widely used.
Due to its higher concentration in our atmosphere, water vapor is the
main actor of the greenhouse effect taking place on Earth. It is not the only
one, however, as many other gases also absorb infrared radiation, such as
carbon dioxide, methane, ozone, and nitrous oxide. We call these greenhouse
gases. In the presence of all greenhouse gases combined, the equilibrium
temperature reached by the lower atmosphere is about 15 ºC, or 33 ºC higher
than it would be without an atmosphere.
To summarize, greenhouse gases in the atmosphere absorb energy in the
form of infrared radiation and maintain Earth’s surface at a higher
temperature than would be the case without an atmosphere. The greenhouse
effect also takes place on other planets, such as Venus, for example, where
the high concentration of carbon dioxide results in extreme temperatures of
750 ºC.
Given how beneficial the greenhouse effect has been for life on Earth,
you may wonder why it is expressed as a problem with respect to global
warming. As we have seen, the greenhouse effect is a natural process by
which Earth has maintained a slowly varying equilibrium temperature
throughout its history, in response to slowly varying greenhouse gas
concentrations. The temperature has sometimes been higher, when
greenhouse gas concentrations were higher (such as in the Eocene), and
sometimes lower, when greenhouse gas concentrations were lower (such as in
the Ice Ages). More recently, human civilization has altered this slowly
changing natural process by increasing the concentration of greenhouse gases
in the atmosphere very rapidly, most certainly over the last two centuries,
since the Industrial Revolution, and possibly for the last 8000 years, with
deforestation and the development of agriculture. Increasing greenhouse
gases means more absorption of infrared radiation, and therefore more heat
and a higher temperature for Earth. This human enhancement of the
greenhouse effect is called global warming, and it is leading to climate
change. Furthermore, it is reinforced by positive feedback processes that may
accelerate the warming, and make the planet warm much faster than the
radiative effect alone would suggest (see Box 4.5 and Chapter 15).

Box 4.5. Feedback Loops

The enhancement of the greenhouse effect is further complicated by

other processes that are altered as a result of increasing temperatures.
In a warmer world, ice and snow start to melt, for example. You
might recall that ice and snow have a high albedo: they are excellent
reflectors of visible light. If the ice and snow coverage decreases, less
sunlight will be reflected, and therefore more sunlight will be
absorbed by Earth’s surface. An increase in the absorption of sunlight
translates into higher temperatures, and therefore even more warming.
This whole chain of events is called a positive feedback loop (Figure
4.5.1) and accelerates global warming in ways that are not always
obvious to model and understand.
Figure 4.5.1. Ice-albedo feedback (example of a positive feedback
loop).

In addition to the ice-albedo feedback just described, we expect

to observe more evaporation from the ocean in a warmer world, and
therefore more water vapor in the atmosphere. Water vapor being a
greenhouse gas, this will lead to even more absorption of infrared
radiation, and therefore even more warming – another positive
feedback loop. In fact, the water vapor feedback is the main positive
climate change feedback (Figure 4.5.2).

Figure 4.5.2. Water vapor feedback.

 A warmer planet, however, will also emit more infrared radiation
(recall Planck’s law), and will therefore lose more energy, leading to
a temperature decrease. In this particular case, warming is not
reinforced by more warming, and the Planck feedback is a negative
feedback (Figure 4.5.3).

Figure 4.5.3. Planck feedback (example of a negative feedback

loop).

The Earth system contains a large number of feedback loops, both

positive and negative, and climate scientists spend a fair amount of
time trying to model and understand which will dominate and control
the magnitude and rate of global warming in decades and centuries to
come.
 4.5 Radiation and Weather
Let us return now to our main topic and discuss how radiative interactions
control temperature, and therefore weather.
 4.5.1 Heat Imbalance
We mentioned earlier the importance of radiative equilibrium, where heat
inputs balance outputs: the amount of incoming solar radiation must be
balanced by outgoing infrared radiation. What happens if a specific region of
Earth is not in radiative equilibrium? What if it receives more energy than it
emits, or vice versa?
In the tropics, for example, the sun is high in the sky, and the sun rays
tend to be concentrated over smaller areas compared to the polar regions,
where the sun is low in the sky and sun rays are spread out over larger areas
(Figure 4.8). Think of two sun beams carrying the same amount of energy:
since that energy is spread out over a larger area at the poles, each unit area is
receiving less radiation. In contrast, since the sun beam is concentrated over a
smaller area in the tropics, each unit area is receiving comparatively more
energy than at the poles. This means that the absorption of solar energy, and
therefore heating, is much more intense at low latitudes – in the tropics – than
at high latitudes – in the polar regions.
 Figure 4.8. A given amount of solar energy is concentrated into a smaller
area in the tropics, whereas it is more spread out at the poles.

In contrast, cooling by emission of infrared radiation is always emitted

“upward” and is not subject to the same angle variation. It is pretty similar all
around Earth, except for the fact that temperatures are higher in the tropics,
and infrared emission is somewhat higher there (according to the Stefan–
Boltzmann law – see Box 4.1). As a result, heating by absorption of solar
radiation is greater than cooling by emission of infrared radiation in the
tropics, and there is net warming. But heating by absorption of solar
radiation is less than cooling by emission of infrared radiation at the poles,
and there is net cooling (Figure 4.9).

Figure 4.9. The emission of infrared radiation by Earth (in red) is

relatively uniform around the globe, while the absorption of visible
radiation from the sun (in yellow) is greater at low latitudes than at the
poles, which results in a net input of energy at low latitudes, and a net
output of energy at the poles. (Note that the yellow arrows at the poles are
shown perpendicular to the surface for clarity, even though the sun rays are
coming from the right of the picture.)

This contrast is enhanced by the fact that sun rays have to travel a longer
distance through the atmosphere to reach the poles (being at a low angle) and
a shorter distance in the tropics (being overhead). On its way to the surface,
sunlight is scattered and much of the scattered energy is lost to space. [This is
somewhat similar to the red sunset example we encountered in Box 4.2.] The
longer the path, the more scattering, and the more the sun beam is depleted of
its energy. Therefore, more energy is being lost at the poles by scattering,
compared to the tropics, which reinforces the heat imbalance between high
and low latitudes.
 Finally, recall that the albedo of Earth is high at the poles, where we find
much ice and snow (Figure 4.10), and low in the tropics, where we find large
regions of ocean. More absorption in the tropics and more reflection at the
poles also contribute to creating a heat imbalance between high and low
latitudes.
 Figure 4.10. Albedo at the poles.

As a result, heat tends to accumulate in the tropics, while there is a

deficit at the poles. Since the whole system tends to return to a state of
equilibrium by natural law, it will evolve in such a way as to transfer
continuously the surplus of tropical heat toward the poles. This is
accomplished by motion of air in the atmosphere and motion of water in the
ocean, through air currents in the atmosphere and water currents in the ocean.
For the atmosphere, a central part of this exchange of heat, executed through
the exchange of warm and cold air masses, is the midlatitude cyclone, to
which we will return in Chapter 10. In the meantime, let us remember that
weather on Earth is largely due to the existence of a heat imbalance between
the tropics and the poles, and that this heat imbalance can be explained by
differences in the absorption and emission of radiative energy at different
latitudes.
 4.5.2 Seasonal Variations
The interaction of solar radiation with Earth also explains why we have
seasons. As shown in Figure 4.11, the sun is higher in the sky in summer, and
lower in winter. Therefore, for similar reasons to those invoked earlier when
describing latitudinal differences, solar radiation is concentrated onto a
smaller area in summer, which results in more energy being absorbed per unit
area, and higher temperatures. In winter, solar radiation is spread out over a
larger area, which results in less energy being absorbed per unit area, and
lower temperatures.
 Figure 4.11. Changes in insolation as Earth orbits the sun. Example given
for the northern midlatitudes.

Moreover, because Earth is tilted toward the sun in summer, a given

location in the summer hemisphere spends more time on the day side than on
the night side (Figure 4.12). In other words, the days are longer than the
nights, which means that such a location is absorbing heat during a longer
time period in summer than in winter. This also contributes to the seasonal
contrast, with higher temperatures in summer and lower temperatures in
winter.

Figure 4.12. Length of day in summer: a location in the midlatitudes

spends more than 12 hours on the day side of Earth (red arrow). Note the
difference between the vertical line separating night from day, and the
dashed line going from pole to pole and separating Earth in two
hemispheres of 12 hours each. Note that the length of day is 12 hours at the
equator, and 24 hours north of the Arctic Circle.
 In summary, seasons are due to the tilt of Earth’s axis with respect to the
ecliptic plane (i.e., the plane of Earth’s rotation around the sun), which brings
the summer hemisphere under more direct solar radiation during longer days,
and the winter hemisphere under less direct radiation during shorter days.
[We note that seasons are not due to changes in the distance between Earth
and the sun during Earth’s revolution, as many people believe. Currently,
Earth is actually slightly farther away from the sun during northern summer.]
 4.5.3 Diurnal Variations
Temperature also changes on a daily basis, which we refer to as daily
temperature variations, or diurnal temperature variations. Obviously, the sun
rises and sets, which means that Earth’s surface receives more solar energy
during the day than during the night. But there is more to the diurnal cycle
than just the times of sunrise and sunset.
Earth also loses energy by emission of infrared radiation, which brings
us back to the concept of dynamic equilibrium (Appendix 3.1). To return to
our bucket analogy, the energy provided by the rising and setting sun is
similar to the water pouring into the bucket from a faucet that is turned on
during the day (increasing after sunrise, and decreasing after noon), and
turned off at night. The energy loss by emission of infrared radiation is
similar to the water flowing out of the hole: it is occurring all the time, night
and day, and is relatively constant (within the temperature range of the
surface – recall that the emission of radiation is a function of temperature).
Therefore, even when the sun rises and starts warming the surface (small
input), the surface is still cooling by emission of infrared radiation (large
output) and might in fact be losing more energy than it is gaining (output >
input). Thus, the temperature may still drop, even though the sun is already
above the horizon. Later in the morning, the incoming solar radiation will be
greater than the outgoing infrared radiation (input > output), and the
temperature starts rising. Because this increase in temperature starts later than
sunrise, we can speak of meteorological sunrise. For similar reasons,
meteorological sunset starts earlier than the actual sunset, when the incoming
solar radiation becomes less than the outgoing infrared radiation.
 To continue with the leaky bucket analogy, we note that the maximum
water depth in the bucket is not reached when the input is greatest. The water
level continues to rise, even when the input starts to decrease, as long as the
input is still greater than the output. The water level is maximum when the
input decreases sufficiently to match the output. Similarly, maximum
temperatures do not normally occur at solar noon, when the input of solar
energy is maximum, but rather close to meteorological sunset, when the
incoming solar radiation is about to become less than the outgoing infrared
radiation.
Figure 4.13 shows an example of the diurnal cycle at Amarillo, Texas, in
February 2014. The temperature can be observed to swing up and down every
day by about 20 °C. While the meteorological sunrise (when the temperature
starts rising) here coincides closely with the actual sunrise (when the blue
shading turns to white), the meteorological sunset (when the temperature
starts falling) can be observed to happen before the actual sunset (when the
white shading turns to blue).

Figure 4.13. Meteogram illustrating the diurnal cycle at Amarillo, Texas,

in February 2014. The red curve indicates temperature. The blue shading
 indicates nighttime, while the white shading indicates daytime (in local
time). The blue dashed lines indicate midnight, while the red dashed lines
indicate noon, for reference.

The combination of conduction, convection, and the diurnal cycle also

explains the changes observed in the temperature profile close to the surface.
Figure 4.14 shows a typical temperature profile (in black) in which the
temperature decreases with height. Recall that heat is transferred primarily by
conduction from the surface to the first few millimeters of air, and then by
convection further up, so that the atmosphere is warmer closer to the source
of heat (the surface). On a sunny day, solar radiation heats the surface, which
heats the first meters of atmosphere by conduction and convection (warm air
rises), as shown by the red temperature profile in Figure 4.14(a). At night, the
surface cools by emission of infrared radiation, which cools the lower
atmosphere by conduction and yields the blue temperature profile in Figure
4.14(b). (Note that convection does not intervene in this case, since cold air
tends to sink. The air tends to settle in layers, with the coldest layers at the
bottom, but there is little mixing, unless there is wind.)
 Figure 4.14. Diurnal cycle in temperature close to the surface. (a) Daytime
profile. (b) Nighttime profile.

Note that, in Figure 4.14, the resulting diurnal cycle is very pronounced
in the first hundred meters of air, but almost non-existent higher up in the
atmosphere. Indeed, conduction does not penetrate very high in the
atmosphere (as we discussed earlier, air is a poor conductor and a very good
insulator). Convection is more efficient, but is constrained by other processes,
as we will learn in Chapter 6. As a result, changes in the temperature of the
surface are felt over roughly the lowest kilometer of the atmosphere. In the
absence of weather systems, the vertical temperature profile changes slowly,
except for a “bottom leg” that sweeps between high and low temperatures
throughout the diurnal cycle.
Note that, in the nighttime temperature profile (in blue), temperature
increases with height in the first hundred meters of atmosphere, which is the
inverse of the average temperature profile (in black). Recall that we call this
feature a temperature inversion. Indeed, a nighttime temperature inversion
often forms when the surface loses a sufficient amount of energy by emission
of infrared energy overnight (radiational cooling). This inversion is most
pronounced close to the surface. We will learn later that such a temperature
inversion also makes the lower atmosphere very stable, which traps pollution
(see Chapter 14).
 4.5.4 The Influence of Clouds
The presence of clouds affects both day and night portions of the diurnal
cycle: clouds prevent incoming solar radiation from heating the surface,
limiting high temperatures during the day (Figure 4.15(a)), and they prevent
infrared radiation from escaping to space, limiting low temperatures at night
(Figure 4.15(b)). As a result, cloudy days are typically cooler than sunny days
and cloudy nights are typically warmer than clear nights. This leads to a
smaller diurnal cycle (i.e., narrower day–night temperature range). The
climate of any given location will reflect this impact of cloudiness. Seattle,
Washington, for example, has a maritime climate characterized by relatively
high amounts of cloudiness. The average low and high temperatures in
January are 3 and 8 °C (37 and 47 °F), respectively: a 5 °C (10 °F) diurnal
cycle. By contrast, Denver, Colorado, has a more continental climate with
much less cloudiness. The average low and high temperatures in January are
–9 and 6 °C (15 and 43 °F, respectively: a 15 °C (28 °F) diurnal cycle.
Figure 4.15. Impact of clouds on the energy budget and the diurnal cycle.
 4.5.5 Land–Ocean Contrasts
Extending the Seattle–Denver comparison, we observe that locations
influenced by the ocean tend to have small diurnal cycles, while continental
locations can experience much larger daily extremes. Why do land masses
cool and heat so easily, while bodies of water (and their surrounding coastal
lands) maintain a comparatively narrower temperature range throughout the
day and throughout the year?
Water is a very special molecule with very interesting properties. The
presence of one oxygen atom and two hydrogen atoms at an angle of 104º
and the asymmetric distribution of positive and negative charges between the
hydrogen and oxygen atoms give unique properties to the water molecule.
One property you might know is the propensity for water molecules to
arrange themselves into a special hexagonal pattern when they freeze, which
explains the formation of delicate ice crystals and snowflakes of definite
shape, and the lower density of ice compared to liquid water. Of more interest
to us is the high heat capacity of liquid water. Heat capacity is the energy
required to increase vibrational energy as the temperature increases. In the
case of liquid water, however, additional energy is required to break
“hydrogen bonds” – weak bonds that exist between the hydrogen atoms
(positively charged) of a water molecule and the oxygen atoms (negatively
charged) of other water molecules when they are in contact. Although weak,
these hydrogen bonds are constantly broken and recreated in liquid water.
They are strong enough that it requires additional energy for water molecules
to speed up, as these bonds need to be loosened continually. This additional
energy corresponds to the higher specific heat of liquid water. These
hydrogen bonds are not broken during the warming of ice, and they are
already broken in water vapor. Therefore, the heat capacity of ice and water
vapor is less than that of liquid water.
To provide a more quantitative comparison, let us consider one gram of
liquid water exposed to one calorie of solar energy (i.e., 4.18 joules). The
gram of water will heat up by one degree Celsius. This is by design, as the
calorie is defined as the amount of energy required to heat up one gram of
water by one degree Celsius. We also say that the heat capacity of water is
equal to:

cp(water)=1cal/(g°C)=4.18J/(g°C)

In contrast, the heat capacity of soil is about 0.2 cal/(g °C):

cp(soil)=0.2cal/(g°C)=0.84J/(g°C)

It takes only 0.2 calorie to heat up the gram of soil by one degree. Therefore,
if we place the soil next to the water and expose them both to 1 calorie of
solar energy, the soil will heat up by 5 degrees – five times more than the
water! That is why, when you go to the beach, dry sand can be burning hot
while wet sand remains relatively cool.
This is one of four reasons why water temperature changes more slowly
than land temperature. For completeness, we list them all here:
As we just described, water has a higher heat capacity than soil.
The energy absorbed by land is concentrated in the first few
millimeters of soil, since conduction is not very efficient at transferring heat
downward. In contrast, the surface of the ocean is stirred by the wind and
mixing redistributes surface water through a much deeper layer. As water is
mixed downward, heat is also transported downward and spread throughout
the layer – it is somewhat similar to convection. A given amount of heat
distributed throughout a thicker layer of water means lower temperatures than
the same amount of heat concentrated in the first millimeters of soil.
Because water is transparent to visible light, solar radiation reaches
below the surface and penetrates through a deeper layer of water than it does
for land. The absorption of energy is distributed through a thicker layer, and,
as a result, heat is more spread out, which also results in smaller temperature
changes.
Part of the solar energy hitting the water surface is used to evaporate
water, rather than to heat it (see Chapter 5). In technical terms, the energy is
used as latent heat of evaporation, rather than sensible heat. We can think of
this latent heat as lost and contained in the water vapor, off to a new
destination in the atmosphere. Since there is less energy available to heat the
water, temperatures do not increase as much as over land.

Figure 4.16. Four factors contribute to land masses heating up much more
than oceans under the sun.
 These effects contribute to creating large temperature contrasts between
land and ocean. In the same way that latitudinal temperature contrasts
between the tropics and the poles drive large-scale weather patterns such as
midlatitude cyclones, land–ocean temperature contrasts drive smaller-scale
circulations such as sea breezes and larger-scale circulations such as
monsoons, which we will discuss in Chapter 8 and 9, respectively.
 Summary
In the atmosphere, heat can be transferred in three different ways:
conduction, convection, and radiation.

► Conduction is a process by which heat is transferred by contact,

collision, and transfer of molecular vibrations, which is slow and
inefficient in the atmosphere – air is a poor conductor.
► Convection is a faster and more efficient process by which air is
displaced vertically, along with its heat content, due to temperature
and density differences – “warm air rises” due to buoyancy. Pockets of
warm air rising by convection are called thermals.
► Radiation carries electromagnetic energy that can be reflected,
transmitted, scattered, or absorbed by the different elements of the
Earth system.

Radiation takes the form of electromagnetic waves characterized by a

wavelength and a frequency. Lower frequency electromagnetic waves, such
as infrared radiation, have longer wavelengths and carry less energy than
visible light. Higher frequency electromagnetic waves, such as ultraviolet
radiation, have shorter wavelengths and carry more energy than visible light.
When electromagnetic radiation is absorbed by a material, it is
converted to heat, and the temperature of the material increases.
The reflective properties of a material are defined by its albedo, which
is the percentage of incoming sunlight that is reflected by the object or
surface. The albedo of snow can be as high as 90%, whereas the albedo of the
oceans and forests is as low as 10%.
 Earth also emits electromagnetic radiation, mostly in the infrared. Earth
is close to radiative equilibrium, by which the amount of incoming solar
energy that is absorbed is equal to the amount of outgoing infrared energy
emitted by Earth.
Certain variable gases of the atmosphere, such as water vapor, carbon
dioxide, and ozone, absorb and emit radiation at specific wavelengths.
By absorbing and reemitting infrared radiation, some of these gases
contribute to the greenhouse effect and maintain Earth’s surface at an
average surface temperature of 15 °C, 33 °C warmer than without the
atmosphere. The enhancement of the greenhouse effect by anthropogenic
increase of certain greenhouse gas concentrations, such as carbon dioxide and
methane, leads to global warming and climate change.
The angle of sun rays at different latitudes and the albedo of snow and
ice at the poles create a net warming at the equator and a net cooling at the
poles. The resulting temperature contrast generates currents in the ocean,
and air motion in the atmosphere to move heat poleward, including cyclones
in the midlatitudes.
The tilt of Earth’s axis exposes each hemisphere to more solar radiation
in summer and less solar radiation in winter, resulting in seasonal variations
in temperature. The tilt of Earth’s axis also exposes each hemisphere to
longer days in summer and shorter days in winter, which enhances the
seasonal contrasts.
The daily rotation of Earth causes diurnal variations in temperature.
The diurnal cycle is most pronounced near the surface, due to the absorption
and emission of radiation by the ground, conduction of heat between the
ground and the atmosphere, and convection. At night, the decrease in
temperature at the surface caused by emission of infrared radiation can create
a temperature inversion, where temperature increases with altitude.
The presence of clouds in the atmosphere limits the amount of solar
radiation that reaches Earth’s surface during the day, and limits the emission
of infrared radiation out to space during the night, which reduces the
amplitude of the diurnal cycle.
Water bodies experience smaller temperature variations than land
surfaces due to turbulent mixing, deeper penetration of sunlight, the
evaporation of water, and a higher heat capacity. Therefore, the overlying air
and coastal locations also experience a smaller diurnal and seasonal cycle
than inland, continental locations.
CHAPTER 5
Water
CONTENTS
5.1 The Water Cycle
5.2 Saturation
5.3 Humidity
5.4 Relative Humidity
5.5 Humidity and Temperature
5.6 Dew Point Temperature
5.7 Applications of the Dew point Temperature
Summary

Water in its different forms creates the possibility of a wide array of

atmospheric phenomena, from clouds to rainbows, and from lightning to
hurricanes. Before we can describe these phenomena in detail, however, we
need to understand how and when water can change phase, from solid (ice),
to liquid (liquid water), and gas (water vapor). To that end, we will explore
different aspects of the water cycle, in particular condensation and
evaporation, leading to the formation of clouds and precipitation. We will
define quantities of great importance to this exploration: saturation, relative
humidity, and dew point temperature.
The presence of water on Earth makes our planet very special. Most
importantly for us, it makes life possible. But the fact that water can exist in
three different phases, or states, in the same environment (solid, liquid, and
gas) makes the Earth system even more special: it provides additional paths
for efficient energy transfer. Thus, to understand the role played by water in
the atmosphere, we first need to understand how water cycles through the
Earth system in these different states.
 5.1 The Water Cycle
On Earth, water can exist under three forms, or phases: solid (ice, hail, ice
crystals, frost, snowflakes), liquid (ocean water, cloud droplets, rain), and gas
(water vapor). These phases differ in the arrangement of water molecules, as
well as in kinetic energy. In solid ice, for example, molecules do not move
around at all, but merely vibrate, which corresponds to a lower energy state.
Moving molecules in liquid water and water vapor contain comparatively
more energy. More importantly, for ice to melt and for water to evaporate, the
hydrogen bonds existing between water molecules (see Chapter 4) must be
broken, which requires additional energy. Therefore, to change state from
solid to liquid or liquid to gas, water molecules need an energy boost: enough
energy to break the hydrogen bonds. This extra boost of energy could be
obtained from the sun, for example. Conversely, when water returns from a
higher energy level to a lower energy level, from water vapor to liquid
(condensation), or from liquid to ice (freezing), this extra boost of energy is
released (Figure 5.1). In other words, it is returned to the surrounding
environment (to other gases, for example, such as nitrogen and oxygen). In
that respect, we can think of water vapor as containing an extra quantity of
energy that allows it to exist in the vapor phase: we call it latent heat, as if it
is in a latent state, waiting to be released. We will see that the release of latent
heat is of critical importance, as it can effectively warm the environment.
Conversely, when water evaporates, or when it changes phase directly from
solid to vapor (sublimation), it takes the necessary latent heat from the
environment, which tends to cool the environment. (It cools the ocean, for
example, in the case of evaporation of water from the ocean.)

Figure 5.1. Intake and release of latent heat as water changes phase
between liquid and gas.

When water returns from the liquid to the solid state (freezing), latent
heat is also released. When it goes straight from the vapor phase to the solid
phase (deposition), as in the formation of frost or snow flakes, an even
greater amount of latent heat is released, since this process is the equivalent
of two phase changes. This gives us the schematics shown in Figure 5.2, in
which all phase changes are summarized, along with the gains or losses of
latent heat.
 Figure 5.2. Phase changes of the hydrologic cycle with corresponding
intake (blue arrows) and release (red arrows) of latent heat. (Alternative
terms used in physics are indicated in gray.)

Let us now investigate the conditions under which these phase changes
are possible.
 5.2 Saturation
If you have ever forgotten a glass of water on a table for several days, you
know that water tends to evaporate over time. After a few days, the glass will
be empty and dry. If you cover the glass with a lid, however, the water will
not completely evaporate. Even if there is only a little bit of water, and a lot
of space in the glass, the water will never completely evaporate to occupy
that space as water vapor. It is quite an important observation, as it suggests
that there is a limit to the amount of water that can be evaporated in a given
space, or volume. Indeed, if we imagine an open water surface under dry air,
the water will first evaporate into the dry volume, as individual molecules
escape the liquid phase to become water vapor (Figure 5.3(a)). As the amount
of water vapor molecules increases, however, some water molecules will start
rejoining the liquid surface underneath, as a result of random motion and
collisions. Thus, condensation is taking place at the same time as evaporation
continues (Figure 5.3(b)). As long as evaporation is greater than
condensation, or, in other words, as long as the net evaporation is positive,
the amount of liquid water will diminish. At some point there will be enough
water vapor above the liquid water surface that the condensation flux will
equal the evaporation flux: a state of equilibrium in which evaporation and
condensation are continuously taking place while the overall amount of liquid
water and water vapor does not change (Figure 5.3(c)). This is yet another
example of dynamic equilibrium (see Appendix 3.1) and is somewhat similar
to our bucket of water with equal input and output, or to a room in a museum
in which the rate of people entering the room equals the rate at which other
people are leaving the room, so that the total number of patrons enjoying the
art in the room remains constant, even though the patrons themselves are
never the same.

Figure 5.3. From evaporation to saturation. (a) Water evaporates into dry
air. (b) Evaporation is greater than condensation, so there is positive net
evaporation. (c) When saturation is reached, condensation balances
evaporation.

When we have reached that state of equilibrium, and therefore the

maximum amount of water vapor that can occupy the space above the liquid
water, we say that the air is saturated with water vapor, or that we have
reached saturation. [In reality, the presence of dry air is irrelevant, as gases
behave independently of each other, and it is really the space, or volume, that
is saturated with water vapor. We would obtain the same result in a vacuum.
But since we always deal with air in meteorology, it is common practice to
say that the air is saturated.]
Note that the notion of saturation is relative to the surface available for
the water vapor to condense upon. In the above example, we considered a flat
surface of water. But water vapor typically condenses upon many different
types of surface. For example, grass provides a useful surface for dew to
form, while in the atmosphere water vapor will condense upon available solid
particles called aerosols, such as dust particles and ice crystals. We will see
shortly that the amount of water vapor in the air at saturation will depend
upon the available surface types.
We can foresee that these concepts will have important implications, as
condensation and cloud formation must depend on that level of saturation. In
fact, lacking a suitable surface to condense upon, water vapor can reach much
higher concentrations than we would find otherwise. Before we go any
further, however, let us introduce another useful notion.
 5.3 Humidity
If we are to discuss the amount of water vapor present in the air, or humidity,
to understand the formation of clouds and rain, we need a way to quantify it,
and therefore a way to measure it. Instruments that measure humidity are
called hygrometers. We can resort to different methods to estimate the
amount of water vapor around us, each leading to a different measure and
different units. We can express it as the mass of water vapor per volume of
air (in which case it is called absolute humidity), as the mass of water vapor
per mass of air (specific humidity), or the mass of water vapor per mass of
“dry” air, i.e., per mass of all other gases except water vapor (mixing ratio).
In our case it will be more interesting to use yet another measure called water
vapor pressure: the pressure exerted in a given space by water vapor alone,
as if all other gases were removed from that space (see Box 5.1). We use the
lowercase letter e, and, as for atmospheric pressure, we will use hectopascals
(hPa) as our unit. In this unit system, water vapor pressure (e) is typically on
the order of ones to tens of hectopascals.

Box 5.1. Dalton’s Law

In 1801, John Dalton (1766–1844) established that, in a mixture of

gases, the total pressure exerted by the gas mixture is equal to the sum
of the partial pressures exerted by the individual gases. We usually
refer to this statement as “Dalton’s law”.
For example, imagine an air parcel containing nitrogen, oxygen,
and water vapor with respective partial pressures pN2, pO2, and pH2O.
 Dalton’s law states that the total air pressure of the parcel is equal to

p=pN2+pO2+pH2O

For many applications, it is useful to differentiate the partial pressure

of dry air (pdry) and the partial pressure of water vapor (usually
written ‘e’):

p=pdry+e

Since air pressure is a variable that we are already using, expressing

the water vapor content of the air as water vapor pressure is
convenient, and simplifies some equations related to weather physics.

These measures of humidity, however, are not necessarily always

helpful to us. How is a water vapor pressure of 3 or 30 hPa relevant to the
formation of clouds and rain? A more interesting concept arose as we were
discussing evaporation and condensation: that of saturation.
 5.4 Relative Humidity
Have you ever noticed that you are really cold when you come out of the
swimming pool, even when the air is warmer than the water? Indeed, when
you come out of the pool, you are usually covered with water, and this water
evaporates into the air, which requires energy. The water needs an extra boost
of energy to leave the liquid phase, so that the molecules of water can detach
and turn into water vapor (Figure 5.2). It needs latent heat of evaporation.
This latent heat is taken from your body, which cools you down.
Similarly, when playing sports, your body cools by sweating. Indeed,
remember that humans have evolved to sweat because evaporation of sweat
into the surrounding air is an efficient cooling mechanism. In fact, it cools
you down 100 to 1000 times faster than if you were to wait for conduction to
transfer your excess heat away by contact, through molecular transfer. [Some
researchers think that the evaporative cooling of sweat is one of the
characteristics acquired by humans two million years ago that allowed us to
hunt other animals by pushing them to heat exhaustion, i.e., endurance
running and persistence hunting.]
Now, if you are playing in a dry environment, your sweat will evaporate
fairly easily, as the air can still contain additional water vapor. If you are
playing in a moist environment, such as on a humid day, for example, the air
already contains a lot of water vapor, and there is less evaporation potential.
If the air is saturated with water vapor, it is especially difficult to evaporate
sweat into it. The maximum has been reached, so to speak. Sweat remains on
your skin, your body keeps heating up, and you might suffer heat stroke.
 We could return to the swimming pool and observe similar extremes. If
you come out of the water to find yourself in a steamy Turkish bath, saturated
with water vapor (as opposed to a hot, but dry, sauna), you will not evaporate
much water, and therefore will not cool effectively. If you come out of an
outdoor pool in a hot, dry place, such as in a desert area, evaporation will be
so strong that you will feel quite cold before the sun starts warming you
again.
Thus, more than the actual amount of water vapor in the air, it is the
amount of water vapor in the air compared to the maximum there can be that
is of interest to us. In other words, how close we are to saturation. If we are
far from saturation, evaporation is stronger. If we are close to saturation,
evaporation is limited – and more importantly for us, as we will see shortly,
condensation is not far from happening, which creates the possibility of
clouds.
This leads us to define a new quantity that will be much more useful to
us. We first define the water vapor pressure that the air would have if it were
saturated: the saturation water vapor pressure (es), where the subscript
stands for “saturation.” And we further define the percentage of water vapor
present in the air, compared to the maximum amount at saturation, as the
relative humidity (RH):

RH=actual water vapor pressuresaturation water vapor pressure=ees

For example, if the maximum amount of water vapor that can be present in
the air is es = 12 hPa (we can think of it as the maximum capacity), and the
actual amount that is present is only e = 9 hPa, then the relative humidity is
 RH=9/12=0.75=75%

Note, also, that the above formula can be manipulated to become

e=RH×es

That is, if we know the relative humidity and the saturation vapor pressure,
we can calculate the amount of water vapor as measured by the vapor
pressure, which can be handy in some circumstances (see section 5.5.1).
Since this measure of humidity is only relative to a maximum capacity,
but not absolute as in the mass of water per volume of air, we call it relative
humidity. This subtlety is important to remember, as the relative humidity
can be high even though there is very little moisture in the air, as we will
illustrate shortly with an example.
Note that, in rare circumstances, the air can contain more than its
maximum capacity: we say that the air is supersaturated. This is the case, in
particular, in the absence of a surface for the water vapor to condense upon.
Recall that we have defined saturation with respect to a flat surface of water.
The surface provides a stable anchor point for the water vapor molecules to
attach to. When there is no surface, however, water vapor molecules bump
into each other, and have nothing to attach to. Arguably, they could attach to
each other, but since they have high kinetic energy, they tend to push each
other apart after each collision. For them to form an initial clump of water
molecules, a much higher RH than 100% is required, so that the water vapor
pressure would be sufficient to keep the water molecules together. While this
process, called homogeneous nucleation, is possible in theory, it rarely
happens in practice, because the atmosphere is full of small, invisible, solid
or liquid particles called aerosols, on which water vapor can readily attach, as
we will see shortly.
 5.5 Humidity and Temperature
We can easily think of circumstances in which the water vapor pressure will
change: if we evaporate water from the ocean, for example, the amount of
water vapor in the air (e) will increase, and the relative humidity (RH) will
increase as a result. But can the saturation water vapor pressure change? Can
the maximum capacity of the air change under any circumstances? Although
this is more difficult to conceive, a little thought experiment will help.
Let us return to our open water surface under saturated air
(Figure 5.3(c)). We had reached a state of equilibrium in which the upward
evaporation flux was balanced by a downward condensation flux. Let us
imagine that the whole system warms for some reason: there is overall more
energy in the water, and more energy in the air. The whole system is at a
higher temperature. Having more energy, the liquid water molecules will tend
to evaporate more easily, which will increase the evaporation flux. Having
more energy overall, these water vapor molecules will tend to remain as
vapor rather than return to the liquid phase, which increases the total amount
of water vapor in the air. Shortly thereafter, the system will reach a new state
of equilibrium in which condensation once again balances evaporation, but
the system can now sustain more water vapor in the gas phase: we have
increased the capacity of the air. (To return to our museum analogy, it is now
a busy weekend, we have more visitors coming into the room, more visitors
walking out, and therefore more visitors in the room at any time.)
Thus, in short, warmer air can contain more water vapor. [Note again
that, strictly speaking, the air itself has nothing to do with it. It is merely the
water that can more easily exist as vapor due to the higher temperature,
independent of the presence of warm air. But it is common parlance to say
that “warmer air can contain more water vapor.”] Conversely, the colder the
air, the less water we can evaporate into it. This is captured mathematically
by a formula derived in the nineteenth century by Émile Clapeyron(1799–
1864) and Rudolf Clausius (1822–1888), and represented graphically in
Figure 5.4.

Figure 5.4. Saturation water vapor pressure with respect to liquid water in
relation to temperature (Clausius–Clapeyron formula).

Note that the increase in saturation water vapor pressure with

temperature is exponential, so that the rate of increase is much greater at high
temperature than at low temperature. For example, a 5 ºC temperature
increase from 35 to 40 ºC corresponds to an increase in saturation water
vapor pressure of 17 hPa, whereas a 5 ºC temperature increase from 0 to 5 ºC
corresponds to an increase in saturation water vapor pressure of only 2 hPa.
This has very important consequences on Earth, where we have warm, humid
tropics and cold, dry poles.
 5.5.1 Relative vs. Absolute Humidity
Two applications come to mind. First of all, temperature changes with
latitude and environmental conditions. Since the saturation water vapor
pressure (es) is a function of temperature, we can also expect it to vary with
latitude and environmental conditions, along with relative humidity (RH),
which contains es in the denominator. For example, if the temperature is –20
ºC in Canada, we can read in Figure 5.4 that the saturation water vapor
pressure es is equal to 2 hPa. If the actual water vapor content is 2 hPa, then
the relative humidity will be

RH=e/es=2/2=1=100%

In contrast, if the temperature is 40 ºC in Death Valley, California, we read in

Figure 5.4 that es = 74 hPa. Even if the relative humidity is only 40%, which
is much less than the case before and very far from saturation, the actual
water vapor content is

e=RH×es=0.4×74=30hPa

In other words, even though the relative humidity is low, the actual water
vapor content (30 hPa) is much higher than in the previous case (2 hPa),
where the relative humidity is 100%. This reinforces the distinction between
absolute and relative humidity. Even though both are measures of humidity,
one can be high while the other is low. They tell us two very different things
about moisture. Absolute humidity tells us about the actual amount of water
vapor in the air, whereas relative humidity tells us how close or far the air is
from saturation.
 5.5.2 Condensation
The second application that comes to mind is of critical importance for
understanding the formation of clouds. Let us imagine a parcel of air at 20 °C
containing 15 hPa of water vapor. It is close to the ground and, as the sun sets
and the emission of infrared radiation by the ground becomes greater than the
absorption of solar radiation, the ground starts cooling. By conduction of heat
from the overlying air to the ground, the air parcel starts cooling as well.
At first, with a temperature of 20 °C, we can read in Figure 5.4 that the
saturation water vapor pressure (es) is equal to 23 hPa, and since the actual
water vapor pressure (e) is 15 hPa, we can calculate that the relative humidity
is

RH=e/es=15/23=0.65=65%

(This and all subsequent results are summarized in Table 5.1.)

Table 5.1. Example of saturation and condensation by temperature decrease.

T (°C) es (hPa) e (hPa) RH (%)

20 23 15 65

19 22 15 68

18 21 15 72

17 19 15 75

16 18 15 83
 15 17 15 88

14 16 15 94

13 15 15 100

12 14 14 100

11 13 13 100

10 12 12 100

As the air parcel cools, however, its water vapor capacity decreases. At
19 °C, for example, the saturation water vapor pressure is only 22 hPa, and
the relative humidity is now

RH=e/es=15/22=0.68=68%

In other words, as temperature decreases, the relative humidity increases.

[Note that es is in the denominator, and therefore we have a relationship of
inverse proportionality.] Table 5.1 shows how the relative humidity slowly
increases to 100% as the temperature decreases to 13 °C as the night
progresses. Pause for a second to ponder this result, as it is critical for our
understanding of cloud formation later on. Note that the water vapor content
(e) of the air parcel is not changing. It remains constant at 15 hPa. Thus, we
are not saturating the air parcel by adding water vapor into it: we are getting
closer to saturation by decreasing the water vapor capacity of the air parcel.
To return to our museum analogy, we could imagine a room occupied by a
few visitors whose number does not change. The room is not full, but the size
of the room decreases, as if the walls were closing in on the visitors. At some
point, when the capacity of the room decreases to the exact number of
visitors, the room will become full (saturated), but it is very different from
adding visitors to the room. Here in our air parcel, we have decreased the
capacity to 15 hPa, which is the exact amount of water vapor we had in the
parcel to start with.
What happens if we decrease the temperature further? The water vapor
capacity will decrease further and will be less than the actual amount of water
vapor in the air parcel. At 12 °C, the capacity decreases to 14 hPa, but the air
parcel contains 15 hPa. We have too much water vapor, and 1 hPa must leave
the gas phase to become either liquid or solid. In other words, 1 hPa of water
vapor must condense (or be deposited as ice if the temperature is below
freezing). Typically, close to the ground, water will condense on blades of
grass, flowers, spider webs etc., to form dew (Figure 5.5). At subfreezing
temperatures, it will form frost (Figure 5.6). As the temperature keeps
dropping throughout the rest of the night, the saturation water vapor pressure
keeps decreasing, and more water vapor has to leave the gas phase to
condense and form dew. This process will be the motivation for introducing
yet another measure of humidity, the dew point temperature – see Section
5.6.
 Figure 5.5. Dew forms on blades of grass as the temperature drops and the
air becomes saturated.

Figure 5.6. Frost forms when the air becomes saturated below 0 °C.

Note that, beyond this point, the actual water vapor pressure (e)
decreases along with es, because more and more water vapor leaves the gas
phase. Therefore, the relative humidity remains at 100%.
Two other interesting phenomena can be observed at this point. First, as
the ground gets colder and conduction proceeds, a deeper layer of air might
cool down to saturation and dew might be observed higher up above the
ground, on tree branches and rooftops. Second, water vapor might start
condensing on aerosols present in the air (suspended solid particles such as
salt, dust, pollen, smoke, pollution, etc.) to form tiny water droplets. In other
words, we might observe fog. This kind of morning fog is very common after
clear nights when the ground has had a chance to cool by emission of infrared
radiation. (Recall the temperature inversion of the nighttime temperature
profile in Figure 4.14). It is typically lying low and close to the ground
(where conduction is most efficient) and is appropriately called radiation
fog.
We will return to the formation of cloud droplets in Chapter 6.
 5.6 Dew Point Temperature
We now have everything we need to introduce our last measure of humidity.
In Section 5.5.2, we demonstrated how condensation could be predicted to
happen if, without changing the pressure or the water vapor content of the air,
we decreased the temperature of the air (by removing heat) down to a point
where the saturation water vapor capacity (es) becomes equal to the actual
water vapor pressure (e). Specifically in our example, we found that by
decreasing the temperature to 13 °C, we reach saturation, and by cooling the
air below 13 °C, condensation starts. Because dew starts forming at 13 °C, we
call that threshold temperature the dew point temperature (Td), or “dew
point,” for short.
Note that the dew point temperature is not a temperature that must be
reached, although it can be reached. It is a theoretical threshold that informs
us about the water vapor present in the air, and that is reached only when and
if the air is cooled down sufficiently at constant pressure. If there is no
change in temperature (or moisture), the dew point temperature is never
reached.
Note further that the dew point temperature is specific to the air parcel
and depends on the specific conditions of the parcel. You can verify for
yourself that, if the air parcel contained 18 hPa of water vapor to start with,
the dew point temperature would be 16 °C. If the air parcel contained only 12
hPa of water vapor, the dew point temperature would be 10 °C. The dew
point temperature, in spite of its name, does not directly tell us anything
about the temperature, or heat content of the air parcel (even though it is
called a “temperature”). It provides useful information about the moisture
content of the air parcel.
Thinking backward, we can take advantage of the previous observation.
If the air parcel, still at 20 °C, is very moist (for example, e = 22 hPa), the
dew point temperature is high and close to the actual temperature (Td = 19
°C, which is close to T = 20 °C). In other words, we need only a small drop in
temperature to reach saturation. If the air parcel is dry (for example, e = 12
hPa), the dew point temperature is low and far from the actual temperature
(Td = 10 °C, which is further from T = 20 °C). In other words, we need a
large temperature drop to reach saturation. This temperature drop to the dew
point is a useful concept in itself, called the dew point depression, i.e., the
drop in temperature necessary to reach saturation, T – Td.
Therefore:

If Td is high and close to T, the air is moist.

If Td is low and far from T, the air is dry.

Or, alternatively:

small dew point depression = moist

large dew point depression = dry.

Thus, the dew point temperature is really a measure of humidity. Like the
relative humidity, it tells us how far we are from saturation, but it
circumvents the need to measure the actual water vapor pressure (e). This has
technological advantages, because measuring e is challenging. Measuring Td,
however, can be done with a trick. By cooling a mirror, we can observe when
the mirror fogs up with condensation, which indicates that we have reached
the dew point. This can be automated in a weather station by shining a beam
of light on the mirror while it is being cooled. As long as the mirror is clear,
the beam will be reflected, which can be detected with a sensor. As soon as
the beam stops being reflected (because of the condensation), we have
reached saturation and Td can be recorded.
Finally, note that the dew point temperature also informs us about what
will happen if the temperature continues to decrease after saturation has been
reached. Recall that latent heat is released when water changes from the gas
to the liquid phase, as dew forms. This release of latent heat warms the air,
often reducing the rate of cooling after reaching saturation. At warm
temperatures, when there is a lot of water vapor in the air at saturation, this
provides a very powerful limit on how cold it can get overnight: the
temperature will drop to near the dew point, and then stop decreasing once
dew forms, due to the release of latent heat. So the dew point temperature can
be a useful guide to morning low temperatures under warm, clear, and calm
conditions. At very cold temperatures, there is so little water vapor in the air
at saturation that there is little latent heat release; as a result, the dew point is
not a good guide to morning low temperatures when the air is very cold.
 5.7 Applications of the Dew point Temperature
The dew point is a common measure in meteorology, and is used in various
situations. Three examples will further reinforce this new concept.
 5.7.1 Surface Weather Maps
Figure 5.7(a) shows the surface weather map corresponding to our February
2014 cyclone, now indicating dew point temperature measurements. It is
common practice to indicate the actual temperature in the upper left corner
and the dew point temperature in the lower left corner of the data for each
weather station. By looking at individual weather stations, we can appreciate
how it is warmer and moister closer to the Gulf of Mexico (southeast), and
colder and drier further west and north. To appreciate the distribution better,
we have colored and contoured the dew point temperature field in
Figure 5.7(b), and reproduced our temperature and pressure surface map in
Figure 5.7(c). We see the cold and warm fronts, the warm sector, and the low
pressure center reflected in the dew point temperature field (see Chapter 2).
Figure 5.7. (a) An example of temperature (in red) and dew point
temperature (in blue) distribution on February 20, 2014, at 18:00 UTC. (b)
Corresponding dew point temperature field, colored and contoured. (c)
Corresponding temperature and pressure field. Fronts and low pressure
center are indicated for reference. The location of Amarillo, Texas (A),
Dallas, Texas (D), and Shreveport, Louisiana (S) are indicated in (c) to aid
 help discussion. See the text for details.

It appears in Figure 5.7(b) that the warm sector (the sector of warm air
delimited by the two fronts) is also particularly moist (high dew point
temperature). It makes sense if we match the measurements with the wind
direction in the warm sector: this warm and moist air is essentially blown
inland from the Gulf of Mexico, i.e., from the warm and moist subtropics. By
contrast, the wind in the cold air mass behind the cold front is northwesterly
and brings dry, cold air from continental Canada. Therefore, the dew point is
much lower and there exists a zone of sharp transition from low to high dew
point where the contours are tightly packed together. In this particular
situation, the zone of transition roughly coincides with the cold front. Thus,
the frontal region, which separates cold and warm air with a strong
temperature gradient, also separates dry and moist air with a strong moisture
gradient.
Let us now explore how all this translates to a single weather station.
 5.7.2 Meteograms
When measured at a station and plotted against time, along with temperature
and wind, the dew point tells us about local changes in humidity, in particular
as a response to the horizontal displacement of moist or dry air.
Figure 5.8 shows a meteogram indicating temperature, dew point
temperature, relative humidity, and wind at Amarillo, Texas, on February 19–
20, 2014, during the passage of our February 2014 cyclone. [The location of
Amarillo is indicated by A in Figure 5.7(c) for reference.] Each wind barb at
the bottom of Figure 5.8 corresponds to a wind measurement. Thus, it is a
time series of wind speed and direction at the weather station over the course
of the two days displayed in the meteogram. The wind can be observed to
switch direction from southwesterly to northwesterly, and to increase in
speed, early on February 20, when the cold front passes through the city (see
blue arrows in Figure 5.8). We will learn in Chapter 10 that extratropical
cyclones typically move from west to east. Thus, if we recall Figure 5.7(a)
and imagine that the whole wind field was further west at an earlier time, and
then swept eastward through Texas, we can see that the weather station is
first in the warm sector, where the wind is southwesterly, then in the frontal
region, and finally behind the cold front, where the wind is northwesterly,
consistently with the meteogram. Finally, we will learn that the wind is
typically stronger around a cold front, which is also consistent with the wind
speed observations in Figure 5.8.
 Figure 5.8. Meteogram of temperature, dew point temperature, relative
humidity, and wind at Amarillo, Texas, on February 19–20, 2014 (local
time). The passage of the cold front is indicated by blue arrows.

Looking at temperature, we again find some signs of the diurnal cycle

discussed in Chapter 4, but the cycle is disrupted by the passage of the cold
front. Before we analyze temperature in more detail, however, let us look at
humidity. First of all, note how the relative humidity is low when T and Td
are far apart, and high when T and Td are close to each other. That is
consistent with our rule of thumb:
 T and Td far apart= far from saturation= lower RHT and Td far
close= close to saturation= higher RH

There is a more interesting story in this meteogram, however. The dew point
temperature can be observed to be relatively constant, except early on
February 20, around 4:00 a.m., when it increases by 10 °C in less than an
hour. That is when the cold front of our February 2014 cyclone moves
through the city (as confirmed by the changes in wind speed and direction).
Recall that a high dew point means more water vapor. Indeed, moisture,
clouds, and rain tend to concentrate along temperature fronts. (See also the
tongue of higher humidity along the cold front in Figure 5.7(b)) Furthermore,
a higher dew point means that we are closer to saturation, and the relative
humidity should increase, which is indeed what we observe in the lower
panel of Figure 5.8 during the frontal passage.
The increase in relative humidity, however, is not only due to the
increase in moisture content. A frontal passage also brings a significant
change in temperature. Recall that the cold front is the boundary of a cold air
mass moving into warmer air. Therefore, as the cold front was approaching
from the west, Amarillo was successively in the warm air mass, then in the
frontal area, and finally in the cold air mass advancing eastward, which
explains the temperature drop observed between 4:00 and 7:00 a.m. The -
temperature drops by more than 10 °C and never fully recovers during the
following diurnal cycle. [Here again, looking at Figure 5.7(c) and imagining
the temperature field (i.e., the cyclone) sweeping from west to east will help
you to visualize the two air masses sweeping over Amarillo and the
temperature dropping when the front moves through.]
As the temperature drops, the saturation water vapor pressure also
decreases, and the relative humidity increases (as in Table 5.1). Thus, the
reason why the relative humidity rises so drastically between 4:00 and 7:00
a.m. (from about 20 to 70%) is twofold: the moisture content increases and
the temperature drops. Later on during the day, northwesterly winds bring in
dry air from the continental interior and the dew point temperature drops by
10°C. Furthermore, the temperature increases slightly (although not as much
as the previous days), and, as a result, the relative humidity drops back down
to 20%.
 5.7.3 Radiosonde Profiles
The contrast between the cold and warm air masses, on each side of the cold
front, can be further illustrated by studying radiosonde profiles at appropriate
locations. Figure 5.9 shows two such profiles, at Dallas, Texas, in the cold air
mass after the frontal passage, and at Shreveport, Louisiana, in the warm
sector before the frontal passage. [The locations of Dallas and Shreveport are
indicated in Figure 5.7(c) for reference.] Figure 5.9 also shows the dew point
temperature, along with wind barbs at various altitudes. Each wind barb is to
be read as you would on a surface map, i.e., from the north when it is
pointing down, and from the south when it is pointing up. Thus, for example,
the surface wind at Shreveport is not pointing upward, but northward, i.e., it
is a southerly wind. Similarly, all upper-level winds are westerly.
 Figure 5.9. Comparison of temperature and wind profiles on both sides of
a cold front. (a) Vertical profiles at Dallas’ Texas, on February 21, 2014, at
00:00 UTC, on the western side of the cold front, after the front has moved
through. (b) Vertical profiles at Shreveport, Louisiana, on February 20,
2014, at 12:00 UTC, on the eastern side of the cold front, in the warm
sector, before the front moves through. Pressure levels (hPa) are indicated
by blue dashed lines.

By applying our now familiar rule-of-thumb, we can see that the air is
much drier in the cold air mass (T far from Td) than in the warm air mass (T
close to Td). Furthermore, the winds are most different in the lower layer of
the troposphere, up to about 800 hPa. They are mostly northerly or
northwesterly at Dallas, and mostly southerly or southwesterly at Shreveport.
Higher up they are mostly westerly throughout the rest of the troposphere.
The lower-level winds are consistent with what we observed in Figure 5.7(a),
as well as with the wind shift observed in the meteogram in Figure 5.8. The
drier profile at Dallas reflects the dry, cold air moving southward from the
interior United States and continental Canada (northwesterly winds), while
the moist profile at Shreveport reflects the warm, moist air moving northward
from the Gulf of Mexico (southwesterly winds).
 5.7.4 Back to Relative Humidity
Let us consolidate our new understanding with one more exercise. We
mentioned that Td was easier to measure than the water vapor pressure (e),
and that the vapor pressure (e) can be recovered from Td, which we shall now
demonstrate.
Let us return to our air parcel and the results of our thought experiment
compiled in Table 5.1. Let us further imagine that we know only the
temperature of the air parcel (T = 20 °C) and its dew point temperature (Td =
13 °C). We already know that we can use Figure 5.4 to deduce the saturation
water vapor pressure (es) from T (es = 23 hPa). What does the dew point tell
us? It tells us that if we cool the air parcel to 13 °C, the new water vapor
capacity will be equal to the actual water vapor pressure (es = e) and the air
parcel will be saturated. We can also use Figure 5.4 to calculate this new
saturation water vapor pressure. At 13 °C, es = 15 hPa. (Note that this is not
the original saturation water vapor pressure at 20 °C. It is a theoretical
saturation water vapor pressure that we would have if we cooled the air parcel
down to 13 °C.) Since es would be equal to e at this new temperature, we can
conclude that the saturation water vapor pressure at 13 °C (the dew point
temperature) is in fact the same as the actual water vapor pressure at 20 °C.
In other words, the dew point temperature (Td) tells us how much water vapor
our air parcel contains.
These two results are summarized in Figure 5.10. A temperature T of 20
°C tells us that es = 23 hPa. A dew point temperature of Td = 13 °C tells us
that e = 15 hPa. Therefore, the relative humidity is

RH=e/es=15/23=0.65=65%
and we were able to calculate RH completely indirectly, without measuring
either e or es.

Figure 5.10. Using T and Td to calculate es and e, and therefore RH = e/es.

 5.7.5 How to Saturate
To conclude, let us summarize and describe three different ways to saturate
an air parcel.
(1) The first way is by putting more water vapor into the air parcel, as
would happen by evaporation over a warm water body. It amounts to forcing
e to increase to equal es. This is the basic principle behind evaporation fog
(see Chapter 6).
The two other ways consist in reducing the saturation water vapor
pressure es by decreasing the temperature. It amounts to forcing es to decrease
and become equal to e by letting the temperature decrease to the dew point.
And we can do so in two different ways:
(2) We can remove heat from the air parcel. This is the basic principle
behind radiation fog, as described in Section 5.5.2, when the temperature of
the ground decreases due to radiational cooling and heat is conducted from
the air parcel to the ground. Such a process, where heat is exchanged between
the air parcel and the environment, is called a diabatic process (from the
Greek “dia” (through) and “batos” (going), as the heat goes through the
envelope of the parcel).
(3) We can also quickly change the volume and pressure of the air parcel
so that no heat exchange between the air parcel and the environment is
possible. Such a process is called adiabatic. Since heat cannot be transferred
in or out of the air parcel, the temperature adjustment is internal. In particular,
as we will see in Chapter 6, the temperature of a rising air parcel will quickly
drop as the pressure of the air parcel adjusts to the decreasing atmospheric
pressure of the surrounding air. If the temperature of the air parcel drops to
the dew point, saturation is reached and a cloud forms.
 Summary
Water can exist in three states, or phases, in the atmosphere: solid (ice
crystals, snow, hail), liquid (cloud droplets, raindrops), and gas (water vapor).
Water absorbs latent heat to change phase from solid to liquid to gas,
which cools the surrounding air in the process. Conversely, water releases
latent heat when changing from the gas phase to liquid to solid, which warms
the surrounding air in the process.
A volume of air is said to be saturated when it contains the maximum
amount of water vapor that can exist in equilibrium. Thus, it is useful to
define the water vapor pressure (e), describing the amount of water vapor
present in the air, and the saturation water vapor pressure (es), describing
the maximum amount of water vapor that the air can contain.
The relative humidity (RH) is defined as e/es × 100 and tells us how far
(low RH) or close (high RH) the atmosphere is to saturation.
The saturation water vapor pressure depends only on temperature, and
increases exponentially with temperature as described by the Clausius–
Clapeyron formula.
The dew point temperature (Td) is the temperature at which saturation
occurs when the air is cooled. If the temperature (T ) decreases to the dew
point (Td), condensation takes place.
Dew forms when condensation takes place on blades of grass and
similar features near the ground. If the dew point is below 0 °C, then we
observe the formation of frost. If condensation takes place on aerosols above
the surface, then we observe the formation of cloud droplets (cloud in the
form of fog).
The difference between the temperature and the dew point temperature
is called the dew point depression. When the dew point depression is small
(temperature and dew point are close), the air is relatively moist. When the
dew point depression is large (temperature and dew point are far apart), the
air is relatively dry.
The dew point can be used to identify humidity fronts (“dry lines”) in
weather maps, the passage of fronts in meteograms, and moist/dry layers of
the atmosphere in radiosonde profiles.
Since the temperature (T ) indicates the saturation water vapor pressure
(es) and the dew point temperature (Td) indicates the actual water vapor
pressure (e), we can use T and Td to calculate the relative humidity (RH).
An air parcel can become saturated due to: (1) evaporation of water into
the parcel at constant pressure and temperature, increasing the vapor pressure
to the saturation value; (2) a decrease in temperature to the dew point by loss
of heat from the parcel to the environment at constant pressure; or (3) a
decrease in temperature by adiabatic expansion as the parcel is brought to
lower pressure through lifting.
From Fog to Clouds
Saturation, condensation, relative humidity, and dew point temperature
provide us with all we need to understand how water vapor condenses to
form clouds.
CHAPTER 6
Cloud Formation
CONTENTS
6.1 Adiabatic Processes
6.2 Adiabatic Processes in the Atmosphere
6.3 Dry Adiabatic Lapse Rate
6.4 Relative Humidity
6.5 Moist Adiabatic Lapse Rate
6.6 Orographic Lifting
6.7 Lifting by Convergence
6.8 Frontal Lifting
6.9 Convection
Summary

We understand the basic mechanism behind condensation and the formation

of dew: decrease the temperature until saturation is reached, and water vapor
will start leaving the gas phase to form dew or frost. But what does it take for
a cloud to form in the atmosphere? Why would the temperature decrease?
What does water vapor condense upon in the absence of blades of grass and
solid objects? In this chapter we will describe the cloud formation process
and the conditions that are conducive to the formation of different types of
clouds.
Clouds are made of water droplets (in addition to ice crystals, which we will
describe in Chapter 7). These water droplets necessarily form by
condensation of water vapor into liquid droplets. We mentioned in Chapter 5
that such droplets could, in theory, form by homogeneous nucleation: water
molecules condensing onto each other, in supersaturated conditions, i.e.,
under very high relative humidity, about 400%. But, in practice, cloud
droplets always form by heterogeneous nucleation. The water molecules
condense onto a solid surface found amongst the many aerosols available in
the atmosphere: solid particles such as salt, dust, pollen, and pollutants from
vehicles and factories. Water vapor uses these aerosols, also called cloud
condensation nuclei, as anchor points, from where further condensation can
take place at 100% relative humidity.
While aerosols provide a seed for the formation of water droplets, we
still need a mechanism for bringing the air to saturation. We concluded
Chapter 5 by describing three ways by which we can saturate an air parcel:
by adding water vapor, decreasing the temperature by removing heat, or
decreasing the temperature internally through an adiabatic process. The first
two processes are effective close to the surface, and lead to the formation of
fog (Box 6.1). Higher up in the atmosphere, although radiative cooling of
moist air is an effective cooling mechanism, especially in the upper
troposphere, clouds mostly form through the third process. So we focus now
on adiabatic processes, and how they lead to different types of clouds.

Box 6.1. Fog

Fog is a special case in the cloud family: it is a cloud that touches the
surface, and therefore does not require lifting and adiabatic cooling.
Recall that we can saturate an air parcel by either adding water vapor
or decreasing its temperature (Section 5.7.5), which can be
accomplished in several ways.
(1) When cold air is in the presence of a warmer water surface,
such as the Gulf Stream in the North Atlantic Ocean or a warm lake at
the end of summer, water evaporates into the colder air and quickly
saturates it, forming water droplets and fog. Recall that the saturation
water vapor pressure is lower at colder temperatures (Figure 5.4) and
little water vapor is required to saturate the air. This type of fog,
usually called evaporation fog (Figure 6.1.1), corresponds to the first
situation described in Section 5.7.5, where e increases to become
equal to es.

Figure 6.1.1. Evaporation fog.

(2) More often, however, warmer air is in the presence of a cold

surface, in which case heat is transferred from the warm air to the
underlying surface and the air temperature decreases. If the
temperature drops to the dew point, water droplets form, resulting in
fog. This typically happens after a clear night, when the ground has
been cooling by emission of infrared radiation. The overlying air
cools down by conduction and a temperature inversion forms close to
the ground (see Figure 4.14(b)). That is why we often refer to this
type of fog as radiation fog (Figure 6.1.2).
Figure 6.1.2. Radiation fog.

(3) However, warm air can also be in the presence of a colder

surface because it moves over colder land or water. Horizontal motion
is called “advection,” and we refer to this type of fog as advection fog
(Figure 6.1.3). For example, when relatively warm marine air from
the Pacific Ocean moves eastward over the colder waters bordering
the coast of California, it cools to the dew point and fog forms. As it
continues eastward, it invades San Francisco Bay and may linger
there for hours or days. A similar situation can occur over the cold
Labrador Current in the North Atlantic Ocean.

Figure 6.1.3. Advection fog.

 Advection fog also forms when warm air moves over a colder
land surface, as is typical in southern England in winter, when
relatively warmer marine air that spent some time over the English
Channel invades the southernmost part of the island and cools when
in contact with the cold surface. The fog in southern England is
notorious for reducing visibility and disrupting both road and air
traffic. In London, which had a long history of burning coal in the
nineteenth and twentieth centuries, fog would often form on particles
of smoke, creating a thick and sickening mixture of smoke and fog
that came to be called smog.
 6.1 Adiabatic Processes
Before we investigate how air can cool to saturation, it proves useful to
consider the opposite process, where we warm an air parcel. We are already
familiar with a simple way to warm the atmosphere: heat the surface by
absorption of solar energy, and transfer the heat to the lower atmosphere by
conduction. The absorbed heat will be quickly transferred upward by
convection and mixing.
When an air parcel is heated by conduction, energy is input directly into
the parcel without actively changing its pressure or volume. We could,
however, obtain a similar result by compressing the air, for example by
quickly changing the pressure and volume of the air parcel (recall Box 1.3
and the ideal gas law). That is what happens when we compress air in a bike
pump. You might have noticed that the pump itself warms to the touch, as the
air inside heats up. The temperature increases because we exert work on the
air. We contribute our own energy (our muscle work) to the air as we impact
the air molecules with the pump piston. As the energy imparted by our
muscle work is converted to heat, the total energy is conserved, in agreement
with the law of conservation of energy (recall Box 1.1). But we must now be
more specific about conservation of energy, which takes the form of the first
law of thermodynamics – see Box 6.2.

Box 6.2. The First Law of Thermodynamics

In the nineteenth century, several scientists formulated what is now

called the first law of thermodynamics, which says that an increment
 of heat added to a system can change the temperature of the system or
be used by the system to perform work. Formally,

ΔQ=cpΔT+W

where Q stands for heat, T for temperature, W for work, cp for

specific heat of the substance being heated, and ∆ indicates a change.
In an adiabatic process, there is no exchange of heat between the
system and its surroundings. Therefore, the total amount of heat in the
system does not change and ∆Q = 0. Mathematically, this means that

cpΔT+W=0

and

cpΔT=−W

In other words, work exerted on the system (W < 0) or by the system

(W > 0) must be compensated by a change in temperature. In the bike
pump example presented in the text, the work exerted by our muscles
is positive for us, and therefore negative for the air (it is received by
the air), therefore –W is positive and ∆T is positive: the air warms
(adiabatic warming).
In the bike tire example, work is exerted by the air rushing out of
the tire and pushing against its surroundings, therefore W is positive
and ∆T is negative: the air cools (adiabatic cooling).

In both cases (conduction and compression), we have increased the heat

content of the air parcel, and therefore its temperature. By inverse reasoning,
we can assume that the heat content of the air can decrease: (1) by conduction
of heat from inside the parcel to the outside environment, which is what
happens at night when the lower atmosphere cools by conduction of heat to
the surface; or (2) when work is exerted by the air such that it loses energy. A
good example of this second situation is the reverse operation of pumping air
into a tire: when the air rushes out of the bike tire after the valve is opened.
Since air in the tire is at higher pressure than the air outside the tire, air rushes
out of the tire and expands to occupy a larger volume than it did in the bike
tube alone. As it does, it must push the surrounding air aside, and this
requires work (Figure 6.1). In other words, the air inside the tire exerts work
on the surrounding air and, in doing so, loses energy, which cools the air. In
fact, if you place your hand in front of the valve, you will feel that the air is
colder than the air you pumped in. Work has an energy cost, and it is paid out
of the heat content of the air doing the work.

Figure 6.1. Adiabatic cooling. (a) At first, the air is in the bike tire at high
pressure. (b) When it rushes out, it expands quickly without mixing or
exchanging heat with the surrounding air. The blue envelope with dashed
lines indicates that the air inside the tire in (a) can be viewed as an isolated
air parcel that expands outside of the tire, increasing its volume and
decreasing its pressure and temperature.
 These temperature changes are possible because the two processes
(compression in the pump and escape out of the tire) are happening very
quickly – so quickly that no exchange of energy is possible between the air
being compressed or released and the environment (i.e., the surrounding air).
To help us visualize this abstract concept from thermodynamics, we can
imagine that the air inside the tire is like an isolated air parcel (Figure 6.1(a)).
In the first case (compression), as we push on the piston, the volume of this
closed system decreases very quickly, and the pressure increases, along with
temperature. In the second case (air escape from the tire), the air in the tire
mostly does not mix with the surrounding air when it escapes, but remains
inside an imaginary envelope that expands out of the tire (Figure 6.1(b)). It
remains mostly a closed system because the expansion is very fast. The
volume of this system increases as the pressure decreases, along with
temperature.
A process in which no energy is being exchanged between the system
and its surroundings is called an adiabatic process. The compression in the
bike pump causes adiabatic warming, and the expansion of air outside the
tire causes adiabatic cooling. A more formal explanation is provided in
Box 6.2.
 6.2 Adiabatic Processes in the Atmosphere
How is all this information about adiabatic processes relevant to cloud
formation? The reason is that air parcels undergo adiabatic expansion
whenever they move upward. Recall that atmospheric pressure decreases
rapidly with height (Figure 1.6). For example, consider an air parcel at the
surface, which you can imagine as a balloon of air with an invisible
boundary. As it rises, the balloon quickly finds itself in an environment
where atmospheric pressure is much less than inside the balloon, which
makes the balloon expand outward against the surrounding air at lower
pressure. Because the pressure adjustment is instantaneous and adiabatic (no
exchange of air, no exchange of heat), the air molecules in the air parcel
perform work on the surrounding environment and lose energy as a result: the
air parcel cools. The reason the air temperature inside the parcel does not
adjust to the outside temperature is because the heat exchange processes,
such as mixing and radiation, are much slower than adiabatic expansion, and
as a result the parcel cools.
Now you probably recognize a familiar scenario: if we lift the air parcel
high enough, its temperature will decrease sufficiently so that the saturation
water vapor pressure becomes equal to the actual water vapor content: the air
in the parcel will have reached saturation and condensation will take place.
We are on our way to making a cloud.
How far does an air parcel need to rise before it reaches saturation?
Should we always observe the formation of a cloud when air rises? What are
the environmental factors that control the formation of clouds? Before we
answer these questions and continue our exploration of cloud formation, let
us have a quantitative look at adiabatic expansion.
 6.3 Dry Adiabatic Lapse Rate
The vertical structure of the atmosphere is relatively well known. It is known
on average (see Figures 1.6 and 3.2, for example), and it can be known in
detail on any given day by releasing a radiosonde, which makes
measurements on its way up through the atmosphere. We mentioned in
Chapter 3 that the environmental lapse rate (the rate at which temperature
decreases with height) is about 6.5 °C/km on average, but that it can be very
different on any given day and at any given location. The environmental
lapse rate is set up by a mix of radiative and convective processes, as well as
horizontal air motion (i.e., advection).
Since we know the rate at which pressure decreases with height, we can
calculate the work that the air parcel does as it is lifted above the surface, and
therefore the rate of temperature decrease (see Box 6.2). The rate of
temperature decrease due to adiabatic lifting and expansion is called the dry
adiabatic lapse rate and is about 10 °C/km (or 5.5 °F/1000 ft) for
unsaturated air. (We will deal with moisture and saturation shortly.)
Note, and this is an important point, that the concept of an adiabatic
lapse rate applies to air inside the parcel, which is different from the concept
of an environmental lapse rate outside the parcel. When establishing an
environmental temperature profile and calculating an environmental lapse
rate, we are simply measuring temperature at each altitude with a
thermometer. Keep in mind that this vertical temperature structure is
determined by several processes, including radiative and convective
processes, as well as advection. When calculating the dry adiabatic lapse rate,
however, all that matters is that the air parcel is rising and losing energy as it
works against the surrounding environment during expansion (Figure 6.2).

Figure 6.2. The temperature inside a rising air parcel decreases at the dry
adiabatic lapse rate (10 °C/km) while the temperature measured by a
thermometer at different altitudes decreases at the environmental lapse rate
(6.5 °C/km on average).

Note further that, by adiabatic expansion, the air parcel cools at a much
faster rate than the surrounding air outside the parcel because that air is not
rising. And because there is no exchange of heat between the parcel and its
surroundings, the air parcel usually becomes colder than the surrounding air.
Therefore, the temperature drops in the air parcel independent of the
temperature of the surrounding air. This difference in temperature between
the air parcel and the surrounding air is critical to understanding cloud
formation, and will occupy us for much of the rest of this chapter.
Finally, note that the adiabatic lapse rate applies in the downward
direction as well: when an air parcel subsides (another word for going down)
by 1 km, it warms by 10 °C. That is because, finding itself in an environment
where the pressure is greater, the air inside the parcel is being compressed.
The surrounding air now performs work on the parcel, which results in an
increase in temperature (Figure 6.3). This is exactly the same process as the
bike pump example, where the pump warms as we increase the pressure in
the tire.

Figure 6.3. The temperature inside a subsiding air parcel increases at the
dry adiabatic lapse rate (10 °C/km) due to adiabatic compression.
 6.4 Relative Humidity
When an air parcel rises, its water content does not change, since there is no
exchange of air, and therefore no exchange of moisture with air surrounding
the parcel: to a first approximation, the water vapor pressure (e) stays the
same. [Note that, strictly speaking, p and e both decrease proportionally,
following Dalton’s law – see Box 5.1 – but, for demonstration purposes, we
will assume that e remains constant.] The temperature, however, decreases
due to adiabatic cooling, and we know that the saturation water vapor
pressure (es), being a function of temperature, also decreases (Figure 5.4). As
a result, the relative humidity (RH) increases. For example, Figure 6.4 shows
an air parcel at 20 °C containing 10 hPa of water vapor. Using Figure 5.4 we
can calculate that es = 23 hPa, and therefore:

RH=e/es=10/23=0.43=43%

After rising 1 km, the air parcel has cooled by 10 °C (at the dry adiabatic
lapse rate) to reach a temperature T = 10 °C, which corresponds to es = 12
hPa, so the new relative humidity is

RH=e/es=10/12=0.83=83%

As the temperature decreases, the relative humidity increases (consistently

with what we found in Chapter 5 for the formation of dew). Note again that
RH does not increase because there is more moisture in the air parcel – only
because the capacity decreases, and there is more moisture in relation to the
maximum capacity.
 Figure 6.4. The relative humidity (RH) inside a rising air parcel increases
due to adiabatic cooling and the decrease in saturation water vapor
pressure.

Again, since the temperature decrease applies to the parcel, and not the
air surrounding the parcel, the increase in relative humidity pertains only to
the air in the air parcel. The water vapor content of the environment at that
altitude (outside the parcel) might be only e = 3 hPa, because most of the
water vapor is concentrated near the surface and e quickly decreases with
height. If the surrounding air temperature is 13.5 °C, which corresponds to
es = 15 hPa, the relative humidity of the surrounding air is:

RH=e/es=3/15=0.20=20%

The relative humidity is much lower outside than inside the parcel. That is
because the air parcel has been lifted with its water vapor, and is therefore
necessarily more moist than its surroundings, where there is very little
moisture.
The reverse is also true: when an air parcel subsides in the atmosphere,
its water vapor content does not change, but its temperature increases due to
compression and adiabatic warming, which increases its saturation water
vapor pressure (Figure 6.5). As a result, the relative humidity decreases. We
will see that these circumstances are typical of anticyclones, as well as the
downwind side of mountains.

Figure 6.5. The relative humidity (RH) inside a subsiding air parcel
decreases due to adiabatic warming and the increase in saturation water
vapor pressure.
 6.5 Moist Adiabatic Lapse Rate
What happens if an air parcel rises and cools sufficiently that it reaches
saturation, but we keep lifting it? We know that some water vapor will leave
the gas phase and condense on aerosols to form cloud droplets. We also know
that the process of changing from the gas to liquid phase releases latent heat
(the additional energy that was allowing the water molecules to be at a higher
level of molecular activity). This latent heat warms the air parcel (see Box 6.3
for a more formal approach). Thus, we are dealing with two competing
effects: adiabatic cooling due to expansion, and latent heating due to the
condensation of water. As a result, a saturated air parcel rising 1 km in the
atmosphere cools by about 10 °C due to adiabatic cooling, but also warms by
about 4 °C due to latent heating (Figure 6.6). The net result is that the air
parcel cools by only 6 °C/km. We call this the moist, or saturated, adiabatic
lapse rate. (In fact, this rate depends strongly on temperature, because it
depends on the saturation vapor pressure, but we will use this value since it
applies broadly.)

Box 6.3. The First Law of Thermodynamics, Revisited

When a rising air parcel is already saturated, some water vapor

condenses to form cloud droplets and latent heat is released. Since
this latent heat is released in the parcel and warms the air inside the
parcel, the process is not adiabatic: the change in heat content of the
parcel, ∆Q, is positive. Formally,
 ΔQ=cpΔT+W>0

and

cpΔT=−W+ΔQ

The change in temperature of the parcel is due both to work exerted

by the system and to the release of latent heat into the system.
Because W is greater in magnitude than ∆Q, the change in
temperature is still negative, but it is less than what it would be
without the release of latent heat.

Figure 6.6. The temperature inside a rising parcel saturated with water
vapor decreases at the moist adiabatic lapse rate (6 °C/km).

Note that, in our example, the air parcel contained 19 hPa of water vapor
(e) at the surface, but now contains only 13 hPa after rising 1 km, because the
saturation water vapor pressure (es) has dropped to 13 hPa and 6 hPa of water
vapor has condensed to form cloud droplets. In other words, e follows es and
they both decrease at the same rate as the air rises – recall that e can never be
greater than es.
Let us apply these new concepts to a real-world situation.
 6.6 Orographic Lifting
We now have all the ingredients to make a cloud through lifting. We know
that, if air rises, it expands adiabatically and cools at the dry adiabatic lapse
rate, and, if lifted enough, can reach saturation. The simplest mechanism to
lift air is a mountain range in the path of horizontal wind. As the air runs into
the mountain range, it rises over the mountain (on the windward side) and
subsides on the other side (the leeward side, also called the “lee” of the
mountain). Because the air is lifted by orographic features (i.e, by the
topography), we refer to this lifting mechanism as orographic lifting.
Let us consider an air parcel at 24 °C, containing 8 hPa of water vapor,
and ascending a 5 km mountain range (Figure 6.7). We can see from
Figure 5.4 that, at 24 °C, the saturation water vapor pressure of the air parcel
is 30 hPa. Therefore, its relative humidity is 8/30 = 0.27 = 27% at the foot of
the mountain.

Figure 6.7. Schematic representation of an air parcel ascending a

 mountain, cooling due to adiabatic expansion, and reaching saturation at an
altitude of 2 km (cloud base).

As the air parcel rises up the mountain slope, the pressure decreases, and
the parcel temperature decreases at the dry adiabatic lapse rate due to
adiabatic expansion: it decreases by 10 °C for each kilometer gained in
elevation. At any elevation, we can calculate the new temperature, the
saturation water vapor pressure, and therefore the relative humidity, as
summarized in Table 6.1.

Table 6.1. Example of changes in T, es, e, and RH as an air parcel rises along
a mountain slope.

Altitude Temperature Saturation Water Relative

(km) (°C) water vapor humidity
vapor pressure (%)
pressure (hPa)
(hPa)

0 24 30 8 27

1 14 16 8 50

2 4 8 8 100

3 –2 6 6 100

4 –8 4 4 100

5 –14 2 2 100
 We can see that saturation is reached at an elevation of 2 km, where the
temperature has dropped by 20 °C, and es has dropped to 8 hPa, becoming
equal to the actual water vapor pressure (RH = 100%). This elevation marks
the altitude of the cloud base and is called the Lifting Condensation Level
(LCL). With additional lifting, the air parcel temperature keeps decreasing,
which means the saturation vapor pressure es decreases, and the excess water
vapor condenses out of the vapor phase onto condensation nuclei to form
cloud droplets: we are now inside the cloud.
Since water vapor condenses (above 2 km), latent heat is released and
the air parcel now cools at the moist adiabatic lapse rate (6 °C/km). As water
leaves the gas phase, the actual water vapor content of the air parcel
decreases – it “follows” the saturation water vapor pressure, down to 2 hPa at
the top of the mountain.
Where does this condensed water go? We have assumed that air parcels
exchange no mass or energy with the surrounding air, but we have to make an
exception once clouds form, since precipitation processes can cause the water
to leave the air parcel permanently (for example, as rain or snow reaching the
ground).
What happens as the air parcel subsides back down the other side of the
mountain range? As soon as it starts subsiding, it is compressed adiabatically
and the temperature increases. With the slightest temperature increase, the
saturation water vapor pressure also increases, and es becomes greater than e:
the relative humidity drops below 100% and the air parcel is no longer
saturated when it subsides. Therefore, the air parcel temperature increases at
the dry adiabatic lapse rate all along the lee slope (Figure 6.8), i.e., much
more than it decreased on the windward side. The whole descent is
summarized in Table 6.2, where the air parcel can be seen to warm to 36 °C,
which is 12 °C greater than its original temperature at the base of the
mountain on the windward side. Moreover, because the air parcel temperature
is high (and therefore so is the saturation water vapor pressure), and because
the actual water vapor content is so low (because the water left the air parcel
in clouds and precipitation on the windward side), the relative humidity is
extremely low. In other words, not only is the air very warm, but also it is
very dry.

Figure 6.8. Schematic representation of the air parcel subsiding on the

other side of the mountain, and warming due to adiabatic compression.

Table 6.2. Example of changes in T, es, e, and RH as the air parcel subsides
back down on the lee side.

Altitude Temperature Saturation Water Relative

(km) (°C) water vapor humidity
vapor pressure (%)
pressure (hPa)
(hPa)
 5 –14 2 2 100

4 –4 5 2 40

3 6 10 2 20

2 16 18 2 11

1 26 34 2 6

0 36 60 2 3

These calculations are useful to explain many observations of orographic

precipitation and subsidence around the world. All along the West Coast of
North America, for example, the prevailing westerly winds are lifted by the
coastal mountain ranges and produce enhanced precipitation, and therefore
lush vegetation cover on the western slopes. The eastern slopes are
comparatively warmer and drier, and are often a desert landscape. That is
where the air subsides and warms by adiabatic compression at the dry
adiabatic lapse rate. This is illustrated in Figure 6.9 for Washington State,
where the westerlies first ascend the Coastal Range and the Olympic
Mountains, then the Cascades, and finally the Rocky Mountains. Note on the
satellite image (Figure 6.9(a)) how the western slopes are relatively green,
while Eastern Washington, in the lee of the Cascade mountains, has the
brown appearance of a desert. The same picture applies to the Sierras of
California and the deserts that exist in their lee. Note also that air parcels do
not always subside back to sea level, but more often to an elevated plateau, so
that the increase in temperature is usually not as great as described in our
idealized example.
 Figure 6.9. Example of orographic precipitation on the windward slopes of
the Coastal Mountain Range, the Cascades, and the Rocky Mountains in
the Pacific Northwest. (a) Vegetation cover and snow as captured by
MODIS on the NASA Terra satellite. (b) Land elevation along the red
dashed line in (a).

The warm and dry winds caused by subsidence on the lee side of
mountain ranges are well known to people living in these locations. The
chinook, for example, results from subsidence on the eastern slopes of the
Rockies that blows into the Great Plains. The foehn results from
Mediterranean air blowing over the Alps and descending into the plains and
valleys of Central Europe. The terms “chinook” and “foehn” are used more
generally to refer to winds that result from orographic lifting and subsidence.
Orographic precipitation and lush vegetation contrasting with warm and dry
lee slopes can be observed in many other places, such as New Zealand, the
southern part of South America, in Chile and Argentina, and Hawaii
(Figure 6.10).

Figure 6.10. The northeastern slopes of Hawaii, exposed to the trade

winds (blue arrow), are lush and green, while the southwestern slopes are
in the lee of the dominant winds, and therefore much drier.

To conclude, let us emphasize that all our deductions were made with
very little information: merely the temperature (T ) and water vapor content
(e) at the base of the mountain. Everything beyond that was based on
knowledge of the adiabatic lapse rates and the mechanism of condensation. It
is noteworthy that we can predict the altitude of the cloud base and the
conditions in the lee of mountains with no information about the atmosphere
above the surface.
Finally, let us summarize and emphasize that cloud formation requires
cooling air to saturation, and that, in most cases, the cooling occurs as a result
of lifting and adiabatic expansion. As for subsidence, we established that, as
soon as the air subsides, it warms due to adiabatic compression. Necessarily,
the saturation water vapor pressure (es) increases and the relative humidity
drops. Consequently, clouds cannot form in sinking air without a source of
moisture. In other words, subsidence inhibits cloud formation. We will make
use of this fact shortly. In the meantime, let us explore additional
mechanisms leading to the formation of clouds.
 6.7 Lifting by Convergence
When air flows from different directions and meets in the same location, it is
said to converge, much like cars on different roadways converging on an
intersection. Although in principle the air density could increase at the point
of convergence, it turns out that air is more likely to move in a different
direction rather than change density. When convergence happens at the
ground, the only option is upward, in which case adiabatic cooling can occur,
along with cloud formation and possibly precipitation (Figure 6.11(a)).

Figure 6.11. Two examples of lifting by convergence. (a) Winds converge

from different directions. (b) Wind speed decreases as the roughness of the
underlying surface increases.
 Convergence can also take place when the wind slows down along its
path, while keeping a higher speed upstream (Figure. 6.11(b)). Although this
is more difficult to visualize, one can see how air would tend to accumulate
and pile up because the air ahead is not flowing fast enough. This is similar to
a car slowing down on the highway, and forcing the cars behind it to get
closer and more compact. While cars have no choice but to slow down as
well, air can continue upward, possibly causing condensation and cloud
formation if saturation is reached due to adiabatic cooling. This slowing
effect can happen when wind blowing over a water surface, such as a lake,
reaches the shore and continues to flow over land. While water is relatively
smooth and allows the wind to blow faster, land is rougher, and friction slows
the wind. In such a situation, clouds are observed just downstream of the
shore, over land. In winter, these clouds can produce snow.
 6.8 Frontal Lifting
You might recall that a cold front occurs where a cold air mass advances
against a warm air mass. We introduced a bird’s eye view in Chapter 2, here
reproduced in Figure 6.12(a). If we switch to a cross-section (Figure 6.12(b)),
we notice that cold air, being denser, tends to sink close to the surface and to
wedge underneath the warm air mass. This causes the warm air to rise, which
can lead to condensation and the formation of clouds. Because this ascent of
the warm air is relatively abrupt, clouds of the cumulus family tend to form,
especially if the atmosphere ahead of the cold front is unstable – see Section
6.9. Cumulus clouds have more vertical development, and a cauliflower
appearance that makes them easily recognizable (see Appendix 6.1).

Figure 6.12. Cloud formation by frontal lifting. (a) Bird’s eye view
showing a cold air mass advancing against a warm air mass. (b) Cross-
 section showing the same cold air mass wedging underneath the warm air
and lifting it up, causing the formation of cumuliform clouds. (c) Bird’s
eye view of a warm air mass advancing toward a cold air mass. (d) Cross-
section showing the warm air overrunning the cold air and reaching
saturation, causing the formation of stratiform clouds.

In contrast, when a warm air mass advances against a cold air mass
(Figure 6.12(c)), it does not wedge underneath, but slides upward and over
the cold air mass, a phenomenon called overrunning (Figure 6.12(d)). As
usual, clouds form as soon as saturation is reached, but since overrunning is
the result of a more gentle uplift than the wedging of the cold air mass
described earlier, clouds of the stratus family tend to form (stratus,
altostratus, and cirrostratus). Stratus clouds have a more horizontal
development and take on the form of extensive sheets of uniform clouds (see
Figures 6.32 and 6.33 in Appendix 6.1).
 6.9 Convection
Last but not least, clouds can form when warm air is lifted by convection -- in
other words, by buoyancy forces, as described by Archimedes’ principle (see
Box 6.4). This lifting mechanism is the most difficult to understand, but
results in spectacular weather in the form of thunderstorms, so we will devote
the rest of this chapter to understanding how it works and how, in general,
buoyancy affects the shape of the clouds that we see in the atmosphere.

Box 6.4. Buoyancy and Archimedes’ Principle

Any object bathed in the atmosphere is experiencing pressure forces

from surrounding air molecules. Since atmospheric pressure
decreases with height, the downward force exerted by air molecules
above the object is less than the upward force exerted by air
molecules below. This pressure differential results in a net force,
called the buoyancy force, that pushes the object upward. When the
weight of the object is exactly balanced by the buoyancy force, the
object is in equilibrium – it is floating (Figure 6.4.1). This would be
the case of a balloon floating at a given height above ground, or a
submarine stable at a given depth in the ocean (the same reasoning
applies in water).
Figure 6.4.1. Force balance between the weight of an air parcel and
the buoyancy force applied to that parcel in the case of equilibrium.

Another way to look at this balance of forces is to see that the

buoyancy force is equal in magnitude to the weight of the floating
balloon – and opposite in direction. If it were not so, the balloon
would either be pushed up (if the buoyancy force were greater than
the weight) or down (if the weight were greater than the buoyancy
force).
Now for a thought experiment: replace the balloon with the
parcel of air that would be there if the balloon were not – in other
words, replace the balloon with the air that was displaced by the
balloon. That air parcel would have the same volume as the balloon,
and, since it would also be floating and stable, it must have the same
weight as the balloon. Therefore, the balloon and the displaced air are
interchangeable, and the buoyancy force, which is equal in magnitude
to the weight of the balloon, is also equal to the weight of the
displaced air. This is, in essence, Archimedes’ principle: “When an
object is immersed in a fluid, it is subject to an upward buoyancy
force equal in magnitude to the weight of the displaced fluid.”
 Let us go one step further. Replace the floating balloon with a
parcel of identical volume, but containing warm air. Recall that warm
air has a lower density, and therefore (the volume being equal) the
warm parcel contains less mass of air than a cold parcel. As a result,
the weight of the warm parcel is less than the weight of the earlier
parcel at environmental temperature. In particular, it is not sufficient
to balance the upward buoyancy force. Therefore, there is a net
upward force that pushes the parcel up: warm air rises Figure
6.4.2(a)). Replace the parcel with cold, dense air, and the weight will
be increased and will overcome the buoyancy force. The net force
will push the air parcel downward: cold air sinks Figure 6.4.2(b)).

Figure 6.4.2. (a) Warm air rises. (b) Cold air sinks.
 6.9.1 Stable Air
Let us consider an atmosphere described by the temperature profile shown in
Fig. 6.13, which could be measured by a radiosonde. (By default, we use red
to show the environmental lapse rate – red like the color of a thermometer). It
starts with a temperature of Te = 20 °C at the surface, and quickly drops with
height. Let us now imagine an air parcel at sea level that is lifted from the
surface. At first, the air parcel temperature is equal to that of the environment
(Tp = 20 °C). As soon as the air parcel is lifted, its temperature decreases at
the dry adiabatic lapse rate, 10 °C/km (let us not worry about humidity right
now). If it rose 1 km, its new temperature would be Tp = 10 °C (see the blue
arrow in Figure 6.13), which we can compare to the temperature of the
environment at that height, outside the air parcel. The environmental
temperature (as indicated by the red curve) is Te = 15 °C. Therefore, the air
parcel would be colder than its surroundings, more dense, and would tend to
sink back down. In these circumstances, air parcels return to where they came
from. We say that the atmosphere is stable, or that we have a
stable temperature profile. (See Box 6.5 for a complete description of
stability.)
Figure 6.13. Example of a stable temperature profile (in red). The blue
arrow represents the decrease in temperature inside an air parcel as it is
lifted from the surface.

Box 6.5. Stability

Lifted air parcels can be either stable, unstable, or neutrally stable,

depending on what happens when they are lifted. To understand
better how the concept of stability applies to the atmosphere, it is
useful to go back to the more general definition of stability in physics,
and to use the classic example of a ball on a curved surface.
In Figure 6.5.1, the ball is at the bottom of a valley. If given a
small push, left or right, the ball will tend to return to its original
position. We say that the ball is in a stable position.
Figure 6.5.1. Stable situation.

In Figure 6.5.2, the ball is at the top of a mountain. If given a

small push, the ball will continue rolling away from its original
position. We say that the position of the ball is unstable.

Figure 6.5.2. Unstable situation.

By analogy, a temperature profile such as shown in Figure 6.13

is described as stable, because a lifted parcel will tend to return to its
original position, whereas a temperature profile such as described in
Figure 6.14 is described as unstable because a lifted parcel will
continue to rise, away from the surface.
In between those two situations, there is the possibility that the
ball is on a flat surface, in which case, if given a small push, the ball
will neither return to its original position nor continue rolling. It will
simply stop at its new position (Figure 6.5.3). We say that the ball is
neutrally stable.
Figure 6.5.3. Neutrally stable situation.

A temperature profile that would exactly follow the adiabatic

lapse rate would be described as neutral because a lifted air parcel
would cool at the same rate as its surroundings. If lifted, the air parcel
would be at the same temperature as the environment and would stop
at its new position.
There exists a final type of situation in which the ball can have a
stable or unstable behavior depending on the circumstances. In
Figure 6.5.4, the ball will be stable if given a small push, but will
become unstable if given a large enough push that it reaches one of
the adjacent summits. We say that the position of the ball is
conditionally unstable.

Figure 6.5.4. Conditionally unstable situation.

In the atmosphere, conditional instability occurs when lifted air

parcels are stable when the lift is small, but become unstable when
the upward displacement is sufficiently large (Figure 6.22).
 6.9.2 Unstable Air and Thermals
By contrast, consider the atmosphere described by the temperature profile in
Figure 6.14, where temperatures close to the surface are higher than in the
previous example, presumably due to the absorption of solar radiation by the
surface and transfer of heat to the lower atmosphere by conduction. Let us
repeat the previous exercise and lift an air parcel 1 km above the surface,
starting at a temperature of 30 °C. Its new temperature is Tp = 20 °C, which is
greater than the environmental temperature at that height (Te = 13°C).
Therefore, even though it has risen and cooled, it is still warmer than its
surroundings. This means that the air parcel is less dense than its
surroundings, and therefore still subject to an upward buoyancy push. We say
that the atmosphere (or at least that particular layer of the atmosphere) is
unstable.

Figure 6.14. (a) Example of an unstable temperature profile (in red). The
 blue arrow represents the decrease in temperature inside an air parcel as it
is lifted from the surface. The air parcel is buoyant up to about 3 km,
where it stops rising. (b) Resulting thermal.

As the air parcel continues to rise, its temperature continues to decrease

at the same rate, until it becomes colder than its surroundings, in which case
it stops rising because its density matches that of the surroundings and the
buoyancy force becomes zero. By following the blue arrow in Figure 6.14(a),
we can see that this happens above 3 km, where the temperature profile is
once again stable.
On a sunny afternoon, many such warm parcels might experience such
an upward buoyancy push. We cannot imagine, however, an entire layer of
warm air close to the surface rising all at the same time. Rising air needs to
be replaced by some fresh air coming from another direction – the void
created by the rising air needs to be filled, in a sense. In practice, the
atmosphere organizes itself in such a way that warm air rises in places, while
cooler air from aloft sinks in between, as shown in Figure 6.15. The pockets
of rising warm air, which provide a natural upward push, are called thermals.
They are invisible to the naked eye, but are perceptible to the senses of birds
and glider pilots, who learn to fly from one thermal to another, and can
remain airborne indefinitely.
Figure 6.15. Organization of convective cells with alternating thermals
and regions of subsidence.
 6.9.3 Stable vs. Unstable
If you pay attention to the environmental temperature profiles in Figures 6.13
and 6.14, you will notice that the dry adiabatic lapse rate, the blue line, does
not change (of course). What determines the stability of the atmosphere is the
particular tilt of the environmental temperature profile, of the red curve – in
other words, the environmental lapse rate. Whenever the environmental
temperature profile is leaning to the left of the adiabatic lapse rate, the
atmosphere is unstable (because, again, a rising air parcel will necessarily
remain warmer than the environment). Whenever the environmental -
temperature profile is leaning to the right of the blue line, the atmosphere is
stable, as summarized in Figure 6.16. If the environmental temperature
profile is exactly aligned with the adiabatic lapse rate, it is neither stable nor
unstable: we say that it is neutral, or neutrally stable. You can verify for
yourself that, with a neutral temperature profile, a lifted air parcel will neither
continue to rise nor sink back down. Being at the same temperature as its
surroundings, it simply stays at its new position (see also Box 6.5).
 Figure 6.16. Various environmental lapse rates (red) in relation to the dry
adiabatic lapse rate (blue). Temperature profiles in the blue sector tend to
be unstable, while they tend to be stable in the red sector.

Certain situations tend to make the atmosphere unstable, e.g., large-scale

lifting and cooling aloft. Conversely, subsidence warming, for example, tends
to make the atmosphere more stable. An interesting case occurs when the
environmental temperature profile is increasing with height (a temperature
inversion), as is the case after a clear night, after the lower atmosphere has
cooled by conduction of heat to a cold surface (Figure 6.17). Since the
atmosphere itself becomes warmer with height, and a rising air parcel can
only cool when lifted, any lifted air will necessarily experience a strong
buoyancy force to sink back down. That is why temperature inversions are
notorious for trapping air at low elevations, preventing the dissipation of fog,
for example, or trapping pollution and causing health hazards (discussed in
more detail in Chapter 14).
Figure 6.17. A temperature inversion is necessarily stable, as a lifted air
parcel necessarily cools down, while the environmental temperature
increases with height.
 6.9.4 Fair-Weather Cumulus Clouds
Let’s add water to our understanding of stability to see how cloud formation
fits in the picture. For example, consider the temperature profile shown in
Figure 6.18 and lift a parcel from the surface. The air parcel has a
temperature Tp = 30 °C and contains e = 12 hPa of water vapor. Using
Figure 5.4, we can calculate that the saturation water vapor pressure at 30 °C
is es = 43 hPa. Therefore, the air parcel is not saturated and cools at the dry
adiabatic lapse rate (10 °C/km) when lifted. After rising 1 km, it has cooled
to 20 °C, which corresponds to es = 23 hPa. The parcel is still not saturated (e
< es), and it is still warmer than the environment (Tp > Te). Therefore, it keeps
rising due to the upward buoyancy force. So far, we are following the
unstable scenario described previously.

Figure 6.18. (a) Example of an unstable temperature profile (in red) with
formation of a fair-weather cumulus. The blue and green arrows represent
the decrease in temperature inside an air parcel as it is lifted from the
 surface: first at the dry adiabatic lapse rate (blue) and then at the moist
adiabatic lapse rate above 2 km (green). The air parcel is buoyant up to
about 3 km, where it stops rising. (b) Resulting cumulus cloud.

After rising one more kilometer, however, the air parcel has cooled
down to 10 °C, at which point es = 12 hPa. We have reached saturation
(es = e, RH = 100%), and we know that any further lifting will lead to
condensation and cloud formation. Recall that this critical threshold is called
the lifting condensation level (LCL) as it is the location where the cloud
base is realized.
We also know that, due to latent heat release, any cooling from the air
parcel rising further will occur at the moist adiabatic lapse rate (6 °C/km).
Since the air parcel is still warmer than its surroundings (Tp > Te), it keeps
rising and, after one more kilometer, will cool to 4 °C. The saturation water
vapor pressure at this new temperature is 8 hPa, which means that 4 hPa of
water vapor have condensed to form the lower portion of a cloud. As for the
air parcel ascending the mountain, the water vapor pressure decreases and
“tracks” the saturation water vapor pressure (e = es = 8 hPa). The relative
humidity remains 100% and the air parcel is saturated at all times.
Is the air parcel still buoyant? Its new temperature is equal to the
environmental temperature (Tp = Te = 4 °C), therefore it stops rising, and the
cloud does not develop much above 3 km (the cloud top). We are in the
presence of a relatively small cloud. The upward motion of air associated
with convection tends to form clouds of the cumulus type (bright white
bundles with a cauliflower appearance and sharply defined boundaries).
Since these clouds tend to form on sunny days when the surface is heated by
absorption of solar radiation and the atmosphere is only mildly unstable, we
often call them fair-weather cumulus (Figure 6.19). In the same way that
thermals are organized within convective cells, fair-weather cumulus clouds
tend to appear within the ascending branch of convective cells and to
alternate with regions of subsidence where there is no cloud (Figure 6.20).
That is why they often appear as isolated clouds distributed at somewhat
regular intervals. Note also that, because the lower atmosphere is often
relatively uniform in temperature and humidity, all air parcels reach
saturation at about the same height (the LCL). That is why cumulus clouds
often have a flat cloud base, and why all cumulus clouds present at a given
time often have their cloud base at around the same height.

Figure 6.19. Fair-weather cumulus clouds.

 Figure 6.20. Formation of fair-weather cumulus clouds within the
ascending branch of convective cells.

If you observe cumulus clouds for a few minutes, you will notice that
they form and dissipate relatively quickly. Due to the turbulent motion
caused by convection, they mix with the surrounding air, especially along
their boundaries. This process is called entrainment. Since the surrounding
air is drier, and not saturated, the entrainment of environmental air into the
cloud tends to lower the relative humidity inside the cloud, causing
evaporation and cooling of the warm buoyant air. In the process, however,
moist air from the cloud is being mixed into the environment, which moistens
the atmosphere. Moisture has been lifted from surface level up to higher
elevations, increasing the relative humidity. Therefore, as cumulus clouds
grow and evaporate, the temperature and humidity profiles evolve. In
particular, the atmosphere can become more unstable due to moistening the
air at higher levels. As the clouds grow in size and number, they start to
occupy, and “congest,” the sky, developing into cumulus congestus
(Figure 6.21). If the atmosphere is very moist and unstable, individual clouds
can grow to great heights, turning into cumulonimbus clouds, which we now
describe.
Figure 6.21. Cumulus congestus.
 6.9.5 Conditional Instability and Cumulonimbus
Consider the temperature profile shown in Figure 6.22, which is a little more
complicated than our previous example, and will allow us to introduce an
important new concept. The lower portion of the atmosphere is stable; there
is even a small temperature inversion that caps the lowest kilometer.
Therefore, air parcels lifted from the surface will sink if released and not
much will happen, if the lift is small. However, if air parcels are pushed
sufficiently high, by a lifting mechanism such as frontal lifting or
convergence, and rise above the temperature inversion, they will become
positively buoyant and continue to rise. In situations such as this, air parcels
are said to be conditionally unstable (unstable on the condition that they are
lifted past a threshold point – see Box 6.5). Conditional instability is an
important criterion to consider for thunderstorm development, as it is the
most common profile that leads to deep convection; the atmosphere is almost
never absolutely unstable over a deep layer.
 Figure 6.22. (a) Example of a temperature profile (in red) leading to the
formation of a cumulonimbus once air parcels have been forced to the
level of free convection (LFC). The blue and green arrows represent the
decrease in temperature inside an air parcel as it is lifted from the surface:
first at the dry adiabatic lapse rate (blue) and then at the moist adiabatic
lapse rate above 2 km (green). The air parcel is buoyant up to about 12 km,
where it stops rising. (b) Resulting cumulonimbus cloud.

To illustrate, let us consider an air parcel with temperature Tp = 16 °C at

the surface, and water vapor pressure e = 5 hPa, as in Figure 6.22. Using
Figure 5.4, we can determine that its original saturation water vapor pressure
is es = 18 hPa. We further know that the air parcel will reach saturation when
the saturation water vapor pressure decreases to es = e = 5 hPa, which will
happen at a temperature of –4 °C (using Figure 5.4 again). Therefore, we
would need a 20 °C temperature drop, from 16 °C down to –4 °C, to start
forming a cloud. Since the temperature first decreases at the dry adiabatic
lapse rate (10 °C/km), a 20 °C temperature drop is achieved with a 2 km gain
in elevation. Thus, saturation would occur at an altitude of 2 km (the LCL).
From then on, further ascent results in the temperature decreasing at the moist
adiabatic lapse rate (6° C/km).
Although a cloud will start forming at 2 km, the air parcel (Tp = –4 °C)
would still be colder than the environment at that location (Te = 10 °C), and
therefore heavier than the surrounding air, and thus subject to a downward
buoyant force. To form a deep cloud, we need to force the air parcels up to
that altitude by some forcing mechanism. If such a forcing mechanism is in
place, however, be it frontal lifting, convergence, or otherwise, and if an air
parcel reaches 3 km, it will become warmer than the environment (see the
green arrow in Figure 6.22(a)), become freely buoyant and will continue to
rise by itself. Forced convection will give way to free convection. This
critical altitude is called the level of free convection (LFC) and constitutes an
important diagnostic of the state of the atmosphere when there is a possibility
of thunderstorms. Indeed, if we follow the ascent of our air parcel, we can see
that it will keep rising until it becomes once again colder than the
environment, which happens at 12 km, near the tropopause. Our cumulus
cloud has grown and developed into a full-fledged cumulonimbus cloud,
with the likely presence of thunder and lightning – a thunderstorm.
Note that our calculations are greatly simplified for the sake of
demonstration, and recall in particular that the moist adiabatic lapse rate is
not a constant – our value of 6 °C/km is only an approximation to illustrate
the basic process. Therefore, not all thunderstorms grow to be 12 km tall.
Nevertheless, fully developed cumulonimbus clouds often span the entire
troposphere and abruptly stop at the tropopause, where the environmental
temperature stops decreasing, giving way to the very stable stratosphere. That
is why these clouds often seem to flatten out at the top, giving them the
appearance of an anvil (Figure 6.23). Just as it takes time to stop a speeding
car, due to its high speed and inertia, some air parcels overshoot the
tropopause and penetrate into the stratosphere, which explains why the cloud
top is not always completely flat, and can show some overshooting
cauliflower heads (“overshooting tops”).
 Figure 6.23. Cumulonimbus cloud flattening against a stable layer, giving
it the appearance of an anvil.

To conclude, we have shown that we can form clouds by lifting air

parcels, essentially cooling the air adiabatically until it reaches saturation.
Stability of the environmental temperature profile affects all lifting
mechanisms, but especially the convective mechanism, which depends on the
buoyancy of air parcels to “lift themselves.” The stability properties of the
temperature profile play a key role in determining the types of clouds that
form, which are summarized in a “cloud family album” in Appendix 6.1.
 Summary
Cloud droplets form by heterogeneous nucleation: when saturation is
reached, water vapor molecules condense onto condensation nuclei – small,
airborne, solid particles (aerosols) such as dust, pollen, pollutants, and salt
particles from the ocean.
Saturation can be achieved by evaporating water and increasing the
water vapor content of the air, as in the case of evaporation fog, or by
decreasing the temperature of the air, and therefore the saturation water vapor
pressure, as in the case of radiation fog, and the formation of cloud droplets
by adiabatic expansion and cooling.
The temperature of a rising air parcel decreases because the parcel
expands to adjust to the decreasing environmental pressure. Because this
expansion is adiabatic (i.e., there is no exchange of heat between the parcel
and the environment), the energy needed to accomplish this expansion comes
from the kinetic energy of air molecules in the parcel, which results in a
decrease in temperature of the air parcel. This adiabatic cooling explains the
formation of many clouds in the atmosphere (except for fog, as well as some
cirrus and stratocumulus clouds).
When air is not saturated, a rising air parcel cools at the dry adiabatic
lapse rate, which is about 10 °C/km. When the air parcel is saturated, water
vapor condenses to form cloud droplets as the air parcel ascends.
Condensation releases latent heat, which warms the air. Therefore, a saturated
rising air parcel cools at the moist adiabatic lapse rate, which is about 6
°C/km.
 The altitude at which saturation is reached is called the lifting
condensation level (LCL). A rising air parcel cools at the dry adiabatic lapse
rate below the LCL, and at the moist adiabatic lapse rate above the LCL.
When air is forced over a mountain (orographic lifting), it tends to
produce clouds and precipitation on the windward side, and clear weather on
the lee side. Because the air cools at the moist adiabatic lapse rate on the
windward side above the LCL, and at the dry adiabatic lapse rate on the lee
side, the subsiding air on the lee side is both drier and warmer than on the
windward side. These lee winds are found in many locations on Earth and
given various names by the local populations (e.g., chinook on the eastern
slopes of the Rocky Mountains in the United States, and foehn in the Alps in
Europe).
Upward motion leading to the formation of clouds can be achieved by
orographic lifting, by convergence, by frontal lifting, and by convection.
An air parcel is said to be stable when it returns to its original position
after being displaced. It is said to be unstable when it accelerates away from
the location from where it was displaced, which means positively buoyant
and rising when lifted, and negatively buoyant and sinking when lowered.
When an air parcel becomes positively buoyant only when it is lifted past a
certain threshold, it is said to be conditionally unstable.
Temperature inversions are always stable: they inhibit vertical motion,
convection, and the formation of clouds.
Lifting of unstable air tends to produce cumuliform clouds, whereas
lifting of stable air tends to produce stratiform clouds.
A cumulonimbus cloud typically forms when rising air parcels reach
the level of free convection (LFC): the level at which the air parcels continue
rising under the sole effect of buoyancy, without any other lifting mechanism.
Cumulonimbus clouds often span the entire depth of the troposphere. Since
the stratosphere is stable (by virtue of the stratospheric temperature
inversion), the rising air parcels in the cloud cannot penetrate into the
stratosphere, and the cloud flattens out at the tropopause, giving the top part
of the cumulonimbus cloud the appearance of an anvil.

Appendix 6.1 A Cloud Family Album

We now know enough about cloud formation to understand the logic behind
cloud classification. First, the cloud formation process tells us about the type
of cloud, and it is convenient to distinguish between cumuliform and
stratiform clouds.
Cumulus clouds (Figure 6.24) form in convective situations, and
relatively rapid vertical motion with entrainment and mixing often gives them
a round, plump, cauliflower appearance. Fair-weather cumulus clouds form
when the atmosphere is only mildly unstable. They have a limited vertical
extent and a flat base. They are separated by areas of subsidence, and do not
produce any precipitation. They are typical of warm, sunny afternoons, when
the upper atmosphere is stable.
 Figure 6.24. Cumulus clouds.

When the atmosphere becomes more unstable and convection

intensifies, or when a lifting mechanism forces the air upward (convergence,
for example, or a cold front), the clouds accumulate and grow into cumulus
congestus (Figure 6.25). These often block the horizon and are noticeable
because of their bright white appearance.

Figure 6.25. Cumulus congestus.

 When the atmosphere is very unstable and an individual cloud grows
vertically, sometimes spanning the entire troposphere and flattening out
against the tropopause, we have a cumulonimbus, and most likely thunder
and lightning. (Technically, “nimbus” means “rain-producing.”) In
Figure 6.26, taken by a NASA astronaut on the International Space Station,
we see a cumulonimbus from above, with an extensive flat top and some
overshooting tops where convection is most intense. The thunderstorm is
surrounded by a number of cumulus clouds, including cumulus congestus.
Some of the rising cloud towers might themselves grow into more
cumulonimbus clouds later during the day.

Figure 6.26. Cumulonimbus cloud.

Stratus clouds have more horizontal than vertical development.

(Stratus, in Latin, means “spread out.”) They form when the air rises more -
gradually – typically, when warm air overruns a cold air mass along a warm
front (see Figure 6.12(d)). These clouds can stretch for hundreds of
kilometers and effectively block the sunlight. Therefore, their underside is
often dark gray (Figure 6.27). When viewed from above, from a mountain or
a plane, they appear as a bright white blanket covering the entire landscape.
Rain-producing stratus clouds are called nimbostratus.

Figure 6.27. Stratus clouds.

Cirrus clouds form high in the troposphere, where the temperature is

less than –40 °C, and are made of ice crystals. These lend a wispy appearance
to the cirrus clouds, with no well-defined boundary (Figure 6.28). Through
mechanisms we will discuss in Chapter 7, the ice crystals grow and
sometimes fall, forming virga. When these falling ice crystals are blown by
upper-level winds, they give the cirrus clouds the appearance of a veil,
of falling hair, or of an animal’s tail. Cirrus, in Latin, means “curl of hair,”
and such drifting ice particles are sometimes called “mare’s tails” or
“fallstreaks.”
 Figure 6.28. Cirrus clouds.

Even though the processes leading to each type of cloud are quite
different, the distinction can sometimes sound academic, as several processes
can be at work at the same time, and some clouds seem to fit different
descriptive types. That is why we introduce another parameter in classifying
clouds: the altitude at which they form. We typically distinguish low clouds,
midlevel clouds, and high clouds. Midlevel clouds are given the prefix
“alto-.” Since high clouds are almost always made of ice crystals, they are
given the prefix “cirro-.”
Thus we can expand our cloud family and distinguish cumulus (a low
cloud), altocumulus (Figure 6.29, a cloud forming in a mildly unstable layer
of the midlevel atmosphere), and cirrocumulus (Figure 6.30), made of ice
crystals. The cloud bundles in altocumulus are closer together than fair-
weather cumulus clouds, and bigger in size than in a cirrocumulus – although
it is sometimes difficult to distinguish them. Altocumulus clouds are also
darker than cirrocumulus, and sometimes share the sky with stratocumulus,
while cirrocumulus often coexist with other cirrus clouds, such as cirrus and
cirrostratus.
 Figure 6.29. Altocumulus clouds.

Figure 6.30. Cirrocumulus clouds.

Similarly, we can expand the stratus family and distinguish low clouds
(stratus and nimbostratus), midlevel clouds (altostratus), and high clouds
(cirrostratus). They all have wide horizontal extent, but differ in several
respects. Stratus clouds are low in the atmosphere, are dark gray and
gloomy, and let little sunshine through. Altostratus are not as thick, and the
sun is usually visible as a faint white disk – although the light is not sufficient
to cast strong shadows. Cirrostratus are clearly distinct: they have a whitish
appearance, the fuzziness typical of cirrus clouds, and definitely appear
higher in the sky. The sun is visible and often surrounded by a halo, as in
Figure 6.31: indeed, as sun rays go through the cloud, they are refracted by
ice crystals at a 22° angle specific to the structure of the water crystal (see
Chapter 7).
 Figure 6.31. Cirrostratus, with halo around the sun.

The three types of stratus can sometimes be observed when a warm front
is approaching, or, in other words, when a midlatitude cyclone is
approaching. If you are standing in the path of the storm, as shown in
Figure 6.32, with the warm front advancing toward you, you will see a series
of clouds such as shown in Figure 6.33, with low stratus on the horizon,
turning into higher altostratus, and finally cirrostratus overhead, possibly
mixed with cirrus and cirrocumulus. As the front advances, you will find
yourself under ever lower and thicker stratiform clouds, and eventually
precipitation, as the front moves through. A thickening cirrostratus cloud is a
good sign of an approaching storm, and a sign that sailors and farmers knew
to recognize when weather forecasting was still empirical.
 Figure 6.32. Typical cloud sequence associated with a warm front.

Figure 6.33. Approaching strom.

Stratus clouds sometimes break up into cumulus slabs, and are then
called stratocumulus (Figure 6.34). They can also be the result of mild
convection being impaired by strong subsidence aloft, as could happen under
an anticyclone. This is the case for large sections of ocean in the subtropics.
 Figure 6.34. Stratocumulus clouds.

In the end, know that clouds can fall between categories, and it is more
important to understand the mechanism behind the formation of the cloud
than it is to get its name right.
Finally, remember that any lifting mechanism can lead to cloud
formation, given the right circumstances. When wind is deflected above a
mountain, for example, a cloud can form if the air is sufficiently close to
saturation. When the atmosphere is stable enough that the flow remains
horizontal (with no vertical cloud growth), a lenticular cloud can form
(Figure 6.35).
Figure 6.35. Lenticular cloud.
CHAPTER 7
Precipitation
CONTENTS
7.1 Warm vs. Cold Clouds
7.2 Collision and Coalescence
7.3 Ice-Crystal Growth
7.4 Precipitation Types
Summary

Why does water fall from clouds? How do cloud droplets turn into raindrops?
Why are there different types of precipitation? These are some of the
questions we will answer in this chapter, as we explore how cloud droplets,
and in particular ice crystals, grow through various processes and turn into
precipitating particles. From rain to hail and snow, we will describe specific
conditions and processes in which various types of precipitation occur.
Why do some clouds produce precipitation, while others don’t? There is a
very large size difference between cloud water droplets (on the order of 10
µm in diameter) and raindrops (on the order of 1 mm in diameter). To use an
analogy, if a cloud droplet were the size of a very small green pea, then a
raindrop would be about the size of a basketball. Thus, a cloud droplet needs
to grow a hundredfold in diameter, and therefore a millionfold in volume, to
become a raindrop. While cloud droplets can grow by condensation of
additional water vapor at first, this is a relatively slow process, and
condensation alone is not sufficient to explain the rapid growth of raindrops
in a precipitating cloud. Indeed, additional processes are at work, and the
details of the processes depend on whether or not ice particles are present in
the cloud.
 7.1 Warm vs. Cold Clouds
To describe the full spectrum of precipitation processes, it is useful to define
two types of clouds: warm clouds and cold clouds. In warm clouds, the
temperature everywhere is above freezing (T > 0 °C), whereas in cold clouds,
there exist parts of the cloud that are below freezing.
Warm clouds only contain liquid water droplets and water vapor,
whereas cold clouds also contain ice particles. In fact, cold clouds can contain
a little of everything depending on altitude and temperature (Figure 7.1).
Where the temperature is above freezing, only liquid water droplets and water
vapor are to be found (except maybe for falling frozen particles that are
melting). Where the temperature is below –40 °C (typically the top of the
cloud, or a very high cloud), only ice crystals and water vapor are to be
found. A high cloud containing exclusively ice particles, such as a cirrus, is
said to be glaciated.

Figure 7.1. Distribution of liquid water droplets (blue), supercooled

droplets (white), and ice crystals (stars) in a cold cloud, as a function of
 temperature.

What should we expect in between? You might think that only frozen
water should be found when the temperature is below 0 °C, but that is not
necessarily the case. Liquid water droplets can exist at temperatures below 0
°C. They are said to be supercooled, and are well known to aircraft pilots,
because they freeze on contact with the surface of the aircraft, or within the
engine, and can cause the wing flaps to freeze or the engine to stall. If you
have ever been on a plane that was “de-iced” prior to takeoff, it was done so
to prevent this from happening. In fact, supercooled liquid water droplets are
what we mostly find just above the freezing level, where the temperature is
less than 0 °C but still greater than –10 °C.
Then there exists a region of the cloud where supercooled liquid water
droplets and ice particles coexist, where the temperature is below –10 °C, but
still above –40 °C. As shown in Figure 7.1, the distribution shifts from a
higher concentration of supercooled liquid water droplets to a higher
concentration of ice particles as we rise in the cloud and the temperature
decreases. A cloud containing both ice particles and supercooled liquid water
droplets is called a mixed-phase cloud.
Fair-weather cumulus clouds are relatively low in the atmosphere, and in
summer are a good example of warm clouds. In contrast, cirrus clouds are
high in the troposphere, and are a good example of cold clouds. They only
contain ice particles, which gives them their wispy appearance.
Cumulonimbus clouds are tall and span the entire range of temperatures
shown in Figure 7.1. Their lower portion contains mostly liquid water
droplets, which gives them their cauliflower-like appearance, with sharply
defined boundaries, whereas their top portion only contains ice particles,
which explains why the top of the anvil is often more wispy and not as
clearly defined.
 7.2 Collision and Coalescence
In warm clouds, droplets grow by colliding with each other and coalescing to
form larger drops. They do so not only due to the presence of turbulent
motions, which make droplets collide randomly, but also because the droplets
fall at different speeds. Bigger drops fall faster than smaller drops, because
friction has less of a dragging effect on bigger drops, relative to their size and
weight. The drag of air molecules against the side of the droplets, as they fall
through the air, is larger for smaller droplets (relative to their size), which
creates greater resistance and slows the fall speed.
If all cloud droplets had the same size and fell at the same speed, they
would never meet, i.e., the collision efficiency would be low. If a large
droplet were falling through much smaller droplets, the large droplet would
fall faster than the very small ones, but the small droplets would be swept
aside by the displacement of air caused by the large falling droplet, i.e., the
collision efficiency would again be low. If the differences in size are neither
too large nor too small, however, the droplets are more likely to collide, i.e.,
the collision efficiency is higher. Because there is typically a wide range of
drop sizes in a cloud, they have many opportunities to collide and coalesce,
i.e., to merge into bigger droplets (Figure 7.2).
 Figure 7.2. Growth of cloud droplets by collision-coalescence as a bigger
droplet collects smaller droplets on its way down.

Ultimately, the drops become so big that their weight overcomes any
updraft and they fall out of the cloud. They can be as small as 1 mm in
diameter (drizzle) or as big as 5 mm in diameter (in a thunderstorm, for
example).
 7.3 Ice-Crystal Growth
In cold clouds, ice particles can form in different ways. They can be the result
of the freezing of supercooled liquid water droplets. The freezing can occur
by homogeneous nucleation: in much the same way as water vapor molecules
would need to come together and survive as a stable clump of liquid water to
form a cloud droplet, liquid water molecules inside a supercooled droplet
need to come together and survive as a clump of ice, an embryo from which
the rest of the droplet can freeze. In practice, because the kinetic energy of
the water molecules must be low enough for this to happen, homogeneous
nucleation occurs preferentially at colder temperatures. For temperatures of
around –35 to –40°C and colder, it happens reliably; there is very little liquid
water at these temperatures.
Most often for temperatures between 0 and –35 °C, freezing is seeded by
the presence of a freezing nucleus inside the supercooled droplet, in
somewhat the same way that cloud droplets are seeded by condensation
nuclei (heterogeneous nucleation). The freezing nucleus, also called more
generally an ice nucleus, provides a surface to which the liquid water
molecules can attach, making the crystal grow and causing the entire droplet
to freeze. The reason liquid water droplets persist at temperatures below
freezing is because, unlike condensation nuclei, which are abundant, ice
nuclei are relatively rare; they require special properties that promote the
formation of the crystalline lattice specific to frozen water.
Cloud droplets can also freeze rapidly by contact with an ice nucleus (in
the same way as you can cause a tub of supercooled liquid water to freeze
rapidly by throwing in a piece of ice).
Once we have an ice crystal, there are several ways that it can grow. One
mechanism for growth is deposition of water vapor molecules onto an ice
crystal. This mechanism was discovered at the beginning of the twentieth
century by Alfred Wegener (who also formulated the continental drift
hypothesis), Tor Bergeron, and Walter Findeisen. It is usually known as the
Bergeron process. The three scientists made use of the fact that the
saturation water vapor pressure over ice (esi) is lower than that over water
(esw). In other words, less water vapor can exist in the presence of ice than
can exist in the presence of liquid water, at the same temperature (Figure 7.3).
Some values of esi and esw are shown in Figure 7.4.

Figure 7.3. Schematic representation of the saturation water vapor

pressure over ice and water at the same temperature. When the air is
saturated and in equilibrium with the underlying surface, less water can
exist as vapor over ice (esi) than over water (esw).
 Figure 7.4. Saturation water vapor pressure over water (esw) and ice (esi) in
relation to temperature.

To grasp fully the significance of this result, let us imagine the cold part
of a cloud where supercooled liquid water droplets coexist with ice crystals,
and let us further zoom in onto a liquid water droplet in the presence of an ice
crystal (Figure 7.5). The temperature is –15 °C and the saturation water vapor
pressure is 1.9 hPa over water and 1.6 hPa over ice. The water vapor
surrounding the droplet and the ice crystal cannot be in equilibrium with both
at the same time. Let’s imagine that it is in equilibrium with the water
droplet – that means the water vapor pressure is 1.9 hPa. Since the vapor
pressure is greater than 1.6 hPa, the air is supersaturated from the point of
view of the ice crystal, and, as a result, some water vapor molecules will be
deposited onto the crystal, contributing to its growth. But in the process,
moisture has been removed from the air, and the water vapor pressure has
decreased below 1.9 hPa. Therefore, the air is no longer saturated from the
point of view of the water droplet and some water molecules will evaporate
from the droplet to saturate the air. Thus, as soon as some water vapor is
deposited onto the ice crystal, some water is evaporated from the droplet to
compensate the loss. In other words, the ice crystal grows at the expense of
the water droplet, which is slowly depleted.

Figure 7.5. Growth of ice crystals at the expense of supercooled liquid

water droplets. (Blue dots represent water vapor molecules.)

Once initiated, this process continues and allows small ice crystals to
grow into large crystals. Additional mechanisms then contribute to the
multiplication and growth of ice crystals. First of all, being relatively fragile,
large ice crystals can break into many small parts through fragmentation,
providing many new ice nuclei that can seed new ice crystals (see later in this
section). Second, supercooled liquid water droplets can freeze onto an
existing ice crystal during a collision – a process called riming, or accretion,
which tends to form conical spongy pellets called graupel. It is thought that
ice particles forming by riming might also fragment into numerous ice
splinters, providing many new ice nuclei. And third, when two ice crystals
collide, they can simply stick to each other and form a larger crystal by
aggregation, which is how large snowflakes are created.
Depending on the temperature, ice crystals of different shapes can form,
as summarized in Box 7.1 after results from controlled laboratory
experiments. Although very different in appearance, they all share the same
six-sided configuration. This is due to the very unique angular structure of the
water molecule, which favors a certain orientation and organization of the
molecules in the crystals. In actual clouds, ice particles appear to be more
irregular in shape. Their structure depends on the environmental conditions
(temperature, moisture content, strength of updrafts, turbulence, etc.), and can
also be determined by the particular ice nucleus that seeded the ice crystal.
The hexagonal shape of ice crystals is key to the formation of certain optical
phenomena (see Appendix 7.1).

Box 7.1. Ice Crystals

 As we mentioned previously, ice nuclei are relatively rare, as they need
to have a very specific structure. The best ice nuclei are in fact ice crystals
themselves, or bits and pieces created when a crystal fragments. Engineers,
however, have tried to manufacture ice nuclei, hoping to seed the atmosphere
and to provoke precipitation, or to affect the size distribution of the
precipitating particles (one of the first attempts at weather modification). In
1946, Vincent Shaefer, a General Electric researcher, discovered that dry ice
(frozen carbon dioxide) could be used to modify clouds. The dry ice creates
ice crystals which then grow at the expense of the liquid water. A short time
later, Schaefer’s colleague, Bernard Vonnegut, discovered that silver iodide
had a structure similar to that of ice and should also modify the distribution
of ice crystals in clouds. Since then, many experiments have been carried out,
but available statistical analyses are inconclusive and the contribution of
silver iodide to the formation of clouds and precipitation remains unclear.
 7.4 Precipitation Types
Why do we experience different types of precipitation? There are a number of
factors that can affect the type and size distribution of precipitating particles.
Temperature, of course, is one such factor, but the dynamics of the weather
system in which precipitation forms play a more important role in the end. In
warm clouds of the convective type (cumulus, cumulus congestus,
cumulonimbus), droplets grow by collision and coalescence and fall as “fat”
raindrops in the form of showers, possibly quite intense, and rarely long
lasting, unless there is a forcing mechanism that maintains the convective
cloud in place. When convection is very intense, especially if the
thunderstorm rotates, separating updrafts from downdrafts (see Section 11.4),
droplets can be pushed up high into very cold regions of the cloud, where
they freeze. As they leave the updraft and fall through the cloud, they
accumulate ice by riming and grow in size, and then cycle back up into the
updraft to freeze before adding another layer. These little balls of ice
eventually become too heavy for the updraft to support them and they fall to
the surface, where we experience them as hail (Figure 7.6).
 Figure 7.6. Hailstones can grow to a variety of sizes as they go through
different regions of the cloud and accumulate ice.

When the updraft is relatively gentle and in shallow clouds, the size
range of droplets is small. As a result, droplet collisions are not as frequent
and raindrops remain relatively small. When raindrops are very small, we
refer to them as drizzle drops.
In the midlatitudes, and especially in winter, the upper atmosphere is
usually much below 0 °C and precipitation often starts as falling ice crystals.
If they remain frozen all the way to the surface, we experience snow. If they
melt, however, as they fall through a deep layer of warmer air, we experience
rain. Sometimes, they melt as they fall through a first layer of warmer air, but
refreeze as they continue down through a deep colder air mass near the
surface. In this case, we observe small pellets of ice, which we call sleet
(Figure 7.7). When the ice crystals melt as they fall through warm air, and
fall through a shallow cold layer only to refreeze when they touch the ground,
we speak of freezing rain – the characteristic feature of ice storms. When on
the road, we also call it black ice, because it is transparent and looks similar
to the black asphalt underneath.
 Figure 7.7. Snow falling off a stratiform cloud along a warm front. Ice
crystals might persist as snow deep into the very cold air mass, but they
might also melt as they traverse the warmer air mass, in which case they
can form rain, or refreeze to form either sleet or freezing rain.

The last type of rain you may have heard about is acid rain. This is not
a different type of precipitation, but regular raindrops containing sulfur-based
and nitrogen-based dissolved gases such as sulfur dioxide (SO2), nitrogen
oxide(NO), and nitrogen dioxide (NO2). These gases, blown by the wind
from the exhaust of power plants located upstream, dissolve in cloud droplets
and make them acidic. When these cloud droplets grow into raindrops and
precipitate, they contaminate the environment, plants, and water sources (see
more on air pollution in Chapter 14).
As for snow, there are three basic ingredients conducive to snow
formation: moisture, cold temperatures, and of course upward motion. If
temperatures are quite low, the snowflakes will be relatively small and dry
and we then refer to the snow as powder. If the conditions are also very
windy, then the snow contributes to a blizzard. On the contrary, if the
temperatures are above 0 °C, the snowflakes begin to melt, which makes
them more “sticky” and facilitates aggregation. Snowflakes can fall about
300 m (1000 feet) in above-freezing conditions before they completely melt,
with surface temperatures as high as about 5 °C.
 Summary
As far as precipitation is concerned, it is convenient to divide clouds into two
types, warm and cold.

► Warm clouds, in which the temperature is everywhere above freezing

(T > 0 °C), contain exclusively liquid water droplets and water vapor.

► Cold clouds contain parts that are below freezing. Cold clouds can
contain liquid water droplets, supercooled droplets (below 0 °C, but
not frozen), ice crystals, and water vapor.
A high cloud containing exclusively ice particles, such as a cirrus, is said to
be glaciated.
To form raindrops, cloud droplets must grow from about one
micrometer to at least one millimeter in size. In a warm cloud, they grow by
collision and coalescence of cloud droplets falling at different speeds. In a
cold cloud, several processes are at work:

► Supercooled liquid water droplets in high clouds, with temperatures

below –40 °C, can freeze by homogeneous nucleation (i.e., from water
molecules alone).

► Freezing can be initiated by the presence of a freezing (or ice)

nucleus.

► Ice particles can then grow by the Bergeron process, in which ice
crystals grow at the expense of liquid water droplets by deposition of
vapor onto the ice crystal due to a difference between the respective
saturation water vapor pressures over ice and water.
 ► Large ice crystals can break into many small parts through
fragmentation, providing many new ice nuclei that can seed new ice
crystals.

► Supercooled liquid water droplets can freeze onto an existing ice

crystal during a collision (riming, or accretion), forming conical ice
pellets called graupel.
► Two ice crystals can collide and stick to each other to form a larger
crystal by aggregation.

Precipitation can take the form of:

► Rain showers, when droplets grow by collision and coalescence,

typically in a convective cloud, and fall in a relatively short amount of
time. They may also originate as snowflakes and melt before reaching
the ground, and continue to grow by collision and coalescence on their
fall through the warm part of the cloud.

► Hail, when droplets are pushed up high into very cold regions of a
convective cloud, freeze, and sweep repeatedly in and out of the
updraft of the cloud, accumulating successive layers of ice by riming.

► Light rain, when the updraft is expansive and relatively gentle, as

happens along a warm front.

► Drizzle, when raindrops are very small.

► Snow, when falling ice aggregates remain frozen down to the surface.
Snow can have a wide range of textures, from small and powdery to
large, wet snowflakes.
 ► Sleet, or ice pellets, when rain freezes in a deep layer of cold air near
the ground.

► Freezing rain, when rain falls through a shallow layer of cold air at
the ground, freezing in contact with features near the surface (the
ground, trees, etc.).

Appendix 7.1 Some Optical Phenomena

The interaction of sunlight with cloud and precipitating particles can give rise
to various optical phenomena, two of which we describe here. We
encountered the first one in Chapter 6: the 22° halo observed around the sun
in the presence of a cirrostratus cloud (see Figure 6.31 in Appendix 6.1).
Such a halo is caused by the double refraction of sunlight as it passes through
hexagonal ice crystals (Figure 7.8). An observer facing the sun will see, in
addition to direct sunlight, those rays that are refracted by ice crystals in such
a way as they reach the observer’s eyes after a 22° refraction (see the red
lines in Figure 7.9), but will not see the sun rays refracted by the other ice
crystals (gray line in Figure 7.9). Therefore, when refraction takes place all
around the sun, the observer sees a ring of light at a 22° angle from the sun,
i.e., a circular halo (Figure 7.10). Halos can also be observed around the
moon, as shown in Figure 7.11. When the sun’s rays are refracted through the
top or base of the ice crystals (rather than the sides), they can cause a 46°
halo, although it is more rare and difficult to observe.
Figure 7.8. Double refraction of light passing through an ice crystal.

Figure 7.9. Sun rays refracted by ice crystals, as seen by an observer.

Figure 7.10. Circular halo resulting from the refraction of sunlight passing
through ice crystals.
 Figure 7.11. Halo around the moon.

When light passes through liquid raindrops, a rainbow can form. In this
case the sun’s rays are first refracted as they enter a droplet (a in Figure
7.12(a)), then are reflected by the back of the droplet (b), before finally being
refracted again on the way out (c). The rays exit the droplet in the reverse
direction that they entered, at a 40–42° angle from the direction of the
original sun ray.

Figure 7.12. Refraction and reflection of light passing through a water

 droplet. (a) Angle combination. (b) Wavelength-dependent angles.

Moreover, sun rays of shorter wavelength are refracted at a greater angle

than those of longer wavelength, which results in a separation of colors as
they exit the raindrop (Figure 7.12(b)). Therefore, an observer standing with
their back to the sun watching raindrops falling will see different colors
emerging out of different raindrops, as shown in Figure 7.13.

Figure 7.13. Different colors emerging out of different raindrops, as seen

by an observer.

Of course, each raindrop also refracts and reflects other colors, but these
leave the raindrop at a different angle and the observer cannot see them. The
band of raindrops that refract and reflect sun rays toward the observer is
relatively narrow and semi-circular, and takes the form of a rainbow
(Figure 7.14 and 7.15).
Figure 7.14. Rainbow resulting from wavelength-dependent refraction of
sunlight passing through raindrops.

Figure 7.15. Rainbow.

CHAPTER 8
Wind
CONTENTS
8.1 Force and Acceleration
8.2 Pressure Gradient Force
8.3 Sea Breeze and Land Breeze
8.4 Coriolis Force
8.5 Geostrophic Wind
8.6 Gradient Wind
8.7 Surface Winds
8.8 Friction
8.9 Topography
Summary

Wind is a manifestation of air in motion, from light breezes by the sea to

powerful jet streams in the upper troposphere. But what causes wind? Motion
is always initiated by forces, and wind is no exception. We will now explore
how pressure differences result in forces that set air in motion, how additional
factors come into play at different scales of motion, and how a given pressure
distribution generates specific wind patterns.
In trying to understand the formation of wind, we must return to the
fundamental laws of physics, and in particular the law of conservation of
momentum, in the form of Newton’s laws of motion. These will allow us to
have both a diagnostic approach (i.e., describing the current state of the
atmosphere) and a prognostic approach (i.e., predicting the future state of the
atmosphere – see Chapter 13).
 8.1 Force and Acceleration
As for any moving object, air follows the laws of motion described by Isaac
Newton (1643–1727) in the seventeenth century. Newton’s first law states
that

An object at rest remains at rest, and an object in motion moves in a

straight line at constant speed unless acted upon by a force.

Newton’s second law states that

The acceleration experienced by an object is equal to the force exerted

on that object divided by its mass.

This is captured by the famous expression F = ma, or a = F/m, where an

object of mass m experiences an acceleration a when a force F is applied to it.
[Bold letters indicate that the quantities are vectors, with both direction and
magnitude.] Here the air has a certain velocity (i.e., speed and direction), and
the acceleration is the rate of change of that velocity, in the same way as the
speed of a car changes when we push on the accelerator. Since the
acceleration is equivalent to the force applied to the air divided by its mass,
understanding wind amounts to understanding the forces applied to the air in
various circumstances. When there is more than one force, Newton’s laws
apply to the net force, which is the sum of all forces.
In describing air motion, it is useful to differentiate vertical from
horizontal motion. Vertical motion is very much constrained by gravity and is
described by buoyancy and stability, as explained in Chapter 6. Horizontal
motion is free of gravity and is determined by a set of horizontal forces: the
pressure gradient force, the Coriolis force, the centrifugal force, and friction,
which we will now describe.
 8.2 Pressure Gradient Force
Consider the following thought experiment, in which a glass of water has
been flipped over on a table top (Figure 8.1(a)). As we lift the glass, water
flows out and spreads over the table top. Why does the water flow outward?
We often hear that water “wants to equalize” and be level, but what is the
force that sets the water in motion to make that happen? For the glass of
water, we know that, right at the counter top, where the water column is
highest, the pressure is also higher than outside the glass because the water
column weighs more than a column of air outside the glass. Therefore, there
exists a pressure difference between the interior and the exterior of the water
column (Figure 8.1(b)). In dealing with wind, we prefer to use the term
pressure gradient, since it is a vector that has both magnitude and direction.
 Figure 8.1. Thought experiment, in which water spreads out over a table
top as it falls from an inverted glass.

Because there is a pressure gradient, more active water molecules at

higher pressure push outward against less active molecules under less
pressure. In other words, the pressure difference creates a force directed
outward from the water column that pushes the water out. We call this force
the pressure gradient force (PGF), and note that it is always directed from
high toward low pressure.
If we change our point of view to a bird’s eye view, we obtain a good
analogy for the mechanism that initiates motion in the atmosphere, as
illustrated in Figure 8.2(a). When there is a region of high pressure
surrounded by lower pressure, the pressure gradient force pushes the air
outward, causing wind to blow. In the same way that the water spreads out
and away from the glass, we can picture the air spreading out and away from
the high pressure center, and emptying the air column. In fact, the analogy
extends to the vertical: in the same way that the water level falls in the glass,
air tends to subside in a high pressure region, to replace the surface air exiting
through the sides.

Figure 8.2. Distribution of pressure gradient force (PGF) around a high

pressure region (a) and a low pressure center (b).
 Although it is more difficult to create such a straight analogy for low
pressure centers with a glass of water (we have to create a partial vacuum to
create a column of air that weighs less than the surroundings), we can use our
imagination and flip the above picture to obtain Figure 8.2(b). The pressure
gradient force pushes inward, from high to low pressure, and wind will blow
toward the low pressure center. Of course, other forces will play their part to
create a slightly more complicated picture, and winds will end up blowing,
not in and out of lows and highs, but around lows and highs, as we will soon
explain. But in the meantime, the pressure gradient force is the first force we
should consider when trying to understand the origin of wind, and indeed the
main one when trying to understand small wind circulations covering short
distances and time periods. To illustrate, we will take the example of a sea
breeze on the coast.
 8.3 Sea Breeze and Land Breeze
Imagine a sunny day at the coast where we can observe the coastline. As we
explained in Chapter 4, land warms and cools much more and much faster
than water (due to the larger heat capacity for water, the redistribution of heat
in water by mixing, the vertical spreading of heating in water due to the
penetration of light to greater depths, and the evaporation of water requiring
latent heat). Uneven heating of the surface carries over to the two air columns
over the land and ocean, so that, during the day, the lower part of the land air
column is warmer than the lower part of the ocean air column. Now recall
that warm air expands, and, in our particular case, the land air column
expands upward to greater heights than the ocean air column.
Let us draw some pressure levels and observe how they change due to
this uneven expansion (Figure 8.3(a)). Pressure surfaces are raised higher
over the land than over the ocean, due to thermal expansion, which makes
them tilt as shown in Figure 8.3(b). Now consider some horizontal level and
assess the pressure gradient. By interpolating between pressure levels, we can
estimate that the atmospheric pressure at point A in Figure 8.3(b) is about 910
hPa, whereas it is about 895 hPa at point B. Therefore, due to the differential
thermal expansion and the tilting of the pressure levels, there exists a pressure
difference, or pressure gradient, that will cause a wind to blow aloft from the
land to the sea.
 Figure 8.3. Formation of a sea breeze on a sunny day.

As air is blown from the land to the sea, the air column over the land is
depleted of air, while air accumulates in the air column over the sea. Since
atmospheric pressure at the surface is tantamount to the weight of the air
column above the surface, this redistribution of mass increases the surface
pressure over the sea and decreases it over the land, as illustrated in
Figure 8.3(c) by comparing the pressure at points C and D. This surface
pressure gradient causes the wind to accelerate, blowing from the sea to the
land. Because this wind is usually quite mild, we call it a breeze, and,
because it blows from the sea, we call it a sea breeze.
We note two important points in Figure 8.3. First, the exchange of mass
between the two air columns is usually accompanied by rising motion over
the land and subsidence over the sea, although subsidence is usually less
concentrated and more spread out over the water. This closes the loop and
constitutes what we call a sea breeze circulation. It is driven by the
temperature contrast between land and water, and it occurs in response to the
corresponding energy imbalance: it redistributes energy from the warm
region to the cold region to bring the system back to a more balanced state.
Note, however, that, even though we indicated a low pressure at ground
level over the land and a high pressure aloft, the surface pressure at D (about
970 hPa) is still much higher than the pressure aloft at A (about 910 hPa).
That is why we used a larger H and L at the surface and smaller ones aloft, to
avoid the temptation to diagnose some air flow from A to D or from C to B.
Values of atmospheric pressure must be compared on the same horizontal
level.
Second, the rising air over the land can lead to the formation of cumulus
clouds, while subsidence over the water inhibits cloud formation. Therefore,
if clouds form, they will line up along the coast, over the land, while the
nearby water will be cloud free – a common situation in summer that can
easily be observed from a plane following a coastline.
What happens when the sun sets? There is no longer an input of solar
energy into the ground or water, but they both continue to lose energy in the
form of infrared radiation, as we explained in Chapter 4. For similar reasons
to those given earlier, the land loses energy much faster than the water, and
the whole circulation is reversed to give offshore flow, called a land breeze
(Figure 8.4). If clouds form, they will also line up along the coast, but over
the water, and they will be more difficult to observe at night.

Figure 8.4. Formation of a land breeze at night.

This type of circulation can form anywhere there exists weak ambient
wind and a temperature contrast – by a lake, for example, or between two
types of land surface where one might be a forest, or an expanse of snow. In
the end, an important message is that a temperature contrast induces a
horizontal pressure contrast, and the resulting pressure gradient causes wind
to blow. Moreover, the stronger the temperature contrast, the stronger the
pressure gradient, and therefore the stronger the wind.
 8.4 Coriolis Force
If the pressure gradient force were the only force acting on the air, winds
would blow from high to low pressure. But simple observation of weather
maps reveals that this is rarely the case. While it is somewhat true over short
distances, such as in the case of a sea breeze, it does not hold over larger
distances. Recall the counterclockwise circulation of winds in Figure 2.1. In
fact, by comparing Figures 2.1 and 2.4, we can see that the wind does not
blow across the isobars (perpendicularly) from high to low pressure, as
suggested above, but along the isobars (parallel), 90° away from what we
might expect based on the pressure gradient force alone. It turns out that,
after a few hours, Earth’s rotation plays a significant role in the wind
direction. Considering two points in the direction of the pressure gradient, the
wind is accelerated from one point to the other due to the pressure gradient
force, but both points are also rotating around Earth’s axis. This creates a
complicated motion that can only be fully described by taking into account
Earth’s rotation (i.e., the rotation of the frame of reference from which we
observe the motion). The traditional approach to this situation consists in
deriving the equations of motion on a rotating sphere, from which we can
infer the existence of an acceleration that deflects the air at a right angle to
the direction of motion, to the right in the northern hemisphere and to the left
in the southern hemisphere. We can think of this effect as either an
acceleration (i.e., the Coriolis acceleration) or a force (i.e., the Coriolis force,
after the French mathematician Gaspard-Gustave Coriolis (1792–1843) who
first investigated the matter), depending on our point of view. From that
second point of view, the air is accelerated as if it were pulled by a force
pointing to the right of the direction of motion (in the northern hemisphere).
Since that force is not a push or a pull in the traditional sense – it exists only
because we view the motion from the rotating Earth – we think of it as an
apparent force. Mathematically, we can also determine that the Coriolis force
varies with latitude (it is zero at the equator and strongest at the poles) and is
proportional to wind speed (although it does not affect the wind speed, only
the wind direction).
While such a calculation is beyond the scope of this book, one can get a
glimpse of the effect of a rotating frame of reference by considering the
following popular analogy. Imagine two persons throwing a ball on a merry-
go-round. If the merry-go-round is not spinning, then the ball follows a
straight line from the thrower to the catcher (Figure 8.5(a)). If the merry-go-
round is spinning (counterclockwise in Figure 8.5), the ball still moves in a
straight line after it is thrown, as Newton’s first law says it must, which is
exactly what we see if we observe it from outside the merry-go-round
(Figures 8.5(b)–(e)). However, if we observe the ball from the merry-go-
round, something strange happens.
Figure 8.5. Apparent motion of a ball thrown from a merry-go-round
spinning counterclockwise. (a) If the merry-go-round is still, the ball (in
red) goes in a straight line. (b) If the merry-go-round is spinning, the
thrower and catcher rotate with it, but the ball follows the original straight
line (dashed) as soon as it leaves the hand of the thrower. (c)–(e) As the
merry-go-round continues to spin, the ball is further to the right of the
thrower, rather than in front of him. (f) If we position ourselves on the
rotating frame of reference and rotate with the thrower and catcher, it
appears as if the merry-go-round is not moving and the ball follows a
curved trajectory to the right.
 If we position ourselves on the merry-go-round, we rotate with it, and it
looks to us as if the platform, the thrower, and the catcher are not moving.
We are in a rotating frame of reference, and from that point of view we are
not moving. If we track the position of the ball from the rotating frame of
reference (Figure 8.5(f)), it looks like the ball follows a curved path, as if it
were pulled by a force to the right of the direction of motion (for this
particular direction of rotation of the merry-go-round, i.e., counterclockwise).
Note the use of the expressions “it looks like” and “as if.” It is only an
apparent deflection of the motion if you observe that motion from the
rotating frame of reference. The fundamental reason is that the circular
rotation of the merry-go-round provides an acceleration to the observer on the
platform: if we direct our arm in a certain direction from the merry-go-round,
the location we point to changes. Thus, the ball, which is not attached to the
rotating merry-go-round any more, appears to be accelerated to the right, but
only because we have introduced this apparent acceleration by chosing a
rotating point of view.
The deflection of a parcel of air moving on Earth is somewhat similar to
the deflection of the ball moving over the merry-go-round. Of course, the
analogy does not fully explain the Coriolis acceleration of air and water on
Earth. For one thing, Earth is a sphere, not a flat disk. Second, air and water
are not free to move out of the rotating frame of reference – if they were, they
would fly out to space. They are held down to Earth by gravity, which exerts
a very strong constraint. The merry-go-round analogy, however, at least
conveys the notion of apparent motion and apparent force when observing
from a rotating frame of reference.
We always observe the atmosphere from this rotating frame of reference
attached to Earth, so we must live with this apparent force. It turns out to play
an important role; that is, the fact that Earth is rotating rapidly is
fundamentally important to the development of weather. In practice,
meteorologists establish basic rules of atmospheric motion that incorporate
the Coriolis force, and then apply them systematically.
What are these rules of atmospheric motion? They are very few, in
reality, and should be memorized for future reference. We will start with a
first set of rules summarized in Box 8.1, and will now examine how the
Coriolis force interacts with the pressure gradient force to create specific
wind patterns.

Box 8.1. The Coriolis Force

Rule set #1
(1a) Objects in motion in the northern hemisphere are deflected at a
right angle and to the right of the direction of motion.
(1b) Objects in motion in the southern hemisphere are deflected at a
right angle and to the left of the direction of motion.
(2) The Coriolis force is zero at the equator, and increases with
latitude, to a maximum at the poles.
(3) The Coriolis force is proportional to the wind speed: the faster the
wind, the stronger the deflection.
(4) Because the Coriolis force is perpendicular to the wind direction,
it only affects the wind direction and does not affect the wind speed.
In other words, it makes the wind veer, but it does not accelerate or
decelerate it.
 8.5 Geostrophic Wind
Let’s consider an air parcel embedded in a pressure gradient and observe
what part the Coriolis acceleration plays in the development of motion
(Figure 8.6). To keep it simple, we choose the midtroposphere and a few
straight isobars, parallel to each other and oriented west–east, so that the
pressure gradient is oriented from south to north. Recall from Chapter 2 that
we can switch interchangeably between pressure contours (at constant height)
and height contours (at constant pressure) to observe and understand the
motion. In Figure 8.6(a), the height contours decrease northward from 5760
to 5580 m, indicating that the pressure gradient force (PGF) is pushing air
from south to north. For the sake of demonstration, let us imagine that the air
parcel is first at rest – a somewhat unrealistic situation, but one that will serve
our purpose. Since a force is applied to the parcel (the pressure gradient
force), it is accelerated (according to Newton’s second law of motion). As
soon as it starts to move, it acquires some speed, which means that, from our
frame of reference, it is also accelerated to the right of the direction of motion
by the Coriolis force (recall that the Coriolis force is proportional to wind
speed). Therefore, rather than moving due north, the parcel starts turning to
the right (Figure 8.6(b)). Since the direction of motion of the parcel has
changed, the Coriolis force changes direction accordingly in Figure 8.6(b) so
that it always remains perpendicular to the direction of motion.
Figure 8.6. Adjustment of an air parcel to geostrophic balance between the
pressure gradient force (PGF) and the Coriolis force (COR).
 As the air parcel continues to move and to accelerate, the Coriolis force
increases and keeps deflecting the parcel to the right (Figure 8.6(c)), until the
air is blowing due east, at which point the Coriolis force is directed opposite
the pressure gradient force and the air parcel has acquired enough speed that
the Coriolis force is equal in magnitude to the pressure gradient force
(Figure 8.6(d)). We have reached a state of equilibrium, called geostrophic
balance, in which there is a balance between two forces equal in magnitude
and opposite in direction, such that the air parcel continues to move at a
constant speed. Note that the net force applied to the air parcel is now zero –
the pressure gradient force and the Coriolis force exactly cancel. Therefore,
there is no acceleration, and the parcel keeps moving in a straight line, at the
same speed, “unless acted upon by a net force”(according to Newton’s first
law).
What will happen, however, if the balance of forces changes? If the
pressure gradient force increases or decreases? Let us consider the situation
represented in Figure 8.7(a), in which the air parcel enters a region where the
pressure gradient increases (i.e., the isobars, or height contours, are closer to
each other). Since the pressure gradient force increases, the air parcel should
be deflected northward (dark blue arrow in Figure 8.7(a)). However, since
there is now a net positive force pushing the air parcel northward, the parcel
is also accelerated in that direction and its speed increases. The Coriolis
force, being proportional to wind speed, also increases and pulls the air parcel
southward (Figure 8.7(c)). Eventually a new equilibrium is reached where the
pressure gradient force and the Coriolis force are both stronger and balanced,
and the wind speed has increased (Figure 8.7(d)).
Figure 8.7. Adjustment of geostrophic balance to a stronger pressure
gradient.
 With this in mind, we can now enunciate two new rules of motion that
will help us quickly identify wind patterns using a pressure map (Box 8.2).
Given a set of isobars (or, equivalently, height contours), we can use these
rules quickly to estimate the direction and relative speed of the wind.

Box 8.2. Geostrophic Balance

Rule set #2
(1) Wind blows parallel to the isobars (i.e., along the isobars) with
higher pressures on the right (in the northern hemisphere).
(2) Wind speed is proportional to the pressure gradient (i.e.,
proportional to the tightness of the isobars).

Of course, isobars are not always straight. Most of the time they are
curved, as wind blows around highs and lows, troughs and ridges. We can,
however, make an approximation and assume that, locally, the wind is
blowing in a straight line for a very short time. It amounts to estimating that
the wind blows parallel to the local tangents to the isobars (Figure 8.8). This
will allow us to formulate a general sense for the wind circulation. We should
remember, however, that geostrophic balance is only an approximation, and
that it works best for straight flow. (See Section 8.6 for a more exact
treatment of curved flow.)
 Figure 8.8. In curved flow, we can make a geostrophic balance
approximation and assume that the wind is blowing along the tangents
(dashed blue lines) to the isobars.

Given a pressure map, then, and assuming geostrophic motion, we shall

follow the steps summarized in Box 8.3, at whatever point we need to
estimate wind speed (in a relative sense) and wind direction. Using this
simple methodology, we will obtain approximate wind maps such as shown
in Figure 8.9. Note how, in this particular example, high pressures (and
heights) are to the south and low pressures are to the north, and how the
pressure gradient force always points from high to low pressure, and is
perpendicular to the isobars. The Coriolis force is pointing in the opposite
direction to the pressure gradient force, and the wind is blowing along the
isobars, first around a ridge of high pressure (point A), and then around a
trough of low pressure (point C).

Box 8.3. Balance of Forces and Geostrophic Wind

 Rule set #3
(1) Identify the direction of the pressure gradient force.
(2) Knowing that the Coriolis force is equal in magnitude and
opposite in direction to the pressure gradient force, draw the Coriolis
force.
(3) Knowing that the Coriolis force is pointing to the right of the
direction of motion (in the northern hemisphere), identify the wind
direction.
(4) Wherever the spacing of the isobars or height contours is wide,
reduce the wind speed; wherever it is narrow (i.e., isobars close to
each other), increase the wind speed.

Figure 8.9. Determining the pressure gradient force (PGF), Coriolis force
(COR), wind speed, and wind direction from a height contour map
assuming geostrophic balance.
 8.6 Gradient Wind
How good is the geostrophic approximation? It is quite good outside of the
tropics when the flow is straight. When the flow is curved, however, the air is
subject to an additional “apparent” force (again, like the Coriolis force, this
force is due to our choice of reference frame, in a curved flow): the
centrifugal force – recall the centrifugal force pulling geostationary satellites
away from Earth (Chapter 1) and the centrifugal force you experience when
you turn at high speed with your car (apparent because, while your car is
turning, your body is trying to move in a straight line). Around a low, the
centrifugal force pulls the air in the same direction as the Coriolis force, and
it is the sum of both forces that balance the pressure gradient force
(Figure 8.10). Therefore, a smaller Coriolis force is required, which
corresponds to a smaller wind speed. As a result, the wind speed is less than
geostrophic balance would suggest: the wind is said to be subgeostrophic.
Compare Figure 8.10 with Figure 8.8 to convince yourself that this is indeed
the case.
 Figure 8.10. Balance of forces around a low when taking into account the
centrifugal force (CENT): the wind speed is reduced, the wind is
subgeostrophic.

In contrast, around a high, the centrifugal force points in the same

direction as the pressure gradient force (Figure 8.11). Therefore, a stronger
Coriolis force is required to balance the sum of both pressure gradient and
centrifugal forces. As a result, the wind speed is greater than geostrophic
balance would suggest: the wind is said to be supergeostrophic.

Figure 8.11. Balance of forces around a high when taking into account the
centrifugal force: the wind speed is increased; the wind is
supergeostrophic.

In practice, this new subtlety gives us slightly different wind speed

patterns as compared to the geostrophic wind, but the wind direction remains
the same: following pressure (or height) contours, with high values on the
right (with the wind at your back) in the northern hemisphere. Figure 8.12
shows the same pressure pattern as in Figure 8.9, but with different wind
speed and Coriolis force once curvature is taken into account. By comparison
with Figure 8.9, note how the wind is supergeostrophic around the ridge
(point A), whereas it is subgeostrophic around the trough (point C). Where
the flow is relatively straight (B), the balance of forces is unchanged from the
geostrophic case. Thus, for curved flow, we add a new rule of thumb,
summarized in Box 8.4, that will prove particularly useful when analyzing
upper-level flow and jet stream patterns.

Box 8.4. Curved Flow

Rule #4
The wind is supergeostrophic around a ridge and subgeostrophic
around a trough.

Figure 8.12. Determining wind speed and direction from a height contour
map assuming gradient wind balance.

Note that, in the example shown in Figure 8.12, we took great care to
keep all pressure gradients the same as in Figure 8.9, for illustration purposes.
In reality, of course, the pressure gradient changes constantly as does flow
curvature, yielding non-trivial results (see, for example, Figure 2.10).
Furthermore, variations in wind speed create regions of convergence and
divergence. These regions are crucial for understanding the development of
rising and sinking air, and midlatitude cyclones, as we will see in Chapter 10.
 8.7 Surface Winds
The gradient wind approximation is a relatively good approximation aloft
(850 hPa and above), and it is tempting to apply it closer to the surface, when
analyzing weather maps at sea level. As we approach the surface, however,
yet another factor comes into play: surface friction, which has a more
significant impact on the wind pattern than the centrifugal force. So we will
put the gradient wind aside for now, and start our discussion of surface winds
by returning to the geostrophic approximation.
Figure 8.13(a) shows an example of a surface pressure map featuring a
high pressure center (a “high”) and a low pressure center (a “low”). Even
though the geostrophic approximation is often inadequate close to the
surface, applying our rules of geostrophic motion at several points around the
high and the low allows us to see a general pattern emerge. Because winds
blow parallel to the isobars with higher pressures on the right, they tend to
blow clockwise around high pressure regions and counterclockwise around
low pressure centers (in the northern hemisphere) – summarized in Box 8.5
and Figure 8.13(b).
Figure 8.13. (a) Balance of forces and wind pattern at the surface
assuming geostrophic balance. (b) Cyclonic circulation around a low vs.
anticyclonic circulation around a high (northern hemisphere).
Figure 8.13. (c) Balance of forces and wind pattern at the surface when
taking friction (Fr) into account. (d) Convergent cyclonic circulation
around a low vs. divergent anticyclonic motion around a high.

Box 8.5. Highs and Lows

 Rule #5
Winds tend to blow clockwise around highs and counterclockwise
around lows (in the northern hemisphere).

In practice, low pressure centers are also the signature of cyclones.

Therefore we also say that the rotation of the winds around a low is cyclonic.
By extension, the rotation of the winds around a high (an anticyclone) is said
to be anticyclonic. Thus, in the northern hemisphere, cyclonic rotation is
counterclockwise, while anticyclonic rotation is clockwise. In the southern
hemisphere, flow is reversed, with cyclonic flow clockwise and anticyclonic
flow counterclockwise. In both hemispheres, cyclonic flow is in the same
direction as the local rotation of Earth when viewed from above the pole
(Earth appears to rotate counterclockwise when viewed from above the North
Pole, and clockwise when viewed from above the South Pole).
This simple rule is useful to form a general sense of the wind
circulation, but a simple look at a surface weather map will reveal that the
wind rarely blows exactly along the isobars. More often than not, it crosses
the isobars at an angle that can be 45° and sometimes more. The deflection is
due to friction between air and Earth’s surface. Surface friction will indeed
add an interesting twist to our picture.
 8.8 Friction
When a moving object rubs against a surface, it experiences resistance in the
direction of motion. The object could be our hand rubbing against a table, a
ski sliding on a blanket of snow, water flowing on a river bed, or air blowing
over a surface, be it ground, water, or a forest. This resistance, friction, is a
force acting upon the object in the direction opposite the direction of motion,
which slows the speed of the object. Thus, if no other force but friction is
acting on a ski in motion, the ski will slow down and eventually come to a
stop. Similarly, surface friction acts to slow down air in motion close to the
surface, and the wind would stop completely if no other force was at work.
When surface friction acts on the wind, the wind speed is reduced.
Therefore, the Coriolis force is also reduced, since it is proportional to wind
speed. In this case the pressure gradient force is balanced by the sum of the
friction and Coriolis forces, so that the wind must blow at an angle to the
isobars, toward low pressure, as shown in Figure 8.14. The greater the
friction effect, the more the wind speed will be reduced, and the greater the
angle at which the wind will blow across the isobars.

Figure 8.14. Comparison of the geostrophic approximation (left) and the

new force balance including friction (Fr) close to the surface (right).
 The impact of surface friction is most pronounced at the surface and
decreases with height, until it is very small in the free troposphere (above
about 850 hPa). The lower layer of the troposphere where surface friction
alters the force balance is called the planetary boundary layer. It can be
very shallow (down to a few dozens of meters in some extreme
circumstances, such as over the Antarctic ice sheet) or very deep, more than
two kilometers in some hot and dry situations where convection constantly
stirs the lower troposphere.
The surface type also plays a large role in determining the effects of
friction at the surface. Some land types are very rough, such as forests and
urban locations, while others are much smoother, such as grassland and
deserts. Oceans are definitely smoother than land and, as a result, generate
less friction. As a rule of thumb, winds tend to be deflected (relative to the
isobars) by an angle of 15° to 30° over oceans, as compared to 30° to 45°
over land.
Rules of motion for surface winds are summarized in Box 8.6 and
illustrated in Figure 8.15, where surface winds have been overlaid on
pressure and temperature. Note in particular how, in this view of our
February 2014 cyclone, surface winds blow across the isobars and into the
low, and outward of the high. This can be observed on surface weather maps
every day, and will be an important part of understanding weather with
surface cyclones (Chapter 10).

Box 8.6. Surface Winds

Rule set #6
 (1) Surface winds tend to blow at an angle to the isobars, pointing
toward low pressure.
(2) Surface winds tend to converge into low pressure centers, and
diverge away from high pressure centers.
(3) Convergence into lows is often associated with upward motion,
cloud formation, and precipitation, while divergence out of highs is
often associated with subsidence and clear skies.

Figure 8.15. Example of surface wind vectors around a high and a low on
February 20, 2014, at 18:00 UTC.

The frictional effect has very important consequences at the surface, as

illustrated in Figure 8.13(c) on p. 137. Since surface winds tend to be
deflected toward low pressure values, they will tend to be deflected away
from high pressure centers and toward low pressure centers. In other words,
winds tend to diverge away from highs, and to converge into lows
(Figure 8.13(d), see p. 137). This is better illustrated with streamlines, as
shown in Figure 8.16. Streamlines are drawn in such a way as to be
everywhere parallel to the wind direction. They convey a sense of flow – in
particular, nodes of convergence and divergence. Note in Figure 8.16 how the
air diverges out of the high, flows southeastward toward the cold front, takes
a sharp turn at the cold front to aim northward, and takes another sharp turn
at the warm front, to continue westward and eventually into the low. Notice
the many streamlines starting in the high pressure region and diverging away,
as well as the many streamlines converging and ending into the low. [Note
that streamlines are different from trajectories, which define the path of air
over a period of time. Streamlines provide a snapshot of motion at one instant
in time.]
 Figure 8.16. Example of surface streamlines diverging out of a high and
converging into a low on February 20, 2014, at 18:00 UTC.

These patterns of converging and diverging air have critical implications

for the mass balance of weather systems. Converging air does not accumulate
into a surface low, but continues on in a new direction. At the surface, since
the ground is rigid, that new direction is upward, which provides a lifting
mechanism that can enhance cloud formation and precipitation. That is one
reason why low pressure centers (i.e., cyclones) are often associated with
clouds and precipitation. Conversely, air diverging away from a surface high
is replaced by air subsiding from aloft. Since subsidence inhibits cloud
formation, high pressure regions (i.e., anticyclones) are often associated with
few clouds or even completely clear skies.
Finally, you will also note in Figure 8.15 that winds tend to be stronger
around lows and weaker around highs (and, correspondingly, isobars tend to
be more closely packed around lows and more loosely spaced around highs).
Indeed, converging winds swirling around a low tend to accelerate.
Somewhat similarly to an ice skater spinning faster and faster as she pulls her
arms and legs in toward the center line of her body, the rotating winds spin
faster and faster around the low as they converge closer to the center of
rotation. As a result, winds can be very strong near the center of deeper
cyclones. Conversely, winds tend to slow down as they diverge away from a
high pressure center, in the same way the ice skater slows down by spreading
out her arms before stopping. That is why winds are usually very mild under
an anticyclone.
With these general rules in mind, we are about ready to study all wind
patterns around Earth. Before we do so, however, let us investigate a few
additional factors that can influence wind development.
 8.9 Topography
The greatest influence on wind, after Earth’s rotation, curvature, and friction,
is that of topography. We have already mentioned the influence of orographic
lifting on cloud formation, precipitation, and vegetation, as air is lifted by
mountain slopes, as well as the impact of subsidence and adiabatic warming
in the lee of mountains, with the formation of chinook winds, for example
(see Chapter 6). Mountains of significant height can greatly alter the direction
and speed of the wind at the surface, of course, but also in the higher
atmosphere, as the deflection can be transmitted quite high above the
mountain top. Mountain barriers such as the Andes, the Himalayas, and the
Rocky Mountains can effectively force the air upward, or channel it at a
completely different angle from that suggested by geostrophic balance. In the
example shown in Figure 8.17, westerly flow from the Pacific Ocean is
deflected by the Olympic Mountains and channeled through the Strait of Juan
de Fuca (between the Olympic Mountains and Vancouver Island), and then is
blocked and deflected by the Cascades. Note in particular how the winds are
southerly in Seattle, Washington, where geostrophic balance would suggest
westerly winds.
Figure 8.17. Blocking and channeling of westerly winds by the Olympic
Mountains, upstream of Seattle.
 8.9.1 Mountain Breeze and Valley Breeze
Mountains and valleys can also induce local wind circulations through
differential heating and buoyancy. During the day, the mountain slopes that
form the “walls” of a valley heat faster than the air over the valley at the same
altitude. Thus, the air parcels closest to the slopes are heated by conduction
and are warmer than the environment. They are upwardly buoyant and tend to
rise (Figure 8.18(a)). Relatively colder air sinks down into the valley and
replaces the rising air along the slope. This causes a local circulation by
which air tends to rise up the valley and flow along the mountain slopes,
creating a valley breeze (blowing from the valley) that is compensated by
subsidence down into the valley floor (Figure 8.18(b)). Such a valley breeze
is usually quite mild, but it can lead to the formation of cumulus clouds above
the top of the mountains and sometimes showers and afternoon
thunderstorms.
 Figure 8.18. Development of a valley breeze during daytime.

The reverse circulation takes place after sunset. Emission of infrared

radiation by the mountains cools the valley walls, which in turn cool the
adjacent air. Becoming colder than air at the same elevation over the valley,
the air is negatively buoyant and sinks down the mountain slopes into the
valley. Being denser than the warm air above, the cold air flows down the
valley, like the rivers that are also found there. In places where the valley
floor is relatively flat, cold air can accumulate, cooling even further as the
night progresses. If the dew point is reached, condensation takes place and
valley fog forms, trapped in the valley by the temperature inversion
(Figure 8.19). Valley fog is very common in autumn, and can be easily
observed in visible satellite imagery, as it follows the valleys and meanders
of the topography (Figure 8.20). It typically dissipates during the day, as the
sun warms the cloud layer and convection mixes the air close to the surface.

Figure 8.19. Cooling by emission of infrared radiation (IR), downslope

flow of cold air at night, and formation of fog in the valley when the
temperature drops below the dew point.
 Figure 8.20. Valley fog in the Central Valley of California.

As a result of this diurnal cycle of heating, in the base of the valley, it is

common to observe a breeze blowing up-valley in the afternoon and down-
valley at night (Figure 8.21).
Figure 8.21. Daytime up-valley flow (a) vs. nighttime down-valley flow
(b).
 8.9.2 Katabatic Winds
When very cold and dense air flows down especially high topography, it can
give rise to some of the strongest and persistent winds on Earth’s surface.
This is the case in Greenland, and especially Antarctica, where cold air drains
off an elevated continental mass covered with ice and snow. The combination
of extremely low temperatures, due to radiational cooling and the absence of
sunlight in winter, results in the development of cold high pressure systems.
Air flows away from these high pressure systems, diverging and accelerating
away from the continent, in very strong winds called katabatic winds. The
air flows downslope toward the edge of the ice-covered continental mass
because it is very dense and therefore negatively buoyant, accelerating under
the effect of gravity to reach a maximum wind speed at the coastline
(Figure 8.22). Even though the air warms up by adiabatic compression, it
remains colder than the environmental air it displaces.

Figure 8.22. Schematic representation of very cold air accelerating

downslope over Antarctica to produce strong katabatic winds at the
 coastline. (The proportions have been greatly exaggerated.)

These cold winds exiting the continental mass continue northward and
feed into the cold sector of midlatitude cyclones developing over the
Southern Ocean. Indeed, winds are not isolated phenomena. They are all
connected within the larger system that is the dynamic atmosphere. Thus,
understanding local winds also means understanding the larger atmospheric
circulation at the scale of the planet, which will be the focus of Chapter 9.
 Summary
Horizontal motion of air is largely determined by the pressure gradient
force and the Coriolis force, with the addition of the centrifugal force in
regions of curved flow, and friction close to the surface.
The pressure gradient force results from horizontal variations in
atmospheric pressure, and, in the absence of other forces, accelerates the
wind from high to low pressure.
In the absence of weather systems, a sea breeze (wind blowing from sea
to land) will form at a coastline during the day when air over land becomes
warmer than the air over the adjacent water surface. At night, when air over
land becomes colder than air over water, a land breeze (wind blowing from
land to sea) forms.
Earth’s rotation creates an apparent force (the Coriolis force) that tends
to deflect the air at a right angle and to the right of the direction of motion in
the northern hemisphere (and to the left in the southern hemisphere). The
Coriolis force is zero at the equator, and increases with latitude. It is
proportional to the wind speed, but only affects the wind direction (i.e., it
does not affect the wind speed).
When the pressure gradient force and the Coriolis force are equal and
opposite, resulting in no net force, the wind is said to be in geostrophic
balance. The wind then blows parallel to the isobars with higher pressures on
the right in the northern hemisphere (on the left in the southern hemisphere),
and the wind speed is proportional to the magnitude of the pressure gradient.
 When the flow is curved, the centrifugal force makes the wind
subgeostrophic around a trough and supergeostrophic around a ridge.
At the surface, winds tend to blow clockwise around highs and
counterclockwise around lows in the northern hemisphere (opposite in the
southern hemisphere), but friction slows the wind speed and changes the
direction so that the wind blows at an angle to the isobars, toward low
pressure. Therefore, surface winds tend to converge into low pressure
centers, causing upward motion, cloud formation, and precipitation, and
diverge away from high pressure centers, causing subsidence and clear skies.
Topography can deflect and channel the wind at a different angle and
speed than those suggested by geostrophic balance (e.g., mountain ranges,
passes, straits).
Differential heating along valley slopes can cause local circulations by
which air tends to rise and flow up the mountain slopes during the day,
creating a valley breeze with afternoon clouds and showers over the
mountains, and to sink down the mountain slopes at night with the formation
of valley fog if the dew point is reached.
Katabatic winds form when very cold and dense air flows down an
elevated plateau, accelerating down the slopes under the effect of buoyancy
and, often, additional factors (e.g., subsidence, pressure gradient), as is
typical in Greenland and Antarctica.
CHAPTER 9
Global Wind Systems
CONTENTS
9.1 The Averaged Atmosphere
9.2 The Single-Cell Model
9.3 The Three-Cell Model
9.4 Some Large-Scale Circulations
Summary

Since they impact our daily life, we tend to think about wind and weather
locally, but they are in fact interconnected with the dynamics of heat transfer
in the atmosphere. We will now step back and apply our concepts of
differential heating, pressure gradient forces, and geostrophic motion at a
global scale, to explore the implications for what is called the general
circulation of the atmosphere, which will set the broader context within
which individual weather systems exist.
Weather is ever changing; however, some features persist through this daily
variability and guide the seasonal rhythms of life. Subsaharan countries such
as Ghana and Burkina Faso, for example, can rely on a yearly rainy season
from about April to November. By contrast, high pressure over the Sahara is
consistent all year round and prevents cloud formation and precipitation,
maintaining a desert. In midlatitudes, the presence of a high pressure system
over the North Pacific Ocean insures consistently mild weather on the West
Coast of the United States in the summer. This suggests that there are
somewhat permanent (“semi-permanent”) features that anchor the weather.
Although daily variations can depart substantially from these average
features, they vary around this semi-permanent state, which provides some
level of consistency.
To isolate the semi-permanent state, we first start by averaging
atmospheric data over long periods of time. This procedure erases the detail
of individual weather events, but reveals persistent patterns that we call the
general circulation of the atmosphere. Then we elaborate a model that
accounts for the observed averaged fields and provides a framework for
understanding variability in weather and climate events.
9.1 The Averaged Atmosphere
 9.1.1 Surface Temperature
Consider the average distribution of air temperature at the surface shown in
Figure 9.1. These maps were created by averaging surface air temperature
over 68 years for January and July, respectively. We recognize familiar
features, such as the warm equatorial region and the cold polar regions,
which we know exist because of the differential heating of Earth and the
resulting accumulation of energy at low latitudes compared to high latitudes
(see Chapter 4). Figure 9.1 also shows that, whereas the isotherms are
relatively widely spaced at low and (to a lesser extent) high latitudes, the
region of transition between the warmer tropics and the colder polar regions,
i.e., the midlatitudes, corresponds to a region of strong temperature gradient.
This is most obvious in the southern hemisphere in winter (Figure 9.1(b)),
because of the presence of Antarctica, where the coldest temperatures on
Earth’s surface are recorded. The temperature contrast between the extremely
cold continent and the warmer ocean creates a very strong temperature
gradient around 60 °S, reminiscent of a temperature front. A similar pattern is
found in the northern hemisphere in winter (Figure 9.1(a)) between 30 °N and
60 °N. The presence of cold land masses complicates the picture in that the
gradient meanders south over land areas and north over the oceans, but the
sharp temperature gradient is clearly visible, both in the color shading and the
tightness of the isotherms. This transitional region is sometimes called the
polar front.
Figure 9.1. (a) Mean surface air temperature for the month of January.
 Figure 9.1. (b) Mean surface air temperature for the month of July.

Note that the temperature gradient is much weaker in the southern

hemisphere in January (Figure 9.1(a)) and in the Northern Hemisphere in
July (Figure 9.1(b)); i.e., in the summer hemisphere. Conversely, the
temperature gradient is strongest in the winter hemisphere. Figure 9.2 shows
schematically why this should be so. The tropics, being more directly under
the sun, are relatively warm independent of the season. The polar region,
however, becomes extremely cold in winter, because it is pointing away from
the sun and receives little sunlight during few hours of daylight, if any, but is
constantly losing heat to space by emission of infrared radiation.
Consequently, the temperature contrast between the tropics and the poles is
much greater in winter than in summer.
 Figure 9.2. The temperature contrast between the tropics and the polar
region is much stronger in winter than in summer.

Figure 9.2 also reminds us that the region most exposed to the sun shifts
north and south over time. It shifts further south in January and further north
in July. We can verify in Figure 9.1 that the band of highest temperatures
indeed shifts to the south in January (see Australia in Figure 9.1(a), for
example) and to the north in July (see northern Africa in Figure 9.1(b)). This
will have important implications later on when we add pressure and winds to
the picture.
The surface temperature distribution has direct consequences on the
pressure distribution around Earth, as we will now explore by first
considering the upper troposphere.
 9.1.2 Upper-Level Heights
First, a map of average 250 hPa heights reveals a broad swath of higher
heights over the equator (recall the equivalence of height values on pressure
surfaces with pressure values on height surfaces), a large area of lower
heights anchored around Antarctica, as well as lower pressures at the North
Pole, although the latter are less intense in the summer (Figure 9.3). In the
northern hemisphere, two prominent troughs are found over Eastern Asia and
the eastern part of North America in winter (Figure 9.3(a)). They are due to
the presence of the two continental land masses, affecting air flow through
land–sea temperature contrasts as well as topography. These upper-level
troughs will play an important role in explaining the concentration of
midlatitude cyclones in those regions (see Chapter 10).
Figure 9.3. (a) Average 250 hPa heights, winds, and wind speeds for the
month of January with schematic representation of higher and lower
pressures, and upper-level troughs. Areas of maximum wind speed indicate
the position of the jet streams.

Figure 9.3. (b) Average 250 hPa heights, winds, and wind speeds for the
month of July, with schematic representation of higher and lower
pressures. Areas of maximum wind speed indicate the position of the jet
streams.
 Figure 9.3. (c) Schematic representation of the semi-permanent jet streams
and upper-level troughs (in winter for each hemisphere).

We also note in Figure 9.3 that the upper-level winds generally blow
from west to east (i.e., westerly winds). We can now explain this observation
if we recall that, at the planetary scale, winds are influenced by Earth’s
rotation and adjust to geostrophic balance. Therefore, winds should blow
parallel to the isobars (or height contours) with higher pressures (heights) on
their right and lower pressures (heights) on their left in the northern
hemisphere, and conversely in the southern hemisphere, which is indeed what
we observe. Most interesting, however, is the fact that these westerlies are
concentrated in narrow regions of very high wind speed in both hemispheres,
where the meridional pressure gradient is most intense, as indicated by the
tightness of the height contours around 30–60 °N and 30–60 °S, i.e., in the
midlatitudes. The blue ribbons of Figure 9.3 are indeed the strong upper-level
jets that we identify as the jet streams, as shown in Figure 9.3(c). They reach
wind speeds higher than 100 knots in these average maps, and much higher
wind speeds are found on individual days. Moreover, we can verify that the
jet streams reach their maximum wind speed not only in the midlatitudes, but
also at the tropopause, i.e., at about 250 hPa (see Figure 5.9 for a nice
illustration). The northern hemisphere and southern hemisphere jet streams
are shown schematically in Figure 9.4(a), where the color-shaded wind
speeds of Figure 9.3(a) have been retained in the background. The jet streams
are shown both as blue arrows indicating westerly motion, as if viewed from
above, and as blue crosses representing the tail of the arrows going into the
page, as if we were looking at the arrows from behind.
Figure 9.4. (a) Schematic representation of the jet streams, both as a lateral
 view (blue arrows) and a rear-end view, as if looking at the tail of the
arrows going into the page (blue crosses). The blue shading indicates wind
speed as in Figure 9.3. (b) Increase in westerly wind speed with height in
the midlatitudes shown on three pressure surfaces. The jet stream reaches
maximum wind speed at the tropopause (250 hPa). (c) Same as (b) after
flattening Earth’s surface.

To explain the presence and the intensity of the jet streams, we need to
return to our average temperature maps (Figure 9.1) and recall that there
exists a strong temperature gradient at the surface, in the midlatitudes
(Figure 9.4(b)). If we lay out this temperature gradient on a flat surface, as if
we were flattening Earth’s surface, or as if we were looking at Figure 9.1(a)
from the west, we obtain the schematics shown in Figure 9.4(c). We can now
add a cross-section of the atmosphere and show a few pressure surfaces.
Recall that, where the atmosphere is warmer (i.e., in the tropics), the
atmosphere is “thicker” and the pressure surfaces are raised compared to the
poles. If we now compare two points at the same height near the tropopause,
we find that the pressure is higher at the top of the warm air column (i.e.,
above the tropics) and lower at the top of the colder air column (i.e., over the
poles). As shown in Figure 9.4(c), and in accordance with geostrophic
balance, the resulting winds are westerly, and into the page.
But Figure 9.4(c) also shows that the slope of the pressure surfaces over
the midlatitude temperature gradient increases with height, i.e., the pressure
surfaces are increasingly steeper with height. In other words, the pressure
gradient increases with height and, according to geostrophic balance, the
wind speed increases. It turns out that it increases in such a manner up to
about 250 hPa. Recall that, above the tropopause, the temperature increases
with height due to warming from ozone absorption of ultraviolet radiation.
Since we cross the tropopause over polar regions at lower altitudes than over
tropical regions, the temperature decrease with latitude becomes smaller with
height above the tropopause and eventually switches direction. As a result,
the temperature gradient, and hence the pressure gradient, decrease above
about 250 hPa. Therefore, the wind speed decreases as well, and we are left
with a jet of maximum wind speed in the middle, right at the tropopause; i.e.,
the jet stream.
In summary, the temperature distribution is such that there exists a
strong temperature gradient in midlatitudes, which induces a pressure
gradient in the upper troposphere, which itself induces a westerly jet. This
allows us to explain yet another subtlety of Figure 9.3. You might have
noticed that the jet stream is strongest in January in the northern hemisphere,
and strongest in July in the southern hemisphere. In other words, the jet
stream is always stronger in the winter hemisphere. By now this should not
come as a surprise, since we know that the jet stream is intimately tied to the
midlatitude temperature gradient. Indeed, recall from Figures 9.1 and 9.2 that
the midlatitude temperature gradient is always stronger in the winter
hemisphere. It stands to reason that a stronger temperature gradient should
induce a stronger pressure gradient and a stronger jet stream. (The seasonal
contrast is less evident in the southern hemisphere than in the northern
hemisphere because of the relative absence of continents relative to open
water and, as a consequence of the thermal inertia of water discussed in
Chapter 4, less seasonality.)
Theory tells us that, given the size and rotation rate of Earth, it is on the
borderline of supporting two distinct jet streams. (Jupiter, for example, is
huge compared to Earth, with much faster rotation, and has many jet
streams). Indeed, if we look carefully, we find that the jet streams meander
and sometimes split. In winter, in particular, we can guess the presence of
two separate jets entering North America and two separate jets entering
Europe (Figure 9.3(a)), as well as two separate jets over the South Pacific
Ocean (Figure 9.3(b)). Indeed, there are often two jets, a polar jet and a
subtropical jet, which sometimes combine to form a stronger, unique jet in
some areas (Figure 9.3(c)): in particular, a Pacific jet stream over the North
Pacific Ocean, east and downstream of Japan; and an Atlantic jet stream over
the North Atlantic Ocean, east and downstream of New England. Recall that
these features are revealed after averaging a large number of daily situations,
and are therefore what is found on average. As such, these jet streams can be
expected to exert a strong constraint on the average weather. These two
regions where the jet stream is strongest are found south of the persistent
upper-level troughs extending south from Eastern Siberia and Eastern
Canada, and correspond to areas of enhanced midlatitude cyclogenesis. We
will see in Chapter 10 that local meanders of the jet stream, short-wave
troughs, are critical for the development of midlatitude cyclones.
 9.1.3 Surface Pressure
At the surface, the pressure field is broken up in closed highs and lows that
more or less follow the distribution of land masses and oceans (Figure 9.5).
In particular, we note the presence of subpolar lows at about 60–70 °N and
60–70 °S, especially in winter, as well as the presence of subtropical highs
at about 30 °N and 30 °S. The subtropical highs tend to form on the eastern
side of the ocean basins. Note that, in the southern hemisphere, amidst the
highs and lows, there persists a continuous stream of midlatitude westerlies at
about 40–50°S, where the ocean is continuous from west to east. These
particularly strong winds were called the “roaring forties” by early sailors,
who knew to sail from west to east to take advantage of the general direction
of the winds, but to avoid the regions of strongest winds, as they are
treacherous and can sink a ship.
 Figure 9.5. (a) Mean sea level pressure for the month of January.

Figure 9.5. (b) Mean sea level pressure for the month of July.

The lows are stronger in winter, and we give them particular names
when describing and forecasting the weather. In the northern hemisphere, the
Aleutian Low is centered over the Aleutian Islands, a chain of islands
extending southwest from mainland Alaska, while we usually find the
Icelandic Low near Iceland (Figure 9.5(a)). Similarly, the subtropical highs
are stronger in summer, and we usually refer to them as the Pacific High and
the Bermuda/Azores High in the northern hemisphere, the latter being
typically found between Bermuda and the Azores Islands (Figure 9.5(b)).
 9.1.4 Precipitation
A map of average precipitation also contains a number of interesting features
(Figure 9.6). The equator features a band of heavy rain, while the polar
regions show little precipitation. In the midlatitudes, between about 30° and
60° north and south of the equator, we recognize rain patterns near the jet
streams over the Pacific and Atlantic Oceans corresponding to precipitation
from midlatitude cyclones (see Chapter 10). We also notice strikingly dry
areas in the subtropics, in particular over the eastern side of the ocean basins
and some land areas, such as Australia and the Sahara desert. (Note the
correspondence with the subtropical highs of Figure 9.5.)

Figure 9.6. Mean annual precipitation.

 Our goal is now to elaborate a sparse model of the circulation of the
atmosphere that articulates all of these semi-permanent features. We start
with the main feature of the energy distribution of the atmosphere: the
temperature contrast between the equator and the poles. Recall that
temperature contrasts create pressure gradients aloft, which set the
atmosphere in motion. This was illustrated with the sea breeze circulation in
Chapter 8. Even though the sea breeze mechanism applies only to local-scale
circulations, the physical discussion that underlies it constitutes a good
starting point for building our first model of the general circulation of the
atmosphere.
 9.2 The Single-Cell Model
Recall that, in a sea breeze circulation, air over the warmer area flows toward
the colder air aloft, increasing the surface atmospheric pressure over the
colder area and decreasing it over the warmer area (Figure 9.7(a)). Air
subsides over the colder area and the sea breeze returns from the cold to
warm area at the surface, closing the circulation.

Figure 9.7 (a) Schematic representation of a sea breeze created by a

horizontal temperature contrast. (b) Sea breeze concept applied to the
temperature contrast between the equator and the polar regions.

If we start by assuming that Earth does not rotate and we view the
temperature difference between the equator and the poles as such a
temperature contrast, but taking place over much larger distances
(Figure 9.7(b)), we can expect the same response such that the air should rise
over the equator and diverge away from the equatorial region aloft (at the
tropopause, or about 250 hPa). As a result, the surface pressure should drop
at the equator, creating a trough of lower pressures. As upper-level air
converges at the poles, the surface pressure should increase and the air should
sink, closing this double circulation (northern and southern hemispheres).
Thus we should expect poleward upper-level winds and equatorward surface
winds.
This circulation, named after George Hadley (1685–1758), who first
proposed it, is called a Hadley cell, and our simple model suggests that there
should be a Hadley cell in each hemisphere (Figure 9.8). This picture is
indeed consistent with aspects of our earlier observations of the mean
atmosphere. At the tropopause, we indeed observe higher pressures over the
equator, and lower pressures at the poles (Figure 9.3). The Hadley cell model
depicts converging winds at the equator, which is also consistent with
observations. Indeed, note in Figure 9.5 how, at low latitudes, the surface
winds tend to have a northeasterly direction north of the equator, and a
southeasterly direction south of the equator. These winds, called the trade
winds, necessarily meet and converge in the equatorial region, to form a
semi-permanent area of convergence called the Intertropical Convergence
Zone (ITCZ). Since converging winds tend to rise (see Chapter 6), the ITCZ
is coincident with convection and precipitation, as confirmed by the band of
higher rain rate visible in the equatorial region in Figure 9.6, and by the
cumulonimbus clouds often observed in satellite imagery (Figure 9.9). This
band of heavy precipitation observed at the equator is consistent with
equatorial convection predicted by the Hadley cell model (refer to
Figure 9.8), while reduced precipitation in the polar regions is consistent with
subsidence. Indeed, low pressures and convection at the equator should
produce clouds and precipitation, while subsidence at the poles should inhibit
cloud formation and precipitation.
Figure 9.8. Single-cell model.

Figure 9.9. Cold cumulonimbus cloud tops indicate the location of the
Intertropical Convergence Zone (ITCZ) in this (GOES-14) infrared
satellite image.
 The bands of precipitation observed in the midlatitudes, however, are
not explained by the simple single-cell Hadley model. We also noted in
Figure 9.3 that the upper-level winds are not blowing from high to low
pressures in the midlatitudes as our two Hadley cells would suggest (i.e., in a
north–south direction), but overall from west to east. Similarly, whereas the
Hadley cell model suggests the presence of only low surface pressures at the
equator and high surface pressures at the poles, we noted that, in reality, the
surface pressure field is broken up in closed highs and lows in midlatitudes.
In summary, our Hadley cell model has some qualities consistent with
observations, such as convergence and precipitation near the equator, and
subsidence at the poles. However, it does not have westerly jet streams or
highs and lows in the subtropics and midlatitudes, which require a more
detailed model.
 9.3 The Three-Cell Model
Our single-cell model neglects one crucial component: Earth’s rotation.
Indeed, as the upper-level winds blow away from the equator and poleward,
they tend to be deflected to the right in the northern hemisphere and to the
left in the southern hemisphere (i.e., eastward in both cases) due to the
Coriolis acceleration (see Chapter 8). Similarly, surface winds blowing
toward the equator tend to be deflected to the right in the northern
hemisphere and to the left in the southern hemisphere: westward in both
cases (Figure 9.10). In other words, this should result in upper-level
westerlies and surface easterlies.

Figure 9.10. Distortion of the Hadley cell due to Earth’s rotation.

It turns out that this configuration is unstable, and the circulation breaks
down into a pattern that is better captured by a three-cell model, as shown in
Figure 9.11. The upper-level flow of the Hadley cell returns to the surface in
the tropics, between 20 and 30 °N in the northern hemisphere, and between
20 and 30 °S in the southern hemisphere, which limits the extent of the
Hadley cells to about 30° from the equator. Similarly, the rest of the single
cell breaks down into a polar cell spanning about 60–90° latitude and an
intermediate region spanning 30–60° latitude. We refer to this midlatitude
region as the Ferrel cell, after William Ferrel (1817–1891), although the
circulation there does not really take the form of a cell. While the Hadley and
polar cells are somewhat representative of the real circulation and are
conceptually useful, midlatitude dynamics are largely defined by the
development and propagation of extratropical cyclones (see Chapter 10), for
which the circulation is far more complex. The Ferrel cell represents the
average effect of these features, and is therefore more of a mathematical
artifact that depends on how the data is averaged. (For example, if, rather
than latitude, we use temperature as a north–south coordinate for averaging,
the Ferrel cell disappears.)

Figure 9.11. Three-cell model.

 Besides capturing the features previously described (ITCZ, convection
at the equator, subsidence at the poles), the three-cell model also suggests the
presence of subsidence in the subtropics, between 20°N/S and 30°N/S, which
explains the presence of the subtropical highs observed in Figure 9.5, as well
as the dry areas observed in Figure 9.6. (Recall that subsidence prevents
cloud formation, and therefore precipitation.) In fact, this latitude band
contains most of the major deserts on Earth (Figure 9.12). (Also note the
location of the rain forests in this figure, consistent with convection and
precipitation at the ITCZ.)

Figure 9.12. Examples of tropical rain forests along the equator (in green)
and subtropical deserts along the tropics (in white).

Combined with Earth’s rotation and the Coriolis acceleration, the three-
cell model also explains the direction of the trade winds converging into the
ITCZ. These winds are northeasterly in the northern hemisphere and
southeasterly in the southern hemisphere, consistently with the deflection of
the surface branch of the Hadley cell (recall Figure 9.10) and with the
direction of the circulation on the equatorward flank of the subtropical highs.
In the end, the three-cell model constitutes a useful framework for
understanding general weather and climate features at the regional scale, as
we will now illustrate with some examples.
9.4 Some Large-Scale Circulations
 9.4.1 West Coast vs. East Coast
Summer weather on the West Coast of the United States is in stark contrast
with that on the East Coast. While Seattle, Wasington, Portland, Oregon and
San Francisco, California are experiencing relatively cool summer
temperatures and dry weather, Georgia and the Carolinas are experiencing
warm and moist conditions, with frequent convective summer rain. These
sharp differences can be explained by considering the average surface
pressure distribution shown in Figure 9.5 and reproduced in Figure 9.13 with
schematic winds. While the Pacific high produces northwesterly to northerly
winds down the West Coast, bringing cool, dry air from the Pacific Ocean
and Canada, the Bermuda High produces southerly winds, bringing heat and
moisture from the tropics toward Florida and the East Coast. Therefore,
although at similar latitudes, and both in a coastal environment with
apparently similar exposure to the ocean, the two regions experience very
different summer weather conditions.

Figure 9.13. Summer prevailing winds on the West Coast and East Coast
of the United States, as determined by the semi-permanent highs over the
Pacific Ocean and Atlantic Ocean.
 9.4.2 Antarctica
We can apply similar reasoning to the southern hemisphere. Recalling that
there is subsidence and high pressure over the South Pole (see Figure 9.8), we
can explain why surface winds diverge toward the surrounding Southern
Ocean (Figure 9.14). These surface winds are the katabatic winds
encountered in Chapter 8. Recall that the Antarctic ice cap is so cold that the
overlying air, cooled by conduction, sinks close to the surface and flows
downslope toward the ocean, channeled by the topography of the glaciers. All
these effects (high pressure, subsidence, divergence, higher density due to
temperature, downslope flow due to negative buoyancy) combine to create
strong, cold air currents that exit the continent and surrounding ice to
continue their course northward.

Figure 9.14. Katabatic winds and midlatitude cyclones resulting from the
general circulation of the southern hemisphere.
 9.4.3 The Sahel
The general circulation is of course not static. It is an average picture that
gives us a general idea of what to expect. On short time scales, weather
systems respond to other forcings that create the variability we observe. The
primary variation we experience throughout the year is due to the seasonal
change in solar insolation. Figure 9.3 illustrates the impact of this seasonal
variation on the upper-level pressure distribution and jet streams, while
Figure 9.5 illustrates its impact on surface pressure patterns and winds.
In the tropics, this seasonal variation directly impacts the position of the
Intertropical Convergence Zone (ITCZ) and associated weather. Recall that
the ITCZ is an area of wind convergence, associated with the formation of
cumulonimbus clouds and heavy precipitation. Convection is driven at least
in part by the absorption of incoming solar radiation by the underlying land
or ocean. Therefore, the position of the ITCZ responds in part to the location
of maximum insolation. As shown in Figure 9.15, this maximum shifts
southward during northern hemisphere winter and northward during northern
hemisphere summer. As a result, the ITCZ is displaced southward and
northward accordingly. Note, however, that other factors control the ITCZ,
such as the presence of land or ocean and the dynamics of the Hadley cells.
Therefore, the displacement is not uniform around the globe, and it is not
symmetric between the two hemispheres. In particular, the northward
displacement is typically larger than the displacement into the southern
hemisphere.
 Figure 9.15. The ITCZ is displaced southward during (northern
hemisphere) winter and northward during (northern hemisphere) summer.

A good example of the impact of this seasonal variation of the ITCZ on

local weather is found in Africa, in the Sahel, a narrow band of semi-arid
grassland at about 15°N, between the equatorial rain forests and the Sahara
desert (Figure 9.16(a)). In winter, the ITCZ is displaced southward, and
northerly winds bring warm, dry air to the Sahel. Recall that the ITCZ is
inherently a region of low pressure; therefore, winds tend to blow from the
subtropical highs toward the ITCZ (Figure 9.16(b)). In summer, by contrast,
the ITCZ is displaced northward, along with the associated convection and
precipitation, as southerly winds bring moisture from the ocean
(Figure 9.16(c)). As a result, the Sahel experiences a few rainy months that
bring enough water to sustain a steppe-like environment and a modest level
of agriculture. Thus, while the equatorial rainforests experience a wet regime
during a large fraction of the year, and the Sahara experiences a desert
climate all year round, dry and wet seasons alternate in the Sahel every year.
 Figure 9.16. Displacement of the ITCZ across the Sahel. (a) Composite
satellite view of Earth showing rain forests and desert areas in Africa. (b)
Southward displacement of the ITCZ in northern winter. (c) Northward
displacement of the ITCZ in northern summer.

Note that climate models predict that, in a warmer world such as

projected due to global warming, desertification will threaten the Sahel and
the local populations it sustains (see Chapter 15).
 9.4.4 The Indian Monsoon
A similar displacement of the ITCZ can be observed in the Indian Ocean, and
greatly affects the weather in India and surrounding countries. It is enhanced
by two additional factors. First, the land–ocean temperature contrast drives a
wind circulation similar to a sea breeze, except on a much larger spatial scale,
and therefore subject to Earth’s rotation and the Coriolis force. Second, the
Himalaya mountains and the Tibetan Plateau to the north, which act as an
elevated heat source or sink, enhance convection or subsidence at different
times of the year (Figure 9.17(a)).

Figure 9.17. Displacement of the ITCZ during the Indian monsoon. (a)
Topography of India and the Tibetan Plateau. (b) Southward displacement
of the ITCZ in northern winter. (c) Northward displacement of the ITCZ in
northern summer.
 In Northern Hemisphere winter the ITCZ shifts southward and a high
pressure region settles over the Tibetan Plateau (Figure 9.17(b)). The
resulting northerly winds bring dry air and reduced precipitation to India
from about November to May. Since these winds descend along the slopes of
the Himalayas, the subsiding air warms, which makes for an often hot, dry
season, especially toward April–May.
In northern hemisphere summer the ITCZ shifts northward and, the
continent being much warmer than the ocean, a continent-scale sea-breeze
type of thermal circulation develops, which enhances the high-to-low surface
winds from the Indian Ocean to the continent (Figure 9.17(c)). This
enhancement moves the ITCZ much farther north than would normally be the
case. Warm temperatures over land also enhance convection, and the
southerly flow tends to push the air against the Himalayas, forcing it to rise.
Since the southerlies originate over the warm Indian Ocean, they contain
abundant water vapor, and convergence along the ITCZ, convection, and
orographic lifting all combine to produce heavy precipitation during the rainy
season, from about July to September. The eastern part of India is notorious
for receiving record amounts of precipitation (as high as three meters) in
short amounts of time.
 9.4.5 El Niño
The best known fluctuation in tropical climate starts off the coast of South
America, where the subtropical high pressure system drives southerly winds
(Figure 9.18). As the wind blows over the ocean, it drags the surface water
along with it. However, since water in the ocean is subject to the same
Coriolis acceleration as air in the atmosphere, it tends to be deflected to the
left (in the southern hemisphere), at a right angle to the wind, as shown
schematically with gray arrows in Figure 9.18. As the surface waters move
away from the coast, they are replaced by deeper water rising from the deep
ocean. This pumping of deep, cold, nutrient-rich water is called upwelling. It
is well known to local fishermen, as the nutrient-rich waters attract many fish.
 Figure 9.18. Formation of the cold tongue. The thin blue arrows represent
the wind blowing northward along the coast of South America and
converging along the equator. The thick gray arrows show the horizontal
displacement of surface water, at a right angle to the wind. The blue cold
tongue results from the upwelling of cold water from the deep ocean.

Along the equator, a similar mechanism is at work. As the easterly trade

winds converge toward the ITCZ, they also drag surface water with them.
The Coriolis force deflects the water northwestward in the northern
hemisphere (to the right of the trade winds) and southwestward in the
southern hemisphere (to the left of the trade winds), resulting in a divergence
of water away from the equator. Surface water pulled away from the equator
is replaced by rising water from the deeper ocean. This upwelling of cold
water forms a cold band, in the middle of otherwise warmer water, that looks
like a tongue extending from Peru and Ecuador out into the open ocean
(shown in blue in Figure 9.18). Scientists commonly refer to this feature as
the equatorial cold tongue.
The easterly trade winds also drag surface waters westward and induce a
higher sea level, by about 40 cm, in the Western Pacific Ocean, as water
accumulates in the western part of the ocean basin. This impacts the upper
layer of warmer ocean water. This upper layer, a few hundred meters thick, is
separated from the deeper cold ocean waters by a transition region called the
thermocline (see Figure 9.19). In normal conditions, due to upwelling of
deep, cold water in the eastern part of the basin, the thermocline slopes
upward as it approaches South America (Figure 9.19(a)). In contrast, the
thermocline sits lower in the western part of the basin, resulting in a deep
pool of warm water in the Western Pacific Ocean called the Pacific warm
pool. (Upwelling also occurs in the western part of the basin, but the mixed
layer of warm water is much deeper, so that the cooling effect is small.)
 Figure 9.19. Schematic representation of El Niño/Southern Oscillation. (a)
Normal conditions: the Walker circulation spans the entire Pacific basin.
(b) El Niño: the cold tongue weakens or disappears, the Walker circulation
weakens, and a secondary circulation forms over the western half of the
Pacific basin.

The alert reader will have recognized a familiar setup. With warmer
water in the Western Pacific and colder water in the Eastern Pacific due to
upwelling, the conditions are ripe for a large-scale circulation driven by the
temperature contrast (Figure 9.19(a)). Indeed, we observe lower pressures
over Indonesia, along with convection, thunderstorms, and precipitation, and
higher pressures over the cold tongue, along with subsidence and primarily
low clouds. This large-scale east–west circulation takes place over thousands
of miles and is referred to as the Walker circulation, after British scientist
Sir Gilbert Walker (1868–1958), who first noticed the correlation between
the Western and Eastern Pacific while analyzing pressure data to understand
and predict the monsoon in India.
If the cold tongue is well known to local fishermen in Peru and Ecuador,
it is also well known that temperatures and weather conditions change
drastically every few years, following an irregular cycle of 2 to 7 years.
Ocean temperatures rise, fish disappear, and heavy rains affect the coastal
regions. In the meantime, Indonesia experiences drier conditions as
convection moves eastward over the ocean. Because these episodes tend to
start around Christmas, fishermen used to refer to them as El Niño, the child
(i.e., the Christmas child, Jesus). We now know that these events are part of a
large-scale oscillation involving both the atmosphere and the ocean, and we
refer to it more generally as the “El Niño/Southern Oscillation” (ENSO).
ENSO involves a positive feedback loop between surface winds and
upper-ocean conditions. The trade winds, for example, are observed to
weaken during an El Niño event (Figure 9.19(b)). As a result, upwelling is
reduced along the equator, which weakens the normal cooling process of
surface water in the cold tongue. Furthermore, with weaker trade winds, the
higher sea level in the Pacific warm pool is no longer sustained (a net force
directed to the east), so that warm water begins to push eastward, further
warming the water in the upper eastern ocean. As a result, the temperature
contrast between Indonesia and the Eastern Pacific, which sustains the
Walker circulation, is also much reduced. In the process, low pressure and
convection move eastward toward the Central Pacific, as Indonesia
transitions to high pressure and subsidence, which explains the dry conditions
experienced there during an El Niño event. The thermocline also levels out,
as upwelling is reduced and warm water invades the Eastern Pacific
(Figure 9.19(b)).
With a weaker Walker circulation, the trade winds, which really
constitute the surface branch of the circulation, also weaken and we are back
to where we started, with ever more weakening. This positive feedback loop,
once initiated, shifts the whole system from normal to El Niño conditions,
and can bring about drastic weather changes over large portions of Earth.
One might ask, when and why does the feedback loop stop? Or why
does El Niño not keep growing indefinitely? To answer that question, we
need to return to the ocean. As the trade winds blow over the Pacific Ocean,
they drag the surface waters along and push them westward. To visualize
how the water might pile up in the Western Pacific, one can think of a
bathtub in which we are pushing the water to one side with our hands. The
water level will rise on that side. Similarly, the water level rises in the
Western Pacific during normal conditions. This increase in sea level height is
subtle (on the order of about 40 cm), but we can now measure it precisely
with satellite instruments called altimeters. In the same way that our bathtub
water will not remain piled up on one side of the tub if we stop pushing, the
warm waters of the Western Pacific Ocean will start sloshing back toward
South America as soon as the trade winds weaken and stop dragging the
surface waters westward. The positive feedback will continue as the warm
water propagates eastward on timescales of several months. But one can see
that there is an inherent limit to the displacement of the warm water, due to
the size and shape of the ocean basin and other dynamical constraints. El
Niño corresponds to the maximum eastward extent of the warm water. When
that natural limit is reached, heat from the warm water that has spread across
the equatorial Pacific is radiated to space, the trade winds start dragging the
warm surface waters westward again, and “recharge” the system, starting
another cycle.
By analogy with El Niño, scientists have named the opposite phase of
the oscillation, when the trade winds are unusually strong, the cold tongue is
at its maximum extent, and warm surface waters are displaced to the west, La
Niña. They have designed a number of indices that indicate the extent to
which the tropical Pacific ocean–atmosphere system is experiencing an El
Niño or La Niña phase. These indices are typically based on measurements of
atmospheric pressure, winds, and sea surface temperature. The most
commonly used ENSO index combines multiple measurements and is shown
in Figure 9.20. Positive values of the index indicate an El Niño phase, while
negative values indicate a La Niña phase. A large positive value such as
occurred in 1982–1983 and in 1997–1998 indicates a very strong, or historic,
El Niño.
 Figure 9.20. Variations of the most commonly used ENSO index since
1950. A positive index (in red) indicates El Niño conditions, while a
negative index (in blue) indicates La Niña conditions.

It is apparent in Figure 9.20 that ENSO follows an irregular cycle, with

some phases being as short as 1 year, while others are as long as 6 years. A
full cycle through El Niño and La Niña takes place over roughly 2 to 7 years,
and is determined by complex ocean–atmosphere dynamics that involve more
than just the tropics, and more than just the Pacific Ocean.
Because ENSO involves both atmosphere and ocean, it is called a
coupled oscillation. Indeed, the displacement of water would not be possible
without the trade winds, and the Walker circulation would not be possible
without the ocean temperature contrast. Thus, both components are needed
for the system to oscillate. Furthermore, we now know that ENSO affects
weather patterns not only in the tropical Pacific, but also worldwide through
what we call teleconnections – patterns of variation in the circulation of the
atmosphere that are persistent over large distances. For example, during an El
Niño event, tropical convection in the central equatorial Pacific Ocean
distorts the jet streams in both hemispheres, producing large-scale waves on
the jet stream. As a result, surface pressures are anomalously low over the
Gulf of Alaska, which brings warmer temperatures than usual to the Pacific
Northwest and Western Canada, while the jet stream is displaced to the south
of its usual position, channeling storms toward California and Florida, where
it is colder and rains more than usual (Figure 9.21(a)). By contrast, during a
La Niña event, surface pressures are anomalously high over the Gulf of
Alaska, which tends to split the jet stream into a northern branch, which
brings cold, dry air to Western Canada, and a southern branch that steers
midlatitude cyclones toward the Pacific Northwest, where it rains more than
usual (Figure 9.21(b)). In the meantime, the Southern United States
experience unusually dry weather.

Figure 9.21. Impact of El Niño (a) and La Niña (b) on the weather of
North America.
 El Niño also affects the frequency and intensity of tropical cyclones.
Indeed, as we will see in Chapter 12, tropical cyclones are powered by the
latent heat released in their core during condensation, the water vapor itself
originating from evaporation of warm ocean water. This might explain why
the warmer Pacific Ocean seems to fuel more intense tropical cyclones during
El Niño events. However, tropical cyclones are also sensitive to the vertical
wind shear (i.e., the change in wind speed with height – see Chapter 12), as
strong shear disrupts their vertical structure and prevents self-organization.
This is why the frequency of hurricanes decreases over the Atlantic Ocean
during El Niño events, since the displacement of the jet stream to the south
causes stronger upper-level winds and a stronger vertical wind shear over the
latitude band where Atlantic hurricanes develop.
Thus, it appears that, through the interconnectedness of the general
circulation, many large-scale weather patterns, and in particular large-scale
oscillations, potentially affect each other through teleconnections. ENSO is
by no means the only oscillation that animates the atmosphere. Other
oscillations exist in the Atlantic Ocean (the North Atlantic Oscillation, NAO),
in the North Pacific Ocean (the Pacific/North American pattern, PNA), and in
the South Pacific Ocean (the Pacific/South American pattern, PSA), among
others. Their mechanisms and interactions are not yet fully understood, and
they remain active research topics in the atmospheric science community.
 Summary
Time-averaged maps of weather variables (temperature, pressure, wind,
precipitation, etc.) reveal semi-permanent features of the general circulation
of the atmosphere, such as the region of strong temperature gradient found in
each hemisphere between 30° and 60° latitude.
The polar front is associated with an increase in the speed of the
westerlies from the surface up to the tropopause, where the westerly winds
are maximum. This wind speed maximum is found in both northern and
southern hemispheres, and is referred to as the jet stream. Two westerly jets
often form in each hemisphere, a polar jet and a subtropical jet, but they
sometimes merge to form a single jet stream, as is often the case over the
North Atlantic Ocean and North Pacific Ocean.
The polar front and the jet stream are strongest in the winter hemisphere,
due to the stronger temperature contrast between the tropics and the poles.
At the surface, average mean sea level pressure maps reveal semi-
permanent low pressure regions in the midlatitudes, such as the Aleutian Low
in the North Pacific Ocean and the Icelandic Low in the North Atlantic
Ocean, and semi-permanent high pressure regions in the subtropics, such as
the Pacific High and the Bermuda/Azores High in the northern hemisphere.
To account for these features, we discussed a “single-cell” model
composed of an equator-to-pole Hadley cell in each hemisphere (i.e., an
overturning vertical circulation spanning the depth of the troposphere). While
this model explains the convergence of the trade winds, as well as cloudiness
and precipitation at the Intertropical Convergence Zone (ITCZ), it does not
account for the midlatitude westerly jets.
The “three-cell” model is composed of three overturning circulations in
each hemisphere: a Hadley cell from the equator to 30° of latitude, a polar
cell spanning 60–90° latitude and an intermediate region spanning 30–60°
latitude, referred to as the Ferrel cell. The three-cell model explains the
presence of subsidence and high pressure regions in the subtropics, where
most of the major deserts are found.
The seasonal displacement of the ITCZ, combined with additional land–
ocean temperature contrasts and dynamical effects, explain the alternating
wet and dry seasons of the monsoon in India, South Asia, and sub-Saharan
Africa.
In the equatorial Pacific Ocean, an interaction between the atmosphere
and the ocean takes place that creates semi-periodic oscillations on timescales
of 2 to 7 years, referred to as El Niño/Southern Oscillation (ENSO).
Upwelling of deep, cold water off the coast of South America and along the
equator creates a region of relatively colder ocean temperatures (the so-called
cold tongue), which, in combination with the Pacific warm pool found in the
western part of the basin, causes an east–west temperature contrast and a
zonal, overturning circulation (the Walker circulation). Roughly every 2 to
7 years, a positive feedback mechanism weakens the Walker circulation,
allowing warm water from the warm pool to slosh eastward, thus shrinking or
replacing the cold tongue – a phenomenon known as El Niño. When the trade
winds become stronger, the cold tongue intensifies, and warm water moves
westward to the warm pool – a situation referred to as La Niña.
Although confined to the equatorial Pacific Ocean, ENSO influences
weather patterns worldwide through teleconnections, as do other large-scale
oscillations in other regions of the world.
CHAPTER 10
Air Masses, Fronts, and Midlatitude Cyclones
CONTENTS
10.1 Air Masses
10.2 Fronts
10.3 Midlatitude Cyclone Development
Summary

Wind, clouds, rain... Most midlatitude weather is a result of the movement of

warm tropical air poleward and cold polar air equatorward as the atmosphere
acts to reduce the strong temperature gradients in the middle latitudes. The
contrast of warm and cold air masses is most pronounced along warm and
cold fronts, where most of the weather (clouds and precipitation) is found.
We will now build a full picture of the midlatitude, or extratropical, cyclone,
the weather system in which this air mass encounter is occurring.
We now have all the pieces of the midlatitude cyclone puzzle. As we get
ready to assemble them and build a full story of cyclone development, let us
review each component carefully.
 10.1 Air Masses
When a large region of air spends several days over a land or ocean surface
with relatively uniform properties, the air acquires the properties of the
surface. In particular, it acquires the local temperature and moisture content
of this source region. This large region of air of relatively uniform
temperature and humidity is called an air mass. After some time spent over
its source region, the air mass may be set in motion by weather systems to
transport its heat and moisture content to remote locations.
It is convenient to classify the main source regions into polar (with
abbreviation P) vs. tropical (T), and continental (with prefix c) vs.
maritime (m), depending on whether the air mass spent time at high or low
latitude, and over land or ocean. Thus, by combination, four main air mass
types can be observed, as illustrated in Figure 10.1 for North America:

► Continental polar (cP): cold, dry, stable air that originates over high
latitude land areas, such as Canada or Siberia. When the air mass
forms at very high latitude, we sometimes refer to it as continental
Arctic (cA).

► Maritime polar (mP): cool, moist, and usually unstable air that
originates over high latitude oceans, such as the North Pacific or North
Atlantic Oceans. Such air masses can have a continental polar origin,
explaining their low temperatures, from where they move over the
ocean and acquire water vapor by evaporation. Recall, however, that
all is relative, as cold air contains only a small amount of water vapor
and quickly reaches saturation.
 ► Maritime tropical (mT): warm, moist, and usually unstable air that
originates over the low latitude oceans, such as the subtropical Pacific
or Atlantic Oceans.
► Continental tropical (cT): warm, dry, stable air that originates over
continental deserts.

Figure 10.1. Main air mass types contributing to midlatitude weather in

North America.

Due to the heat contrast between the equator and the poles, and as a result of
the natural tendency of the atmosphere to redistribute energy around the
planet to reduce heat imbalance, tropical air masses tend to move poleward
overall, while polar air masses tend to move equatorward overall. (In other
words, warm air moves toward colder locations and cold air moves toward
warmer locations.) This exchange of air masses takes place in the
midlatitudes in the form of a large weather system called a midlatitude
cyclone, in which the two air masses spin around a central region of low
pressure. The rotation is counterclockwise in the northern hemisphere and
clockwise in the southern hemisphere, and is said to be cyclonic. Note that
the rotation is consistent with geostrophic balance and the wind direction
around the low pressure center (Figure 10.2).

Figure 10.2. Cyclonic rotation of the cold and warm air masses, as well as
associated fronts, around the low pressure center of a midlatitude cyclone.
 10.2 Fronts
Where air masses are drawn into the circulation associated with the cyclone,
there exist regions of sharp transition where temperature and humidity
change abruptly. Temperature might increase from a low value characteristic
of a polar air mass to a high value characteristic of a tropical air mass.
Similarly, the dew point temperature might increase abruptly from a low
value characteristic of a continental air mass to a high value characteristic of
a maritime air mass. These regions of transition between air masses are called
fronts, or frontal regions, and can be associated with intense weather activity
within midlatitude cyclones. On a surface map, fronts are located on the
warm edge of the region of maximum temperature gradient.
The “front” terminology came about in Europe in the early twentieth
century, when the continent was at war and armies were fighting along a
“front line”. By analogy, the region where cold polar air was perceived to
“advance against” warmer air was called a cold front, while the region where
warm tropical air was perceived to “advance against” cooler air was called a
warm front (Figure 10.2). On a weather map, cold fronts are represented
with blue lines and blue triangles pointing in the direction of motion of the
cold air, while warm fronts are represented with red lines and red semi-circles
pointing in the direction of motion of the warm air.
 10.2.1 Stationary Fronts
Fronts develop where two air masses come in contact. This can happen in the
midlatitudes where warm tropical air lies next to cold polar air, with small
movement in the air masses. We refer to this contrast as a stationary front
and represent it on a weather map with alternating blue triangles and red
semi-circles facing in opposite directions (Figure 10.3). If the two air masses
are set in motion and start rotating cyclonically around an incipient low
pressure center, the stationary front evolves into a different structure. It is
then useful to distinguish three types of fronts.

Figure 10.3. Representation of a stationary front on a surface map.

 10.2.2 Cold Fronts
A cold front is formed where cold air moves into warmer air (Figure 10.4(a)).
Cold fronts are characterized by a sharp contrast in temperature and moisture.
The wind direction shifts abruptly from the cold to the warm air, which
corresponds to a pressure trough (Figure 10.4(b)). This sharp turn in the
isobars provides good guidance as to the position of the cold front on a
surface map. Cold fronts are usually narrow and coincide with clouds and
precipitation, typically of the convective type. Note how the cloud band is
concentrated along the cold front in Figure 10.4(a). Clouds form as warm air
is forced upward ahead of the cold front and tend to produce a band of
showers ahead of the front (Figure 10.4(c)). In some cases, clouds can also be
observed behind the surface cold front on a satellite image, in which case
they can produce precipitation behind the cold front.
Figure 10.4. Anatomy of a cold front in a midlatitude cyclone. (a)
Cyclonic rotation of cold and warm air around a low pressure center,
superimposed on satellite imagery. (b) Pressure trough and wind shift
corresponding to the cold front. (c) Vertical cross-section through the cold
front along line AB in (b).
 10.2.3 Warm Fronts
A warm front is formed where warm air moves into cooler air
(Figure 10.5(a)). The pressure trough and wind shift are usually not as
pronounced at the warm front as they were at the cold front, but the pressure
trough still provides a good indication as to the location of the warm front.
 Figure 10.5. Anatomy of a warm front in a midlatitude cyclone. (a)
Cyclonic rotation of warm and cooler air around the low pressure center,
superimposed on satellite imagery. (b) Warm sector and wind shift
corresponding to the warm front. (c) Vertical cross-section through the
warm front along line CD in (b).

Warm fronts also tend to move more slowly than cold fronts. There are
several factors that control the speed of movement of fronts, including the
evolution of the wind field, which may squeeze or pull apart isotherms in a
way that affects the speed of the front. Ignoring that process, which is more
advanced than we can discuss here, we may regard the front simply as a
moving air mass boundary. In that case, at the warm front, warm air is
ascending over the cold air mass (overrunning), which means that, at the
surface, the front moves from erosion of cold air from above; this retreat of
cold air is a slow process. In contrast, the cold air behind the cold front is
advancing, and not retreating. It wedges underneath the warm air, easily
displacing the warm air due to the density difference between these two air
masses. The sector delimited by the two fronts, called the warm sector
(Figure 10.5(b)), often shrinks with time as the air masses “wrap up” around
the low center, and as a result we say that the warm sector closes.
As the warm air rises gently over the cold air mass, it creates a more
extensive area of clouds and precipitation with a milder slope than cloud
structures encountered along a cold front. Stratiform clouds tend to form,
producing lighter and longer-lasting precipitation (Figure 10.5(c)). They form
an extensive cloud deck, which is easily discernible on satellite images, ahead
of the warm front. The combination of an extensive cloud deck with a narrow
tail of convective clouds along the cold front gives the cyclone the
appearance of a comma in satellite imagery.
 10.2.4 Occluded Fronts
After an incipient low has formed (Figure 10.6(a)), the midlatitude cyclone
develops (Figure 10.6(b)) and deepens (Figure 10.6(c)), provided the
appropriate upper-level conditions are present (see Section10.3.2). As the
cold and warm air wrap around the low pressure center, the warm sector
narrows and becomes separated from the low pressure center, now isolated in
the cold air (Figure 10.6(d)). At this mature stage, the warm air near the
cyclone has been pushed up above the surface. This is better seen on a cross-
section, such as that shown in Figure 10.7. The advancing cold air, being
typically less stable than the warmer air ahead, slides over it, producing a
frontal surface slanting forward from the occluded front at the surface. It is
not uncommon for the low pressure center to continue deepening after
occlusion has started and to produce extensive clouds and a wide range of
precipitation types.
 Figure 10.6. Four stages of the life cycle of a midlatitude cyclone: (a)
incipient, (b) deepening, (c) mature, and (d) occluded.

Figure 10.7. Frontal structure at the occluded stage. (The cross-section

refers to line AB in Figure 10.6(d).)

The configuration depicted in Figure 10.7 is also called a warm

occlusion, because the overruning is reminiscent of the overrunning of warm
air over cold air observed along a warm front. In theory, a cold occlusion
may also occur, in which the advancing cold air is more stable and the
occluded frontal surface slopes rearward with height; however, cold
occlusions are rare and very few have been documented.
 10.2.5 Large-Scale Influences on Cyclone Structure, and the T-bone
Model
The previous description of air masses and fronts is based on an early model
of cyclone development by Norwegian meteorologists at the beginning of the
twentieth century (see Appendix 10.2). While providing a generally good and
useful description of midlatitude weather, this model is not universal and
does not account for factors that affect the structure of cyclones and fronts.
For example, the structure of the jet stream has a strong influence on the life
cycle of cyclones. Changes in the jet stream winds in the east–west direction
can elongate or shrink the fronts in that direction, and this differs
systematically for cyclones developing at the eastern and western ends of
ocean basins. For example, at the western end of the basin, where the jet
stream winds increase toward the east, warm fronts tend to be more extensive
than at the eastern end, where the jet stream winds tend to decrease toward
the east. Changes in jet stream winds in the north–south direction have been
found to control how the fronts wrap up around the cyclone, with cyclones on
the north side of the jet stream getting more intense “wrapping” than those to
the south. There is also evidence that some cyclones, especially over the
ocean, lack a clear occlusion as described above, and atmospheric scientists
developed alternative models. One such alternative is the Shapiro–Keyser
model, named after Dr. Mel Shapiro and Professor Daniel Keyser, and
sometimes also referred to as the T-bone model. In this model, the cold front
tends to be at a right angle to the warm front, producing a structure
reminiscent of a T (Figure 10.8). This model better depicts situations for
strong oceanic cyclones, in which the air masses are wrapped completely
around the low pressure center, isolating a core of warm air in the middle of
colder air (“warm core seclusion”), while the warm front bends around the
low pressure center (“bent-back front”). Sting jets (especially damaging
winds descending from the midtroposphere) are sometimes associated with
the end of the bent-back front, while cloud bands can be observed to spiral
into the low pressure region on satellite images.

Figure 10.8. “T-bone” model: an alternative to the Norwegian model of

cyclone and frontal development.
 10.3 Midlatitude Cyclone Development
How do all these elements fit in the same picture? How do midlatitude
cyclones come about? To answer these questions, we need to understand the
full development of a cyclone, from the initial disturbance to decay, as a
three-dimensional weather system spanning the entire depth of the
troposphere.
 10.3.1 The Life Cycle of a Midlatitude Cyclone
We first need to return to the notion of heat imbalance. With a constant gain
of heat near the equator and loss of heat near the poles, the atmosphere would
not be a balanced system in the absence of a mechanism to redistribute heat
from the source region to the sink region. The natural tendency of any system
out of balance is to seek to return to equilibrium, and, in the case of the
atmosphere, midlatitude cyclones are a primary mechanism that maintains
equilibrium by moving heat poleward. Since the heat imbalance is constantly
recreated by the uneven absorption of solar radiation around Earth, there is a
persistent driving force creating midlatitude cyclones.
As we discussed in Chapter 9 (Sections 9.2 and 9.3), due to Earth’s
rotation, the poleward redistribution of heat is not accomplished by a single
Hadley cell, but rather takes place in midlatitude cyclones and the exchange
of air masses that occurs with these features. Because of Earth’s rotation, the
westerly jet streams are intrinsically unstable: they become wavy, and
eventually break down into individual cyclones. Each cyclone goes through a
series of phases generally described as the life cycle of midlatitude cyclones.
Most often, these cyclones originate from disturbances in the westerly
jet streams, troughs of low pressure (low heights), called short-wave troughs
(or “short waves”), that cause the pressure to fall at the surface, as we will
discuss later in this chapter. Often there is no pre-existing surface front when
this happens, and fronts develop along with the cyclone. In some cases,
however, the life cycle begins with a wave cyclone on a pre-existing surface
front, which is the sequence described by the Norwegian cyclone model. For
the sake of simplicity, we proceed with a description of the Norwegian
model, but note that the case without an initial front is described by stages (2)
through (5) after the formation of a surface low associated with an upper-
level disturbance:
(1) Wave cyclone: A weak pressure disturbance appears as warm air
starts to move poleward and upward, while cold air starts moving
equatorward and downward (Figure 10.6(a)). Convergence and upward
motion cause a patch of clouds to form poleward of the developing warm
front, which is often the first hint of the developing cyclone on satellite
imagery.
(2) Deepening cyclone: The central pressure falls and the low is now
delimited by several closed isobars (Figure 10.6(b)). The cold and warm
fronts are far apart and the warm sector is broad.
(3) Mature stage: The cyclone is fully developed with well-defined
cold and warm fronts and a relatively low central pressure (Figure 10.6(c)). It
has an easily recognizable comma shape on satellite imagery.
(4) Occluded stage: Cold air wraps around the low pressure center,
which becomes isolated in the cold air (Figure 10.6d). The warm sector is
increasingly separated from the low pressure center, but a vestigial axis of
relatively warm air remains along the occluded front.
(5) Decaying stage: The cyclone decays as the air column fills in and
the central pressure increases. Clouds persist but gradually dissipate as they
wrap around the low center.
This life cycle can take from about 2 to 6 days, depending on the time of
year, the strength of the midlatitude temperature gradient, and environmental
conditions. The cyclone, being embedded in a mean midlatitude flow that is
westerly overall, moves eastward while going through the life cycle. In the
southern hemisphere, midlatitude cyclones also move eastward, but their
structure is a mirror image of their northern hemisphere counterparts with
respect to the equator (see Appendix 10.1).
This model of cyclone development was first proposed by
meteorologists in Bergen, Norway, at the beginning of the twentieth century,
and is usually referred to as the Norwegian model from the Bergen school, or
the Bjerknes model, after Jacob Bjerknes (1897–1975), who wrote a seminal
article about polar front theory in 1922. As general as it is, this model gives
us some level of predictability as to how a pressure disturbance will evolve
over the days following initial development. It does not, however, tell us
where a pressure disturbance will appear and which of these pressure
disturbances will evolve into fully developed cyclones. To answer these
questions, we need to make some additional observations and return to the
upper levels.
 10.3.2 Vertical Structure of Cyclones
A comparison of the pressure distribution at various levels of the atmosphere
is very instructive. It reveals that the pressure minimum characteristic of a
deepening cyclone is not found at the same location on different pressure
levels. The upper-level minimum is found westward of the surface minimum
during development. In other words, if we picture the cyclone as a spinning
vortex with a core of low pressure, the core of a deepening cyclone is not
vertical, but tilts westward with height (Figure 10.9). By contrast, the core of
a cyclone that has reached the decaying stage is vertically aligned, or tilted
eastward with height. This suggests that we need to investigate the upper-
level flow, and in particular the alignment of upper-level troughs with surface
lows, to understand the timing and location of cyclone development.

Figure 10.9. Deepening lows tilt westward with height (a), whereas
occluded lows are vertically aligned (b).

Figure 10.10 shows a schematic representation of a 300 hPa trough

flanked by two ridges. You might recall from Chapter 8 that the flow is
supergeostrophic around a ridge and subgeostrophic around a trough, as
shown in Figure 10.10 by arrows of different sizes. If we now follow the air
flow along a “channel” delimited by two isobars, we note that the air enters
the channel with greater speed, and decelerates as it reaches the trough. This
causes a convergence of air between the ridge and the trough (i.e., there is
more air entering the channel than leaving it). Conversely, further
downstream, east of the trough, the air accelerates as it flows into the next
ridge. This causes a divergence of air (i.e., there is more air leaving the
channel than entering it). This dual pattern of convergence–divergence has
two major effects:

Figure 10.10. Areas of convergence and divergence within a 300 hPa

trough.

(1) Net convergence adds air to the air column, which means that the
pressure underneath rises. Conversely, net divergence removes air from the
air column, which means that the pressure underneath falls.
(2) For reasons similar to those invoked in earlier chapters, air cannot
accumulate indefinitely where there is convergence and must take a new
direction. Since we are at 300 hPa, at the tropopause, with a strong
temperature inversion above and a very stable stratosphere preventing
upward motion, the converging air cannot rise, and so it sinks downward.
Therefore, the region of convergence is a region of subsidence as well as
rising surface pressure. Conversely, diverging air is replaced by air rising
from below, and the region of divergence is a region of upward motion as
well as falling surface pressure.
There we have a good hint as to why low pressure tilts westward with
height for developing cyclones, and why the precursors for development are
troughs in the jet stream. If the surface cyclone is to deepen, it needs to be
under the region of upper-level divergence, east of the upper-level trough.
And that is indeed what we observe, as shown schematically in Figure 10.11.
As the upper-level trough, moving from west to east, approaches the surface
low from the west, the region of upper-level divergence is ideally positioned
to cause upward motion (producing clouds) and decrease the central pressure
of the surface low. In other words, it causes the surface low to deepen.

Figure 10.11. Superposition of upper-level convergence and divergence

patterns over a surface anticyclone and a deepening cyclone.
 Surface winds are consistent with this picture. Recall that, due to
friction, surface winds do not blow parallel to the isobars, as suggested by
geostrophic balance, but flow toward lower pressure into the low, which
results in surface convergence. Thus, in concert with the upper-level flow,
surface winds tend to converge and to reinforce upward motion.
West of the upper-level trough, the inverse situation can be observed.
Upper-level convergence leads to subsidence, which coincides with the
surface divergence typically observed in anticyclones, as surface winds tend
to flow away from the high pressure (toward lower pressure).
As the cyclone deepens and matures, the upper-level trough moves over
the surface low (Figure 10.12) and the region of upper-level divergence,
which has moved further east, can therefore no longer sustain upward motion
and decreasing pressure in the cyclone. Intensification stops, and the cyclone
begins to decay when the upper-level trough continues eastward and the tilt
reverses from what it was during cyclone development. Since the converging
air at the surface no longer has an upper-level outlet, convergence adds air
mass to the air column and the surface pressure increases. We say that “the
low fills in”. Friction gradually spins down the surface winds.
 Figure 10.12. Displacement of the upper-level trough over the surface low
as the midlatitude cyclone becomes occluded and decays.

Note that upper-level divergence is necessary for a midlatitude cyclone

to deepen and intensify, but it is not sufficient. If the atmosphere is too stable,
it resists upward motion, even in the presence of upper-level divergence.
Therefore, an additional condition for cyclone development is a relatively
weak atmospheric stability in the region where the incipient low is forming.
This helps explain why we find cyclones developing preferentially over the
oceans, where during winter the relatively warmer water heats the air near the
surface, which produces weak stability (see Section 3.4).
In the end, from wave cyclone to decay, the cyclone has gone through a
full life cycle and has effectively exchanged warm and cold air across the
midlatitude temperature gradient, which results in a net flow of heat toward
the poles.
 10.3.3 The February 2014 Cyclone
We can now verify how all these theoretical concepts apply to a real case: the
February 2014 cyclone we have been studying throughout the book.
On February 20, 2014, at 06:00 UTC, a surface low pressure center has
formed over the Texas panhandle. A small cyclonic circulation has started to
develop, with small cold and warm fronts (Figure 10.13). We have a nice
example of an incipient wave. Another area of low pressure is found over
North Dakota, with one center of low pressure along the border with
Minnesota, in the location favored for development to the east of a weak
trough at 500 hPa. Some snow is found north of our incipient wave along
with the weak system over North Dakota. The infrared satellite image
suggests little cloudiness along the cold front at this stage, and some
stratiform clouds north of the warm front, corresponding to the rain showing
in the surface weather station measurements (Figure 10.13(d)). The surface
low is ideally located underneath an upper-level trough, east of the trough
axis, where we expect upper-level divergence to enhance upward motion and
deepen the surface low, as we just discussed.
Figure 10.13. Incipient wave on February 20, 2014, at 06:00 UTC. (Top
left) 500 hPa heights with winds and wind speed. The position of the
 surface low is indicated by a black dot. (Bottom left) Surface pressure and
temperature. (Top right) Infrared satellite image with surface isobars and
the position of the surface low, with cold and warm fronts. (Bottom right)
Surface pressure, surface winds, and precipitation. (Note that there is more
rain and snow in Canada than suggested in the figure because there are
fewer weather stations in Canada than in the United States. See
Figure 1.21.)

Twelve hours later (Figure 10.14), the main cyclone has indeed
deepened significantly. We can see more closed isobars, and stronger winds
(i.e., tighter isobars). The fronts are better established, with a clearly defined
warm sector, and a sharp turn of the wind at both the cold and the warm
fronts (i.e., a sharper turn, or kink, in the isobars). Clouds are more extensive,
with rain reported south of the Great Lakes, and an extensive area of snow
over Minnesota, Iowa, and Missouri, where cool air from the east ascends
over very cold air from Canada. Snow is observed at the surface even though
temperatures are above freezing in many locations. Recall that snowflakes
can fall through about 300 m of above-freezing air before completely
melting. The surface low is still ideally located underneath and east of the
upper-level trough, which will lead to further deepening. The second low to
the north has become subsumed into the main cyclone, but still appears in the
surface isobars as a trough extending north from the main cyclone into
Minnesota.
Figure 10.14. Deepening cyclone on February 20, 2014, at 18:00 UTC.
 Details are same as for Figure 10.13.

On February 21, 2014, at 06:00 UTC, the cyclone has reached the
mature phase (Figure 10.15). The central pressure has dropped below 980
hPa, and an occluded front has started forming where the cyclone is
separating from the warm sector. Notice the axis of relatively warm air that
extends along the occluded front from the low center to the “triple point,”
which is defined by the point of intersection of the warm, cold, and occluded
fronts. A thick band of clouds is aligned with the cold front, producing rain
ahead of the front, consistent with our cold front picture, in which the cold air
lifts the warmer air ahead and causes the formation of clouds and
precipitation (Figure 10.4). Rain continues ahead of the warm front, where air
from the warm sector is overrunning colder air to the north, as well as snow
west and north of the Great Lakes. The surface low is nearly underneath the
upper-level trough axis. Since it is leaving the area of upper-level divergence,
we anticipate the end of the developing phase of the cyclone.
Figure 10.15. Mature cyclone on February 21, 2014, at 06:00 UTC.
Details are same as for Figure 10.13.
 Twelve hours later (Figure 10.16) we find that a closed low has formed
at 500 hPa, and that this upper-level low and the surface low are now stacked
on top of each other, as described in Figures 10.9(b) and 10.12. The central
pressure of the surface cyclone is still below 980 hPa, which is similar to
what it was 12 hours previously. Surface winds now circle completely around
the low, along with cold air and snow over the Great Lakes. The cyclone is
now fully occluded, with an axis of warm air connecting the low center to the
warm sector at the triple point, and, from then on, the cold front will continue
its eastward course detached from the main low.
Figure 10.16. Decaying cyclone on February 21, 2014, at 18:00 UTC.
 Details are same as for Figure 10.13.

This cyclone was one of the strongest cyclones to affect the area during
the winter of 2014. It produced everything from heavy rain, to sleet, to
freezing rain, and, of course, snow. The strong pressure gradients produced
very strong winds, with up to 65 mph wind gusts. It spawned several
thunderstorms ahead of the cold front, and even a couple of tornadoes were
reported in Illinois.
 10.3.4 Where do Cyclones Form?
We have established that midlatitude cyclones form on the midlatitude
temperature gradient, in a region of relatively weak static stability downwind
of a trough in the westerly jet stream. There are certain regions, however,
where cyclogenesis (i.e., the genesis of cyclones) is enhanced: regions where
the combination of factors leading to cyclone intensification is more optimal
than others. You might recall, for example, the presence of semi-permanent
upper-level troughs east of Asia and east of North America in our mean
upper-level pressure maps (Figure 9.1). Even though the upper-level flow
changes and evolves from one day to the next, and there is no stationary
trough on a day-to-day basis, the fact that there exists a trough on average
reveals that the occurrence of upper-level troughs is more pronounced in
those two regions. It is largely due to the presence of major mountain ranges
(Tibetan Plateau and Rocky Mountains) and land–ocean temperature
contrasts associated with the Eurasian and North American continents.
Since ocean waters are relatively warm in these regions, the atmosphere
is also relatively warm and moist, and therefore only weakly stable. This
combination of factors makes these two regions prone to midlatitude cyclone
development. Cyclones tend to form over the Western Pacific and West
Atlantic and cross the ocean basin as they deepen and mature. We refer to
these two regions of enhanced cyclonic activity as the storm tracks
(Figure 10.17(a)).
Figure 10.17. 500 hPa heights overlaid on 7 years of midlatitude cyclone
 trajectories (in blue). (a) In the northern hemisphere, storms tend to form
southeast of the semi-permanent upper-level troughs (dashed lines) and to
follow the Pacific and Atlantic storm tracks. (b) In the southern
hemisphere, storms form all around Antarctica.

In the southern hemisphere, land masses are not as prominent, and the
Southern Ocean offers a continuous channel around Antarctica. Therefore,
cyclogenesis is more uniformly distributed, with a relatively higher incidence
of cyclones east of South Africa, southeast Australia, and the southern tip of
South America (Figure 10.17(b)).
Cyclogenesis is also enhanced in the lee of mountain barriers such as the
Rockies in North America and the Andes in South America, where prevailing
westerly winds produce downslope warming and a trough of low pressure at
the surface. In fact, our February 2014 cyclone was one such example of a
low that formed in the lee of the Rockies.
Similarly, the low pressure trough corresponding to the cold front of a
mature or decaying cyclone can serve as the environment for the birth of a
new cyclone. The pressure is already locally lower and the trough has the
characteristics of a front: if an upper-level trough comes along and produces
divergence above the trough, it can produce an incipient wave, called a
frontal wave. Given the right conditions, this incipient wave can grow into a
full-fledged cyclone. We often see this behavior at the eastern end of the
storm tracks, where mature cyclones leave long trailing cold fronts, which
provide a fertile environment for upper-level troughs to begin the process
anew.
In the end, knowing these areas of enhanced cyclogenesis allows us to
recognize familiar patterns and to forecast better the formation and
development of midlatitude cyclones.
 Summary
Midlatitude cyclones form on temperature contrasts near the jet stream.
Winds circulating around the cyclone distort and concentrate the temperature
contrasts into fronts. The frontal region where cold air moves into warmer air
is called a cold front and is characterized by a sharp temperature and
moisture contrast along a pressure trough, a shift in wind direction, and
usually a band of clouds and precipitation, often of the convective type.
The frontal region where warm air moves into cooler air is called a
warm front and is also characterized by a pressure trough and a shift in wind
direction, although usually not as pronounced as for the cold front. As the
warm air rises gently over the cold air mass (overrunning), stratiform clouds
tend to form, producing lighter and longer-lasting precipitation.
As the fronts become stretched and elongated, the low center separates
from the warm sector, creating an occluded front. An axis of relatively warm
air is found along the occluded front, connecting the low center to the triple
point where all three fronts meet. Aloft, air that was behind the cold front has
moved over the warm front due to the fact that the air below the warm front is
relatively more stable (and therefore the air behind the cold front is easier to
lift).
The midlatitude cyclone goes through a typical life cycle by which:

► a first wave cyclone, or incipient low, forms;

► the low deepens (i.e., the central pressure falls), and the cyclone
intensifies (stronger winds, more clouds, and precipitation);
 ► it reaches a mature stage where it is fully developed and the rate of
deepening often begins to slow;
► the low separates from the warm sector, the low pressure center
migrates back into the cold air under the upper-level trough, and an
occluded front forms; and
► the cyclone eventually decays (i.e., the central pressure increases).

When a surface low pressure center is located east of an upper-level trough

axis, upper-level divergence makes the surface pressure fall and the cyclone
intensify. Conversely, upper-level convergence causes subsidence and the
development of high pressure at the surface, to the west of the upper-level
trough axis. When the upper-level trough moves over the surface low,
intensification stops and the cyclone may begin to decay.

Appendix 10.1 Southern Hemisphere Midlatitude Cyclones

Recall that, in the southern hemisphere, the Coriolis force deflects air parcels
to the left, our rules of geostrophic motion are reversed, and cyclones rotate
clockwise. Therefore, we can flip all the schematics in Figure 10.6 upside
down: they are really a mirror image of the northern hemisphere case with
respect to the equator (see Figure 10.18). However, remember that the
southern hemisphere jet stream is also composed of westerly winds (Figure
10.19) – do not flip the direction of the jet stream! Moreover, the upper-level
waves that make the jet stream undulate also propagate eastward in the
westerly flow; therefore, midlatitude cyclones overall progress from west to
east, as in the northern hemisphere.
 Figure 10.18. Life cycle of a southern hemisphere midlatitude cyclone:
incipient (a), deepening (b), mature (c), and occluded (d) stages.

Figure 10.19. Example of a southern hemisphere midlatitude cyclone on

February 20, 2014, at 18:00 UTC. (a) 500 hPa heights and winds. (b)
Infrared image with surface fronts and general air mass circulation.

Appendix 10.2 The Bergen School of Meteorology

The Norwegian model of cyclone development was developed at the

beginning of the twentieth century by a group of meteorologists working
under the supervision of Vilhelm Bjerknes (Figure 10.20) and later his son,
Jacob Bjerknes (Figure 10.21), who wrote the seminal paper “On the
structure of moving cyclones” in 1919. Father and son laid the foundation for
the Bergen school of meteorology, a group of meteorologists based at the
University of Bergen, Norway, and developed a school of thought that has
inspired much of modern meteorology.

Figure 10.20. Vilhelm Bjerknes (1862–1951).

Figure 10.21. Jacob Bjerknes (1897–1975).

 Several of the Bergen school scientists have made significant
contributions to the field of meteorology. Tor Bergeron (Figure 10.22)
proposed the mechanism explaining the growth of ice crystals at the expense
of liquid water droplets in cold clouds, known as the Bergeron process (see
Chapter 7). Sverre Petterssen (1898–1974) is remembered for his D-Day
forecast during the Second World War, and encouraging General Dwight
Eisenhower to postpone the landing in Normandy from June 5 to June 6,
1944. Carl-Gustaf Rossby (Figure 10.23) founded the Chicago school of
meteorology and advanced our understanding of large-scale atmospheric
motions. In particular, he identified and characterized the undulations of the
jet stream, now referred to as Rossby waves.

Figure 10.22. Tor Bergeron (1891–1977).

Figure 10.23. Carl-Gustaf Rossby (1898–1957).
CHAPTER 11
Thunderstorms and Tornadoes
CONTENTS
11.1 Ordinary Thunderstorm
11.2 Severe Thunderstorm
11.3 Lightning and Thunder
11.4 Supercells
11.5 Tornadoes
Summary

Upwardly buoyant air parcels can produce powerful thunderstorms towering

cumulonimbus clouds producing rain, hail, lightning, and thunder. When
these thunderstorms rotate, they can also spawn tornadoes and cause some of
the most severe weather on Earth. We will now describe the different stages
of thunderstorm development, the formation of lightning and thunder, and
will explain the circumstances and mechanism by which tornadoes can form
at the base of a supercell thunderstorm.
Mild convection often produces innocuous fair-weather cumulus clouds. We
recall that air parcels, if they are warmer than the environment, experience an
upward buoyancy force that can lift them above the lifting condensation level
(LCL), where their temperature drops below the dew point temperature,
causing condensation and the formation of cloud droplets. When atmosphere
is quite unstable, however, continued convection can produce tall cumuliform
clouds along with hail and rain in the form of heavy showers. When the wind
shear (i.e., the change in wind speed with height) is weak, what are referred
to as ordinary storms develop. Such storms generally are not accompanied by
severe weather, but exceptions can occur in environments that are extremely
unstable. When significant wind shear is present, highly organized storms can
develop that can take the form of supercells or squall lines and produce more
extensive severe weather. We will now describe each type of thunderstorm in
detail.
 11.1 Ordinary Thunderstorm
The development of a thunderstorm is usually preceded by the formation of
one or several cumulus clouds that grow and dissipate, moistening the air in
the process, and prime the atmosphere for subsequent storm development
(Figure 11.1). The convective process being quite turbulent, the air parcels
delimiting the cloud and containing a high level of moisture and cloud
droplets are constantly mixing with drier air parcels outside the cloud. This
mixing of cloud parcels with dry air is called entrainment. The entrainment
of dry air into the cloud lowers the dew point temperature and causes some of
the cloud droplets to evaporate. Thus, in the end, moisture has been
transported upward by convection, from the surface to a higher region of the
atmosphere, where it has increased the relative humidity. The first cumulus
cloud might dissipate, but its net effect has been to make the atmosphere
more moist, which reduces the amount of evaporation the next cloud
experiences from entrainment. In summer, several such cumulus clouds
might form throughout the afternoon, each one getting larger than its
predecessor as the ground warms (Figure 11.2(a)).
 Figure 11.1. Moistening of the atmosphere by formation and evaporation
of cumulus clouds.

Figure 11.2. Developmental stages of an ordinary thunderstorm.

At some point, the cumuliform cloud might become tall enough that
collision and coalescence of cloud droplets within the cloud start producing
precipitation, often made of big raindrops (Figure 11.2(b)). The whole
process is usually relatively fast, which produces convective precipitation in
the form of concentrated showers, sometimes heavy (Figure 11.3). If the
cloud reaches altitudes where temperatures are below about –10 °C, ice
crystals form, which will be critical for the physical processes that lead to the
formation of lightning (see Section 11.3). When the cumulus cloud reaches
the stable stratosphere, or a stable layer of the troposphere, it flattens out
against that stable layer. (Recall that the stratosphere is very stable and that
the temperature inversion prevents further upward motion.) The cloud has
turned into a cumulonimbus, and its flattened top gives it the familiar
appearance of an anvil (Figure 11.4). Strong updrafts might overshoot the
level of neutral buoyancy but are quickly brought back down by the stability
of the layer they overshoot into.

Figure 11.3. Concentrated rain shower under a growing thunderstorm.

 Figure 11.4. The anvil of a cumulonimbus cloud.

As precipitation falls in subsaturated air below the cloud base, it

evaporates. We recall that evaporation requires latent heat that, being taken
from the environment, has a cooling effect. Cold air being more dense, it
tends to sink back down toward the surface, underneath the cloud. This
creates cold downdrafts which compete with the convective updrafts. Friction
between falling raindrops and the surrounding air also contributes to dragging
the air downward and enhances the downdrafts.
When it reaches the surface, the sinking cold air forms a cold pool and
tends to spread outward, creating a gust of cold wind that you might have
experienced if you have been in the vicinity of a thunderstorm in the summer.
Since full-blown thunderstorms often develop at the end of a warm afternoon,
the cold temperatures accompanying the wind gust are often in sharp contrast
with the summer temperatures.
When the cold pool of air becomes so extensive that it cuts off the
supply of warm, humid air to the storm, the thunderstorm decays and
dissipates (see Figure 11.2(c)). The whole formation and dissipation cycle
takes on the order of an hour.
Meanwhile, the sinking and spreading cold air advances against warmer
air, much in the same way as a cold air mass creates a cold front when it
moves into a warm air mass. The boundary between the rushing cold air and
the warmer surrounding air is called a gust front. Similarly to a cold front, it
wedges underneath the warm air and forces upward motion, which can cause
the formation of small cumulus clouds. In some cases, these cumulus clouds
themselves grow and develop into a new thunderstorm. It is not uncommon to
observe thunderstorms developing in series by triggering each other through
cold pool expansion.
 11.2 Severe Thunderstorm
An ordinary thunderstorm is eventually defeated by its own downdrafts,
which kill convection and remove the initial source of energy. When the
cumulonimbus cloud is sheared, however, downdrafts are separated from the
updrafts and can coexist symbiotically for extended periods of time,
enhancing each other’s intensity and the likelihood of severe weather. This
can happen when the wind speed increases with height and the upper-level
winds displace the top of the cloud sideways (Figure 11.5). The thunderstorm
can then turn into a much more long-lived and severe weather system.

Figure 11.5. In a severe thunderstorm, the cumulonimbus is sheared

horizontally and the downdrafts become separated from the updrafts.

Where the anvil extends out, small precipitation particles can be blown
outward with the updraft and slowly settle out of the cloud, occasionally
producing round cloud pockets hanging down from the anvil. These
protruding cloud elements are called mammatus clouds; they form as the
dense air (cooled by evaporation), entrains cloud particles from the anvil
downward (Figure 11.6). This is a rare example of the case where subsiding
air can remain saturated, because precipitation particles are available to
maintain saturation through evaporation.

Figure 11.6. Mammatus clouds forming on the underside of a

thunderstorm anvil.

Cool air returns to the surface in the form of downdrafts further

downstream from the updrafts, and convection can continue unhindered.
Thus, severe thunderstorms can last more than 3 hours and produce
extremely intense precipitation, including large raindrops, hail, as well as
lightning and thunder. The wind gusts can reach 50 knots or more, and
isolated, short-lived, but very intense downdrafts called microbursts can
reach up to 100 knots. These downdrafts pose a severe hazard for aircrafts.
Since the whole system is in motion (from left to right in Figure 11.5),
someone on the ground will first observe a sudden drop in temperature
corresponding to the gust front. The passage of the gust front is usually also
accompanied by an increase in wind speed, and possibly also a change in
wind direction. After the gust front has passed, rain and possibly hail come,
and possibly lightning and thunder (although these can also be evident long
before the gust front, especially at night).
In summary, the conditions for the growth of a thunderstorm are:

► a conditionally unstable atmosphere (e.g., cold air aloft, over a warm

and moist lower troposphere)
► a lifting mechanism (e.g., convergence along a front, or free
convection forced by hot ground)

with the addition of:

► strong vertical wind shear
for a severe thunderstorm.
Sometimes, thunderstorms organize into a mesoscale convective system
(MCS), i.e., a self-organized system made of multiple thunderstorms
covering several hundred kilometers. MCSs can last several hours and
produce large amounts of precipitation. New cumulonimbus clouds develop
along the leading edge of the cold pool created by the downdrafts and the
sinking cold air, sustaining the MCS. As the cold pool expands and the cold
air displaces warmer air aloft, a shelf cloud can form over the gust front
(Figure 11.7).
 Figure 11.7. Shelf cloud forming over an advancing gust front.

When convection is triggered by an advancing cold front, as often

happens in the southern and eastern United States, thunderstorms may
become organized along the front and form a frontal squall line, provided
that wind shear is again available to separate the updrafts and downdrafts.
They are easily identified in infrared satellite imagery as a thin line of tall
(white) clouds, typically along the tail of a cold front associated with a
midlatitude cyclone, and in radar imagery as a line of heavy precipitation (see
Figure 2.12).
 11.3 Lightning and Thunder
Cumulonimbus are often associated with lightning and thunder, when
convection is violent and precipitation intense. Lightning forms when an
electric field builds up an electric differential that exceeds the capacity of the
air, i.e., an electric charge difference between two regions of the atmosphere
that grows so large that a violent discharge occurs in the form of a spark
(lightning), which reduces the electrical charge difference. Lightning can
occur between two clouds, between parts of a cloud (in which case the cloud
seems to light up from inside), or between a cloud and the ground, which is
the type of lightning stroke we usually observe, because it is clearly visible
and it can impact us directly – even though cloud-to-ground strokes account
for less than 20% of all lightning strokes.
Charge separation occurs when larger ice particles, such as graupel and
hailstones, fall through the cloud and collide with smaller ice particles
ascending in the updrafts. The larger falling particles become negatively
charged while the smaller lofted particles become positively charged. This
tends to concentrate negative charges in the lower part of the cloud and
positive charges in the upper part of the cloud (Figure 11.8(a)).
Comparatively, the ground is charged positively. And when the charge
difference between the cloud and the ground becomes too great for the air to
sustain, negative charges start moving downward toward the ground.
 Figure 11.8. A sequence of steps leading to cloud-to-ground lightning.

They do not do so all at once, however. They penetrate through the air in
successive pokes, advancing ever further down along a path, extending the
path by a discrete step at each new poke. This advancing discharge is called a
stepped leader and it moves extremely rapidly – on the order of
microseconds – choosing the path of least resistance (Figure 11.8(b)). As it
approaches the ground, it induces a concentration of positive charges in the
“closest”elements of the landscape, i.e., elements that stand out and reach up
in the atmosphere, in particular pointy objects such as trees, electric poles,
and metallic objects such as found at the top of church steeples
(Figure 11.8(c)). That is why it is recommended to stay away from trees
during a thunderstorm.
The positive charges eventually move up from these protruding elements
to meet the stepped leader, connecting and opening the path for further
electrons to flow down: this subsequent flow of electrons is the real lightning
stroke, a much larger discharge of electricity, called the return stroke, that
heats up the air to an incredible 30 000 K and shows as a bright white path of
light (Figure 11.8(d)). [Recall the relationship between temperature and the
emission of radiation – here, the emission of visibe light by heated air
molecules.]
The rapid increase in temperature in the narrow channel of air is
accompanied by a sudden increase in pressure. In other circumstances, the air
would adjust by expanding as the pressure increases, as air parcels do when
they rise or subside in the atmosphere. The discharge of electricity is so
violent, however, and the build-up of pressure so rapid, that the channel
expansion creates a shock wave. The resulting sound wave travels outward
from the channel, and we eventually hear it as thunder.
Because the flash of light created by the return stroke travels at the
speed of light (i.e., 300 000 000 m/s), we see it almost instantaneously. The
sound wave, however, travels at the speed of sound, which is closer to 340
m/s, or about one kilometer in three seconds (i.e., about a mile in five
seconds). Therefore, we usually hear thunder some time after we see the
return stroke, and, by counting how many seconds have elapsed between the
two of them, we can estimate the distance between us and the cumulonimbus.
For example, if 6 s have elapsed, we know the thunderstorm is 2 km away. If,
some time later, a new stroke produces thunder only 5 s later, we know that
the thunderstorm, or at least the lightning-producing part of the cloud system,
is in motion, and it is approaching.
Our lightning story does not end with the return stroke. Once the
channel is established, more electricity can flow through it, if the cloud still
contains pockets of negative charges (Figure 11.8(e)). The new flow of
electrons traveling the established channel is called the dart leader and
causes more return strokes, all happening in less than a second. That is why
lightning often appears to flash very rapidly and repeatedly when it occurs.
These new return strokes all borrow the main channel, the one path that was
selected when it attached to a positively charged element at the surface
(Figure 11.8(f)). That is at the expense of other paths that did not connect to
the ground – failed attempts, as it were, as the stepped leader was probing
and penetrating down through the air. These branches can be seen in
Figure 11.9, surrounding the much brighter channel traveled by the return
stroke.

Figure 11.9. Bright white return strokes from three different areas of
lightning discharge.
 11.4 Supercells
In certain circumstances, intense supercell storms can develop. Supercells
are long-lived thunderstorms with rotating updrafts, sometimes embedded in
a larger weather system, such as a midlatitude cyclone. Supercells might be
triggered and sustained by an advancing cold front, for example. The rotating
dynamics induce a lower pressure core within the updraft called a
mesocyclone.
Supercells acquire their rotation from the vertical wind shear associated
with the increase in wind speed, and/or change in wind direction, with height.
You can convince yourself that wind shear creates rotation by imagining a
spinning wheel embedded in the wind shear, depicted schematically in
Figure 11.10. Since the wind is blowing faster at the top than at the bottom of
the wheel, the wheel will start spinning clockwise (in our example). By a
slightly bigger stretch of imagination, we can imagine air parcels spinning
clockwise, and in fact entire horizontal “rolls” or tubes of air (going into the
page) spinning clockwise, as if rolling a carpet (Figure 11.11(a)). So far,
these rolls are parallel to the ground, but, in the presence of a strong
convective updraft, parts of these tubes are pushed upward, distorting the roll
as shown in Figure 11.11(b). Two vertical segments appear on each side of
the convective center, spinning in opposite directions.
 Figure 11.10. Vertical wind shear creates horizontal rotation.

Figure 11.11. Transformation of horizontal rotation (a) into vertical

rotation (b) in the form of two counter-rotating vortices, as convection
distorts the rotating tube.

The fate of these two counter-rotating vortices is better captured in a

bird’s eye view of the evolving convective cell (Figure 11.12). The cell splits
into two individual rotating systems. The anticyclonic vortex, moving to the
left, commonly weakens and decays, while the cyclonic vortex moves to the
right and intensifies, due to an enhanced buoyancy associated with the
development of low pressure in the right-moving part of the storm. In fact,
this part of the storm turns into a mesocyclone characterized by a well-
defined low pressure center.

Figure 11.12. Convective cell splitting into a decaying anticyclonic vortex

and a growing mesocyclone.

As this new system intensifies, updrafts and downdrafts become

organized (Figure 11.13). As warm, moist air swirls into the convective cell
to form the rotating updraft that, counter to intuition, is so strong that it is
relatively free of precipitation, downdrafts bring cool, dense air back down to
the surface, along with precipitation, at the front and rear of the cell. The cool
air spreads out and forms two gust fronts, on the rear and forward flanks of
the convective cell. Rotation can develop very near to the ground as a result
of the outflow spreading beneath the updraft, and occasionally the rotation
can be intensified to form a tornado, a fast-spinning vortex of much smaller
scale than the thunderstorm itself. This is most likely to happen when the
outflow is not excessively cold, and when the overlying thunderstorm updraft
is particularly intense within about 1 km of the surface.

Figure 11.13. Structure of a tornadic supercell, with preferred location (the

“hook”) for the formation of a tornado (T).

Within the rear-flank downdraft there often appears an area of

precipitation that wraps around the incoming warm air and takes the form of
a hook in radar images. Meteorologists use the formation of a hook in radar
imagery as an indicator of the location where a tornado might form. While
hook echoes are a good indicator of rotation associated with a supercell, only
a subset of supercells produce tornadoes (roughly 20%).
11.5 Tornadoes
 11.5.1 Description
Tornadoes are rapidly rotating columns of air with extremely low central
pressure that suck small particles, dust, sand, dirt, debris, and sometimes
larger objects into their core. The debris rises and spins around with the wind,
which often gives the tornadoes a dark appearance (Figure 11.14).

Figure 11.14. Dirt and debris stirred by an F3 tornado.

Tornadoes vary in size from tens to hundreds of meters, but are always
much smaller than a thunderstorm. They are primarily spawned by supercells.
The intensity of a tornado is determined by the strength of its winds using a
scale designed by Tetsuya Fujita in the 1970s (see Box 11.1). Tornadic winds
can reach more than 250 knots. The tornado is normally a single vortex, but
may also spawn suction vortices, which are smaller and rotate around the
main vortex.

Box 11.1. The Fujita scale

 Tetsuya Fujita (1920–1998) designed a tornado intensity scale based
on damage surveys in the 1970s that was soon adopted by the entire
atmospheric science community (and was recently replaced by the
Enhanced Fujita Scale). He ranked the intensity of tornadoes based on
the destruction caused by the winds and using the strength of the
winds as follows:

F0 35–62 knots

F1 63–97 knots

F2 98–136 knots

F3 137–179 knots

F4 180–226 knots

F5 227–276 + knots

In the USA, where 75% of the world’s tornadoes occur, the strongest
tornadoes tend to occur from April to June, and in so-called Tornado Alley
(see Section 11.5.3).
 11.5.2 Tornado Development
A precursor to the tornado is usually the formation of a wall cloud hanging
from the cloud base and slowly rotating around the mesocyclone
(Figure 11.15). The wall cloud is lower than the cloud base because the air
entering the wall cloud is cool and humid. A funnel cloud then emerges from
the wall cloud or the base of the thunderstorm (Figure 11.16). As the inflow
of air accelerates by conservation of angular momentum and the vortex spins
up – recall the ice skater spinning faster as she pulls her arms and legs inward
– the pressure decreases inside the vortex and the funnel cloud extends
downward, sometimes down to the surface (Figure 11.17). Meanwhile, the air
is rising rapidly in the vortex, due to intense “suction” from above, near the
cloud base. The vortex is typically elongated between the base of the cloud
and the ground. This stretching concentrates the rotation (technically, angular
momentum), much like water approaching a bathtub drain can spin up a
stronger vortex. Thus, the tornado can continue to intensify for minutes, and
sometimes tens of minutes, all the while moving with the convective cloud,
leaving a swath of destruction in its wake.
Figure 11.15. Wall cloud hanging below the cloud base of a thunderstorm.

Figure 11.16. Funnel cloud nearly reaching the ground beneath a supercell
thunderstorm.
 Figure 11.17. The vortex spins up and turns into a full-blown tornado.

Eventually, the overlying storm updraft weakens, or the tornado is

displaced laterally from the overlying updraft, which reduces the “suction”
from above, and therefore reduces upward motion and the convergence of
angular momentum. Alternatively, the air feeding the tornado can become too
cold, which reduces buoyancy and upward motion. During the dissipation
phase, tornadoes are sometimes stretched into a long, thin cloud
“rope”(Figure 11.18). For lack of a rotating updraft and warm, moist air
inflow, the vortex eventually dissipates, but the supercell can spawn more
tornadoes further down its path, giving birth to a tornado family. Some
tornadoes can grow in radius and then give birth to suction vortices – smaller,
but incredibly fast-spinning, vortices with winds up to 250 knots – that form
on the outskirts of the main tornado.
 Figure 11.18. Stretching and distorsion of a tornado.

Scientists have a pretty good understanding of the fact that the

convergence of angular momentum is needed to make a tornado. This,
generally, is most likely to happen when the outflow is not too cold and the
wind shear in the environment is very strong, because this shear promotes the
“suction” effect that is key to pulling cool air upward and angular momentum
inward near the surface. Scientists, however, are not yet able to predict
if/when these processes will suddenly occur within a supercell storm, or what
other processes might accelerate tornado formation.
 11.5.3 Tornado Alley
Why do tornadoes occur in the US Midwest more than anywhere else in the
world? The Great Plains are a wide open corridor for air masses to circulate
and clash. Cold air masses from Canada meet warm, moist air masses from
the Gulf of Mexico and form midlatitude cyclones. These cyclones are
associated with upper-level troughs in the jet stream, which provide a lifting
mechanism to bring air to the level of free convection (LFC; see Chapter 6).
Furthermore, lifting can be enhanced along boundaries such as a cold front,
which provides a lifting mechanism to trigger convection and the formation
of thunderstorms in the warm, unstable air ahead (Figure 11.19). Sometimes,
the lifting is provided by the boundary between continental dry air advancing
against moist air from the Gulf – a boundary called a dryline. A strong
upper-level jet provides the shear necessary to create horizontal rotation,
which is turned into vertical rotation by convection, as we explained earlier
(Figure 11.11). We have all the ingredients for a supercell, and eventually the
possibility of tornadoes.
 Figure 11.19. Synoptic situation conducive to the formation of supercells
and tornadoes.

But the region provides yet another crucial ingredient. Hot, dry air from
the Mexican plateau flows into the basin and flows over the warm, moist air
from the Gulf of Mexico. This creates a temperature inversion right over the
inflow of warm and moist air, at about 850 hPa. Figure 11.20(a) shows how
buoyant air parcels lifted from the surface (blue arrow) cannot rise higher
than the inversion, at first. As the lower atmosphere warms up, however, due
to daytime heating, the temperature of the buoyant parcels increases, so that
at some point, they can rise up to 850 hPa and still be warmer than the
maximum temperature of the inversion (Figure 11.18(b)). Being still buoyant,
they can continue their ascent, and a cumulonimbus cloud can form. Most
importantly, while the lower atmosphere was slowly heating during the day,
it was storing a lot of potential energy, which is the main role of the capping
inversion. A good analogy is a pressure cooker in which the temperature and
pressure increase, but the air can neither expand nor rise. If we remove the
lid, the air will rush out with tremendous speed and energy. Another analogy
could be a car driver pressing on the accelerator of the car while holding the
brake pedal down at the same time. As soon as the brake pedal is released,
the car experiences a huge acceleration. Similarly, from the moment the
inversion cap is “punctured,” air parcels accelerate rapidly due to the very
large buoyancy force, and the supercell thunderstorm intensifies very rapidly.

Figure 11.20. Typical temperature profiles in (a) the morning, as the

capping inversion first prevents convection and (b) in the afternoon, after
daytime heating allows buoyant air parcels to overcome the inversion.

Our understanding of tornado dynamics has greatly improved with the

installation of a network of Doppler radars and the deployment of mobile
Doppler radars during dedicated field campaigns. The exact location and
timing of tornado formation remain difficult to forecast, but the factors that
are conducive to tornado formation can be assessed with observations and
numerical weather prediction models. Forecasters routinely estimate the
stability of the atmosphere, the level of free convection, and the potential for
violent thunderstorms to develop. They estimate the vertical wind shear in
critical areas where supercells are known to form, such as along frontal
boundaries. And they detect the presence of capping inversions in vertical
soundings. When the conditions are ripe, forecasters concentrate on Doppler
radar imagery to identify the presence of a hook echo and any cyclonic
rotation that could be the signature of a mesocyclone. When a tornado is
forming, the weather service may have a few minutes to issue a tornado
warning for the local population to seek shelter.
 Summary
Thunderstorms form by convection in an unstable or conditionally unstable
atmosphere, with varying degrees of intensity and severity depending on the
environmental conditions.
Ordinary thunderstorms grow in the absence of wind shear. Cold
downdrafts created by evaporational cooling cause the formation of a cold
pool that tends to spread outward and to create a gust of cold wind that
advances against warmer air. The resulting boundary, or gust front, can
spawn new convective clouds and new thunderstorms.
When the cold pool of air becomes so extensive that it cuts off the
supply of warm, humid inflow to the storm, the thunderstorm decays and
dissipates.
In the presence of vertical wind shear, updrafts and downdrafts can
coexist symbiotically, and the storm can be long lasting and severe, with
intense precipitation and strong isolated downdrafts called microbursts.
Long-lived thunderstorms with rotating updrafts, called supercell
thunderstorms, can induce a lower pressure core within the updraft called a
mesocyclone. They acquire their rotation when horizontal rotation associated
with vertical wind shear is reoriented by convection. In some supercells, a
much smaller region of rotation can develop very near to the ground and,
occasionally, intensify to form a tornado – a fast-spinning vortex of much
smaller scale than the thunderstorm itself.
The American Midwest is a prime location for tornado formation due to
the absence of barrier for cold air masses from Canada and warm air masses
from the Gulf of Mexico. The air mass encounter provides all the necessary
ingredients for tornado formation:

► lifting by a cold front or a dryline to trigger convection and the

formation of thunderstorms in the warm, unstable air;
► vertical wind shear provided by the upper-level jet, to create
horizontal rotation, later turned into vertical rotation by convection;
and
► inflow of hot, dry air from the Mexican plateau, which creates a
temperature inversion above the inflow of warm and moist air from
the Gulf of Mexico. This allows the surface air to store convective
potential energy before the inversion is punctured, and the buoyant air
swiftly rises to form a cumulonimbus.

This region of the USA, referred to as “tornado alley,” is carefully monitored

by weather forecasters, who use observations, Doppler radars, and numerical
weather prediction models to detect the factors and conditions leading to the
formation of tornadoes, and warn local populations in time.
CHAPTER 12
Tropical Cyclones
CONTENTS
12.1 Facts and Figures
12.2 Tropical Cyclone Structure
12.3 Tropical Cyclone Development
12.4 Conditions for Tropical Cyclone Development
Summary

Formidable heat engines of the tropics, tropical cyclones garner enormous

quantities of energy from the ocean and unleash some of the most destructive
winds on Earth. They confront us with some of the most compelling
dynamics of the atmosphere and still challenge our weather prediction
systems. In this chapter, we describe the structure and development of
tropical cyclones, and explain the mechanisms and factors that contribute to
their intensification, or demise.
Tropical cyclones are the most powerful and the most destructive weather
systems produced by the atmosphere. Years later, local populations still
remember and talk about Hurricane Andrew in Florida in 1992 (thousands of
destroyed homes, 65 dead, and billions of dollars of damage) and Hurricane
Mitch in Central America in 1998 (record-breaking rainfall, catastrophic
flooding, 11 000 killed). Typhoons and tropical cyclones cause even more
damage in Asia. We often cite the historical tropical cyclone that killed 500
000 in Bangladesh in 1970, but more recently, and in spite of all the
technological advances in forecasting typhoons and warning local
populations, Super Typhoon Haiyan killed 6300 in the Philippines alone in
2013.
 Before we try to understand how tropical cyclones come about and
develop, let us gather some facts and figures about them from the different
observation systems available to us.
 12.1 Facts and Figures
Tropical cyclones are tropical storms with sustained winds greater than 64
knots. [Recall that, if the wind is 64 knots when averaged over one minute,
wind gusts can be much higher.] We tend to use the term hurricane when
they form over the Atlantic Ocean or Eastern Pacific Ocean, typhoon when
they form over the Western Pacific Ocean, and tropical cyclone when they
form over the Indian Ocean, although they all share the same dynamics
(Figure 12.1). They are given names to facilitate discussion (e.g., among
weather forecasters) as well as research, because atmospheric scientists often
refer to old cases to illustrate or investigate specific aspects of tropical
cyclone development. A list of names is created for each ocean basin, starting
with the letter A and alternating male and female names in alphabetical order
(e.g., Arlene, Bret, Cindy, etc. for 2017 Atlantic hurricanes). When a tropical
cyclone is particularly devastating and becomes a reference case in the
scientific literature, its name is “retired” from future lists, so that it becomes
uniquely associated with that historical event. Such is the case of Hurricanes
Katrina and Rita, for example, during the record-breaking 2005 Atlantic
hurricane season.
 Figure 12.1. (a) About 150 years of hurricane tracks, color coded by
intensity according to the Saffir–Simpson hurricane intensity scale
(tropical depression (TD), tropical storm (TS), and category 1–5). (b)
Overall direction of motion of the different tropical cyclones, along with
the names used in each ocean basin.

Tropical cyclones are ranked based on maximum sustained wind speed,

first as tropical disturbance, tropical depression, and tropical storm up to 64
knots, and then by category 1 to 5 according to the Saffir–Simpson scale for
hurricane intensity (see Table 12.1, where central pressures are indicated for
reference, based on an algorithm). Category 5 tropical cyclones are the most
intense, but are relatively rare – in fact, before Hurricane Irma in 2017, the
last category 5 hurricane in the Atlantic basin was Hurricane Felix in 2007.

Table 12.1. The Saffir–Simpson hurricane intensity scale

Central pressure Winds (knots)

(hPa)

Tropical disturbance <20

Tropical depression 20–34

Tropical storm 34–64

Cat. 1 >980 64–82

Cat. 2 965–979 83–95

Cat. 3 945–964 96–113

Cat. 4 920–944 114–135

Cat. 5 <920 >135

Tropical cyclones have a low pressure center, much in the same way as
midlatitude cyclones, although their dynamics are very different. The lowest
central pressure ever recorded is 870 hPa in Typhoon Tip in 1979. Cyclonic
winds blow around the low pressure center (clockwise in the southern
hemisphere and counterclockwise in the northern hemisphere), which
suggests that the Coriolis force plays a role in explaining the formation and
development of tropical cyclones – a point to remember.
Although tropical cyclones share some similarities with midlatitude
cyclones, they differ in many important ways. Tropical cyclones are about
300 to 700 km in diameter, and therefore smaller than midlatitude cyclones.
They have no fronts, and are more symmetric than midlatitude cyclones.
They tend to last longer, to have stronger maximum winds, and to cause more
damage. Their central core is warmer than their surroundings, i.e., the
opposite of midlatitude cyclones, which usually have a cold core. Indeed,
tropical cyclones and midlatitude cyclones have very different dynamics and,
in particular, draw their energy from different sources. Midlatitude cyclones
draw their energy from the pole-to-tropics temperature gradient, whereas
tropical cyclones draw their energy from the underlying ocean, as we shall
see shortly.
A map summarizing the geographic distribution of tropical cyclones
(Figure 12.1) tells us that they occur mostly between about 5° and 25° of
latitude. Figure 12.1 also shows us a number of interesting features:

► Tropical cyclones form over the ocean, and not over land. In fact, they
either die or turn into a midlatitude cyclone after making landfall.
► Tropical cyclones clearly do not form at the equator, nor do they form
in the midlatitudes.
► Tropical cyclones rarely form in the South Atlantic and do not form in
the Eastern South Pacific.
► Although their individual tracks can be somewhat erratic (see
Figure 12.2, for example), tropical cyclones follow a mostly westward
trajectory, and sometimes curve poleward before dying or turning into
 a midlatitude cyclone. They seem to follow the flanks and curve
around the subtropical highs.
► The Western Pacific Ocean experiences the most intense tropical
cyclones, and in particular many category 5 storms.

Figure 12.2. Hurricane tracks of the record-breaking 2005 Atlantic

hurricane season. The size and color of the dots indicate hurricane
intensity. The start and end points of Hurricanes Dennis (D), Emily (E),
Katrina (K), Rita (R), and Wilma (W) are indicated. The background
temperature field is constructed from AMSR-E sea-surface temperature
measurements on June 1, 2005. The red contour corresponds to 28 °C.

A comparison of the 2005 Atlantic hurricane tracks with ocean temperature at

the beginning of the season (officially, June 1) in Figure 12.2 suggests that
tropical cyclones tend to form and strengthen where the ocean is warm, a
good hint that tropical cyclones are indeed using energy drawn from the
ocean to develop and grow. Note how the four tropical cyclones that reached
category 5 that year (a record-breaking hurricane season by all measures),
Emily, Katrina, Rita, and Wilma, all reached their highest intensity over the
Caribbean and the Gulf of Mexico, where ocean waters are the warmest.
Before we make sense of all these observations, let us observe some
tropical cyclones from space in more detail.
 12.2 Tropical Cyclone Structure
What can we learn from satellite imagery? In the example shown in
Figure 12.3, we can see that the top of a tropical cyclone is made of cirrus
clouds punctured by tall convective clouds: cumulonimbus clouds
corresponding to thunderstorms. These thunderstorms are organized in
spiraling bands that circle around the eye, a tight area at the center of the
storm, 40 to 60 km in diameter, that is often free of high clouds. The eye is
surrounded by a wall of intense convection called the eyewall, where the
strongest winds and heaviest rain are found.

Figure 12.3. Satellite view of Hurricane Katrina on August 27, 2005. A

close-up of the eye reveals spiraling bands of cumulonimbus clouds.

These observations hint at how tropical cyclones come about and grow.
They are essentially made of spiraling convective clouds drawing their
energy from the underlying warm ocean. As the converging surface winds
evaporate water from the ocean, latent heat is taken from the ocean, focused
into the core of the tropical cyclone, and carried upward by convection
(Figure 12.4). As the water vapor condenses higher up to form cloud droplets
and raindrops, the latent heat is released and warms the core of the tropical
cyclone, which lowers the surface pressure. In turn, the lower pressure pulls
even more surface air into the core of the tropical cyclone, accelerating the
surface winds and the evaporation of water. We are back to step one and have
a positive feedback loop that continues as long as the tropical cyclone is in a
favorable environment over warm ocean water and until it reaches its
maximum intensity. Therefore, a tropical cyclone over water is a self-
sustained heat engine that keeps intensifying as long as the environmental
conditions are favorable.

Figure 12.4. Structure and transversal wind circulation in a tropical

cyclone.
 Aloft, as the air reaches the tropopause, it spreads outward in the form of
extensive cirrus clouds, because it cannot penetrate into the very stable
stratosphere due to the temperature inversion (similar to the anvil top of a
thunderstorm). It then subsides further out, around the tropical cyclone,
which explains why the surroundings of a tropical cyclone are usually free of
low clouds. (Recall that convection causes cloud formation, while subsidence
inhibits cloud formation).
While most of the upper air flow is directed outward, some air returns
down into the eye, where it subsides, which explains why the eye is mostly
free of mid- and high-level clouds. Moreover, subsiding air warms due to
adiabatic compression, which explains why the core of a tropical cyclone is
exceptionally warm and comparatively dry (i.e., low relative humidity).
This basic mechanism provides a good indication as to why tropical
cyclones can form only over the ocean, and why they tend to form over warm
oceans: they are essentially fueled by latent heat release in the convective
towers. The latent heat originates from evaporation of warm ocean water by
the strong surface winds. Evaporation of warm water is crucial, and the
tropical cyclone quickly dies as soon as it finds itself over land, or over
cooler water. This explains why tropical cyclones do not form over the
eastern South Pacific ocean: the water is too cold.
If we apply the same reasoning as we did for the sea breeze circulation,
we find that warming from latent heat release in the core of the tropical
cyclone also expands the central air column and lowers the surface central
pressure (see Chapter 8). This gives a pressure distribution such as shown in
Figure 12.5. Surface pressure is lowest in the eye of the tropical cyclone, and
surface winds blow cyclonically around the low while converging into the
core of the tropical cyclone, due to surface friction (Figure 12.6). At the
tropopause, however, pressure is highest in the center and drives upper-level
winds outward. As they diverge away, the upper-level winds create a
spiraling cirrus top easily identifiable on satellite imagery. The two
circulations are well captured in Figure 12.7, where we can contrast the
converging streamlines at the surface (Figure 12.7(a)) with the diverging
streamlines at 250 hPa, i.e., at the tropopause (Figure 12.7(b)).

Figure 12.5. (a) Pressure distribution and (b) corresponding surface and
upper-level circulation in a tropical cyclone.
Figure 12.6. Surface pressure and winds on August 29, 2005, as Hurricane
Katrina approaches New Orleans, Louisiana. The black triangle shows the
location of NDBC buoy 42007. [The pressure gradients and winds are an
output from a numerical weather prediction model at intermediate
resolution and are greatly underestimated.]
 Figure 12.7. (a) Surface streamlines overlaid on surface pressure and wind
speed on August 29, 2005. (b) Corresponding 250 hPa streamlines overlaid
on GOES-12 infrared satellite imagery.

Surface winds tend to accelerate as they converge inward, by partially

conserving angular momentum. Recall the ice skater spinning faster as she
pulls her arms in, and the surface winds accelerating in a midlatitude cyclone
as they converge into the low pressure center. Similarly, in a tropical cyclone,
the strongest winds are found close to the eye (see the blue ring around the
eye in Figures 12.6 and 12.7(a), which is the location where inward spiraling
air stops moving toward the eye and abruptly rises into the eyewall).
Thus, if we were to experience the passage of a tropical cyclone, we
would observe pressure and wind speed changes such as shown in
Figure 12.8 for buoy 42007 from the National Data Buoy Center (NDBC) at
the National Oceanic and Atmospheric Administration (NOAA). The location
of buoy 42007 is indicated with a black triangle in Figure 12.6 for reference.
It is anchored off the coast of Mississippi, east of New Orleans, along
Hurricane Katrina’s path before making landfall. To understand Figure 12.8,
imagine that the pressure and wind fields shown in Figure 12.6 are shifting
northward over the black triangle. As the hurricane approaches, the buoy first
recorded a decrease in pressure down to a minimum of 927 hPa, followed by
an increase in pressure as the storm continued northward. Consistent with
these observed changes in pressure, the wind speed first increased to a
maximum of 54 knots as the ring of strongest winds around the eye
approached, and then decreased as the storm progressed northward. Note,
however, the “lull” around 11:00 a.m. when the buoy was close to the eye of
Katrina and was experiencing relatively weaker winds for an hour or so.
 Figure 12.8. (a) Pressure and (b) wind speed observations as Hurricane
Katrina passes over NDBC buoy 42007 on August 29, 2005.

The vertical structure of the tropical cyclone shown in Figure 12.5

suggests that the vertical “stacking” of surface low pressure and upper-level
high pressure in the warm core is necessary for the eyewall to exist and the
overall circulation to continue. Contrary to thunderstorms, which become
more severe weather systems when they are sheared and tilted, allowing
downdrafts to become separated from updrafts, tropical cyclones weaken
when there is a strong wind shear with height. That is why tropical cyclones
only form and grow in the tropics, where vertical wind shear is weak, and not
in the midlatitudes, where the presence of the upper-level jet stream implies a
strong vertical wind shear. In fact, when tropical cyclones venture into the
midlatitudes, their structure changes and they either dissipate or turn into a
midlatitude cyclone.
 12.3 Tropical Cyclone Development
Now that we have a good picture of tropical cyclone structure at the mature
stage, let us explore the mechanism that causes them to form in the first
place. A natural starting point involves retracing the steps of a hurricane back
to the eastern side of the Atlantic Ocean basin.
 12.3.1 Tropical Easterly Wave
We recall that the Intertropical Convergence Zone (ITCZ) is a region of
pronounced convection, where the converging trade winds force upward
motion and the formation of tall cumulonimbus clouds. Most of the time,
these thunderstorms are disorganized. They grow and dissipate, following the
easterly flow for a few hours or days. However, the subtropical surface
easterly flow is wavy and contains troughs and ridges, much in the same way
as the midlatitude jet stream. In the Atlantic Ocean, for example, the
undulations of the easterly flow form over the Ethiopian Highlands of East
Africa, before moving westward across North Africa.
We also recall that the wind tends to be subgeostrophic around troughs
and supergeostrophic around ridges (see Box 8.4 and the blue arrows in
Figure 12.9). As a result, the easterly flow contains regions of convergence
and regions of divergence, on either side of the trough axis. Compared to
troughs in midlatitudes, notice that the pressure gradient is reversed: whereas
pressure overall decreases northward at upper level in the northern
midlatitudes, from subtropical high pressures toward the polar trough (see
Figure 9.1), pressure overall increases from the equator to the subtropical
highs at the surface (see Figures 9.8 and 9.11). Therefore, the trough pattern
is rotated by 180°.
 Figure 12.9. Tropical easterly wave embedded in the surface trade winds.

Thunderstorms tend to congregate east of a trough axis, where

convergence enhances convection. These convective patches can be
identified on satellite imagery and are usually followed closely by weather
forecasters, because they can later develop into a stronger system, and, for a
small subset, into a hurricane. At this stage, we refer to this early precursor as
a tropical easterly wave. Such a wave appears in Figure 12.10 off the coast
of Africa on September 1, 2009. The intense wave and cloud activity further
upstream over Africa will later evolve into another easterly wave and
Hurricane Fred over the eastern Atlantic.

Figure 12.10. Visible satellite imagery and 700 hPa streamlines on

September 1, 2009, at 12:00 UTC, with three easterly waves indicated by
red arrows.
 12.3.2 Tropical Depression
We recall that, in a thunderstorm (i.e., in a cumulonimbus), the updrafts
compete with downdrafts bringing cold air down to the surface. These
downdrafts tend to stabilize the lower atmosphere and to suppress further
development. As the cold and dry air comes in contact with the ocean
surface, it removes heat the lower atmosphere had gained from the upper
ocean, and the whole convective process has to start again with warming and
moistening the air near the surface. For a cluster of cumulonimbus to evolve
into a tropical depression, convection needs to persist so as to keep warming
and moistening the lower and middle troposphere. As more and more heating
occurs, the updrafts start winning over the cold downdrafts, and the net
heating becomes overall positive and penetrates deeper into the troposphere.
Each new convective cell transports more moisture aloft, and the heating
accelerates. This process is enhanced when the air in the middle troposphere
moves with the region of convection; that is, when the moistening middle
tropospheric air stays with the region of convection. This tends to happen
when vertical wind shear is weak; strong shear tends to introduce dry air
continually into the middle troposphere, which enhances the formation of
downdrafts.
As shown in Figure 12.5, net heating of the air column lowers the
surface pressure. Surface winds start to converge and to accelerate by partial
conservation of angular momentum, which enhances evaporation, the
transport of moisture into the forming warm core, convection, etc.: our
positive feedback loop is initiated. The warming also raises the pressure
surfaces and creates an upper-level high (see Figure 12.5 again), which forces
air to flow out of the area, consistent with lowering the surface pressure. With
lower pressure near the surface, rotation increases under the influence of the
Coriolis force. At this point, the disorganized convection has turned into an
organized tropical depression (TD), with sustained winds of 20 to 34 knots.
An example of a tropical depression can be seen in Figure 12.10 east of
the Caribbean Islands, along with the corresponding easterly wave. It evolved
into Tropical Storm Erika over the following days.
 12.3.3 Tropical Storm
With a continuation of the processes described in Section 12.3.2, the system
intensifies and turns into a tropical storm (TS) with sustained winds of 34 to
64 knots. It is steered westward by the overall easterly flow, i.e., the southern
flank of the subtropical high (see Figure 9.13).
 12.3.4 Tropical Cyclone (Hurricane)
As the storm intensifies, an eye forms and becomes well defined, with an
eyewall that concentrates the strongest convection and rainfall. The storm
technically becomes a tropical cyclone when sustained winds reach 64 knots.
It is then ranked by category from 1 to 5 according to the Saffir–Simpson
scale for hurricane intensity (Table 12.1). Tropical cyclones typically grow in
intensity as they move westward and cross the ocean basin, but they can also
decrease in intensity if they encounter lower sea surface temperatures, for
example, or different environmental conditions, as might be the case in the
Caribbean or the Gulf of Mexico.
 12.3.5 Tropical Cyclone Decay
Tropical cyclones typically weaken when they move over cooler water (lack
of heat), over land (lack of moisture), or into a region of strong vertical wind
shear, such as the midlatitudes. When moving over land, they usually
dissipate within hours after making landfall, for lack of fuel (latent heat of
evaporation), and due to friction with the land surface. This is typically the
case of hurricanes encountering Central America or the southeastern USA,
and typhoons encountering the Asian continent. However, the western flank
of the subtropical highs has a southerly wind component that sometimes
steers the hurricanes and typhoons northward (see Figure 9.13, for example,
and compare it with Figure 12.1). Since ocean waters are usually colder at
higher latitude, northward motion often announces the demise of the storm.
Yet, when the tropical cyclone starts drifting poleward and its position is
coincident with an approaching upper-level trough in the midlatitude jet
stream, the tropical cyclone can morph into a midlatitude cyclone. It then
develops a warm sector, cold and warm fronts, and its dynamics change to
resemble those of midlatitude weather systems.
Tropical cyclones cause a lot of damage in different ways. Extremely
high winds can destroy poorly built structures, damage the electric grid, and
topple trees over a large area. Extreme rainfall often saturates the soil and
causes extensive flooding. Landslide can ensue, which causes more
problems, including failure of the electrical grid, breaking of water pipes,
roads and houses. And the destruction of infrastructure can have unexpected
consequences, such as even more flooding, the absence of potable water for
days or weeks, and the spread of diseases.
 One of the more dangerous aspects of a tropical cyclone is the storm
surge, which is an increase in sea level following the storm (Figure 12.11).
Indeed, the low pressure at the center of the tropical cyclone allows the
underlying water to rise above its normal level. Recall that the ocean surface
is under atmospheric pressure. If you picture the atmosphere pushing more on
some areas (high pressure) and less on others (low pressure), you can
visualize that the ocean will be depressed under higher pressure, and bulging
upward under lower pressure. Since the atmospheric pressure is extremely
low at the center of a tropical cyclone, it allows the water to rise by more than
one meter for larger storms. While this does not seem much in itself, it
amounts to an enormous amount of water when integrated over the size of the
tropical cyclone. The addition of strong winds and wave action interacting
with the coastline can greatly enhance this effect to more than 8 m (26 ft),
primarily to the right of the storm track where the wind is directed toward the
shoreline in the northern hemisphere (Figure 12.11(d)). Strong, fast moving,
and larger storms have the highest storm surge.
Figure 12.11. (a–c) Storm surge propagating westward under a tropical
cyclone. (d) Bird’s eye view showing the winds to the right of the storm
pushing the surge water against the coastline in the direction of motion of
the storm.
 Storm surges often cause more damage than the wind and rainfall
associated with the tropical cyclone. When they hit a flat coastal area, in
particular, the rushing water can penetrate deep inland and inundate a large
swath of coastal land, as was the case in Galveston Bay, Texas, in 1900 (8000
killed) and in Bangladesh in 1970 (500 000 killed). Similarly, flooding and
landslides resulting from heavy tropical cyclone rainfall can kill many
people, as was the case for Hurricane Mitch in 1998 (7000 killed in Honduras
alone) and Super Typhoon Haiyan in 2013 (6300 killed in the Philippines
alone).
 12.4 Conditions for Tropical Cyclone Development
In summary, the conditions for tropical cyclone development are:

► A warm, conditionally unstable atmosphere. That is often true at low

latitudes over a warm ocean, where there is high evaporation and a
moist lower atmosphere.
► Warm sea surface temperatures (greater than about 27 °C).
► Sufficient deep ocean heat to replenish the surface as it is depleted by
vertical mixing due to wind and wave action.
► Some rotation, and therefore some Coriolis force, i.e., tropical
cyclones can only form north of about 5°N and south of about 5°S.
Waves in the easterlies provide an initial source of rotation over the
Atlantic Ocean, as can the monsoon trough and disturbances in the
ITCZ over the Pacific Ocean.
► Weak vertical wind shear, i.e., tropical cyclones cannot form in the
midlatitudes, where the presence of the upper-level jet stream causes
strong vertical wind shear.

These conditions constrain tropical cyclones to form between about 5 and 15°
latitude over the warmest sections of the tropical ocean.
The temperature of the ocean is now well monitored by satellite, and
regions of high sea surface temperatures can be detected well in advance
during the hurricane season. As for vertical wind shear, weather forecasters
can also obtain some fairly good advance notice from the large-scale
structure of the atmosphere, by observing the evolution of planetary-scale
waves. Thus, even though we cannot predict the details of an evolving
tropical depression, we can predict the conditions that are favorable to its
further development into a tropical cyclone. Forecasting tropical cyclone
track has improved greatly in the past 20 years. In some locations, like the
North Atlantic Ocean, routine reconnaissance aircraft make detailed
measurements of storm location, structure, and intensity, which complements
satellite observations. These improvements in track forecasting allow the
weather services to prepare local populations and to issue appropriate
warnings, and allow local governments to enforce evacuation when
necessary.
 Summary
Tropical cyclones (hurricanes, typhoons) are tropical storms with sustained
winds greater than 64 knots. Tropical systems are ranked as tropical
disturbance, tropical depression, and tropical storm up to 64 knots, and then
as tropical cyclone by category from 1 to 5 according to the Saffir–Simpson
scale for hurricane intensity. They are about 300 to 700 km in diameter, and
are composed of cumulonimbus clouds spiraling in around the central eye.
Tropical cyclones have a warm central core, due to latent heating
associated with condensation in clouds near the eye. They draw their energy
from the underlying ocean, as the converging surface winds evaporate water
from the ocean (latent heat intake) and transport it into the core of the
cyclone, where it condenses to form cloud droplets and raindrops (latent heat
release). This warms the core of the tropical cyclone, lowers the surface
pressure, and causes winds to spiral inward toward the center.
Surface winds converge cyclonically into the warm core of the tropical
cyclone and accelerate until they reach the eyewall. At the tropopause, upper-
level winds diverge outward, creating a spiraling cirrus top with the reverse
circulation from the surface. Vertical alignment is necessary for the integrity
of tropical cyclones, which explains why they cannot develop when the
vertical wind shear is too strong.
In the North Atlantic, hurricanes first appear as tropical disturbances
made of cumulonimbus clouds east of the axis of troughs, where convergence
enhances convection – early precursors called tropical easterly waves. For a
small fraction of easterly waves, convection is able to moisten the atmosphere
sufficiently so that latent heating warms the air in the system, resulting in
lower pressure. This in turn enhances convergence of air into the low
pressure, which accelerates the wind, forming a tropical depression. This
process may continue, with the storm intensifying into a tropical storm, and
possibly even a tropical cyclone, when the eye often first appears.
Tropical cyclones typically die when they move over cooler water (lack
of heat), over land (lack of moisture), or into a region of strong vertical wind
shear, such as the midlatitudes.
In addition to damage caused by their extremely high winds, rainfall,
and sometimes tornadoes, tropical cyclones can cause landslides, as well as
flooding and destruction by the storm surge, where sea level rises near the
storm due to low pressure and strong wind piling water onto the coastline.
CHAPTER 13
Weather Forecasting
CONTENTS
13.1 Weather Forecasts and Uncertainty
13.2 Prognostic Equations
13.3 Ensemble Forecasting
13.4 Chaos and Weather Prediction
13.5 From Forecast Grids to Reliable Forecast Values
13.6 Making a Forecast
Summary

How is it possible to predict, often with great accuracy, upcoming weather?

Is a 5-day forecast as good as a 2-day forecast? Can we predict the weather
out to the distant future? How is a forecast made? Here we will explore the
basic principles behind numerical weather forecasting, as well as the limits of
numerical models in accurately predicting weather beyond a certain time. We
will discuss the capabilities and limitations of computer models of the
atmosphere, methods to address uncertainty, tools to make weather model
forecasts more useful, and the role of human forecaster in analyzing and
interpreting computer-generated weather predictions.
One of the things that makes science a uniquely human endeavor is its
predictive power. Humans have designed elaborate methods for predicting
such things as the motion of planets, tides, and the propagation of diseases,
using physics and mathematics. Yet, while we do not question our ability to
predict, for example, tides, even though those forecasts have errors in timing
and amplitude, we sometimes think of weather as “unpredictable,” and we
tend to remember poor forecasts even though these are relatively rare. In fact,
a close look and a more rigorous assessment of weather forecast accuracy
shows that weather forecasters are actually correct much more often than they
are wrong, and have been steadily getting better over the past 50 years.
Turning weather prediction from a qualitative art to a quantitative science is
one of the major human accomplishments of the twentieth century. In this
chapter we explore the basis for the quantitative approach to weather
forecasting, which uses the knowledge we have accumulated in the past 12
chapters. This enterprise involves the physical conservation laws we have
discussed, a method to solve these equations into the future, powerful
computers, and, of course, observations of the atmosphere.
 13.1 Weather Forecasts and Uncertainty
We use weather forecasts to make decisions every day. Often those decisions
are just about what clothes to wear and whether or not to bring an umbrella.
Although the information in those forecasts is mostly deterministic (a single
number, such as for the expected afternoon high temperature), this is
beginning to change, and there are some aspects that are mostly
communicated in the form of probabilities. For example, the probability of
precipitation, or POP, might be 30%. This says that there is a 30% chance of
measurable precipitation (usually defined as 1 mm or 0.01") at a specific
location during a specific period of time. It does not say that 30% of the
geographical area will receive precipitation, which is a very common
misperception. If it rains, does that mean 30% was a bad forecast? What if it
does not rain? The way we know if it is a good forecast is by looking at the
statistics of many such forecasts and verifying the outcome. We say such a
forecast is reliable if, for all of the 30% POP forecasts, measurable rain
occurred in 30% of the forecasts. Similarly, a POP of 90% is reliable if, for
all past 90% POP forecasts, measurable rain occurred 90% of the time. Note
that, despite the fact that 90% sounds almost like a “sure thing,” we should
expect it not to rain 10% of the time if this is a reliable forecast.
This example illustrates that weather prediction is a fundamentally
probabilistic endeavor. However, the decisions we base on that information
often are not: we make yes/no decisions. A familiar example to illustrate how
we use probabilities to make decisions, which we don’t even think about,
involves our behavior when driving a car at different times of day. When
driving on a major highway at 3 a.m., we may take few precautions when
changing lanes: there are very few cars on the road, and a quick glance in the
mirror is sufficient to confirm that it is safe to move. In contrast, during the
evening rush, when the roads are full, we evaluate changing lanes much more
carefully. There is a higher probability of something going wrong when the
road is full of cars. Here we are also using a rule for our risk tolerance –
consequences in an accident are high, so when probabilities for mistakes
increase, even a little, our decisions reflect that we are more cautious.
A weather-specific example involves a transportation manager who
faces a snowfall forecast and has to schedule overtime pay (expensive) for
drivers to treat the highways. If she does not schedule the drivers and heavy
snow indeed falls, then there is a high price to pay in the form of a disruption
to transit and commerce, which reflects poorly on her job performance. In
contrast, if she schedules drivers and it does not snow, there is a price to pay
in the form of scheduled, but idle, workers. The snow accumulation forecast
can be highly uncertain, say anywhere from 2 to 20 cm, but she still has to
decide, yes or no, whether to schedule the drivers, despite the uncertainty.
Suppose that she knows that, in addition to the wide range of possible snow
accumulation, there is a 30% chance of less than 2 cm, but a 50% chance of
more than 10 cm snow accumulation. She can then make a decision based on
a risk analysis of the likely cost of both options; in other words, quantifying
the uncertainty with probabilities makes her decision easier, and, like the
POP example earlier, she will save money over many cases by making
decisions based on reliable probabilities.
So, uncertainty and probabilities are fundamental to weather forecasts.
Before we explain how we evaluate that uncertainty, we discuss the way
individual forecasts are made, which will expose the sources of uncertainty
and the methods that have been developed to deal with them.
 13.2 Prognostic Equations
Although our daily weather seems complex and ever changing, the
atmosphere is a fluid that obeys the laws of physics, in particular the laws of
motion, as enunciated by Isaac Newton, the laws of thermodynamics, and
mass conservation, as we have established throughout this book. These laws
can be cast in mathematical form and used to calculate the future state of the
fluid body, given its current state. This can be illustrated using Newton’s
second law of motion, starting with the familiar form

F=ma (13.1)

Where F is the sum of all forces applied to an air parcel (bold indicates
vectors), a is the acceleration (change in velocity) experienced by the parcel,
and m is its mass. We can rewrite it as

(13.2)
a=Fm

Now recall that an acceleration is a change in velocity ∆v occurring over an

interval of time ∆t:

(13.3)
a=ΔvΔt=Fm

Thus, we can also write that the change in velocity

(13.4)
Δv=ΔtmF
Since the change is equal to the difference between the future and the current
state,

Δv=vfuture−vcurrent (13.5)

we obtain that

(13.6)
vfuture=vcurrent+ΔtmF

In other words, if we know all that is on the right side (the current velocity of
the air parcel, its mass, the sum of all forces applied to it, and an interval of
time of interest, ∆t), then we know the future velocity of the air parcel. We
can repeat a similar operation with the other conservation laws.
Thus, if we divide the atmosphere into individual air parcels and choose
a time interval ∆t of 10 minutes, for example, we could, in theory, calculate
the future properties of all the air parcels 10 minutes from now. And if we
repeated the calculation by 10 minute intervals, we could extend our forecast
to 1 hour, 6 hours, or any number of days. In order to make this calculation,
or to use other approaches that do not involve following air parcels, we
evaluate the evolution of individual variables (such as velocity, temperature,
humidity) at points on a grid that covers a region of interest (Figure 13.1).
Coding these equations into a computer at each point of our grid and letting
the application run provides a deterministic forecast: a single forecast, using
a single estimate of the current state of the atmosphere. As we will see, we
cannot stop there, because we have not yet measured the uncertainty in the
forecast. Considering how errors emerge in forecasts will emphasize why this
is so important.
Figure 13.1. Three examples of grids at increasingly higher resolution. (a)
The whole globe covered with a 10° grid. (b) The USA with a 1° grid. (c)
Washington State with a 0.1° grid.
 Why, then, is the weather forecast for tomorrow sometimes wrong, and
the forecast for a month from now completely unreliable? In part, because we
do not know exactly all that is on the right side of equation (13.6): the initial
conditions. We either do not know it with enough precision (e.g.,
instrumental errors) or we do not know it at all in certain locations (e.g., over
the oceans, where we have few direct observations). Thus, a small
imprecision in the quantities on the right side translates into a small error in
the calculation of the future velocity on the left side. Basically, these small
errors grow in time to become large errors. Even if we measure the
atmosphere thoroughly and accurately, small errors will always be present,
grow in time, and degrade the forecast. The limit of forecast skill is thought
to be about two weeks, even if we knew the initial state almost perfectly –
although there are situations where that skill can be extended to much longer
timescales, such as global warming forecasts (see Section 13.4).
In addition to uncertainty in the initial state of the atmosphere, the model
representation of the atmosphere is also imperfect. One problem is that we
cannot represent all scales of motion exactly; the finest grid cells in
Figure 13.1 have a finite size. A second problem is that certain processes, like
cloud microphysics, occur at the molecular scale and cannot be resolved
explicitly. Electromagnetic radiation is important to the evolution of
temperature, but we cannot simulate individual photons. Our grid does not
resolve small-scale eddies that contribute to turbulent mixing in the boundary
layer. These and other unresolved processes are parameterized in terms of
the scales of motion we can resolve with the chosen grid. This approximation
translates into a significant level of uncertainty in the model. These model
errors come in addition to errors in the initial conditions. In summary, we can
see that weather models are complex computer algorithms subject to
uncertainty in their construction, in addition to the uncertainty in the initial
data for the forecast.
The amplification of errors in the characterization of the atmosphere is
illustrated in Figure 13.2, where three different forecasts are made with
slightly different initial conditions and the temperature at some weather
station is plotted against time. In this example, the predicted temperatures are
almost identical on the first day, and relatively similar on the second day,
although they start to depart from each other at the end of the day. By the end
of day 3, the difference is as large as 3 °C, while beyond the fourth day the
temperatures span a range of more than 10 °C. Therefore, in this case we can
have confidence in our 1-day forecast, and even our 2-day and 3-day
forecasts to some extent, because in spite of the errors that might be present
in our initial data, the actual weather will most likely evolve within the
envelope of the three curves and the actual temperature will fall within a 3 °C
window. Beyond 3 days, however, the actual temperature might be anywhere
within the 10°C window suggested by the different forecasts, and our
confidence is greatly reduced.

Figure 13.2. Three temperature forecasts with slightly different initial

conditions.
 This decreasing accuracy of weather forecast models beyond a few days
is illustrated in a different way in Figure 13.3. Rather than positioning
ourselves at time 0 and watching a forecast evolve into the future, we now
choose a time of interest (here, our midlatitude cyclone on February 21, 2014,
at 00:00 UTC), and look back at how different forecasts performed, starting
from different points in time. Since the actual weather event has occurred, we
have the observed pressure and temperature field (Figure 13.3(d)) against
which we can verify the different forecasts.
 Figure 13.3. (a)–(c) Comparison of different forecast runs for the same
valid time, February 21, 2014, at 00:00 UTC. (d) The observed pressure
and temperature fields on February 21, 2014, at 00:00 UTC.

Figure 13.3(a) shows a 5-day forecast computed on February 16, 2014,

at 00:00 UTC. Obviously, the forecast did not perform very well, since the
cyclone is completely missing, except for some sign of a low pressure trough
extending south between two highs.
The 3-day forecast shown in Figure 13.3(b) is significantly improved.
Our mature cyclone is present with fully developed low and fronts. Yet, the
warm front is positioned too far north, and the cyclone is not showing any
sign of occlusion. In other words, the predicted storm development is too
slow: the actual cyclone will develop and occlude faster than predicted.
Not surprisingly, the 1-day forecast is much more accurate
(Figure 13.3(c)), and the differences with the control analysis are in the fine
details. Indeed, current models perform extremely well within 24 to 48 hours,
and certain large-scale aspects of the weather are predicted well beyond a
week, as we will see.
Now that we have the basis for making a quantitative weather forecast,
and the sources of error and uncertainty in that forecast, we will return to how
to quantify that uncertainty and provide the probabilities we need to make
decisions.
 13.3 Ensemble Forecasting
As we have seen, if we run a number of weather forecasts with slightly
different initial conditions that reflect our uncertainty about the initial data,
we can compare all the forecasts and gain confidence where they mostly
agree, while exercising caution where they mostly disagree. Similarly, we can
adjust the parameterizations of the unresolved processes in the model, and
make new forecasts to assess how that source of uncertainty affects the
forecast. The group of forecasts that results from this suite of computations is
called an ensemble, and each individual forecast is called a member of the
ensemble. Ensemble forecasting is the modern method by which weather
forecasts are made and probabilities are estimated.
To continue with the previous example in Figure 13.2, where we ran an
ensemble forecast with three members, we could issue a forecast of 6 °C after
day 1 with confidence, 7–8°C after day 2, 5–8°C after day 3, while we would
suggest 6–13°C and 3–13°C for days 4 and 5, respectively. A more
mathematical way of doing this would be to calculate the statistical
uncertainty associated with the forecast. Here, we cannot do statistics with
three members, but, for the sake of illustration, the 1-day forecast might be 6
± 1 °C (i.e., 6 °C plus or minus 1 °C), while the 4-day forecast might be 8 ±
5°C (i.e., a greater uncertainty).
Returning to our 21 February 2014 cyclone, Figure 13.4 shows that the
uncertainty in forecasts of this event is unusually large. The two outlier
ensemble members are chosen to illustrate why looking at a single model
forecast can be very misleading (Figures 13.4(a) and (b)). One ensemble
member has a large and intense midlatitude cyclone, whereas the other has a
large and strong anticyclone. A forecaster relying solely on one of these
forecasts would make drastically different forecasts just 5 days into the
future. Figure 13.4(c) shows the average over all of the ensemble members,
which is the best forecast we have because it averages over all of the random
differences in the ensemble members and leaves the main signal that they
have in common. In this case, the uncertainty is unusually large and the
ensemble mean shows a large area of high pressure across the region where a
major cyclone would occur in just 5 days. Some of the ensemble members
had the right idea of a major cyclone developing, but they were in the
minority, and there is no way to know in advance which ensemble members
are the ones to base a forecast upon.
Figure 13.4. (a), (b) Two examples of particularly different 5-day forecast
ensemble members for the same valid time as in Figure 13.3, February 21,
2014, at 00:00 UTC. (c) The 5-day ensemble mean for the same forecast
time.
 Another way to represent the forecast uncertainty from an ensemble is to
plot a particular contour for all ensemble members, which results in what are
called “spaghetti” plots. This is common on upper-level maps, and an
example for a single 500 hPa contour for our cyclone case is shown in
Figure 13.5. In this particular example, the 5240 m contour (at 500 hPa) is
shown in black for different forecast ensemble members with slightly
different initial conditions. We can see that the spread is relatively small in
the 1-day forecast (Figure 13.5(a)), but increases as we go to 3 days
(Figure 13.5(b)) and 5 days (Figure 13.5(c)). In the 5-day forecast we notice
that all ensemble members agree relatively well in places, like the North
Pacific Ocean (narrow spread), but disagree in other places, like the North
Atlantic Ocean and the Norwegian Sea (wide spread). It tells us that we
should be cautious about drawing conclusions and predictions that are
strongly dependent upon the position of upper-level features in the North
Atlantic, such as the position and intensity of a surface low.
 Figure 13.5. Comparison of different ensemble forecasts for the same
valid time, February 21, 2014, at 00:00 UTC. (a)–(d) The 5240 m contour
of the 500 hPa surface is shown in black for 11 members. The red contour
shows the ensemble mean. (e) The 10-day ensemble mean 500 hPa height
field. (f) Observed 500 hPa height field on February 21, 2014, at 00:00
UTC. In (e) and (f), the 5240 m contour is highlighted in bold, and colors
indicate the height anomaly with respect to the 30 year mean 500 hPa
height field for the month of February.

One drawback with spaghetti diagrams is that the lines tend to spread
out more in regions where the gradient in the field is small (lines spaced far
apart) as compared to regions where the gradient is large (lines spaced close
together). This simply indicates that small changes in the value move the
contour a lot, which is not indicative of large uncertainty.
The 10-day forecast seems relatively useless at first glance
(Figure 13.5(d)). The 10-day ensemble mean, however, contains some useful
information at larger scales. This is shown in Figure 13.5(e), where the 10-
day ensemble mean 500 hPa height field is overlaid on the “anomaly” with
respect to the climatological mean for February: that is, the difference
between the ensemble mean and the climatological mean. Blue areas indicate
lower heights than average for February, and red areas indicate higher heights
than average for February. We can see that the 10-day ensemble mean indeed
contains some indication of an upper-level trough over the western half of the
USA (blue in Figure 13.5(e)), and of an upper-level ridge over New England
(red in Figure 13.5(e)). This pattern roughly corresponds to the observed
synoptic situation and anomalies on February 21, 2014, at 00:00 UTC (Figure
13.5(f)), so the 10-day forecast contains useful information about general
changes in the pattern, although it lacks crucial details about individual
weather features, such as our midlatitude cyclone.
In conclusion, even though we lose specific details about individual
weather features for forecasts out to 10 days, there is still useful information
about large-scale patterns that we can glean from studying the ensemble
members, the ensemble mean, and the anomaly of the mean with respect to
the climatology.
 13.4 Chaos and Weather Prediction
Will we ever be able to make a perfect deterministic weather forecast if, one
day, we can observe the atmosphere in its smallest details and build a
computer that can resolve the smallest scales? Probably not. It is generally
accepted that it is technically impossible to predict details of the weather
beyond about two weeks. And that is not due to human limitations in
observing or modeling the atmosphere; it is due to the nature of the
atmosphere itself. In the end, the amplification of initial errors is due to the
chaotic nature of the atmosphere, which is described by chaos theory. Note
that, in this context, chaos does not mean “total disorder,” as we often imply
in daily parlance. Rather, chaos means that the prediction of the state of the
atmosphere at future times depends sensitively on the details of the
atmosphere at the starting time of the forecast, as we saw quite dramatically
in Figures 13.4 and 13.5. In other words, our inability to eliminate all error
from the initial state inevitably leads to the growth of those errors that
ultimately degrade the forecast.
For the atmosphere, the largest scales of motion, for example planetary-
scale waves on the jet stream (Figure 13.6(a)), have the longest predictive
timescale, and the smallest scales of motion (Figures 13.6(e) and (f)) have the
shortest predictive timescale. This is due partly to the fact that small spatial
scales naturally have short timescales. For example, gusts of wind may last a
few seconds whereas changes in the wind in planetary waves may take place
over many days. A second fact is that the largest scales have much more
energy than the small scales; the total energy in all of the small wind gusts is
dwarfed by the energy in the jet stream. As a result, errors of a given size take
much longer to “saturate” at the larger energy values found at planetary
scales compared to small scales. Therefore, while small-scale errors would
eventually grow and become significant at larger scales in a longer-term
forecast, they remain relatively small compared to larger scales in a 3 to 5-
day forecast.
Figure 13.6. Atmospheric motion organized at various scales: small spatial
scales have shorter time scales and less energy than large scales of motion.
 Although we have painted a gloomy picture of improving weather
forecasts beyond about two weeks, there is an important exception to this
rule. If something occurs that “forces” the atmosphere, then predictive skill
may result if the forcing is known or may be predicted. This forcing is
usually something that occurs on longer timescales than those that dominate
weather variability in the atmosphere. Two examples are global warming and
the seasonal prediction of the El Niño/Southern Oscillation (ENSO)
phenomenon.
For global warming, all forecasts of specific weather events (like a
particular cyclone) are still subject to the two-week limit, but the warming
itself, which is essentially a known forcing from greenhouse gases, can be
well forecast out to decades and centuries on the global scale (see Chapter
15). This leads to changes in the jet stream and, consequently, in the location
of storm tracks and weather events. Again, we cannot predict specific
weather events, but these global warming forecasts can inform us about the
change in the statistics of weather events. For example, if the jet stream shifts
poleward, then regions that will experience the effects of extratropical
cyclones will also shift poleward.
For ENSO, changes in the temperature of the tropical Pacific upper
ocean force changes in the atmosphere through teleconnections related to
planetary waves. These planetary waves are excited by convection that occurs
over unusually warm water. Since the evolution of the upper ocean
temperatures is a process taking place over 6–12 months, and since it often
does so in a predictable way, it provides some predictive skill to weather
averages on these timescales. Again, we cannot predict individual weather
events, but the statistics of weather features, such as temperatures and
precipitation, can often be reliably forecast to be higher or lower than normal
for El Niño and La Niña events.
 13.5 From Forecast Grids to Reliable Forecast
Values
Once the computation for an ensemble forecast has been completed, the raw
result is a grid of numbers for temperature, pressure, wind, precipitation, etc.
While we could use those numbers for making a weather forecast, it turns out
that another important step is usually taken: statistical post-processing. The
“post” part of the processing reflects the fact that this step takes place after
the computer simulations have been performed. It turns out that some of the
errors in the forecast are not random, but systematic, and we can increase the
skill of our forecasts if we remove the systematic error. One source of
systematic error is called bias. We often find that weather forecasts from the
computer models are systematically too warm or too cold for some locations;
for others, there may be too much or too little precipitation. Often these
biases result from a complex set of processes in the models that are very
difficult to correct. For example, errors in the treatment of clouds can produce
these biases in temperature and precipitation, but since the physics of clouds
depends on many parts of the model, including those that are parameterized
(e.g., cloud microphysics and radiation), it is very difficult to fix these biases
without making other aspects of the forecast worse.
Bias may not be obvious in any individual forecast because other
sources of error may be much larger, but it becomes apparent when we
average the errors over many forecasts. By estimating bias in this way, and
removing it from the raw computer forecasts, we can produce forecasts that
are more accurate. This is an example of forecast calibration. Another
example of forecast calibration involves the uncertainty estimate derived
from the ensemble forecast. We often find that the predicted uncertainty from
the range of forecasts in the ensemble is too small. This is due to the fact that
some errors are deterministic, which reduces the range of uncertainty that the
ensemble should capture. For example, if there are no errors coming from the
land surface below the atmosphere, that tends to reduce ensemble variability
in near-surface fields. Again, as with bias, if the ensemble forecasts are
verified over a large number of forecasts, that systematic error can be
corrected, resulting in calibrated ensembles.
Another very successful approach in statistical post-processing is called
model output statistics (MOS). Rather than making forecasts directly from
the model grids, the MOS approach involves using a set of predictors from
these grids to estimate the quantities of interest. For example, we want daily
high and low temperature forecasts at specific locations, and we could just
use the raw model output, or its bias-corrected version, directly. Instead,
MOS uses a wide range of quantities that are known to be important and
directly related to high and low temperature forecasts. For example,
temperatures, winds, and clouds throughout the lower atmosphere are known
to be important predictors of surface temperature. Using a large sample of
historical forecasts of these predictors, and the actual high and low
temperatures that were observed, MOS builds a calibrated statistical
relationship that allows us to predict the high and low temperature better.
These relationships are custom to each location and season, which insures
they are optimal for each location. One reason this method works is that it
provides an efficient way to remove bias from a large number of fields that
may contribute to temperature forecasts, but it also accounts for the fact that
the points for which we want a forecast are rarely at model grid points.
 MOS provides an important contribution to actual weather forecasts, and
it is difficult for humans to intervene and improve upon these forecasts. One -
situation where humans can, in fact, beat MOS forecasts is during extreme
events. When the forecast extends beyond where MOS has been calibrated, it
can lead to unusually large errors. For example, record high temperature
forecasts mean that the value has not been observed previously. It is in these
situations, which often have a high impact on society, that the human
forecaster plays an outsized role in the forecast. Another place where humans
play an important role is in very short-range forecasts, which we call
nowcasts. If observations reveal that the most recent computer forecasts are
in error, the human forecaster can correct for that. Newer techniques, based
on machine learning, are also under development and promise to improve
further our ability to post-process weather forecast information.
Once we have completed statistical post-processing of the forecast
information, it is time to make the forecast. As we have indicated, it is
difficult for the human forecaster to improve upon this forecast information,
with important exceptions. Instead, the main role of the human forecaster is
to understand why the forecast is as it is, and to explain that to society. To do
so, human forecasters often follow a method of building a conceptual
understanding of the forecast, and we now discuss a common method that
you can use as well.
 13.6 Making a Forecast
To conclude this chapter, let us survey some of the tools meteorologists use
to issue their forecasts. As a general rule, when we prepare a weather
forecast, we usually start with the largest scales and then zoom in on a
particular region of interest. Thus, it is common to obtain first a view of the
entire hemispheric circulation, commonly using the 500 hPa height field, for
example, as shown in Figure 13.7(a). It is helpful to look at a loop going back
a few days in the past, to follow where the troughs and ridges are coming
from, and extending a few days into the future, to see where these features are
going. This overview helps us to appreciate the larger-scale context in which
our local weather is occurring, and becomes increasingly important as we
consider forecasts further in the future.
Figure 13.7. Some of the maps and images a weather forecaster might
analyze in time-lapse animations. (a) Hemispheric map of 500 hPa heights.
(b) Infrared satellite imagery. (c) 500 hPa heights and mean sea level
pressure.
 Satellite imagery is a good complementary resource, as it provides us
with a hemispheric view that we can use to assess quickly the amount of
cloud and water vapor associated with the 500 hPa troughs that we have
previously identified. Midlatitude cyclones at different stages of their life
cycle are evident, because we know their cloud signature (Figure 13.7(b)). By
comparing visible and infrared imagery, we can identify fog, low stratus,
and cumulus versus high cirrus, as well as thunderstorms.
With this broad picture in mind, we can now zoom in on a particular
region of interest and study the surface pressure and temperature fields in
more detail. We often combine them in different ways to obtain a synthetic
view. For example, combining 500 hPa heights with surface pressure
facilitates comparison between the position of upper-level troughs and
surface lows (Figure 13.7(c)). At this point we will focus on specific features
that we know to be locally important. For example, if a front is moving
through during the forecast period, we will look at weather station data to
characterize the strength of the front in terms of temperature contrast and
wind, and the type and intensity of the precipitation field.
If there is a chance of snow in the forecast, we will want to know the
exact position of the low, because that tells us the location of the overrunning
warm air, precipitating into the colder air mass underneath. Recall from
Figures 10.13–10.16 that it often snows north to northwest of the low
pressure center. Thus, if the low is passing to our south, we might be in the
snowfall area. If it is passing to our north, we might be spared, or experience
a situation where snow changes to rain. Therefore, we will pay close attention
to the numerical model forecasts of the low’s path, and to the temperature
patterns both at the surface and aloft. Recall that, even if it is below freezing
at the surface, warm air aloft can melt snow and result in freezing rain or ice
pellets. Forecasting frozen precipitation therefore increases the importance of
the temperature profile, which we monitor closely through radiosonde
profiles and other measurements.
Temperature profiles are also useful when forecasting the probability of
fog and how long it will last before burning off. If there is a temperature
inversion and the dew point is high with little wind, fog is more likely to
remain trapped. This is especially true when the sun is low in the sky, such as
during winter. Similarly, temperature profiles are critically important when
forecasting the probability of thunderstorms, as they tell us about the stability
of the atmosphere and how much buoyancy is available for the development
of deep cumulonimbus clouds. So we look not only at radiosonde profiles,
but also at temperature profiles produced by numerical model forecasts,
which can provide a guide to the timing and location for convection to
develop. This problem remains, however, one of the most difficult high-
impact weather events to forecast accurately.
When precipitation is forecast, radar images are very instructive, as
they tell us about the exact location of the rain patterns, their width and
structure, and the speed at which they are moving. Specific signatures in the
radar echoes provide information that can indicate the presence of hail, or the
level in the atmosphere where snow is melting.
Once the forecaster has developed a good familiarity with the features
that will affect the forecast, including the observed history and projections
into the future from numerical model forecasts, it is time to produce a specific
forecast. Common elements of a forecast consist of quantitative predictions
for low temperature, high temperature, and probability of precipitation.
Additional factors such as wind speed and cloud cover are often also
provided. If MOS is available for the forecast location, this is typically the
best starting point for such forecasts, as it is hard to beat on average. As we
discussed, though, there are situations where the forecaster can and should
adjust MOS predictions, such as when observations show the model forecast
is off track, during extremes, or in certain situations where MOS itself is
known to have a bias. (MOS only removes longer-term bias, not bias
conditional on specific weather features.)
Perhaps the most important thing that a forecaster provides, and this
cannot be automated, is a discussion of the forecast that provides the context
and physical basis for the forecast, an assessment of forecast uncertainty (or
confidence in the forecast), and an indication of what can go wrong. This is
often done by broadcast meteorologists, as well as being routinely found in
the Area Forecast Discussion issued by local offices of the US National
Weather Service. A frequent, high-impact, example is winter snowstorms,
where small changes in the track of the low center, for example, can have a
large impact on snow-amount forecasts at particular locations. In that case, as
discussed earlier, the forecaster should review the information coming from
ensemble forecasts, which sometimes shows that the uncertainty is skewed in
a particular direction. For example, although there may be uncertainty in the
track of the low center, the ensemble may show a preference for a southern
track as compared to a northern track, and that is useful information to
communicate in the forecast.
 13.6.1 Medium to Long-Range Forecasting
Even though it becomes difficult to forecast the details of upcoming weather
beyond about 5 days, weather services often issue a longer-term forecast
called the medium-range forecast, typically one to two weeks in advance,
on more general aspects, rather than the details found in a short-term forecast.
Instead of trying to predict exact temperatures, precipitation amounts, or
snow depth, medium-range forecasters look for critical features that are
persistent in the model forecasts, from run to run, and from model to model.
As such, ensemble forecasts are the primary tool for medium-range forecasts.
These forecasts usually associate large-scale patterns with changes in
climatological (historical average) probabilities for what normally happens
at the particular time of year. A persistent upper-level ridge, for example, is
often indicative of a greater chance of clear skies, and weaker winds. The
presence of an upper-level trough, however, may suggest a greater chance of
clouds and precipitation. Forecasters might mention that chances for
precipitation are higher or lower, without explicitly detailing the features that
will produce the precipitation. Figure 13.8 shows such an example of a 6–10-
day outlook and an 8–14-day outlook issued by the NOAA Climate
Prediction Center, and corresponding to our February cyclone.
 Figure 13.8. Example of a 6–10-day outlook ((a), (c), (e)) and an 8–14-day
outlook ((b), (d), (f)) from the NOAA Climate Prediction Center
corresponding to the four-day average conditions centered on February 21,
2014, at 00:00 UTC. (a), (b) 500 hPa heights (solid green contours) and
anomalies (dashed red contours). (c), (d) Temperature probability. (e), (f)
Precipitation probability.

These longer-term forecasts are becoming more accurate and valuable as

computer power increases and computer models improve our ability to
perform ensemble forecasting.
 13.6.2 Seasonal Outlook
Similarly, forecasters can use the departures from climatology based on long-
term indicators such as teleconnections to issue a seasonal outlook several
months in advance. The main source of predictive skill at these timescales is
associated with ENSO. For example, increasing sea surface temperatures in
the Central Pacific Ocean, announcing the return of an El Niño event, can be
the basis for predicting higher chances of rain and flooding for California in
late winter (i.e., higher chances than average). Here again, forecasters are not
trying to forecast the weather on a specific day three months in advance using
the techniques described for short-term forecasting. They are mainly using
statistical techniques to determine if weather variables will be average
(climatology), higher than climatology, or lower than climatology (i.e, the
seasonal-average conditions over a long period of time, typically 30 years).
These seasonal outlooks can be valuable for farming, energy planning,
recreational activities, and risk preparedness, for example. Again, with
increases in computer power, numerical models are now being used, coupled
to ocean prediction models, to make seasonal forecasts.
 Summary
The atmosphere is a fluid that obeys the laws of physics, and the
mathematical form of these laws can be encoded for computers to solve,
yielding a future prediction given the current state of the atmosphere. This
involves calculating the evolution of individual variables (such as velocity,
temperature, humidity) at certain points on a grid that covers a region of
interest. Weather observations provide the starting point for the forecast.
Our ability to forecast future weather is limited because:

► the observations used as initial conditions cannot be known exactly;

► certain small-scale processes are not resolved by the model grid and
must be parameterized, which translates into uncertainty and model
errors; and
► small errors in the initial conditions grow in time to become large
errors, ultimately completely contaminating the forecast.

The limit of weather forecasting skill is thought to be about two weeks, even
if we know the initial state almost perfectly. Ensemble forecasts, where
many forecasts are generated from slightly different initial conditions, and by
different forecast models, provide the tool by which we measure the loss of
skill in the forecast. Typically, small differences among the different
forecasts grow quickly, so by 5–10 days we cannot usually speak with
confidence on details in the forecast, but often there are broader patterns of
agreement for which we do have forecast confidence. Persistent features
allow weather forecasters to issue medium-range forecasts out to two
weeks.
 In spite of the chaotic nature of the atmosphere, forecasting its future
state is possible on longer timescales because there exist forcings that affect,
or constrain, the dynamics of the atmosphere and therefore the statistics of
weather events (e.g., global warming and ENSO). These forcings allow
forecasters to issue a seasonal outlook several months in advance and to
make forecasts of climate for several decades. Again, individual weather
features cannot be predicted on these timescales, but we can often make
useful predictions about how the weather, on average, will depart from
historical averages (climatology).
CHAPTER 14
Air Pollution
CONTENTS
14.1 Pollutants
14.2 Wind and Stability
14.3 Large-Scale Patterns
14.4 Topography
Summary

Among all the gases and aerosols constituting the atmosphere, some are
particularly detrimental to human health and to the environment. Their
concentration varies greatly from land to ocean, and from city to countryside.
Weather phenomena and the stability of the atmosphere play a large role in
mixing or concentrating these pollutants and also influence their formation.
Thus, in describing and understanding the distribution of these polluting
gases and aerosols, we will review and apply some of the fundamental
principles encountered in this book.
The atmosphere is composed of a number of gases in various concentrations
and particles of various sizes. We recall that some gases have a relatively
stable concentration (e.g., permanent gases such as nitrogen and oxygen),
while others can vary greatly in space and time (e.g., variable gases such as
water vapor and carbon dioxide). Among these variable gases, some are
harmless (e.g., water vapor) while others can impact the environment (e.g.,
CFCs) or human health (tropospheric ozone). Similarly, some aerosols do not
significantly affect human health (e.g., salt particles), while others, small
particles in particular, are very detrimental. After describing the most
common of these pollutants, we will explore how their distribution and
concentration are affected by weather.
 14.1 Pollutants
When describing the interaction between solar radiation and atmospheric
gases in Chapter 4, we encountered several pollutants of major importance:
chlorofluorocarbons (CFCs), which destroy the ozone layer in the
stratosphere and allow more ultraviolet radiation from the sun to penetrate
through the troposphere, causing sunburn and skin cancer, among other
ailments; and greenhouse gases such as carbon dioxide (CO2), methane
(CH4), and nitrous oxide (N2O), which play a major role in the greenhouse
effect (see more about global warming in Chapter 15). To these we should
add a number of gases and particulates that directly affect our health.
 14.1.1 Gases and Compounds
Most important among pollutants in gaseous form is carbon monoxide (CO),
which results from the incomplete combustion of carbon when there is not
enough oxygen (i.e., when the chemical concentrations are not in the optimal
ratio). This is typically the case in gasoline and diesel engines, and even more
so in older cars. Carbon monoxide is also released during forest fires, and in
the combustion of coal and biofuels.
Carbon monoxide is quickly absorbed by our blood and prevents the
absorption of oxygen. It is all the more dangerous that it is odorless and
colorless, and readily accumulates in enclosed spaces such as tunnels and
garages. Prolonged exposure to carbon monoxide can lead to death.
Sulfur dioxide (SO2) is released during the combustion of fossil fuels
containing sulfur, in particular oil and coal. It is often encountered
downstream of power plants and refineries, and is oxidized in the atmosphere
to form sulfuric acid (H2SO4). The new particulates then serve as
condensation nuclei for droplets to grow (see primary and secondary
pollutants in Section 14.1.2). Precipitation of these (mildly) acidic water
droplets is referred to as acid rain. Power plants and traffic exhaust also
release nitric oxide (NO) and nitrogen dioxide (NO2), more generally
referred to together as NOx. Nitrogen dioxide can also be oxidized in the
atmosphere to form nitric acid (HNO3) and can contribute to acid rain.
Nitrogen dioxide also contributes to the formation of ozone, through
complex photochemical reactions. As much as stratospheric ozone is
beneficial to life on Earth because it absorbs ultraviolet radiation,
tropospheric ozone (O3) in high concentrations causes eye irritation and
respiratory disorders. Ozone is also the main constituent of photochemical
smog, typical of cities such as Los Angeles, Beijing, and Mexico City (see
Section 14.1.3).
In recent years, a lot of attention has been given to a class of
hydrocarbons called volatile organic compounds (VOCs). While many
VOCs are not harmful in themselves, some are carcinogenic (e.g., benzene),
while others contribute to photochemical smog.
 14.1.2 Particulates
In addition to gases, our atmosphere contains particulate matter (PM) from
various sources. Some sources are natural, like volcanoes (ash and dust),
forest fires (smoke), deserts (dust), and vegetation (pollen). But large
quantities of particles are emitted by industries and vehicles (e.g., gasoline
and diesel engines). These particulates are harmful in themselves, since they
concentrate in our lungs and cause a number of respiratory problems. In
addition, certain gases, referred to as primary pollutants, can form liquid
particulates in moist conditions (secondary pollutants), on which water
vapor can readily condense, forming water droplets and smog – a
concatenation of the two words “smoke” and “fog.” Smog is frequent in
humid industrial areas, even when the relative humidity is less than 100%.
A lot of attention was first given to particulate matter smaller than 10
µm in diameter (PM10), which was identified to impact human health as early
as the 1980s. In 1997, however, and later, in 2006, the US Environmental
Protection Agency (EPA) created new standards to regulate ambient
concentrations of fine particles of diameter less than 2.5 µm (PM2.5), now
considered to be even more detrimental to human health, because they can
pass through the lungs and penetrate deeper into human tissue.
 14.1.3 Photochemical Smog
Certain gases and aerosols can take part in complex chemical reactions in the
presence of sunlight. The resulting mixture of gases and particles is referred
to as photochemical smog, where “photo” stands for “light.” In this context,
however, the term “smog” does not refer to the presence of water droplets, as
photochemical smog is usually a dry-air phenomenon. It refers to the
combination of poor or reduced visibility, aggressive odors, and breathing
problems. While ozone is considered the most important constituent of
photochemical smog by the EPA, many other chemicals contribute to a
significant level, in particular NOx and VOCs. Since photochemical reactions
depend on sunlight, photochemical smog is most common in sunny, dry
areas. It is enhanced by subsidence in anticyclonic conditions, and often
follows a diurnal cycle, with a maximum toward mid-day to late afternoon
and a minimum at night.
Recall that aerosols are ultimately removed from the atmosphere by dry
deposition, where turbulence mixes aerosols to the surface, or by
precipitation. In this latter case, the removal of aerosols occurs both because
they serve as condensation nuclei, and because precipitating raindrops
“scavenge” the pollutants on their way down. Small aerosols can remain
airborne for weeks and can pollute places far removed from their original
source.
The main vehicle for these pollutants is wind, which invites us to
explore the role played by weather in concentrating, transporting, and
dispersing pollution.
 14.2 Wind and Stability
Pollution that has been emitted by a point source (e.g., factory chimney, car
exhaust) will not travel far from the source by molecular diffusion alone,
which is very slow. The main mechanism for transporting pollutants is wind:
gases and particulate matter travel along and with air motion. Thus, a good
rule of thumb is that stronger winds tend to disperse pollution faster.
Additionally, local pollution from a point source can be quickly transported
vertically by convection and upward motion in midlatitude cyclones, to be
picked up and carried along by the jet stream. Once particulate matter and
gases reach the upper troposphere, they can travel large distances around the
hemisphere (Figure 14.1).

Figure 14.1. Long-range transport of pollution by vertical motion and by

the jet stream.
 Pollution sometimes remains at an intermediate level and is transported
and mixed by the local, synoptic-scale circulation, as illustrated in
Figure 14.2 with a Sahara desert sand plume transported over the North
Atlantic.

Figure 14.2. Desert sand plume blowing off the coast of West Africa over
the Atlantic Ocean on February 26, 2000.

Another more important factor to consider, however, in understanding

the dispersion of pollutants is mixing. The atmosphere is intrinsically
turbulent and, once in motion, is constantly stirred by swirling eddies on
many scales. Recall, for example, the turbulent eddies created by wind shear
close to the surface (Figure 1.14). As polluted air becomes mixed with clear
air by these turbulent eddies, pollution is redistributed over larger volumes of
air, both horizontally and vertically.
 The rate of mixing and the scales on which it takes place, however, vary
with atmospheric conditions. In particular, the stability of the atmosphere
determines whether convective overturning will quickly mix polluted air with
clear air aloft (unstable temperature profile), or polluted air will remain
confined to the lower layers of the troposphere (stable temperature profile).
To illustrate the effect of atmospheric stability on a plume, consider the
three situations depicted in Figure 14.3. Figure 14.3 (a) shows a typical
unstable temperature profile (the dry adiabatic lapse rate is indicated in blue
for reference). Recall from Chapter 6 that, in these conditions, rising air
parcels remain warmer than their environment and continue to rise. Exhaust
fumes from chimney stacks are typically much warmer than the surrounding
air. Therefore, they rise and, due to turbulent eddies at the edge of the plume,
mix with clear air. The plume of smoke is concentrated at the chimney exit
and becomes rapidly more diffuse farther up and away from the chimney
(Figure 14.3(b)).
 Figure 14.3. Dispersion of pollution in different conditions of atmospheric
stability. (a), (b) In unstable, convective conditions, the pollutants are
quickly mixed and transported upward. (c), (d) In stable conditions, the
pollutants spread out horizontally. (e), (f) If the atmosphere is unstable
under a subsidence inversion, the layer of pollution is trapped and mixed
downward. (a), (c), (e) Temperature profile shown in red.

In Figure 14.3 (c), the temperature inversion makes the lower

atmosphere stable. Therefore, the smoke neither rises nor sinks. Indeed, recall
that, in such conditions, displaced parcels tend to return to their level of
origin. Thus, the plume of smoke tends to flatten out horizontally in the
direction of the wind (if any). Figure 14.3(d) shows an example of a chimney
emitting smoke into the inversion layer. The presence of layered clouds
faintly visible in the background also suggests that the atmosphere is very
stable. (To be precise, however, the lower smoke plumes suggest that the
very bottom of the atmosphere might be mildly unstable, as described in the
next example.)
In Figure 14.3(e) large-scale subsidence has created an inversion layer,
but the warm surface has created a somewhat unstable layer underneath.
Therefore, the smoke plume can, at first, rise through the unstable layer, as
described in Figure 14.3(a). When it reaches the inversion, however, it
flattens out as described in Figure 14.3(c). Further downstream from the
chimney, turbulent motion in the unstable layer mixes some of the smoke
downward, which distributes the pollutants over the whole depth of this
lower layer and can cause a serious health hazard for people living in the
area, especially when the anticyclonic conditions responsible for the large-
scale subsidence are long lasting (Figure 14.3(f)).
 We refer to the lower part of the atmosphere where turbulent eddies are
present as the mixed layer. The depth of the mixed layer typically varies over
time. It decreases as subsidence intensifies, and increases as turbulent eddies
penetrate higher into the troposphere. In particular, the mixed layer deepens
in the afternoon, as the surface temperature increases and turbulent eddies
become more energetic. But it becomes more shallow at night, as subsidence
and the temperature inversion strengthen. In the early morning, when the
mixed layer is most shallow but industrial and city life resumes, the
pollutants are most concentrated – until turbulent mixing resumes and
deepens again.
Such pollution episodes typically end when a low pressure system
moves into the area and displaces the anticyclone. The vertical motion
associated with a frontal passage mixes the lower atmosphere and destroys
the temperature inversion. If the front brings precipitation, it helps dissipate
the pollutants, as they are scavenged (‘‘washed’’) by falling raindrops.
 14.3 Large-Scale Patterns
Our last example illustrates the role played by large-scale weather patterns in
controlling the fate of pollution. Anticyclones are often associated with a
stable atmosphere (i.e., no vertical transport) and light to no winds (i.e., no
horizontal transport). Therefore, pollutants tend to stagnate and to increase in
concentration if anticyclonic weather persists. That is often the case in Los
Angeles, California, for example, due to the presence of the Pacific High. If a
subsidence inversion forms, with some downward mixing of pollutants as
described in Figures 14.3(e)–(f), the layer of pollution can become trapped
and extend to the surface. In anticyclonic conditions, the atmosphere is also
often free of clouds, which allows sunlight to penetrate and cause
photochemical reactions, leading to the formation of photochemical smog.
Thus, the combination of all these factors can be conducive to a persistent
health hazard, as is well known of cities such as Los Angeles and Mexico
City (Figure 14.4).

Figure 14.4. Photochemical smog in Los Angeles, California.

 In contrast, low pressure systems are associated with convergence and
vertical motion, as well as stronger pressure gradients and wind. Therefore,
pollutants tend to be mixed efficiently and transported upward out of the
mixed layer, where they join the jet stream and may be quickly transported
away from the source (recall Figure 14.1). Additionally, clouds block
sunlight and reduce the potential for photochemical reactions, and
precipitation scavenges aerosols. Therefore, stagnating pollution and
photochemical smog are typically less of an issue in cyclonic weather
conditions.
 14.4 Topography
The concentration of pollutants in the lower troposphere is sometimes made
worse by topography. In particular, recall that mountain slopes cool by
emitting infrared radiation (or by emitting more infrared radiation than they
absorb visible light). As the adjacent air layer also cools (by conduction), it
becomes relatively denser and slides downward toward the valley bottom
(Figure 8.19). This cold air tends to accumulate in the valley, especially on
cold, clear nights, and even more so during anticyclonic conditions. In
valleys harboring industries and urban areas, pollutants follow this
downward, converging air motion and tend to accumulate at the bottom with
the cool air in the valley (Figure 14.5). This often contributes to the formation
of photochemical smog in Los Angeles (surrounded by the San Gabriel and
Santa Monica Mountains) and Mexico City (surrounded by Las Cruces and
Ajusco Mountains and the Sierra Madre), for example.

Figure 14.5. Concentration of pollution at the bottom of a valley.

 During the day, the mountain slopes warm by absorption of solar
radiation, and the adjacent air, warmed by conduction, moves upslope,
draining the pollutants out of the valley, as it were. But this only happens if
the warmer air is sufficiently buoyant to break the temperature inversion.
 Summary
Some chemicals and aerosols are particularly detrimental to human health
and the environment.

► CFCs destroy the ozone layer in the stratosphere and allow more
ultraviolet radiation from the sun to penetrate through the troposphere.
► Greenhouse gases such as carbon dioxide (CO2), methane (CH4),
and nitrous oxide (N2O) play a major role in global warming.
► Carbon monoxide (CO) prevents the absorption of oxygen in the
blood, which can lead to death.
► Sulfur dioxide (SO2) can be oxidized in the atmosphere to form
sulfuric acid (H2SO3) and acid rain. Nitric oxide (NO) and nitrogen
dioxide (NO2), more generally referred to together as NOx, can also
contribute to acid rain.
► Tropospheric ozone (O3) in high concentrations causes eye irritation
and respiratory disorders, and contributes to photochemical smog.
► Some volatile organic compounds (VOCs) are carcinogenic, while
others contribute to photochemical smog.

To this list of gases we add solid particles (aerosols) emitted by combustion.

Particulate matter of diameter less than 10 µm (PM10) concentrates in our
lungs and causes a number of respiratory problems. Fine particles of diameter
less than 2.5 µm (PM2.5) can pass through our lungs. Both are regulated by
the US Environmental Protection Agency.
Certain pollutants take part in complex chemical reactions in the
presence of sunlight to produce a noxious mixture of secondary pollutants
called photochemical smog, containing ozone, NOx, and VOCs, among
others.
Gases and particulate matter are transported and dispersed horizontally
by the local wind circulation, or vertically by convection and upward motion
in midlatitude cyclones, up to the jet stream, and around the hemisphere.
Several factors affect the dispersion of pollutants:

► The rate and scale of mixing vary with the stability of the atmosphere.
► When the atmosphere is unstable, exhaust plumes rise and mix with
clear air, and diffuse higher in the atmosphere.
► In the presence of a temperature inversion, exhaust plumes tend to
flatten out horizontally in the inversion layer.
► When the inversion layer caps an unstable layer underneath, the
pollution is mixed downward by turbulent motion, which distributes
the pollutants over the whole depth of the lower layer (i.e., the mixed
layer).

In anticyclonic weather (i.e., stable conditions), pollutants tend to stagnate

and to increase in concentration, and the formation of photochemical smog is
common in cities and industrial areas. In cyclonic weather (i.e., strong winds
and vertical motion), pollutants tend to be mixed more efficiently.
Pollution tends to be drained down and accumulate into valleys by the
sinking of cold, dense air along the slopes at night, and to be more mixed,
and thus diluted, during the day.
CHAPTER 15
Climate Change and Weather
CONTENTS
15.1 Past and Future
15.2 Changing Composition
15.3 A Warmer World
15.4 An Altered Water Cycle
15.5 Changing Global Wind Systems
15.6 Midlatitude and Tropical Cyclones in a Warmer World
15.7 Beyond Weather
15.8 The Forecast
Summary

How will weather change as our planet warms and our climate changes? Will
there be more droughts? How will precipitation patterns be affected? Will
there be more or fewer storms? Will tropical cyclones be more, or less,
intense? Will all regions of the world be affected? In this last chapter, we will
put in perspective all that we have learned about the atmosphere and apply it
to the pressing issue of global warming.
As we close the exploration of our atmosphere and weather, we must pause
and ask ourselves how they will evolve in the near future. Will weather be
any different on a warmer planet?
Throughout this book, we have come to understand that weather results
largely from temperature contrasts and energy transfer. Thus, since weather is
ultimately about energy redistribution, and global warming involves an
increase in the energy of the atmosphere, it is legitimate to wonder how this
additional energy will affect existing weather patterns. Although it is beyond
the scope of this book to provide a complete description of global warming
and its impacts, we will use this topic to apply what we have learned to
understand current thinking about the most likely evolution of our weather in
a warmer world.
 15.1 Past and Future
Earth’s climate has changed in the past and will continue to change in the
future. Although the time period over which humans have measured climate
directly with scientific instruments is limited to about the last 150 years, the
history of Earth’s climate is recorded indirectly, or by proxy, by living
organisms and geological processes. For example, analysis of the isotopic
composition of water molecules in Antarctic ice cores reveals that Earth’s
climate has gone through many cold and warm periods over at least the past
800 000 years (Figure 15.1). We know that these temperature variations are
due in part to cyclical changes in Earth’s orbit around the sun, as well as its
own rotational axis, producing changes that take place over thousands of
years. Earth’s path on the ecliptic plane, for example, does not always have
the same shape. It is at times more elongated (i.e., more elliptical), and at
other times more circular, following a cycle of about 100 000 years. As a
result, Earth is at times closer and at other times farther away from the sun
and receives more or less solar energy in different seasons. The tilt of Earth’s
axis with respect to the ecliptic also changes, with the North Pole pointing at
times more directly toward the sun, resulting in the summer hemisphere
receiving more sunlight, and at times less, with a cycle of about 40 000 years.
And Earth’s position and orientation after one revolution around the sun (i.e.,
after a year) also change over time with a cycle of about 26 000 years,
resulting in the seasons starting at different points of the ellipse throughout
the cycle. Because all these cycles happen over different periods of time, they
combine in a complex way to produce a varying input of energy into the
Earth system and, importantly, varying seasonal temperature contrasts
between the two hemispheres.

Figure 15.1. Temperature and CO2 variations over the last 800 000 years
reconstructed from the analysis of an Antarctic ice core.

Even though they are complex, however, orbital changes happen with
cyclical regularity and can be computed accurately (something the Serbian
mathematician Milutin Milankovitch did as early as 1920). It turns out that
orbital changes alone are not sufficient to explain the observed temperature
variations of the last 800 000 years. To explain the 8–10°C temperature
variations observed in Figure 15.1, an additional process needs to take place
that enhances the temperature increase or decrease associated with orbital
changes, i.e., a positive feedback mechanism.
We now know that this positive feedback is in part a greenhouse effect
feedback due to the release and intake of CO2 initiated by the temperature
variations. You can verify in Figure 15.1 that the CO2 concentration indeed
increases when the temperature increases, and decreases when the
temperature decreases. Once the positive feedback loop is initiated, of course,
each effect reinforces the other, as we described in Box 4.5.
Thus, the CO2 concentration responds to changes in temperature, and
both can vary in response to a number of natural phenomena, such as orbital
changes and volcanic eruptions: that is what we call the natural variability
of the climate system. But temperature can also respond to changes in CO2 -
concentration, and, most relevant for global warming, to the anthropogenic
changes that began with the onset of the Industrial Revolution (Figure 15.2).
To understand better this forced variability of the climate system, we
develop numerical models of the climate system to gain insights into future
climate changes resulting from increased greenhouse gas concentrations and
global warming.
 Figure 15.2. (a) Temperature variations (brown), 50 year running average
(red), 5th and 95th percentile (gray) over the past 1000 years. (b) Variations
in CO2 concentration over the same period.

We will now explore these predicted changes in relation to the topics we

have explored throughout this book.
 15.2 Changing Composition
We recall from Chapter 3 that the atmosphere is composed of permanent
gases (nitrogen, oxygen, and a number of trace gases) and variable gases such
as water vapor, carbon dioxide, and ozone (among others). Some of these
variable gases contribute to the greenhouse effect, such as water vapor,
carbon dioxide, methane, nitrous oxide, tropospheric ozone, and CFCs. Many
of these variable gases have increased in concentration over the last 200
years, in varying amounts. While the CO2 concentration, for example, has
increased by 40% compared to pre-industrial levels, from about 280 to 400
ppm, the concentration of methane has increased by 150% from about 700 to
1800 ppb (parts per billion).
If we place 400 ppm in Figure 15.1, for example, it is clear we are now
seeing CO2 concentrations that Earth has not experienced in the last 800 000
years. We are heading toward a future for which we cannot find an analog in
the recent climate archive.
 15.3 A Warmer World
We saw in Chapter 4 that an increase in greenhouse gas concentration results
in an increase in the heat content of the atmosphere through the enhancement
of the greenhouse effect. The global average temperature has already
increased by about 1 °C over the past 150 years (Figure 15.2), and current
climate models suggest that, depending on what decisions we make about our
use of fossil fuels and alternative energies, the planet will probably warm by
another 1–4 °C on average before the end of the twenty first century
(Figure 15.3). The Intergovernmental Panel on Climate Change (IPCC)
considers four different representative concentration pathways (RCPs), i.e.,
four different greenhouse gas concentration trajectories expressed in terms of
radiative forcing (net energy gain by Earth). An optimistic scenario,
assuming declining greenhouse gas emissions for the rest of the twenty first
century (RCP2.6, for a 2.6 W/m2 radiative forcing), suggests that the global
temperature increase could be kept at 1 °C or below. A pessimistic scenario,
assuming continuing greenhouse gas emissions (RCP8.5, for an 8.5 W/m2
radiative forcing), suggests that Earth could warm by 4 °C or more by the end
of the century.
 Figure 15.3. Prediction of the future global temperature increase over the
next century under four different scenarios. The red line represents the
mean over various climate models, while the blue shading represents the
standard deviation from the mean. See the text for a description of the
representative common pathways (RCPs).

Even though the overall projected temperature increase might seem

small in comparison to the wide temperature variations we can experience on
a daily or weekly basis (i.e., tens of degrees), it is important to realize that
this projection is an average for the entire globe. The temperature increase is
expected to be much higher in some places, such as the high latitudes and
drier regions of the tropics. It is also thought that a warmer world will feature
many more extreme heat events, such as the infamous European heat wave of
August 2003 that killed thousands.
Since land masses tend to heat up more than the oceans, which have a
greater thermal inertia, we can expect the energy distribution on Earth to
change. In fact, temperature changes observed during the twentieth century
give us an idea of what to expect in the next 100 years. When the temperature
was increasing globally by about 1 °C over the last 150 years, it was warming
much more in the northern polar region than at lower latitudes
(Figure 15.4(b)). By contrast, during the global temperature decrease
happening between 0 and 1850 AD, the northern polar region was cooling
much more than the rest of the planet (Figure 15.4(a)).

Figure 15.4. Spatial distribution of temperature changes between (a) 0 and

1850 AD, and (b) 1850 and 2000 AD.
 By extrapolation, we can expect the distribution and magnitude of
temperature contrasts to change even more over the next century, as the
planet continues to warm by 1–4 °C. Different temperature contrasts will lead
to different transfers of heat to redistribute the energy. Since these transfers
of energy are executed in large part by the atmospheric circulation, we can
expect global wind systems to be affected, and, as a consequence, daily
weather.
 15.4 An Altered Water Cycle
We learned in Chapter 5 that the saturation water vapor pressure (es) of air
increases with temperature, following the Clausius–Clapeyron relationship
depicted in Figure 5.4. As a rule of thumb, this relationship indicates that a 1
°C increase in temperature results in about a 7% increase in saturation water
vapor pressure. Assuming that the global-average relative humidity does not
change, which appears to be a good approximation, the evaporative flux from
the ocean will increase in a warming world, causing the water vapor content
of the lower atmosphere to increase. [Recall that RH = e/es. Therefore, for
RH to remain constant, e must increase in proportion to es. In other words,
the actual water vapor content will increase in proportion to the saturation
water vapor pressure in a warmer world.] This increase in atmospheric water
vapor has several important consequences.
First, recall the water vapor feedback loop (Box 4.5), which tells us that
more water vapor means more greenhouse effect, and therefore more
warming than from the other greenhouse gases alone. The increase in water
vapor will therefore contribute substantially to global warming.
Second, more water vapor in the atmosphere might enhance the
formation of clouds. This is subject to high uncertainty in climate models, as
cloud formation involves a wide range of scales from aerosols to global wind
systems, all of which cannot be resolved in detail. Processes at scales smaller
than the climate model grid box (on the scale of tens to hundreds of
kilometers) are parameterized in terms of larger-scale features, and these
approximations result in uncertainties about the type, number, and
composition of clouds. Low clouds, for example, are good reflectors of
visible light and good absorbers and emitters of infrared radiation (being
lower in the atmosphere, and therefore warmer), whereas high clouds are
poor reflectors of visible light and, being higher in the atmosphere, and
therefore colder, they emit less radiation than low clouds, while still
absorbing infrared radiation effectively. In other words, an increase in high
clouds would contribute to global warming (positive feedback), whereas an
increase in low clouds would tend to counteract global warming (negative
feedback) – although the strength and sign of these cloud feedbacks is still
debated in the climate community. In the end, capturing the exact distribution
of cloud types is essential in climate modeling, and that is currently an area
where climate models differ in details of their forecasts. Moreover, in
addition to changes in water vapor, the aerosol content of the atmosphere has
also been altered by human activity. In addition to altering the absorption and
scattering of radiation by the Earth system (the “direct effect” of aerosols),
increasing numbers of small aerosols due to combustion of fossil fuels are
changing how water is distributed in clouds. Specifically, the condensed
water in clouds is contained in a larger number of smaller droplets, which
makes clouds more reflective to solar radiation (the “first indirect effect” of
aerosols). Furthermore, changes in the droplet size affect precipitation: more
of the smaller cloud droplets and fewer of the larger ones makes collision and
coalescence less likely, and in particular less likely to produce precipitating
droplets (see Chapter 7). Reduced precipitation is expected to increase cloud
lifetime (the “second indirect effect” of aerosols).
Third, more water vapor and more energy in the system might result in
changes in precipitation patterns. Climate models suggest a global-average
increase of about 2% in precipitation for a 1 °C warming of the surface
temperature, which is notably lower than the expected 7% increase for water
vapor for such warming (which measures an average amount of water in the
atmosphere, whereas precipitation measures the rate at which water moves
through the atmosphere). The difference appears to involve global energy
constraints, where latent heat associated with precipitation must be
approximately balanced by atmospheric radiative cooling to space (which is
less effective for increasing greenhouse gas concentrations). It appears that
this radiative constraint limits the amount of latent heating, and hence
condensation, to precipitation increases of about 2%, rather than 7%. Local
changes in precipitation may be much larger, and, in particular, it appears that
for extreme precipitation events, such as with hurricanes, precipitation
intensity will increase at a rate closer to the 7% increase in water vapor.
Finally, the water cycle will be affected by changes in the general
circulation of the atmosphere. Recall from Chapter 9 that the general
circulation is driven by the equator-to-pole heat imbalance: low latitudes
receive an excess of radiative energy compared to the high latitudes, which
drives an overall atmospheric circulation transferring energy from the
equatorial and tropical regions to the midlatitudes and polar regions. This
takes the form of the Hadley circulation in the tropics and storm tracks in
middle latitudes. The temperature contrasts driving the Hadley circulation,
and therefore the Hadley cells themselves, are expected to weaken in a
warming world. However, even with a weaker circulation, the increase in
water vapor is expected to result in more precipitation in the Intertropical
Convergence Zone (ITCZ) – which is already relatively rainy. In the
subtropics, which are relatively drier, evapotranspiration (i.e., the loss of
water from the surface to the atmosphere by evaporation and plant
transpiration) associated with warmer temperatures will increase and lead to
additional drying.
Thus, overall it appears that precipitation will increase in rainy regions
and decrease in dry regions (Figure 15.5). Hence the often used expression,
“wet get wetter, while dry get drier.” Although somewhat of a simplification,
this statement captures the overall direction of the water cycle in a warmer
world.

Figure 15.5. Projected changes in precipitation by the end of the twenty

first century.
 15.5 Changing Global Wind Systems
The Hadley cells will not only become weaker in a warmer world, but also
they will expand. This poleward expansion of the Hadley circulation is less
well understood, but a robust feature of climate model simulations and in
observations during the late twentieth century to the present day. In
particular, dry regions associated with the poleward edge of the Hadley
circulation in the subtropics appear to be moving slowly poleward.
Additionally, the whole system comprising the two Hadley cells separated by
the ITCZ will shift northward. Recall from Chapter 9 that the position of the
ITCZ is determined not only by the maximum in insolation, but also by the
distribution of land masses, as well as dynamical factors. Since the northern
hemisphere contains a larger fraction of land masses (as compared to the
larger fraction of ocean in the southern hemisphere), the ITCZ is displaced
farther north in summer (e.g., monsoons) than it is displaced southward in
(northern) winter. By the same reasoning, the northern hemisphere will warm
more in the future, increasing the heat contrast between the two hemispheres,
which will displace the ITCZ further north. In other words, the Hadley cells
and ITCZ are displaced toward the hemisphere that is warming more, as the
atmospheric circulation attempts to redistribute the excess energy from the
northern to the southern hemisphere.
Due to similar changes in temperature contrasts, monsoon circulations
are expected to intensify, in particular in India and Southern China, with even
more precipitation on land.
 At higher latitudes and higher altitudes, recall that the strength of the jet
stream is determined by the pole-to-equator temperature contrast: as the
temperature contrast increases in winter, the jet stream strengthens; as it
decreases in summer, the jet stream weakens. Again, the absolute values of
temperature do not matter so much as the temperature difference between the
equator and the pole. In a warmer world, the Arctic region is expected to
warm significantly more than the rest of the planet. Consequently, even
though the overall temperature will increase, the pole-to-equator temperature
difference will actually decrease.
To illustrate, if the temperature at low latitudes is Tlow = 28 °C and the
temperature at high latitudes is Thigh = 4 °C, then the temperature difference
is

ΔT=Tlow=Thigh=24∘C.

Now, if the low latitudes warm up by 2 °C, but the high latitudes warm up by
4 °C, then the new temperature contrast is given by

ΔT=30−8=22∘C.

The temperature contrast has decreased, and therefore the jet stream will be
weaker. What this means for midlatitude cyclones is still unclear at present,
in part because cyclones are fueled not only by the pole-to-equator
temperature contrast, which is decreasing, but also by latent heat release,
which is increasing, as we now describe.
 15.6 Midlatitude and Tropical Cyclones in a
Warmer World
A weaker jet stream might be associated with weaker midlatitude cyclones;
more generally, with a weaker pole-to-equator temperature contrast, with less
heat to transfer poleward, we could think that midlatitude cyclones should
weaken. However, while extratropical cyclones result largely from the
unstable temperature contrast existing in the midlatitudes, they also draw a
fair amount of energy from the latent heat released during condensation of
water vapor into cloud droplets. As the latent heat is released, it warms the
air, warming the column and reducing the surface pressure, causing a
stronger storm. One additional complication is that cyclones move heat in the
form of water vapor (latent heat), as well as temperature, and, with more
water vapor in the atmosphere, storms will have the ability to transport more
heat. Therefore, a number of storm processes are affected by global warming,
and the impact of the combined effects remains uncertain.
Similarly, in the tropics, tropical cyclones are expected to change in a
warming world. We saw in Chapter 12 that the primary energy source for
these storms is heat content in the upper ocean, which increases in a warming
world. Theoretical studies show that the maximum potential intensity of
hurricanes, i.e., the maximum amount of energy they can release in optimal
conditions, increases with ocean temperature. Thus, we expect that hurricanes
forming over warmer water should be able to draw more energy from the
ocean by evaporation and to carry more water vapor and more latent heat
upward. In turn, latent heat can be released in the atmosphere through
condensation, which can fuel more convection and cause more intense
hurricanes. However, climate models also project an increase in the vertical
wind shear in the subtropics, which, as we recall from Chapter 12, inhibits
hurricane formation. Thus, two opposing effects will compete in a warmer
world, and it is difficult to predict which one will prevail on purely physical
arguments. Current expectations are that we might observe fewer hurricanes
overall as a result of the increased wind shear, but more of the strongest ones
as a result of the higher ocean heat content.
 15.7 Beyond Weather
Global warming is predicted to have many other impacts. As warmer oceans
expand, for example, and as land glaciers melt, the sea level is rising. Over
the past century, the global sea level has risen by about 10–20 cm. About half
of that increase is due to the thermal expansion of the oceans. This effect has
been amplified locally, such as near the northeast coastline of the US, where
weakening ocean currents have led to changes in the slope of the ocean
surface.
The two main reservoirs of ice on Earth are Greenland and Antarctica. If
the Greenland ice were to melt completely, sea level would rise
approximately 6 m, whereas if Antarctica melted completely sea level would
rise approximately 60 m. Obviously, such changes in sea level would be
catastrophic for human civilization, but neither outcome appears likely for the
next 100 years given current climate model simulations. However, polar ice
is melting at a faster rate than anticipated by climate model projections from
just 5–10 years ago. Although natural variability may explain part of this
discrepancy, if it is in fact more directly linked to global warming, then future
projections may also underestimate sea level rise. Given that roughly 80% of
all humans live within about 100 km of the ocean coastline, changes in sea
level are expected to have a dramatic impact on many of the world’s major
cities. Low-elevation coastal cities, such as Miami in Florida and New
Orleans in Louisiana are particularly vulnerable, because they are
occasionally affected by storm surge from hurricanes on top of the
anticipated rise in sea level.
 There is a long list of other climate change impacts. Permafrost (soil that
is frozen year round) at high latitude, for example, is thawing. As oceans
absorb more CO2, they are becoming more acidic, which affects their
chemistry and biology. Changing temperature and precipitation patterns put
plant and animal species under stress. While these changes to the
environment do not obviously involve weather directly, they contribute to
making climate change a complex issue, and the details of how climate
change will proceed under global warming are not yet fully understood.
 15.8 The Forecast
In Chapter 13 we discussed how the laws of physics, observations of the
atmosphere, and computer programs can be used to predict future weather.
We also discussed the limits of that endeavor due to the fundamental
uncertainty in observing the atmosphere completely with absolute precision,
and how it is important to measure that uncertainty as part of the forecast
process. A similar approach is used to make climate model forecasts, but
there are some important differences between weather and climate forecast
uncertainty that we will discuss here.
Current climate models predict a 1–4 °C temperature increase over the
next century (Figure 15.3). Why is there such a large range of uncertainty?
One unknown, of course, is our future use of fossil fuels and the
corresponding release of carbon dioxide into the atmosphere. Future fossil
energy use, however, is not the only source of uncertainty in the forecast. In
that regard, the climate forecasting problem shares similarities with, and
differences from, the weather forecasting problem. One clear similarity is that
both weather forecasting models and climate forecasting models are
physically based. They both resolve an initial-value problem and are sensitive
to uncertainties in the initial state, as well as uncertainties in the description
of key processes in the model. While uncertainty in weather forecasts derives
mostly from the initial conditions, in climate forecasts these uncertainties are
also linked to many physical processes, but especially to changes in clouds
and aerosols, both of which involve processes that are not well represented in
the forecasting models and involve the feedback processes discussed earlier.
Two key differences between weather and climate forecasting concern the
role of the ocean and forcing, both of which affect longer timescales of
climate forecasts much more than in weather forecasts.
Climate forecasting requires that we take into account both the
atmosphere and the ocean, because they interact on the longer timescales
considered in climate modeling, through exchanges of energy, water,
momentum, and gases (in particular, greenhouse gases). Therefore, we need a
coupled model. In contrast, atmospheric motion happens on much shorter
timescales than ocean dynamics, and the ocean surface can be described as a
static boundary, except in special circumstances, such as near hurricanes.
Including the ocean in climate models is, however, difficult because the
ocean – especially the deep ocean – is not nearly as well observed as the -
atmosphere. Ocean circulations take decades to centuries to redistribute heat,
and these slow processes affect the evolution of features in the atmosphere,
such as storm tracks and jet streams, on these longer timescales. Thus, the
ocean variability simulated by climate models is highly variable, which
contributes to uncertainty in climate forecasts, particularly during the first
few decades of the forecast. Uncertainty tends to decrease as one moves from
local effects to averages over larger areas. For example, there is often
relatively large uncertainty in changes in temperature over local land regions
strongly influenced by the ocean, such as West Coast cities, but over larger
areas that uncertainty decreases because the local effects become dominated
by the main global warming signal.
Forcings are very important for climate models because they can
increase predictability on long timescales. Recall from Chapter 13 that, even
though we cannot predict weather beyond a couple of weeks due to the
chaotic nature of the atmosphere (i.e., sensitivity to the details of the initial
state for the forecast), we can predict climate on very long timescales because
there exist forcings that influence the atmosphere on longer timescales than
those that dominate weather. In the case of global warming, the forcings from
greenhouse gas emissions are known within a range of certainty, and
therefore the effects on climate can be predicted.
Among all the factors contributing to climate change, fossil fuel
consumption is one that we can control by policy and behavior. If we
continue with “business as usual”, we are likely heading for the higher end of
the 1–4 °C range of temperature increase. If we decide to move away from
fossil fuels, however, we might limit the warming to the lower end of the
temperature range – although, given the amount of CO2 already emitted, it is
estimated that we will not be able to limit the temperature increase to 2°C, as
was proposed in 2009 at the Copenhagen Conference of the Parties (COP) of
the United Nations Framework Convention on Climate Change(UNFCCC).
 Summary
Earth’s climate has gone through many cold and warm periods in the
geologic past. Some of these changes are due in part to cyclical changes in
the orbit of Earth around the sun and the tilt of Earth’s axis, and to a
greenhouse effect feedback due to the release and intake of CO2 and water
vapor initiated by the temperature variations. These effects contribute to the
natural variability of the climate system.
More recently, Earth’s temperature has increased due to anthropogenic
changes in greenhouse gas concentrations that began with the onset of the
Industrial Revolution (global warming). The CO2 concentration, for
example, has increased by 40% compared to pre-industrial levels, from about
280 to 400 ppm, while the concentration of methane has increased by 150%.
As a result, the global average temperature has increased by about 1 °C
over the past 150 years, and current climate models suggest that the planet
will probably warm by another 1–4°C on average before the end of the
twenty first century. Moreover, the temperature increase is expected to be
much higher in places such as the high latitudes. Extreme heat events are
expected to become more frequent, and different temperature contrasts will
lead to modifications of weather systems.
The water vapor content of the lower atmosphere will increase in a
warmer world, which will lead to:

► an enhanced greenhouse effect, leading to more warming than from

the increase in CO2 alone;
 ► changing cloud patterns, although this is subject to high uncertainty in
climate models, in particular when it comes to capturing the exact
distribution of cloud types and the impact of increasing numbers of
small aerosols;
► more precipitation: climate models suggest a global-average increase
of about 2% in precipitation for a 1 °C warming of the surface
temperature, and more for extreme precipitation events, such as with
hurricanes; and
► weaker Hadley cells and drying in the subtropics, even though
precipitation at the ITCZ will increase due to the increase in water
vapor content – “wet get wetter, while dry get drier.”

The Hadley cells will also expand, shifting the dry subtropical regions
poleward, and the whole system made of the two Hadley cells separated by
the ITCZ will shift northward, affecting monsoon circulations, for example.
As the pole-to-equator temperature contrast decreases, the jet stream will
be weaker, which may weaken midlatitude cyclones. However, the increase
in latent heat release inside storms due to the formation of more cloud
droplets (higher water vapor content and higher condensation rate) may cause
stronger storms. Thus, the future of midlatitude cyclones remains unclear.
Similarly, a higher ocean heat content will be more favorable for the
intensification of hurricanes, but the increase in vertical wind shear will be
less favorable for the formation of hurricanes in the first place. Therefore, we
might observe fewer hurricanes overall, but more of the strongest ones.
Global warming is predicted to have many additional impacts, from sea
level rise and ocean acidification to the melting of Arctic sea ice, permafrost,
and the ice sheets in Greenland and Antarctica, among many others.
 There are uncertainties in current climate forecasts, linked to physical
processes that are not well represented in the forecasting models (e.g., clouds
and aerosols), to feedback processes, and to uncertainties in the
representation of the ocean, which play a larger role in climate forecasting
than in weather forecasting. But it appears that, if we continue consuming
fossil fuels at the current rate, we are heading for the higher end of the 1–4 °C
temperature range for the global average temperature increase predicted by
climate models.
 Glossary
absolute humidity a measure of the water vapor content defined as the mass
of water vapor per volume of air.
acid rain precipitating water droplets containing sulfuric acid or nitric acid
formed by dissolution of sulfur dioxide (SO2), nitric oxide (NO) and
nitrogen dioxide (NO2), typically released during the combustion of fossil
fuels such as oil and coal.
adiabatic cooling temperature decrease experienced by an expanding air
parcel as a result of the work exerted by the air parcel molecules on the
surrounding air during the expansion process.
adiabatic lapse rate the rate of temperature decrease (increase) experienced
by an air parcel ascending (descending) in the atmosphere, as a result of
adiabatic expansion (compression). For unsaturated air, the dry adiabatic
lapse rate is 10 °C/km; for saturated air, the moist adiabatic lapse rate is
about 6 °C/km.
adiabatic process a process during which a system experiences internal
energy changes due to work, but no transfer of energy or mass takes place
between the system and the surrounding environment.
adiabatic warming temperature increase experienced by a descending air
parcel as a result of the work exerted on the air parcel by the surrounding
air molecules as the air parcel is being compressed.
advection horizontal transport of a quantity (e.g., temperature) by air motion
(in the atmosphere) or water flow (in the ocean).
advection fog surface cloud formation by horizontal displacement of
relatively warmer and moister air (such as maritime air) over a colder
surface (such as cold land or a cold ocean current).
aerosol airborne solid particle of small size, such as contained in smoke,
dust, pollen, or pollution.
aggregation process by which two ice crystals collide and stick together to
form a larger crystal (as in the formation of snowflakes).
air mass volume of air of large horizontal extent, characterized by relatively
uniform temperature and humidity, typically determined by prolonged
interaction with the underlying surface (e.g., land or ocean surface at
particular latitudes).
air parcel imaginary volume of air with uniform characteristics that does not
exchange heat or mass with its surroundings; used especially in
comparisons with the surrounding environment when determining the
stability of the atmosphere.
albedo percentage of the incident visible light that is reflected by a surface.
Aleutian Low semi-permanent low pressure center located over the Aleutian
Islands (between Alaska and Siberia, in the North Pacific Ocean) in
average surface pressure maps, especially in winter.
altocumulus a type of cloud forming at mid-elevation in the troposphere
(i.e., between 2000 and 6000 m above ground level), and characterized
by cumuliform patches separated by clear areas. The patches are typically
smaller than in a stratocumulus cloud.
altostratus a type of cloud forming at mid-elevation in the troposphere (i.e.,
between 2000 and 6000 m above ground level), and taking the form of a
relatively uniform gray sheet. The sun can usually be detected through an
altostratus as a faint feature.
aneroid barometer a type of barometer in which atmospheric pressure is
measured through the volume change of a metallic bellows chamber,
which contracts or expands as a result of increasing or decreasing
pressure.
angular momentum a quantity defined, in scalar form, as the product of the
mass (m), radius of rotation (r), and tangential speed of the rotating body
(v): m ×r × v.
anticyclone a high pressure area characterized by weak to no cloud
formation, as well as weak to no wind – in particular in the center.
Around the anticyclone, winds blow clockwise in the northern
hemisphere and counterclockwise in the southern hemisphere.
Archimedes’ principle a law of physics indicating that, when submerged in
a fluid (partially or totally), a body experiences an upward force equal in
magnitude to the weight of the fluid that was displaced by the body. That
 upward force is also referred to as the buoyancy force.
assimilation (data) method by which observations of the atmosphere (made
by weather stations and, mostly, satellites) are incorporated (i.e.,
assimilated) into the mathematical representation of the atmosphere in a
numerical weather forecast model, before the model is run for a new
forecast cycle.
atmospheric pressure pressure exerted on a horizontal surface by a column
of air extending above that surface, due to the effect of gravity on that air
column. Atmospheric pressure is typically expressed in hectopascal,
millibars, millimeters of mercury, inches of mercury, kilograms per
square centimeter, or pounds per square inch.
atmospheric window a frequency band of the electromagnetic spectrum in
which the gases constituting the atmosphere absorb little to no radiation
(and therefore transmit it).
bar a unit of atmospheric pressure corresponding to 1000 hPa and roughly
one atmosphere.
barometer instrument used to measure atmospheric pressure.
Beaufort wind scale a wind scale used in marine meteorology, originally
based on observations of the effects of the wind on the sails of a ship, and
later extended to include observations of the sea state and land-based
effects.
Bergeron process classically called the Wegener–Bergeron–Findeisen
process, by which ice crystals grow at the expense of liquid droplets, in
the mixed-phase section of clouds, where both ice crystals and
supercooled liquid droplets coexist. The process derives from the fact that
the saturation vapor pressure over ice is lower than over water, so that air
saturated with respect to water is supersaturated with respect to ice.
Bermuda/Azores High a semi-permanent subtropical anticyclone found in
the North Atlantic Ocean and centered between the Azores and Bermuda.
black body an idealized body that absorbs all radiation at all wavelengths of
the electromagnetic spectrum. It also emits the maximum possible
electromagnetic radiation at all wavelengths according to Planck’s law.
black ice a thin layer of ice forming when liquid water covering a surface
(typically, roads) freezes due to a decrease in temperature, or when liquid
precipitation falls upon a surface at subfreezing temperatures and turns to
ice. Because the ice is transparent, we can typically see the black asphalt
 through it, hence its name.
blizzard a snowstorm, characterized by winds greater than 56 km/h (35
mph), which produces blowing and drifting snow that greatly reduces
visibility.
buoyancy force experienced by a body submerged, partially or totally, in a
fluid, and due to the difference between the pressure exerted by the fluid
at the top and the bottom of the body, as described by Archimedes’
principle.
carbon cycle transfer of carbon through various reservoirs of the Earth
system, such as Earth’s crust, plants, the ocean, the atmosphere, etc. by
various physical processes, such as photosynthesis, respiration, erosion,
and volcanic eruptions.
carbon dioxide a variable gas of the atmosphere made of one carbon atom
and two oxygen atoms (CO2) and contributing to the greenhouse effect.
Celsius (degrees) a unit of temperature named after Anders Celsius (1701–
1744), who created the first temperature scale based on the freezing point
and the boiling point of water as fixed points.
centrifugal force an apparent force experienced by a body in a rotating frame
of reference, directed outward from the center of rotation.
chinook wind a relatively warm and dry wind blowing downslope from a
mountain or elevated plateau, in which the subsiding air is warmed by
adiabatic compression. The chinook is a type of foehn, but the term
“chinook” is more often used in the USA and Canada, in reference to the
warm and dry winds blowing down from the Rocky Mountains into the
Great Plains to the east.
chlorofluorocarbons (CFCs) manufactured chemical compounds widely
used as refrigerants, solvents, and in aerosol sprays, and contributing to
the breakdown of stratospheric ozone and the so-called ozone hole over
Antarctica.
cirrocumulus a type of high cloud taking the shape of a patchy layer, in
which the patches are smaller in size than those found in an altocumulus.
cirrostratus A type of high cloud taking the form of a continuous layer
through which the sun can be seen, often surrounded by a 22° halo.
cirrus a type of high cloud (found above 5000 m and made of ice crystals)
taking the form of thin, wispy strands, sometimes producing virga in the
 form of falling ice crystals called “mares’ tails.”
cloud airborne mass of liquid water droplets or ice crystals.
cold front leading edge of a transitional region between a cold and a warm
air mass, characterized by a strong temperature gradient and a
displacement of the cold air toward the warm air.
cold tongue a region of cold surface water temperatures along the equator
extending westward from Peru in the equatorial Pacific Ocean, due to
upwelling of deep, cold water.
condensation a phase change from gas to liquid, used in particular when
water vapor turns to liquid water when in contact with a surface or a
condensation nucleus.
condensation nucleus a small solid particle (of size ~1 µm) that can serve as
a “seed” for water vapor molecules to condense and form a water droplet.
conditional instability atmospheric temperature profile for which moist air
parcels are unstable only for sufficiently large upward displacement; for
smaller displacements, the parcel is stable.
conduction transfer of heat by contact, as kinetic energy is transferred
between particles of matter through collision.
conservation of angular momentum (law of) a law of nature according to
which the angular momentum of a rotating body remains unchanged,
except when acted upon by a torque (a rotational force). When the mass
of a rotating object, such as an ice skater, moves toward the axis of
rotation, no torque is applied, angular momentum is conserved, and as a
consequence the rotation rate increases.
conservation of energy (law of) a law of nature stating that energy cannot be
created or destroyed in a system: it can only enter or leave the system, or
be transformed into other types of energy inside the system.
conservation of mass (law of) a law of nature stating that mass cannot be
created or destroyed in a system: it can only enter or leave the system, or
be transformed inside the system.
conservation of momentum (law of) a law of nature stating that the total
momentum of a system, where momentum is defined as the product of
the mass (or density) and velocity of the bodies constituting the system,
remains unchanged unless acted upon by a force, as implied by Newton’s
laws of motion.
convection vertical displacement of fluid parcels through the fluid due to
 buoyancy, and, by extension, the vertical transfer of heat, as the internal
energy of the parcels is transported with the parcels themselves.
Coriolis force an apparent force experienced by a body in motion in a
rotating frame of reference, directed perpendicular to the direction of
motion.
cumulonimbus a convective cloud of large vertical development, that
typically spans much of the troposphere, flattens out against the
tropopause or against a stable layer, and may produce one or more of the
following: heavy rain, hail, lightning and thunder, and severe weather,
such as tornadoes.
cumulus a low-level convective cloud of relatively small size, forming in a
mildly unstable atmosphere, and typically presenting a flat base and a
puffy, cauliflower-like appearance.
cumulus congestus a convective cloud of significant vertical development,
forming in a moist, unstable atmosphere, and typically merging with
surrounding convective clouds to obstruct the horizon.
cyclone a circulation developing around a low pressure region of the
atmosphere, and taking the form of a counterclockwise (in the northern
hemisphere) or clockwise (in the southern hemisphere) vortex.
density a measure of the concentration of a substance defined as its mass per
unit volume.
deposition a phase change by which matter changes state from gas to solid,
used in particular when water vapor turns to ice.
deterministic forecast a type of weather forecast based on the computational
solution of prognostic equations describing the future evolution of the
atmosphere based on physical principles.
dew formation of water droplets by condensation of water vapor onto solid
surfaces, typically onto low-lying elements such as grass.
dew point depression difference between the actual (measured) temperature
and the dew point temperature.
dew point temperature temperature at which an air parcel becomes
saturated if the air is cooled at constant pressure.
diurnal cycle a cyclical pattern of 24 hour period. Although it applies to
many quantities, it is typically used to describe the variation in
temperature due to the daily cycle in incoming solar radiation.
Doppler radar a type of radar using the frequency shift measured in the
 return signal to infer the speed of motion of precipitating particles relative
to the radar.
drizzle a type of precipitation characterized by water droplets smaller than
0.5 mm in diameter.
dry line boundary between two air masses of different moisture content as
defined by the dew point.
dynamic equilibrium a form of equilibrium in which the contents of the
system might be changing in time, but not in overall amount, because the
inputs are balanced by the outputs and there is no net change in the
system.
El Niño/Southern Oscillation (ENSO) a semi-periodical variation in the
coupled oceanic and atmospheric circulation taking place in the
equatorial region of the Pacific Ocean.
electromagnetic wave oscillation of the electric and magnetic fields that
propagates through space at the speed of light, characterized by its
frequency, or wavelength, and corresponding to different types of
radiation.
emission spectrum intensity of electromagnetic radiation emitted by a body
at each frequency, or wavelength.
ensemble forecast a type of forecast in which more than one computer
simulation is made in order to estimate the uncertainty in the forecast.
Each simulation is called an ensemble member, and consists of small
changes to the initial conditions and/or the computer model.
entrainment mixing of dry, unsaturated, environmental air into the saturated
regions of a cloud, resulting in the evaporation of cloud droplets.
environmental lapse rate rate of decrease of temperature with altitude,
measured in the atmosphere at a given time and location.
evaporation a phase change by which matter changes state from liquid to
gas, used in particular when liquid water turns to water vapor at a water–
air interface, such as Earth’s surface (ocean, lake, river, etc.) or the
surface of a water droplet (cloud droplet, raindrop).
evaporation fog a type of fog occurring when water evaporates from a warm
water surface into colder air and brings it to saturation.
extratropical cyclone a midlatitude weather system in which winds circulate
air around a low pressure center and transport energy from lower to
higher latitudes through the equatorward movement of cold air and
 poleward movement of warm air associated with cold and warm fronts,
respectively.
extratropics latitudinal band found poleward of the subtropics, also referred
to as the midlatitudes.
eye central region of a tropical cyclone, surrounded by the eyewall and
characterized by subsidence, relatively little middle and high cloud, and
relatively lighter winds.
eyewall ring of convective clouds surrounding the eye of a tropical cyclone.
Fahrenheit (degrees) a unit of temperature named after Gabriel Fahrenheit
(1686–1736), who designed the first mercury thermometer and invented a
temperature scale based on three fixed points: 0 in a mixture of ice, water
and sea salt; 32 in water and ice; and 96 “in the mouth or armpit of a
healthy man.”
Ferrel cell a feature of the general circulation of the atmosphere describing
the meridional overturning circulation in the midlatitudes, between the
Hadley cell and the polar cell.
foehn a warm and dry wind resulting from the flow of air over a mountain
and down the leeward slope, as the air subsides and warms at the dry
adiabatic lapse rate.
fog formation of cloud droplets in the lower atmosphere, near the surface,
due to a temperature decrease (radiation fog and advection fog) or an
increase in water vapor content to saturation (evaporation fog).
fragmentation process by which large ice crystals break into many small
parts that can then serve as ice nuclei to seed new ice crystals.
freezing nucleus a particle that can initiate the growth of an ice crystal from
supercooled liquid water droplets.
freezing rain rain that falls as liquid water raindrops, but freezes upon
impact with a cold surface at a temperature at or below 0°C, and forms a
thin layer of ice (typically “black ice” on roads).
friction See surface friction.
front a transitional region between a cold and a warm air mass, characterized
by a strong temperature gradient, a pressure trough, and a wind shift.
funnel cloud a rotating cloud emerging from the wall cloud at the base of a
cumulonimbus and extending downward toward the ground in the
process of tornado formation.
Fujita scale a tornado intensity scale designed by Tetsuya Fujita (1920–
 1998) that ranks the intensity of tornadoes using wind speed and the
damage caused by tornadoes.
general circulation large-scale, time-average organization of air currents
around Earth.
geostationary satellite a satellite in orbit around Earth, in the equatorial
plane and at such a distance (35, 786 km) that the satellite orbits at the
same angular velocity as Earth, so it remains positioned above the same
location on Earth.
geostrophic balance state of equilibrium in which the air moves in a straight
line along the isobars (height contours) due to a force balance between
the pressure gradient force and the Coriolis force, which are equal and
opposite, and directed at right angles to the direction of motion.
geostrophic wind conceptual wind speed and direction assuming the
pressure gradient and Coriolis forces balance, whereby the wind flows
parallel to isobars with a speed proportional to the magnitude of the
pressure gradient. Useful for quickly estimating the actual wind from
isobars and height contours.
global warming enhancement of the greenhouse effect due to the
anthropogenic increase in greenhouse gases such as carbon dioxide and
methane.
GOES US Geostationary Operational Environmental Satellite system that has
provided satellite imagery to support weather forecasting and research
since the 1970s.
gradient wind balance state of equilibrium in which the air moves in a
straight line along the isobars (height contours) due to a force balance
between the pressure gradient force, the Coriolis force, and the
centrifugal force due to the curvature of the flow, which are all directed at
right angles to the direction of motion.
graupel a type of precipitation in which typically conical, “spongy” ice
pellets form by riming of supercooled liquid water droplets freezing upon
impact as the pellets fall through the cloud.
greenhouse effect radiative effect by which Earth’s surface is at a higher
temperature than it would be in the absence of greenhouse gases (water
vapor, carbon dioxide, methane, ozone, etc.), due to the absorption and
emission of infrared radiation by these gases.
gust front leading edge of the outflow of cold, dense air spreading downward
 and outward from a thunderstorm.
Hadley cells a feature of the general circulation of the atmosphere describing
the average circulation of air in the subtropics, as warm, moist air rises
near the equator, flows poleward near the tropopause, subsides in the
subtropics, and returns to the equator near the surface.
hail a type of frozen precipitation formed by the freezing of successive layers
of ice as the ice pellet travels repeatedly through the updraft of a
cumulonimubus cloud collecting water droplets, cloud droplets, snow,
and ice crystals in layers with each pass through the updraft.
halo optical phenomenon appearing as a ring of light around the sun due to
the refraction and reflection of light by ice crystals, such as found in a
cirrostratus. Halos form at 22° and 46° angles from the sun, and the 22°
halo is most common.
haze suspended, dry, solid particles, such as dust, smoke, and pollution,
obscuring the sky and reducing visibility.
heat amount of energy that moves between bodies due to temperature
contrasts. On average, heat transfers energy from warm to cold objects.
heat capacity amount of energy required to raise the temperature of a mass
of matter by one degree Celsius.
hectopascal a unit of pressure, used in particular to express atmospheric
pressure. For example, the average sea level pressure on Earth is 1013
hPa.
heterogeneous nucleation formation of a liquid water droplet by
condensation of water vapor onto a surface, such as provided by a
condensation nucleus (i.e., an aerosol).
homogeneous nucleation formation of a liquid water droplet by
condensation of water vapor molecules onto themselves, without seeding
by an aerosol or a collecting surface.
hurricane See tropical cyclone.
hygrometer instrument used to measure the water vapor content of the
surrounding air.
ice nucleus a small particle that can serve as a seed for an ice crystal to grow.
Icelandic low a semi-permanent low pressure region found near Iceland and
most pronounced in winter.
ideal gas law equation of state describing the relationship between the
pressure, volume, mass (or density), and temperature of an “ideal” gas
 (where molecular collisions are perfectly elastic), typically expressed as
pV = nR*T, or p = ρRT, where p is the pressure, V is the volume, n is the
number of molecules in volume V, T is the temperature, ρ is the density,
R* is a universal constant that applies to any gas, and R is a gas constant
specific to the atmosphere.
Indian monsoon a tropical atmospheric circulation resulting from the
displacement of the ITCZ and Hadley cells in the Indian Ocean, and
enhanced by local geographical and dynamical features. Importantly, the
Indian monsoon alternates between a wet season (when the winds are
onshore) and a dry season (when the winds are offshore).
infrared a frequency band of the electromagnetic radiation spectrum
corresponding to wavelengths in the 1–1000 µm range (shorter
wavelengths than microwaves, but longer than ultraviolet).
infrared window a frequency band in the infrared region of the
electromagnetic radiation spectrum in which atmospheric gases do not
significantly absorb.
initial conditions initial state of a numerical weather forecast model before it
is integrated forward in time, in which the most recent observations have
been assimilated.
Intertropical Convergence Zone (ITCZ) a region of the atmosphere found
near the equator, characterized by convergence of the trade winds and
convection.
inversion (temperature) increase in temperature with height, appearing as a
positive slope in a vertical temperature profile, and therefore as a
deviation from the normal decrease of temperature with height in the
troposphere. By extension, the term also refers to the layer in which the
temperature increase with height is occurring.
ionosphere region of the atmosphere where atoms and molecules are ionized
(i.e., lose electrons) by solar radiation, thus affecting the propagation of
radio waves.
isobar a line of constant pressure.
isotherm a line of constant temperature.
jet stream upper-level westerly air current found near the tropopause,
between 30 and 60 degrees of latitude, in the northern and southern
hemispheres. The jet stream typically has meanders and locally maximum
wind speeds, and sometimes splits into two jets: a polar jet stream and a
 subtropical jet stream.
katabatic winds a dense, negatively buoyant air current flowing downslope
from a mountain or an elevated plateau under the effect of gravity. The
term is typically used to describe fast, cold, dense air currents flowing
down from the elevated ice sheets of Antarctica and Greenland.
kelvin a unit of temperature named after Lord Kelvin (1824–1907). The
Kelvin scale starts at absolute zero (–273.16°C) and increases with
intervals of the same magnitude as the Celsius scale (i.e., 1 K = 1 °C).
kinetic energy energy associated with motion that forms the basis for
defining temperature as a measure of the average kinetic energy of a large
number of air molecules.
knot a unit of wind speed, defined as one nautical mile per hour: 1 knot =
1.852 km/h = 1.151 mph.
land breeze a wind blowing from land to sea when the water is warmer than
the adjacent land, due to an overturning circulation caused by the
temperature contrast.
lapse rate rate of temperature decrease with height.
La Niña “cold” phase of El Niño/Southern Oscillation corresponding to the
maximum extent of the cold tongue in the equatorial Pacific Ocean.
latent heat amount of energy absorbed or released when matter changes
phase between the solid, liquid, and gas states while maintaining constant
temperature.
lenticular cloud elongated cloud, shaped like a lens, forming over the top of
a mountain as the horizontal flow of moist air is deflected upward by the
summit, causing adiabatic cooling and bringing the air to saturation.
level of free convection (LFC) altitude at which a lifted air parcel becomes
warmer than its environment and rises by the sole action of the buoyancy
force.
lifting condensation level (LCL) altitude at which a rising air parcel reaches
saturation due to adiabatic expansion and cooling.
lightning a discharge of electricity resulting from the separation of electrical
charge occurring when hailstones collide with smaller ice particles in a
cumulonimbus cloud.
mammatus cloud a type of cloud forming on the underside of a
cumulonimbus anvil, as dense air, cooled by evaporation, entrains cloud
particles from the anvil downward. A rare example of saturated descent
 of air.
mean sea level average level of the sea surface, used as a reference to
measure, for example, atmospheric pressure, so that all measurements are
comparable.
medium-range forecast a weather forecast established for the 3–10-day time
range, as opposed to short-range forecasting for the next 3 days.
mercury barometer instrument used to measure atmospheric pressure, made
of an inverted glass tube containing mercury up to the level where the
weight of the column of mercury is balanced by the atmospheric pressure
applied to the surface of the mercury in the reservoir.
mesopause transition region between the mesosphere and the thermosphere.
mesosphere layer of the atmosphere from about 50–90 km (1 hPa to 0.01
hPa), in which the ozone concentration decreases with height, along with
temperature, up to the mesopause.
meteogram a graph showing the time evolution of variables measured at a
weather station (temperature, dew point, atmospheric pressure, relative
humidity, wind, etc.).
methane a variable gas of the atmosphere made of one carbon atom and four
hydrogen atoms (CH4) and contributing to the greenhouse effect.
microburst isolated, short-lived, but very intense downdraft occurring under
a severe thunderstorm. These downdrafts can reach up to 150 knots, and
pose a serious hazard for aircraft during take-off and landing.
midlatitudes The 30°–60° latitude region found between the Hadley cell and
the polar cell in each hemisphere.
midlatitude cyclone see extratropical cyclone.
millibar a unit of atmospheric pressure equal to one hectopascal. For
example, the average sea level pressure on Earth is 1013 mb = 1013 hPa.
mixing ratio a measure of the water vapor content of the air, defined as the
mass of water vapor per mass of dry air.
momentum a quantity defined as the product of the mass and velocity of a
body in motion.
monsoon a seasonal, thermally driven atmospheric circulation due to uneven
heating of continents and ocean, associated with changes in surface wind
direction, precipitation, and temperature.
neutral stability a state of the atmosphere in which displaced air parcels
 neither rise nor sink, but remain at the same height.
nimbostratus a rain-producing stratus cloud.
nitric oxide gas, rapidly oxidized to nitrogen dioxide (NO2) in the atmosphere
and contributing to the formation of acid rain; nitric oxide (NO) and
nitrogen dioxide (NO2) are jointly referred to as NOx.
nitrogen a permanent gas made of two nitrogen atoms (N2) and constituting
about 78% of the atmosphere.
nitrogen dioxide gas, oxidized to nitric acid (HNO3) in the atmosphere and
contributing to the formation of acid rain; nitric oxide (NO) and nitrogen
dioxide (NO2) are jointly referred to as NOx.
nitrous oxide a variable gas of the atmosphere made of two nitrogen atoms
and one oxygen atom (N2O) and contributing to the greenhouse effect.
occluded front boundary between two relatively cold air masses, resulting
from the closing of the warm sector in a midlatitude cyclone and lifting of
the warm air mass aloft. Appears on a surface map as a trough in the
pressure field and an axis of warm air extending from the triple point to
the low center.
orographic precipitation precipitation resulting from the ascent of air along
the slopes of a mountain or mountain range.
overrunning ascent of warm air over a relatively cooler air mass along the
warm front of a midlatitude cyclone.
oxygen a permanent gas of the atmosphere made of two oxygen atoms (O2)
and constituting about 21% of the atmosphere.
ozone a variable gas of the atmosphere made of three oxygen atoms (O3).
Tropospheric ozone is detrimental to human health and can cause severe
respiratory problems. Stratospheric ozone, however, absorbs ultraviolet
radiation, and thus protects us from the damaging effects of ultraviolet
energy (in particular, skin cancer).
ozone hole region of the stratosphere, above Antarctica, where the
destruction of the ozone layer by chlorine compounds over the last 50
years has been most pronounced.
ozone layer region of the stratosphere where the concentration of ozone
increases and reaches a maximum, at an altitude of about 50 km.
Pacific High a semi-permanent high pressure region found over the North
 Pacific Ocean near Hawaii.
Pacific warm pool western region of the equatorial Pacific Ocean,
characterized by higher temperatures than found in the eastern basin (i.e.,
in the cold tongue).
parameterization mathematical characterization of small-scale processes,
which cannot be fully resolved in a numerical model, in terms of larger-
scale processes and variables.
permanent gases gases whose relative concentration (relative to each other)
does not vary much over time and space, up to about 80 km. For
example, oxygen always represents about 21% of atmospheric gases.
phase a state of matter: solid, liquid, or gas.
photosynthesis chemical transformation of water and carbon dioxide into
organic matter, given the appropriate amount of radiative energy
(sunlight).
polar cell overturning circulation occurring between about 60 o of latitude
and the pole in each hemisphere, as part of the general circulation of the
atmosphere.
polar front pronounced meridional temperature gradient found in the
midlatitudes in each hemisphere.
polar-orbiting satellite a satellite orbiting Earth from pole to pole at about
800 km altitude.
ppm parts per million, a unit used to express gas concentration.
precipitation liquid or solid water particles falling faster than the updrafts in
which they are embedded.
pressure a quantity characterizing the frequency and intensity of average
molecular collision, defined as the net force per unit area.
pressure gradient force force resulting from the existence of a pressure
gradient that, in the absence of other forces, acts to accelerate air from
high to low pressure.
prognostic equations equations from the laws of physics describing the
evolution of a set of variables such as temperature, humidity, and wind,
which allow us to calculate (i.e., make a prognosis of) the future state of
the atmosphere.
radar instrument that estimates the location and size of precipitating solid or
liquid water particles based on the timing and intensity of returning
electromagnetic waves after they have been emitted by the radar and
 reflected by the particles.
radiation oscillation of the electromagnetic field that manifests itself as
waves of various wavelengths propagating at constant speed (i.e., the
speed of light).
radiation fog formation of cloud droplets due to the decrease in temperature
resulting from a loss of energy by emission of infrared radiation near the
ground (synonymous with “ground fog”).
radiative equilibrium a state whereby a body emits the same amount of
radiative energy as the amount it absorbs – although not necessarily at the
same wavelength.
radiosonde instrument that measures variables of the atmosphere as it
ascends with a helium-filled balloon, and radios the measurements back
to a receiving station.
rain precipitating liquid water particles.
rain gauge instrument that measures the amount and, for certain designs, the
rate of liquid precipitation.
reflection process by which the direction of propagation of electromagnetic
waves reverses at the incident angle, without absorption, when the waves
encounter an obstacle.
relative humidity a measure of the water vapor content defined as the
percentage of water vapor present in the air, compared to the maximum
amount at saturation (i.e., actual water vapor pressure divided by
saturation water vapor pressure, e/es).
respiration a chemical reaction by which organic matter is broken down into
water and carbon dioxide in the presence of oxygen, and energy is
released.
ridge (pressure) axis of relatively high values in a pressure field without
reaching a local maximum, analogous to a mountain ridge on a
topographic map.
riming a process by which supercooled liquid water droplets freeze onto an
existing ice crystal during a collision – a process also called accretion,
which tends to form conical pellets called graupel.
Saffir–Simpson hurricane wind scale a scale used in tropical meteorology
to indicate the intensity of hurricanes (i.e., tropical cyclones) based on
their maximum sustained wind speed.
saturation a state of equilibrium in which the amount of water vapor
occupying a given space is maximum.
saturation water vapor pressure partial pressure of water vapor at
saturation; it is proportional to temperature according to the Clausius-
Clapeyron equation.
scattering reflection of electromagnetic radiation in multiple directions.
sea breeze a wind blowing from sea to land when the land is warmer than the
adjacent sea, due to an overturning circulation caused by the temperature
contrast.
seasonal cycle cyclical variation of period one year of the atmospheric
variables due to the changes Earth experiences in its exposure to solar
radiation as it revolves around the sun.
selective absorber a molecule that absorbs radiation at specific wavelengths,
but not at others – in contrast to a black body, which absorbs all incident
radiation.
shelf cloud a cloud forming over the leading edge of an advancing gust front,
as warmer air is lifted by the cool air spreading outward underneath a
thunderstorm.
sleet (ice pellets) precipitating frozen water drops, which are in liquid form at
some point during their fall, but freeze before touching the ground as they
fall through a colder layer.
smog formation of cloud droplets (fog) in a particularly polluted
environment, i.e., in the presence of numerous aerosols and chemicals,
such as those produced by industries (historically, smoke in London).
snow precipitating ice crystals and ice-crystal aggregates that formed in the
cold part of a cloud and did not melt before reaching the surface.
specific humidity a measure of humidity defined as the mass of water vapor
per mass of moist air.
squall line a line of thunderstorms usually associated with an advancing
front.
stable (air, atmosphere) a state of the atmosphere in which displaced air
parcels return to their original altitude when released.
station model a set of conventional symbols used to synthesize the
measurements and observations made at a weather station or buoy.
stationary front transitional region between a cold and a warm air mass
when neither mass is advancing against the other.
storm tracks preferential locations for midlatitude cyclone development, in
particular over the Western Pacific and West Atlantic oceans, due to a
combination of factors (a strong meridional temperature gradient,
prominent land masses and land–ocean temperature contrasts, and a
weakly stable atmosphere).
stratocumulus a type of low cloud characterized by an overall stratiform
appearance, but broken up in patches, typically due to the presence of
mild convection or turbulence.
stratopause boundary between the stratosphere and the mesosphere.
stratosphere stable layer of the atmosphere found between about 10 and 50
km where the temperature increases due to the presence of ozone and the
absorption of ultraviolet radiation.
stratus a type of cloud forming in the lower troposphere (i.e., below 2000 m
altitude), and taking the form of a uniform gray sheet.
subgeostrophic wind wind occurring when the flow is curved and around a
low pressure region, or trough, and so that the wind speed is less than the
geostrophic wind due to flow curvature.
subpolar low a low pressure region forming on the poleward side of the
midlatitudes, typically between 50° and 70° of latitude, as in the Aleutian
Low and the Icelandic Low observed in average pressure fields.
sublimation phase change by which matter changes state from solid to gas,
without going through the liquid state (in meteorology, from ice to water
vapor).
subsidence downward motion of air, accompanied by adiabatic compression
and warming.
subtropical high a semi-permanent high pressure region found between 20°
and 40° of latitude and over the ocean, and strongest in summer.
subtropics latitudinal belt found in each hemisphere on the poleward side of
the tropics, and considered to be between about 20° and 40° of latitude,
with some geographical variations in places.
supercell thunderstorm a long-lived rotating thunderstorm with updrafts
separated from downdrafts. Hail often occurs with these storms, and
occasionally tornadoes.
supercooled (water) water cooled below 0 °C without a change of state from
liquid to solid (i.e., without freezing).
supergeostrophic wind wind occurring when the flow is curved and around
 a high pressure region, or ridge, and so that the wind speed is greater than
the geostrophic wind due to flow curvature.
supersaturation a state in which the air contains more water vapor than the
maximum capacity allows at that temperature, i.e., when the water vapor
pressure is greater than the saturation water vapor pressure (e > es), which
means that the relative humidity is greater than 100%.
surface friction drag on air motion due to roughness elements at the surface,
which tends to reduce the speed of the wind close to the surface, among
other effects.
sustained wind speed of the wind averaged over a standard time period, such
as one minute.
synoptic scale a spatial scale encompassing weather events on the order of a
few hundreds to a few thousands of kilometers (e.g., typical high and low
pressure systems in the midlatitudes).
teleconnection remote influence of weather events on distant locations in the
atmosphere.
temperature a measure of the heat content of air, therefore related to the
kinetic energy of air molecules.
thermal cloud-free ascending motion occurring in a convective cell, typically
on sunny afternoons over open terrain.
thermocline region of strong temperature gradient separating the well-mixed
and relatively warmer surface layer of the ocean from the deeper and
colder ocean.
thermometer instrument used to measure temperature.
thermosphere upper layer of the atmosphere, above about 100 km, in which
air density is extremely low and the motion of individual air molecules
extremely fast.
thunder a sound shock wave resulting from the rapid heating and expansion
of the channel of air through which electricity flows during a lightning
strike.
thunderstorm a cumulonimbus cloud characterized by strong updrafts and
downdrafts, and usually producing heavy showers, lightning, thunder,
and possibly hail.
tornado a rapidly rotating vortex occurring with cumulonimbus clouds, tens
to hundreds of meters in diameter, characterized by extremely low central
 pressure and high winds of up to 250 knots and sometimes more.
Tornado Alley a region of the US Midwest where environmental conditions
and atmospheric dynamics are favorable for the formation of tornadoes.
trade winds surface winds blowing at low latitude as part of the lower
branch of the Hadley cells. On average, trade winds are northeasterly in
the northern hemisphere, southeasterly in the southern hemisphere, and
converge at the ITCZ.
tropical cyclone tropical weather system developing over the ocean, deriving
its energy from the evaporation of ocean water, and taking the form of a
vortex of strong winds (sustained winds greater than 64 knots) blowing
cyclonically around a low pressure center (the eye), with spiraling bands
of cumulonimbus clouds producing heavy rain.
tropical depression a precursor to a tropical storm in the Saffir–Simpson
hurricane intensity scale, with 20–34-knot winds.
tropical disturbance a precursor to a tropical depression in the Saffir–
Simpson hurricane intensity scale, with 0–20 knot winds.
tropical easterly wave a trough of low pressure in the tropical easterly flow
in which thunderstorms congregate on the east side of the trough axis due
to convergence and enhanced convection. Under favorable circumstances,
a tropical easterly wave can evolve into a tropical storm.
tropical storm a precursor to a tropical cyclone in the Saffir-Simpson
hurricane intensity scale, with 34–64-knot winds.
tropics latitudinal belt delimited, broadly, by the Tropic of Cancer in the
northern hemisphere and the Tropic of Capricorn in the southern
hemisphere.
tropopause boundary between the troposphere and the stratosphere.
troposphere first layer of the atmosphere extending from the surface to the
altitude where temperature stops decreasing with height (the tropopause).
Essentially all weather occurs in this layer.
trough (pressure) axis of relatively low values in a pressure field without
reaching a local minimum, analogous to a valley on a topographic map.
typhoon name given to a tropical cyclone when it forms over the western
North Pacific Ocean.
ultraviolet a frequency band of the radiation spectrum corresponding to
wavelengths in the 0.005–0.4 µm (5–400 nm) range (shorter wavelengths
than visible radiation, but longer than x-rays).
uncertainty range of possible values (error) when a variable is measured
(due to instrumental errors) or predicted (due to errors in the initial
conditions and model errors).
unstable (air, atmosphere) a state of the atmosphere in which air parcels
that are displaced accelerate in the direction of the displacement due to a
buoyancy force. For upward displacement, this means an upward
buoyancy force and upward acceleration.
upwelling vertical displacement of water from the deep ocean to the surface
due to the divergence of surface waters.
valley breeze warming and upward displacement of air along the slopes of a
valley during the day, when the slopes are warmed by the absorption of
sunlight.
valley fog formation and concentration of fog at the bottom of a valley, due
to the downslope drainage of cold, dense air into the valley at night, when
the slopes cool by emission of infrared radiation.
variable gases gases whose concentration varies in the atmosphere over
space and time, such as water vapor and carbon dioxide.
virga precipitating particles that evaporate or sublimate before reaching the
surface.
Walker circulation a large-scale circulation of the atmosphere above the
equatorial Pacific Ocean due to the temperature contrast between the
Pacific warm pool and the equatorial cold tongue.
wall cloud a rotating cloud at the base of a cumulonimbus (i.e., a
thunderstorm) preceding the possible formation of a funnel cloud and
tornado.
warm front warm edge of a transitional region between a warm and a cold
air mass, characterized by a strong temperature gradient, a pressure
trough, wind shift, and the overrunning of warm air over cold air
producing clouds and precipitation.
warm sector region delimited by the cold and warm fronts in a midlatitude
cyclone.
water vapor water in the gas phase.
water vapor pressure a measure of the amount of water vapor in the
atmosphere defined as the pressure exerted by water vapor alone, as if all
other gases were absent.
wavelength distance between two equal phases of a wave, such as two crests
 or two troughs.
weather station assemblage of various weather instruments at conventional
heights or, more generally, the location where these meteorological
observations are made.
wind gust a local, sudden, and brief increase in wind speed due to the
transport of higher momentum air from a different altitude by turbulent
eddies.
wind shear a change in the wind in a given direction, often referring to a
change in wind speed in a direction normal to the direction of the wind,
e.g., a vertical increase in horizontal wind, or a meridional increase in
zonal wind.
wind vane instrument used to measure wind direction.
 References
Dee, D. P., Uppala, S. M., Simmons, A. J. et al. (2011). The ERA-Interim
reanalysis: configuration and performance of the data assimilation
system. Quart. J. Roy. Meteorol. Soc., 137, 553–597, doi:10.1002/qj.828.
Frank, D. C., Esper, J., Raible, C. C. et al. (2010). Ensemble reconstruction
constraints on the global carbon cycle sensitivity to climate. Nature, 463,
527–532, doi:10.1038/nature08769.
Hakim, G. J., Emile-Geay, J., Steig, E. J. et al. (2016). The Last Millennium
Climate Reanalysis Project: framework and first results. J. Geophys. Res.
Atmos., 121, doi:10.1002/2016JD024751.
Jouzel, J., Masson-Delmotte, V., Cattani, O. et al. (2007). Orbital and
millennial Antarctic climate variability over the past 800,000 years.
Science, 317(5839), 793–797, doi:10.1126/science.1141038.
Lemon, L.R. and Doswell, C.A. (1979). Severe thunderstorm evolution and
mesocyclone structure as related to tornadogenesis. Month. Weather Rev.,
107, 1184–1197.
Lord, S. D. (1992). A new software tool for computing Earth’s atmospheric
transmission of near- and far-infrared radiation. NASA Technical
memorandum 103957.
 Credits
Satellite images, land weather station, and buoy measurements were obtained
from the National Climate Data Center ([Link]) at the National
Oceanic and Atmospheric Administration (NOAA). Radiosonde
measurements were obtained from the Earth System Research Laboratory
([Link]) at NOAA. ERA-Interim numerical analyses were obtained
from the European Centre for Medium-Range Weather Forecasts (ECMWF)
after Dee et al., The ERA-Interim reanalysis: configuration and performance
of the data assimilation system, Quarterly Journal of the Royal
Meteorological Society (2011). Infrared transmission spectra were obtained
from the Gemini Observatory website ([Link]) after Lord (1992).
NASA images were obtained from the Earth Observatory
([Link]).
All maps were created by the authors using Python version 3.4
([Link]). All photographs have been provided by the authors,
except for those in the list below.

Introduction

p. xi Image from Caroline Planque

Chapter 1
Title Image from Gareth James Foreman / Freerange Stock / Public
banner Domain

1.2 Image from CambridgeBayWeather / Wikimedia Commons /

Public Domain

1.3 Data from NOAA

1.4 Image from NOAA / Public Domain

1.18 Image from Famartin / Wikimedia Commons / Public Domain

1.19 Image from CambridgeBayWeather / Wikimedia Commons /

Public Domain

1.20 Image from NOAA / Public Domain

1.21 Data from NOAA

Chapter 2

Title Image from Pixabay / Public Domain

banner

2.1 Data from NOAA

2.2 Data from ECMWF

2.4 Data from ECMWF

2.2.1(a), Data from NOAA

(b)

2.7 Data from ECMWF

2.10 Data from ECMWF

2.12 Courtesy of the College of DuPage NEXLAB

2.13 Image from NASA / Public Domain

2.15 Image from NASA / Public Domain

2.18 Image from NASA / Public Domain

2.20 Image from NASA / Public Domain

2.22 Courtesy of the Space Science and Engineering Center

Datacenter, University of Wisconsin-Madison

2.25 Image from NASA / Public Domain

2.26 Image from NASA / Public Domain

2.27 Image from NASA / Public Domain

2.28 Image from NASA / Public Domain

2.29 Image from NASA / Public Domain

Chapter 3

Title Image from NASA / Public Domain

banner

3.1 Data from NOAA, after Frank et al. (2010)

3.2 Image from NASA / Public Domain

Chapter 4

Title Image from Splitshine / Public Domain

banner
4.1.1 Library of Congress / Wikimedia Commons / Public Domain

4.1.2 Image from K. Schoenbauer / Smithsonian Libraries / Public

Domain

4.1.3 University of Frankfurt / Wikimedia Commons / Public Domain

4.1.4 Image from Nobel Media AB 2014 / Wikimedia Commons /

Public Domain

4.1.5 Image from Wikimedia Commons / Public Domain

4.2.1 Image from NASA / Public Domain

4.2.2 Image from Caroline Planque

4.6 Data from the Gemini Observatory after Lord (1992)

Chapter 5

Title Image from Caroline Planque

banner

5.5 Image from Freerange Stock / Public Domain

5.6 Image from Aleksandar Momirovic / Freerange Stock / Public

Domain

5.7 Data from ECMWF

5.8 Data from NOAA

5.9 Data from NOAA

Chapter 6
Title Image from Tim Mossholder / Pexels / Public Domain
banner

6.1.1 Getty Images

6.1.3 Image from Caroline Planque

6.9(a) Image from NASA

6.10 Image from NASA

6.21 Image from Caroline Planque

6.23 Getty Images

6.26 Image from NASA

6.29 Image from Wikimedia Commons / Public Domain

6.30 Image from Caroline Planque

6.31 Image from Caroline Planque

6.35 Getty Images

Chapter 7

Title Image from Caroline Planque

banner

7.6 Image from Caroline Planque

7.11 Image from Reid Wolcott

7.15 Image from Caroline Planque

Chapter 8

8.15 Data from NOAA

8.16 Data from ECMWF

8.20 Image from NASA

Chapter 9

Title Courtesy of the Museum of the History of Science, Oxford, UK

banner

9.1 Data from NOAA

9.2 Data from NOAA

9.3 Data from NOAA

9.4 Data from NOAA

9.5 Data from NOAA

9.6 Data from NOAA

9.7(b) Data from NOAA

9.9 Image from NASA

9.20 Data from NOAA

Chapter 10

Title Image from NASA

banner
10.4(a) Image from NASA

10.5(a) Image from NASA

10.13 Data from ECMWF and NASA data

10.14 Data from ECMWF and NASA data

10.15 Data from ECMWF and NASA data

10.16 Data from ECMWF and NASA data

10.19(a) Data from ECMWF

10.19(b) Image from NASA

10.20 Image from the American Meteorological Society

10.21 Image from the American Meteorological Society

10.22 Image from the American Meteorological Society

10.23 Image from the American Meteorological Society

Chapter 11

Title Image from Caroline Planque

banner

11.3 Getty Images

11.4 Image from Uli Feuermeister / Wikimedia Commons / Public

Domain

11.6 Image from skeeze / Pixabay / Public Domain

11.7 Image from Sensenmann / Wikimedia Commons / Public

 Domain

11.9 Image from Jared Davidson / Freerange Stock / Public Domain

11.13 Image adapted from Lemon and Doswell (1979)

11.14 Getty Images

11.15 Getty Images

11.16 Getty Images

11.17 Getty Images

11.18 Getty Images

Chapter 12

Title Image from NASA

banner

12.1(a) Image from NASA

12.2 Data from NOAA and Remote Sensing Systems

12.3 Image from NASA

12.6 Data from ECMWF

12.7(a) Data from ECMWF

12.7(b) Data from ECMWF and NASA data

12.8 Data from NOAA

12.10 Data from ECMWF and EUMETSAT data

Chapter 13

Title Courtesy of the Department of Atmospheric Sciences at the

banner University of Washington

13.3 Data from NOAA

13.4 Data from NOAA

13.5 Data from NOAA

13.6 Images from NASA

13.7(a) Data from ECMWF

13.7(b) Image from NASA

13.7(c) Data from ECMWF

13.8 Image from NOAA

Chapter 14

Title Image from Foto-RaBe / Pixabay / Public Domain

banner

14.2 Image from NASA

14.3(b) Image from czu_czu_PL / Pixabay / Public Domain

14.3(d) Image from David Karich / Pixabay / Public Domain

14.3(f) Image from Igors Jefimovs / Wikimedia Commons / Public

Domain

14.4 Getty Images

Chapter 15

Title Image from Caroline Planque

banner

15.15 Data from NOAA, after Jouzel et al. (2007)

15.2(a) Data from Hakim et al. (2016), Last Millennium Climate

Reanalysis Project

15.2(b) Data from NOAA, after Frank et al. (2010)

15.3 Data from the IPCC, courtesy of the Institute for Atmospheric
and Climate Science, ETH Zürich

15.4 Data from Hakim et al. (2016), Last Millennium Climate

Reanalysis Project

15.5 Data from the IPCC, courtesy of the Institute for Atmospheric
and Climate Science, ETH Zürich
 Index
Locßators in bold refer to glossary entries
Locators in italic to figures/tables

absolute humidity 77, 240

absorption, radiation 57, 63
electronic transition 60
selective absorbers 63
tropics vs. polar regions 66, 67
vibration and rotation 62
absorption spectrum, visible window 60
accretion (riming), ice crystals 120, 240
accuracy, weather forecasts 209, 212
acid rain 122, 224, 240
acidification of oceans, climate change 237
adiabatic compression 100
adiabatic cooling 92, 93, 240
adiabatic lapse rate 240,
see also dry adiabatic lapse rate;
moist adiabatic lapse rate
adiabatic processes 88, 240
cloud formation 92–94
definition 93
adiabatic warming 240
advection 240
advection fog 91, 240
aerosols 43, 49, 77, 79, 240
as anchor points 90
impact of human activity 235
role in air pollution 225
role in cloud formation 90
aggregation, ice crystals 120, 240
air masses 21, 240
continental Arctic 168
continental polar 167
continental tropical 168
maritime polar 168
maritime tropical 168
midlatitude cyclones 167–168
source region 167
air parcels 2, 240
air pollution 223, 229
aerosols 225
concentration in valley bottoms 228
concentration under temperature inversion 228
dispersion 225–226
dispersion by wind 225, 226
effect of atmospheric stability 225–226, 227
large-scale patterns 228
mixed layer 226
mixing 225
 topography 228
air temperature 3,
see also temperature
aircraft, use in wind measurement 13
albedo 58, 67, 240
definition 58
polar regions 68
tropics vs. polar regions 67
typical examples 58
Aleutian low 151, 240
altocumulus 114, 240
altostratus 115, 240
Andes Mountains 140, 181
anemometers 11
aneroid barometers 9, 240
angular momentum 240
conservation of 192, 242
animals, impacts of climate change 237
Antarctic ice sheet 138, 157
Antarctica 146
katabatic winds 143
technically a desert 151
anticyclones 96, 228, 241
anticyclonic wind rotation 136
Archimedes’ principle 53, 102, 103, 241
Area Forecast Discussion, US National Weather Service 220
Atlantic jet stream 151
Atlantic storm tracks 180
atmosphere xi, 42–50
aspect 42
averaged 146
composition 42–43
effects of climate change 232–233
future evolution 45–47
general circulation 145
permanent gases 42–43, 49
origin and evolution 43–45
single-cell model of circulation 152–154
synoptic view 25
thinness of 42, 49
three-cell model of circulation 154–155
variable gases 49
vertical structure 47–49
atmospheric pressure 6–10, 241
adjustment to mean sea level 24
barometers 8–9
definition 7
measurement 16
pressure maps 22, 24
station model 18
units 9
useful numbers to remember 10
vertical distribution 7–8, 47–49
Automated Surface Observing Systems (ASOS) 15
Automated Weather Observing Systems (AWOS) 15
automated weather stations 14, 15
averaged atmosphere 146
precipitation 151–152
surface pressure 151
surface temperature 146–147
upper-level heights 148–151

balloons, weather, see radiosondes

barometers 8–9, 241
bars (pressure units) 9, 241
Beaufort wind scale 12, 12, 241
Beijing, smog 224
Bergen school of meteorology 173–174, 183
Bergeron, Tor 119, 183
Bergeron process 119, 183, 241
Bermuda/Azores High 151, 155, 241
biological impacts of climate change 237
Bjerknes, Jacob 173, 183
Bjerknes, Vilhelm 183
Bjerknes model (Norwegian cyclone model) 173–174
black body 60, 61, 241
black ice 241, see also freezing rain
blizzard 122, 241
Boltzmann, Ludwig 56
boundary layer 10
buoyancy force 103, 104, 241
buoys 15, 15

carbon cycle 45, 241

carbon dioxide (CO2) 43, 44, 241
 absorption of radiation 60, 64
annual variations 43
concentration 232
increase since Industrial Revolution 43, 45, 49, 232
local variations 43
natural variability 232
pollutants 223
carbon monoxide (CO) 224
Celsius (Centigrade) temperature scale 4, 241
CFCs, see chlorofluorocarbons
chaos theory 216
Chapman reactions, ozone layer 46
charge separation, ice particles 188
chinook wind 241
Rocky Mountains 99, 140
chlorofluorocarbons (CFCs) 46, 223, 242
circulation model of atmosphere 145
impact of climate change 235
Coriolis force 154
single-cell 152–154
three-cell 154–155
cirrocumulus 114, 242
cirrostratus 115, 242
cirrus clouds 114, 118, 242
in hurricanes 200, 202
Clapeyron, Émile 80
Clausius, Rudolph 80
Clausius–Clapeyron formula 80, 233
climate, definition xii
climate change 66, 230, 238–239
atmosphere 232–233
effects beyond weather 237
forecasting 237–238
global wind systems 235–236
midlatitude and tropical cyclones 236–237
past and future 231–232
warming effects, see global warming
water cycle 233–235
climatology 222
cloud condensation nuclei, see aerosols
cloud-to-ground lightning 188, 189
dart leader 189
return stroke 188, 189
stepped leader 188
cloud types
altocumulus 114
altostratus 115
cirrocumulus 114
cirrostratus 115
cirrus 114, 118
cumulonimbus 109, 113, 118
cumulus 102, 113
cumulus congestus 113
fair-weather cumulus 106–109, 118
lenticular 116
mammatus 186, 187
 nimbostratus 114
stratocumulus 116
stratus 102, 113
clouds 242
adiabatic process 92–94
aerosols 90
altitude 114
classification 113
convection 102–112
dry adiabatic lapse rate 94
formation 90–92, 112, 138, 139
frontal lifting 101–102
glaciated 118
heterogeneous nucleation 90
homogeneous nucleation 90
impact of climate change 235
influence on diurnal cycle 71–72, 71
inhibited by subsidence 100
lifting by convergence 101
mixed-phase 118
moist adiabatic lapse rate 96, 106
orographic lifting 97
relative humidity 95–96
result of saturation 77, 80, 88
satellite information 29
station model 18
tropopause temperature inversion 49
warm vs. cold 117–118
 water distribution 235
water droplets 117
CO2, see carbon dioxide
coalescence, water droplets 119
cold clouds 117, 118
cold fronts 22, 22, 29, 101, 242
midlatitude cyclones 169
cold pool, in thunderstorms 185
cold tongue 160, 161, 242
collision and coalescence, precipitation 118–119
collision efficiency, water droplets 118–119
comets, as source of water 44
condensation 77, 77, 80–82, 242
conditional instability 107, 109–112, 242
conduction 3, 51–52, 242
confidence, in weather forecast 212
conservation laws, general xii, 1, 17
conservation of angular momentum 192, 242
conservation of energy (law) 1, 2, 57, 92, 242
conservation of mass (law) 6, 211, 242
conservation of momentum (law) 10, 11, 126, 242
continental Arctic (cA) air mass 167
continental climate 72
continental polar (cP) air mass 167
continental tropical (cT) air mass 168
convection 52–53, 242
cloud formation 109–112
fair-weather cumulus clouds 106–109
 stable air 102–104
stable vs. unstable air 104–105
unstable air and thermals 104
convective precipitation 185
convective systems 36, 40
converging winds 138, 139
causing cloud formation 138, 139
cyclone development 174
Coriolis, Gaspard-Gustave 130
Coriolis force 130–131, 242
influence on circulation model 154
role in formation of hurricanes 199
coupled model of ocean and atmosphere 238
cumuliform 102, 112
cumulonimbus 109–112, 113, 118, 242
in hurricanes 200
lightning and thunder 188
thunderstorms 185
cumulus 102, 113, 242
cumulus congestus 108, 109, 113, 242
cyclones 171–172, 242
influence of jet streams 171
satellite images 39
triple point 178
cyclonic systems, see low pressure areas
cyclonic wind rotation 136, 137

Dalton, John 78
Dalton’s law 78, 95
dart leader, cloud-to-ground lightning 189
day length, summer 70
density of atmosphere 6
deposition, water vapor molecules 119, 243
Descartes, René xii
deterministic forecasts 210, 211, 243
dew 77, 81, 243
dew point depression 83, 243
dew point temperature 82–83, 243
definition 82
as a measure of humidity 83, 87
in meteograms 85–86
radiosonde profiles 87
station model 18
in surface weather maps 83–84
diabatic process 88
diurnal cycle 69–70, 71, 85, 243
land–ocean contrasts 72–73
temperature variations 68–71
diverging winds 138
causing clear skies 138, 139
cyclone development 174
tendency to slow down 139
Doppler radar 29, 243
drizzle 122, 243
dry adiabatic lapse rate 94, 95
dryline 243
Tornado Alley 194
dynamic equilibrium 50, 60, 68, 77, 243

Earth’s orbit, effect on climate 231

Earth’s tilt, effect on seasons 68
El Niño/Southern Oscillation (ENSO) 159–165, 163, 243
forecasting 216–218
seasonal outlook 222
electromagnetic waves 53, 58, 62, 63, 243
emission spectrum 55, 55, 243
energy, law of conservation of 1, 2, 57, 92, 243
ensemble forecasting 213–216, 243
ensemble mean vs. climatological mean 216
medium-range forecast 220
spaghetti plots 214, 215
statistical uncertainty 213–216
entrainment 109, 243
in thunderstorms 185
environmental lapse rate 94, 243
Environmental Protection Agency (EPA) 229
equator, weather patterns 34
equatorial regions, precipitation 151
evaporation 77, 77, 78–79, 243
evaporation fog 88, 91, 243
extratropical cyclones 243, see also midlatitude cyclones
extratropics 35, 243
extreme heat events 233
eye of hurricane 200, 201, 243
eyewall of hurricane 200, 243
Fahrenheit temperature scale 3, 4, 243
fair-weather cumulus clouds 106–109, 118
Federal Aviation Administration 15
feedback loops, climate 65, 231
Ferrel cell 154, 243
Findeisen, Walter 119
first law of thermodynamics 92, 93, 96
foehn (Alps) 99, 244
fog 82, 91, 244
advection 91
evaporation 88, 91
forecasting 220
radiation 88, 91
force units, newtons 9
forcings, climate change 232, 238
forecasting 209–210, 222
accuracy 209
chaos theory 216–218
climate change 237–238
confidence 212
decreasing accuracy 212
deterministic 210, 211
ensemble 213–216
initial conditions 212
limiting factors 212
medium-range 220–222
parameterization of processes 212
preparing a forecast 219–220
 probabilistic 210
prognostic equations 210–213
role of human forecaster 218, 219, 220
seasonal outlook 222
statistical process 218–219
uncertainty 210
fossil fuel use, impact on climate change 238
fragmentation 244
role in forming ice crystals 120
freezing nucleus 244, see also ice nucleus
freezing rain 14, 122, 244
frontal cyclones, see midlatitude cyclones
frontal lifting, cloud formation 101–102
frontal regions, see fronts
frontal squall line 187, 244
fronts 244, see also cold fronts; occluded fronts; stationary fronts; warm
fronts
frost 81
Fujita, Tetsuya 192
Fujita scale, tornadoes 192, 244
funnel cloud 192, 244
funnel rain gauges 13

general circulation 145, 244

Geostationary Operational Environmental Satellite system (GOES) 30, 34,
244
geostationary satellites 30, 33–35, 244
geostrophic approximation 134
geostrophic balance 132–136, 244
geostrophic wind 131–134
glaciated clouds 118
global satellite images 34, 36
global warming 45, 66, 230, 233, 244, see also climate change
effect of hydrofluorocarbons 47
extreme heat events 233
forecasting 216–217
impact on daily weather 233
impact on global wind systems 233
positive feedback loop 65
predicted global temperature increase 233
since Industrial Revolution 232
to-date global temperature increase 233
water vapor feedback 65, 234
global wind systems 145, 165–166
effect of climate change 233, 235–236
pole-to-equator temperature difference 236
gradient, temperature 21
gradient wind 134–135
gradient wind balance 134, 244
graupel 120, 244
greenhouse effect 45, 61, 65, 244
feedback 65, 231
greenhouse gases 66
beneficial for life 66
impact on climate change 238
increase since Industrial Revolution 66, 232
pollutants 223
 variation in concentration 66
Greenland, katabatic winds 143
gust front, thunderstorms 185, 244

Hadley, George 152

Hadley cells 152, 154, 244
impact of climate change 235
hail 13, 122, 244
halo 115, 124, 245
haze 245
heat 2–3, 245
heat capacity 72, 245
heat imbalance 51, 66–68
heat transfer 51, 73–74
conduction 3, 51–52, 242
convection 52–53, 242
radiation 53, 66–73, 248
radiative interactions 55
hectopascals (pressure units) 9, 78, 245
hemispheric map 219, 219
heterogeneous nucleation 90, 245
HFCs, see hydrofluorocarbons
high pressure areas 22, 25, see also anticyclones
ridges 23, 25
subtropical 151, 249
highs and lows, wind rotation rule 137
Himalayas 140, 159
homogeneous nucleation 79, 245
cloud formation 90
 ice crystal formation 119
hook echo, tornadoes 195
humidity 77–78, 79–80
measurement 77, 83, 87
Hurricane Fred 204
Hurricane Katrina 201, 202, 203
Hurricane Mitch 207
hurricanes 197–199, 207, 245
comparison with midlatitude cyclones 199, 204, 206
comparison with thunderstorms 204
conditions for development 207
damage caused by 206
decay 206–207
definition 197
development 204–207
effect of El Niño 164
eye 200, 201, 243
eyewall 200, 243
geographical distribution 199
structure 200–204
hydrofluorocarbons (HFCs) 47
hygrometers 77, 245

ice-albedo feedback 65
ice crystals 119–121
ice heat capacity 72
ice nucleus 119, 245
ice storms 122
Icelandic Low 151, 245
ideal gas law 6, 7, 27, 52, 245
ideal radiator, see black body
incipient waves, cyclone development 176, 177
Indian Monsoon 158–159, 245, see also monsoons
Indian Ocean 159
infrared images 31–32
infrared radiation (IR) 64, 245
infrared window 60, 245
initial conditions 245
use in forecasting 212
insolation changes, annual 69
Intergovernmental Panel on Climate Change (IPCC) 233
Intertropical Convergence Zone (ITCZ) 152, 154, 155, 156, 245
El Niño 159
hurricanes 204
impact of climate change 235, 236
Indian Monsoon 158–159
Sahel 157–158
inversion 245, see also temperature inversion
ionosphere 49, 245
isobars 22, 245
isotherms 20–21, 245

jet stream 10, 25, 148, 149, 148–151, 245

Atlantic 151
caused by temperature gradients 150
influence on cyclone structure 171
Pacific 151
polar 151
 role in forecasting 216
subtropical 151

katabatic winds 143, 246

three-cell model of circulation 157, 157
Kelvin temperature scale 4, 246
Keyser, Daniel 172
kinetic energy 2, 51, 246
Kirchhoff, Gustav 56
Kirchhoff’s law 56, 57, 61
knots (wind speed units) 12, 246

La Niña 163, 246

land breeze 128–130, 246
land–ocean contrasts, diurnal cycle 72–73
landfall, hurricanes 206
lapse rate 48, 49, 246,
see also dry adiabatic lapse rate; moist adiabatic lapse rate
latent heat 78, 83, 246
role in hurricanes 200, 201
water vapor 76, 76
law of thermodynamics, first 92, 93, 96
laws of conservation, see conservation laws
LCL, see lifting condensation level
lenticular clouds 116, 246
level of free convection (LFC) 111, 246
LFC, see level of free convection
lifting, cloud formation 97, 101–102
lifting condensation level (LCL) 97, 106, 109, 246
lightning 188–189, 246, see also thunderstorms
charge separation 188
cloud-to-ground lightning 188, 189
cumulonimbus clouds 188
supercell storms 189–193, 192
tornadoes 191, 192–195
limiting factors, forecasting 212
Los Angeles, photochemical smog 224, 228
low pressure areas 23, 25
subpolar 151, 249
troughs 25, 25, 250

mammatus clouds 186, 187, 246

maps, see weather maps
maritime climate 71
maritime polar (mP) air masses 168
maritime tropical (mT) air masses 168
mass, law of conservation of 6, 211, 246
mean sea level 24, 246
medium-range forecast 220–222, 246
ensemble forecasts 220
example 221
mercury barometers 8, 9, 246
mesocyclone, supercell storms 190
mesolow
mesopause 48, 48, 246
Mesoscale Convective System (MCS) 187
mesosphere 48, 246
meteograms, dew point temperature 85–86
meteorological sunrise/sunset 69
methane (CH4) 43, 223, 246
increase since Industrial Revolution 232
Mexico City, photochemical smog 224, 228
micro-bursts 187, 246
midlatitude cyclones 20, 24, 25, 36, 68, 151, 167, 181
air masses 167–168
climate change 236–237
cloud pattern 32, 34
comparison with hurricanes 199, 204, 206
convergence and divergence regions 135
decaying stage 173, 179
deepening cyclone 173, 177
definition 168
development 172–181
example 176–180
fronts 168–172
height 42
intensification 180
jet stream 48
life cycle 172–174
location 180–181
mature stage 173, 178, 178
meteogram 85
Norwegian cyclone model 173–174
occluded stage 173, 179
southern hemisphere 182
three-cell model of circulation 154, 157
 short-wave troughs 172–174
vertical structure 174–176
wave cyclone 173
midlatitude temperature gradient 150
midlatitude westerlies 151
midlatitudes 34, 246
Milankovitch, Milutin 231
millibars (pressure units) 9, 247
mixed-phase clouds 118
mixing ratio 77, 247
model output statistics (MOS) 218, 220
moist adiabatic lapse rate 96, 97, 106
moisture gradient 84
molecular motion 2, 2
momentum 247
momentum, law of conservation of 10, 11, 126, 247
monsoons 73, 158–159, 236, 247
mountain barriers 140
mountain ranges
cloud formation 97
warming effects 97–100

neutral temperature profile 105

Newton, Isaac 126
Newton’s laws of motion 11, 126–127, 210
newtons (force units) 9
nimbostratus 114, 247
nitric oxide (NO) 224
nitrogen 44, 58–59, 247
nitrogen dioxide (NO2) 224
nitrous oxide (N2O) 43, 223, 247
NOAA Climate Prediction Center 222
North Atlantic Oscillation (NAO) 165
northern hemisphere warming, climate change 236
Norwegian cyclone model 173–174
nowcasts 218
NOx, see nitric oxide (NO); nitrous oxide (N2O)
nucleation, see heterogeneous
nucleation; homogeneous
nucleation

occluded fronts 29, 170–171, 247

ocean acidification, impact of climate change 237
ocean latent heat, role in hurricanes 200, 201
ocean prediction models 222
ocean variability, in climate models 238
Olympic Mountains 140, 141
optical phenomena
22° halo 124
46° halo 124
rainbow 124
orographic lifting, cloud formation 97
orographic precipitation 98, 100, 247
orographic subsidence 97–100
overrunning 102, 169, 171, 179, 247
oxygen 44, 46, 58–59, 247
ozone (O3) 247
absorption of radiation 60, 64
 Chapman reactions 46
chlorofluorocarbons 46
stratospheric 10, 43, 45, 46, 247
tropospheric 224
ozone hole 45, 46, 47, 49, 247

Pacific High 151, 155, 247

Pacific jet stream 151
Pacific storm tracks 180
Pacific warm pool 160, 247
Pacific/North American pattern (PNA) 165
Pacific/South American pattern (PSA) 165
parameterization 213, 247
particulates 224
PM10 224
PM2.5 224
pascals (pressure units) 9
permafrost thawing 237
permanent gases 42–43, 49, 247
Petterssen, Sverre 183
photochemical smog 224–225
photosynthesis 43, 45, 247
Planck, Max 55, 56, 57
Planck feedback 65
Planck’s law 55, 55, 61, 65
planetary boundary layer 138
plants, effects of climate change 237
PNA (Pacific/North American pattern) 165
polar cell 154, 247
polar front 146, 247
polar jet 151
polar regions, low precipitation 151
polar-orbiting satellites 35–37, 37, 247
pole-to-equator temperature difference 236
pollutants 223–225, see also air pollution
carbon dioxide 223
carbon monoxide 224
chlorofluorocarbons 223
gases and compounds 224
greenhouse gases 223
methane 223
nitric oxide 224
nitrogen dioxide 224
nitrous oxide 223
photochemical smog 224–225
primary and secondary 224
smog 224
sulfur dioxide 224
volatile organic compounds 224
positive feedback loops, global warming 65
precipitation 13–14, 117, 123, 248
averaged atmosphere 151–152
collision and coalescence 118–119
effects of climate change 235
equatorial regions 151
ice crystal growth 119–121
measurement 13, 16, 28
 orographic 98, 100, 247
polar regions 151
vertical structure 28
warm vs. cold clouds 117–118
precipitation types 121–122
black ice 122
drizzle 122
freezing rain 14, 122, 244
hail 13, 122, 244
ice storms 122
rain 119, 122, 248
showers 121
sleet 13, 14, 122, 249
snow 13, 122, 249
pressure 6–7, 248, see also atmospheric pressure
pressure gradient force 127–128, 128, 202–204, 248
primary pollutants 224
probabilistic forecasting 210
prognostic equations 210–213, 248
PSA (Pacific/South American pattern) 165

radar 28–29, 248, see also Doppler radar

radar images, use in forecasting precipitation 220
radiation 44, 53, 58, 248, see also absorption, radiation
conversion to heat 57
diurnal variations 68–71
heat imbalance 66–68
influence of clouds 71–72
land–ocean contrasts 72–73
 seasonal variations 68
temperature 61
radiation fog 82, 88, 91, 248
radiative interactions 55
absorption 57
equilibrium 58–60, 248
greenhouse effect 61–66
reflection 57–58
scattering 58–59
selective absorbers 60
visible and infrared windows 60–61
radiosondes 4–5, 5, 13, 94, 248
dew point temperature 87
use in forecasting 220
rain 119, 122, 248, see also precipitation
rain forests 154, 156
rain gauges 13, 14, 248
rainbows 124
reflection 57–58, 248
relative humidity (RH) 78–79, 87–88, 248
cloud formation 95–96
meteograms 85
related to latitude 80
relative vs. absolute humidity 80
representative concentration pathways (RCPs), IPCC 233
return stroke, cloud-to-ground lightning 188, 189
ridges, high pressure 23, 25, 248
riming, ice crystals 120, 248
roaring forties (midlatitude westerlies) 151
rocks, chemical weathering 44
Rocky Mountains 140
chinook 99
enhanced cyclogenesis 180, 181
Rossby, Carl-Gustaf 183
Rossby waves 183

Saffir–Simpson hurricane intensity scale 198, 206, 248

Sahel
alternate dry and wet seasons 157–158
threat of desertification 158
three-cell model of circulation 157–158
satellite images
cloud cover information 29
cloud signatures 2, 39–40, 219
global 34, 36
infrared 31–32
hurricanes 200
use in forecasting 219
visible 30–31, 31
water vapor 32–33
satellites 29–30
geostationary 30, 33–35, 244
polar-orbiting 35–37, 37, 247
TIROS (Television Infrared Observation Satellite Program) 29
use in wind measurement 13
saturated adiabatic lapse rate (moist adiabatic lapse rate) 96, 97, 106
saturation 76–77, 248
 cloud formation 77, 80, 88
in mammatus clouds 186
saturation water vapor pressure 79, 248
increased by global warming 234
lower over ice than water 119
related to latitude 80
related to temperature 79, 87
variation with height 98
scattering of radiation 58–59, 248
scientific method xii
sea breeze 73, 128–130, 151, 248
circulation 129, 152
formation 128–129
sea level rise, impact of climate change 237
seasonal cycles 68, 248
seasonal outlooks, forecasting 222
climatology 222
ocean prediction models 222
value of 222
secondary pollutants 224
selective absorbers 248
severe thunderstorms
frontal squall line 187
mammatus clouds 186, 187, 246
micro-bursts 187
shelf cloud 187, 188, 248
wind shear 184, 187
Shaefer, Vincent 121
Shapiro, Mel 172
Shapiro–Keyser model (T-bone model) 172
shelf cloud 187, 188, 248
short-wave troughs 172–174
showers 121
silver iodide, use in seeding clouds 121
single-cell model of circulation 152–154
sleet 13, 14, 122, 249
smog 91, 224, 249, see also photochemical smog
snow 13, 122, 249
snowflakes, formation of 120
solar radiation, see radiation
solar system formation 44
southern hemisphere, midlatitude cyclones 182
spatial representations, see weather maps
specific humidity 77, 249
squall line 187, 249
stable air 102–104, 249
stable temperature profile 104
stable vs. unstable air 104–105, 106, 107
station model 17–20, 20, 249
stationary fronts 169, 249
statistical process of forecasting 218–219
forecast calibration 218
machine learning 218
model output statistics 218, 220
nowcasts 218
removing bias 218
 statistical post-processing 218
Stefan, Jožef 56
Stefan–Boltzmann law 56, 57, 60, 67
stepped leader, cloud-to-ground lightning 188
sting jets 172
storm intensification, hurricanes 206
storm surges 206–207
storm tracks 180, 249
stratiform 112, 113
stratocumulus 116, 249
stratopause 48, 48, 249
stratosphere 10, 249
temperature inversion 48, 49
stratus 102, 113, 249
streamlines, wind 138, 139
subgeostrophic wind 134, 249
subpolar lows 151, 249
subsidence 34, 97–100, 249
inhibiting cloud formation 100
subtropical deserts 154, 156
subtropical highs 151, 249
subtropical jet 151
subtropics 34, 151, 249
sulfur dioxide (SO2) 224
sunrise/sunset, meteorological 69
Super Typhoon Haiyan 207
supercell storms 189–193, 192, 249
mesocyclone 190
 mesolow
wind shear 190
supercooled water 118, 249
supergeostrophic wind 134, 249
supersaturation 79, 90, 249
surface friction 136, 138–140, 249
surface maps 20–25, 26, 146
dew point temperature 83–84
temperature 21
surface pressure
averaged atmosphere 151
use in forecasting 219
surface temperature 3
averaged atmosphere 146–147
effect on pressure distribution 147
use in forecasting 219
surface winds 135–137
convergence into lows 138
divergence from highs 138
hurricanes 202–203
sustained wind speed 12, 249
symbols, weather maps 18
synoptic scale 225, 250
synoptic view of atmosphere 25

T-bone model, cyclone structure 172

teleconnections 164, 165, 217, 250
temperature 1, 3, 250
contrast between tropics and poles 147, 236
 diurnal cycle in temperature profiles 69–70
fronts 21–22
gradients 21, 146, 150, 167
and heat 2–3
and humidity 79–80
maps 20–21, 21
measurement 2, 16
and pressure 6, 25, see also atmospheric pressure
and radiation 55
scales 3–4
station model 18
surface 3, 146–147, 219
vertical distribution 47, 48
temperature inversions
close to ground 91
night-time 70
stratosphere 48, 49
trapping air 105, 142
tropopause 49
thawing of permafrost 237
thermal expansion 52
thermals 53, 54, 250
in unstable air 104
thermocline 160, 162, 250
thermodynamics 51, 92, 93, 96, 211
thermometers 3, 250
thermosphere 48, 250
three-cell model of circulation 154–155
 extratropical cyclones 154, 157
Ferrel cell 154
Intertropical Convergence Zone 154
katabatic winds 157, 157
polar cell 154
Sahel 157–158
USA, West Coast vs. East Coast 155, 157
thunder 188, 250
thunderstorms 102, 109, 111, 113, 184, 195–196, 250
lightning and thunder 188–189
ordinary 184–186
severe 186–187
Tibetan Plateau, enhanced cyclogenesis 180
time series, weather variables 16
tipping bucket rain gauges 13, 14
TIROS (Television Infrared Observation Satellite Program) 29
topography, influence on wind 140–143
Tornado Alley 192, 194–195, 250
dryline 194
tornado family 193
tornadoes 250
development 192–194, 194
Fujita scale 192
funnel cloud 192
hook echo 195
lightning and thunder 191, 192–195
mesolow
spawned by supercells 192
 vortex 193
wall cloud 192
Torricelli, Evangelista 8
trace precipitation 14
trade winds 152, 155, 250
triple point, cyclone development 178
tropical cyclones 198, 250
climate change 237
tropical depressions 198, 204–205, 250
tropical disturbances 198, 250
tropical easterly waves 204, 250
tropical rain forests 154, 156
Tropical Storm Erika 205
tropical storms 198, 205, 250
tropopause 5, 48, 48, 49, 250
air motion near 48
temperature inversion 49
troposphere 5, 10, 25, 28, 47, 48, 250
troughs, low pressure 25, 25, 250
turbulent eddies 12
typhoons 198, 250

ultraviolet radiation (UV) 45, 53, 250

uncertainties 213–216, 238, 250
United Nations Framework Convention on Climate Change (UNFCCC) 238
unstable air 104, 251
unstable temperature profile 104, 105
upper-level heights
averaged atmosphere 148–151
 jet stream 149
upper-level maps 25–28
upper-level troughs 149, 151
upper-level winds, westerly 148
upwelling 159, 251
US National Weather Service 15, 220

valley bottoms 228

valley breezes 140, 142, 251
valley fog 142, 251
variable gases 49, 251
variables, weather 1
atmospheric pressure 6–10
precipitation 13–14
temperature 1
weather stations 14–15, 14, 15, 16
wind 10–13, 16
vectors, wind 127, 139
vertical atmospheric profiles 13, 16, 24
vertical fluid motion, see convection
vertical structure, cyclones 174–176
visible window, absorption spectrum 60
volatile organic compounds (VOCs) 224
volcanic eruptions, role in atmosphere formation 44
Vonnegut, Bernard 121
vortex, see cyclones; hurricanes; tornadoes

Walker, Sir Gilbert 161

Walker circulation 161, 251
wall cloud 192, 251
warm clouds 117
warm fronts 22, 22, 29, 251
midlatitude cyclones 169–170
warm occlusion (occluded fronts) 29, 170–171, 247
warm pool, Pacific 160, 251
warm sector 25, 169, 251
warm vs. cold clouds 117–118
water
atmospheric 75
heat capacity 72
phases, definition 75
phases, difference in energy 75–76, 76
water cycle 75–76, 233–235
water droplets, cloud vs. rain 117
water vapor 43, 251
absorption of radiation 60, 64
capacity 81
concentration 43
effects of climate change 234
feedback loop 65, 234
heat capacity 72
heating effect 64
images 32–33
latent heat 76, 76
mixing ratio 77
pressure 77, 251
weather (general information) xi–xii
weather balloons, see radiosondes
weather forecasting, see forecasting
weather maps 14, 16, 17, 26
radar 28–29
satellites 29–30
station model 17–20
surface maps 20–25
symbols 18
upper-level maps 25–28
weather modification 121
weather stations 14–15, 14, 15, 16, 251
use in forecasting 219
weather variables, see variables, weather
Wegener, Alfred 119
West Coast vs. East Coast (USA) 155, 157
Wien, Wilhelm 57
Wien’s displacement law 56
wind 126, 143–144
converging 138, 139
Coriolis force 130–131, 134
dispersion of pollution 225, 227
diverging 138
effect of surface friction 136, 138–140
force and acceleration 126–127
geostrophic 131–134
global systems 145, 165–166, 233, 235–236
gradient 134–135
gusts 12–13, 251
 influence of topography 140–143
katabatic 143, 157, 157, 246
measurement 11–12, 13
pressure gradient force 127–128
reporting 12–13
sea breeze and land breeze 128–130
streamlines 138, 140
subgeostrophic 134, 249
supergeostrophic 134, 249
surface 135–137
vanes 11, 251
vectors 127, 139
weather variables 10–13
wind direction 11, 16
station model 18, 19
wind rotation
anticyclonic 136
cyclonic 136, 137
rule, highs and lows 137
wind shear 251
hurricanes 206
severe thunderstorms 184, 187
supercell storms 190
wind speed 11–12, 16
barbs and pennants 19
Beaufort wind scale 12
station model 18, 19
sustained 12, 249