MSc Sustainable Energy Systems

EUB_7_133
Renewable Energy Technologies 1

The Solar Resource

Archimedes uses a solar death ray to defend Syracuse during the Second Punic Wars

Department of Urban Engineering

2013/14

Alan Dunn 1 25/5/13

Alan Dunn 2 25/5/13
Contents
1.0 Nomenclature ........................................................................................................................... 6
2.0 Introduction.............................................................................................................................. 8
3.0 The Sun ..................................................................................................................................... 8
3.1. Earth / sun relationship .............................................................................................................. 8
3.2. Extraterrestrial Radiation ......................................................................................................... 11
3.2.1. Solar Constant ................................................................................................................... 11
3.2.2. Variation in extraterrestrial radiation ............................................................................... 11
3.2.3. Spectral distribution of extraterrestrial radiation ............................................................. 12
4.0 Time........................................................................................................................................ 13
4.1. Local Apparent Time (LAT) ....................................................................................................... 13
4.2. Local Mean Time (LMT) ............................................................................................................ 13
4.3. Local Standard Time ................................................................................................................. 14
4.4. Daylight Saving Time (DST)....................................................................................................... 14
4.5. Conversion between LAT and LMT ........................................................................................... 15
5.0 Solar Geometry....................................................................................................................... 15
5.1. Co-ordinate systems ................................................................................................................ 16
5.2. Latitude, and longitude, .................................................................................................... 17
5.3. Solar declination, .................................................................................................................. 17
5.4. Hour angle, ........................................................................................................................... 18
5.5. Angle of Incidence on an Inclined Surface ............................................................................... 18
5.6. Sunrise and Sunset ................................................................................................................... 18
5.6.1. Sunrise and sunset on an inclined surface ........................................................................ 19
5.6.2. Horizon File ....................................................................................................................... 19
6.0 Terrestrial Radiation ............................................................................................................... 20
6.1. Extraterrestrial radiation on a horizontal plane ....................................................................... 20
6.2. The Earth’s Atmosphere........................................................................................................... 21
6.3. Balance of Atmospheric Energy ............................................................................................... 22
6.4. Clear Sky Radiation .................................................................................................................. 23
6.4.1. Relative optical air mass, m ............................................................................................... 24
6.4.2. Rayleigh optical depth, δr .................................................................................................. 24
6.4.3. Linke Turbidity, TL.............................................................................................................. 24
6.4.4. Clear sky diffuse irradiance on horizontal surfaces ........................................................... 25
6.4.5. Clear sky global irradiance on horizontal surfaces ............................................................ 25

Alan Dunn 3 25/5/13

 6.4.6. Clear sky global irradiation on horizontal surfaces ............................................................ 25
6.4.7. Terrestrial Radiation – Clear Sky Spectral Distribution ...................................................... 26
6.5. Total sky radiation .................................................................................................................... 26
6.5.1. Clearness Index ................................................................................................................. 26
6.5.2. Split of beam and diffuse .................................................................................................. 27
Beam and diffuse components of monthly radiation .............................................................. 27
Beam and diffuse components of daily radiation ................................................................... 27
Beam and diffuse components of hourly radiation ................................................................. 27
7.0 Total Radiation on a Surface ................................................................................................... 28
7.1. Direct Beam Irradiance ............................................................................................................ 28
7.2. Sky diffuse irradiance ............................................................................................................... 28
7.3. Ground reflected radiation ...................................................................................................... 29
7.4. Perez Sky Diffuse Model........................................................................................................... 29
8.0 Algorithmic Chains .................................................................................................................. 31
9.0 Radiation Measurements ....................................................................................................... 32
9.1. Ground based radiation measurements .................................................................................. 32
9.2. Satellite based radiation measurements .................................................................................. 33
10.0 Sources of radiation data...................................................................................................... 33
10.1. Interpolation .......................................................................................................................... 33
10.2. Generation of synthetic hourly radiation data ....................................................................... 34
10.2.1. Markov Transitional Matrices ......................................................................................... 34
10.2.2. Time series prediction ..................................................................................................... 34
11.0 Data for simulation ............................................................................................................... 36
11.1. Measured hourly data ............................................................................................................ 36
11.2. Frequency distributions ......................................................................................................... 36
11.3. Typical day of the month ....................................................................................................... 37
12.0 References ............................................................................................................................ 37
13.0 Appendix: Derivation of mean monthly daily global extraterrestrial irradiation .................. 40
14.0 Appendix: Transitional Markov Matrices .............................................................................. 41
14.1. Markov Chain ......................................................................................................................... 41
14.2. Multiple Transitions ............................................................................................................... 41
14.3. Determining transition probabilities from data ..................................................................... 42
15.0 Appendix: Markov transitional matrices............................................................................... 44
16.0 Appendix: Generating synthetic daily radiation data ........................................................... 50

Alan Dunn 4 25/5/13

 17.0 Appendix: Generating synthetic hourly radiation data ......................................................... 51
18.0 Appendix: Generation of monthly mean hourly data ........................................................... 53
19.0 Appendix: Statistical Evaluation of Models........................................................................... 55
19.1. Regression.............................................................................................................................. 55
19.2. Coefficient of determination, ............................................................................................ 55
19.3. Coefficient of correlation, ................................................................................................... 55
19.4. Student’s t-distribution .......................................................................................................... 55
19.5. Mean bias error, MBE ............................................................................................................ 56
19.6. Root mean square error, RMSE .............................................................................................. 56
19.7. Skewedness, S ........................................................................................................................ 56
19.8. Kurtosis, K .............................................................................................................................. 56
19.9. Outlier analysis....................................................................................................................... 56
19.10. Student’s t-distribution ........................................................................................................ 57

These notes make reference to worksheets within an accompanying Excel Workbook

Solar_Resource.xlsm which is available for download from the VLE.

Download a copy of
this document onto
your smart phone
(1.1MB)

Front cover image is taken from the Latin edition of Alhazen's Thesaurus Opticus, showing how
Archimedes set on fire the Roman ships before Syracuse with the help of parabolic mirrors.

Alan Dunn 5 25/5/13

1.0 Nomenclature
Please note that there is no standard nomenclature for solar energy studies and you will find
differences between that used in these notes and ASHRAE and CIBSE publications

These notes largely follow CIBSE which in turn follows the notation used in the 4th edition of the
European Solar Radiation Atlas.

Symbol Units Definition Equation

W∙m-2
W∙m-2 Daily irradiation
W∙m-2 Daily extraterrestrial irradiance
Hourly irradiation
Hourly extraterrestrial radiation
( ) monthly mean daily radiation
( ) monthly mean daily extraterrestrial radiation
Daily beam

W∙m-2 Extraterrestrial irradiance

K Absolute temperature
- Eccentricity of the Earth’s orbit
- Day number
Radians Day number
minutes Equation of Time
0 or 1 hour1 Correction for daylight saving
- Clearness Index

Greek
( ) 0 to 90o Angle of incidence on an arbitrary surface
nm Wavelength
0 to 180o Longitude
0 to 180o Reference longitude for time zone
-
0 to 90o Solar altitude
0 to 90o wall-solar azimuth
0 to 180o Solar Azimuth
0 to 180o Wall azimuth
0 to 90o Surface tilt angle
0 to 90o Latitude
0 to 23.5o Solar declination
- Ground albedo
Hour angle
Sunset hour angle
0 to 90o Zenith Angle ( )

1
There have been examples, particularly in war time, of double daylight saving where the clocks were
advanced by two hours in summer.

Alan Dunn 6 25/5/13

Note: A – in the unit column indicates a dimensionless variable

Constants
1,367 W∙m-2 Solar Constant
5.67 10-8 Stephan-Boltzmann constant

Abbreviations
WMO World Meteorological Organisation
NASA National Aeronautics and Space Administration
ASTM American Society for Testing and Materials
LAT Local Apparent Time, or Solar Time
LMT Local Mean Time or ‘Clock Time’ see §4.2
GMT Greenwich Mean Time (Local Mean Time at Greenwich)
BST British Summer Time (GMT + 1) see §4.3
UTC Coordinated Universal Time

Learning Outcomes

After studying this topic you should be able to:

 describe the motion of the earth around the sun;

 convert between solar time and clock time;
 calculate the position of the sun at any time at any place;
 calculate the times of sunrise and sunset;
 calculate the angle of incidence on an arbitrary inclined surface;
 calculate the sunrise and sunset times on an arbitrary inclined surface;

Alan Dunn 7 25/5/13

2.0 Introduction
In the design of any solar energy application the greatest Terminology
(by far) uncertainty lies in the prediction of the available
solar radiation at a specified geographical location. Many It is important to clearly understand
simplified approaches are in use that mask the complexity the difference between the following
of the real problem, and perhaps explain why the actual two terms:
performance of many solar installations does not meet
Irradiance
the design predictions.
The instantaneous flux – flow of
The aim of these notes is to provide a methodology by energy per unit time per unit area
which to predict typical hourly values of radiation falling W·m-2
on a surface having a specified orientation and
Irradiation
geographical location. Also, to predict the percentage of
The energy received per unit area
beam and diffuse radiation and the spectral distribution.
integrated over a stated period of
time (hour, day, month or year) in
MJ·m-2 or W·h·m-2

3.0 The Sun

The Sun is a star around 4.5 billion
years old and it will take another 4.5
billion years until it turns into a red
giant when its diameter will increase
from the present 1.39 109 m to more
than 1.5 1011 m completely
engulfing Earth and destroying all life,
as we know it Jim.

The mean distance of the Sun from is

149,597,890 km, which is the
definition of an Astronomical Unit
(AU).
Figure 1 The Sun (NASA)

The sun behaves like a Black Body at a temperature of 5,800 K, thus from Stefan’s Law the total
flux at the sun’s surface is given by:

Where the Stefan-Boltzmann constant,

( )

3.1. Earth / sun relationship

The Earth travels around the sun in an elliptical orbit, during which the distance of the Earth
from the Sun varies from 147,500,000 km at the perihelion, to 152,500,000 km at the aphelion.

Alan Dunn 8 25/5/13

Although this gives rise to a variation of 6% in the radiation received (see §3.2.2), this is not the
cause of seasonal variations on Earth. Looking from above (our North Pole), the rotation of the
Earth, and the movement of the Earth around the Sun, is in an anticlockwise direction.

Figure 2 The earth's orbit around the Sun

The eccentricity2 , of the Earth's orbit today is 0.0167 but over time, the eccentricity of the
Earth's orbit slowly changes from nearly 0 to almost 0.05 as a result of gravitational attractions
between the planets.

The tilt of the Earth’s axis is not perpendicular to the plane of the ecliptic , but inclined at a fixed
angle of 23.5o, called the obliquity of the ecliptic. As the Earth orbits the Sun its axis continues to
point in the same direction relative to the celestial sphere – towards the pole star.

Figure 3 Solstice and equinox

This tilt gives rise to the annual change in the length of the day, and to the variation in the
seasons.

The Summer Solstice, June 21/22 is the longest day in the Northern hemisphere.

2
If the orbit is circular, for the orbit is parabolic. (see: [Link] ) For elliptical
orbits, a simple proof shows that yields the projection angle of a perfect circle to an ellipse of eccentricity .

Alan Dunn 9 25/5/13

Figure 4 The Summer Solstice

The Winter Solstice, December 21/22 is the shortest day in the Northern hemisphere.

Figure 5 The Winter Solstice

Vernal Equinox (March 21/22) and Autumnal Equinox (September 22/23) are days of equal
length.

Figure 6 The Equinox

Alan Dunn 10 25/5/13

 Other minor variations in the Earth’s motion which do not really concern us, but preoccupy the
thoughts of geophysicist, astronomers and astrologers alike, are:

 Precession of the Earth’s axis: rather like the wobble of a slowing spinning top and for the
earth a rotation occurs once every 25,800 years.
 Nutation is a wobble superimposed on the wobble of the precession (so if you don’t care
about precession, care less about nutation).
 Polar motion caused by: movements in the Earth’s core and mantle; the melting of glaciers
which causes land to rise; tectonic movements; all of which have redistributed weight since
1900 causing a change in the point where the Earth’s rotational axis intersects its surface of
20 metres.

3.2. Extraterrestrial Radiation

The solar radiation arriving at the outer limit of the Earth’s atmosphere

3.2.1. Solar Constant

The solar constant ( ) is the value of the extraterrestrial irradiance normal to the beam at the
mean solar distance. The WMO agreed value is:

3-1

In 2002 it was calculated that the Earth received as much solar energy in one hour as it consumed in
one year

3.2.2. Variation in extraterrestrial radiation

The eccentricity of the earth’s orbit causes a variation in extraterrestrial irradiance given by

( )
[ ] 3-2

1.04

1.02

1.00
εj

0.98

0.96
0 30 60 90 120 150 180 210 240 270 300 330 360
Day Number, J

Figure 7 Annual variation in extraterrestrial radiation

The Earth is closest to the sun (perihelion) at the beginning of January – the exact date varies slightly
each year between 2nd and 4th (hence the 3 in the equation).

Alan Dunn 11 25/5/13

The maximum variation is ±3.34% but further variations of up to 1% may be caused by changes in
solar activity and is related to the sun spot cycle.

3.2.3. Spectral distribution of extraterrestrial radiation

Figure 8 Spectral distribution of extraterrestrial solar irradiance

• The solar spectral distribution is close to that of a black body at 5,800 K.3
• Fine structure due to absorption in the cool peripheral solar gas (Fraunhofer lines).
• The area under the curve is 1367 W·m-2 i.e. the solar constant

The solar spectrum may be divided into three regions:

Region Radiation Range % Irradiation eV

I ultra-violet 290 nm < λ < 400 nm ~5%
II visible light 400 nm < λ < 700 nm4 ~43%
III infra-red 700 nm < λ < 2,600 nm ~52%
Figure 9 Regions of the solar spectrum

Wavelength is important for thermal systems and eV for photovoltaics

3
The spectral distribution of a black body is described by Planck’s Law
( ) ⁄( )
4
The maximum spectral sensitivity of the human eye under daylight conditions occurs at a wavelength of
555 nm; this is the mid-point of the visible range and the wavelength having the peak terrestrial
irradiance. Evolution has optimised the human eye to have the best efficiency on the Earth’s surface.

Alan Dunn 12 25/5/13

4.0 Time
There are a number of different time systems in use and we need to be able to convert between
them. In particular, solar position calculations are more easily performed using Local Apparent
Time (otherwise known as solar time); whereas building load profiles are expressed in Local
Mean Time (clock time)..

4.1. Local Apparent Time (LAT)

The time shown on a sun dial is the Local Apparent Time (or solar time), and noon occurs at that
location when the sun is due south. This was the principal method of measuring time in the UK
up until the 19th century.

4.2. Local Mean Time (LMT)

In the 19th century accurate mechanical clocks came in to widespread use. Clocks run at a
constant rate whereas Local Apparent Time does not. Clocks are designed to show the Local
Mean Time, which is the average of the Local Apparent Time. The variation in Local Apparent
Time is due to the eccentricity of the earth’s orbit the velocity is not constant, speeding up as
the earth gets closer to the sun5; and the obliquity of the ecliptic. The difference between the
Local Mean Time and the Local Apparent Time is given by the Equation of Time (EOT) which is an
empirical equation obtained from fitting observed data: therefore several version exist and the
two most common are given below:

(CIBSE, 2002)
( ) ( ) 4-1

(ASHRAE, 2009)
(
( ) ( ) ( )
( ))
4-2
Where

Where is the Day Number (January 1 = 1, February 1 = 32, March 1 = 57 in a non-leap year and
58 in a leap year), and is the Day Number expressed in degrees:

( )

5
Kepler first predicted that planets in an elliptical orbit sweeps out the same area in unit time.

Alan Dunn 13 25/5/13

 Figure 10 Equation of Time Figure 11 EOT as sum of two sine waves

Figure 10 is a plot of the CIBSE and ASHRAE equations of time and they are indistinguishable.
Figure 11 shows how the equation of time is made up from two sine waves: (a) one with a
period of one year caused by the variation in speed along the ecliptic due to the eccentricity of
the orbit, and (b) the other with a period of six months and an amplitude of 9.87 minutes due to
the obliquity of the ecliptic.

4.3. Local Standard Time

Local Mean Time as described above was really local. Clocks in each town were set according to
their Local Apparent Time and were therefore slightly different to clocks in towns at different
longitudes. This was not serious when people travelled slowly by foot or by horse, but became a
problem when people started to travel by train. In the UK 1879 it was proposed that the world
should be divided into 24 time zones, each spanning 15 o of longitude (15o ≡ 1/24 of a day = 1
hour) and that all clocks within a time zone be set to the same time, based on the reference
longitude for that time zone.

Time zones do not strictly follow lines of longitude, but are modified to fit geographical and
political boundaries (UK Hydrographic Office, 2011).

Today the ‘local’ Local Mean Time is no longer used and any reference to Local Mean Time
means the local mean time at the reference meridian. In short it is the time shown by clocks in a
particular time zone.

4.4. Daylight Saving Time (DST)

In the UK known as British Summer Time (BST), this is the practice of advancing the clocks by one
hour during the summer. Originally introduced to reduce wasted morning daylight, but the
benefits today are debateable.

Since 1998 the UK has adopted the Ninth European Parliament and Council Directive on Summer
Time Arrangements in which it states that summer (or daylight saving) time will be kept between
the last Sunday in March to the last Sunday in October. The changes take place at 01.00 GMT.6

6
In the past dates of changes were variable; in particular changes were made to avoid clocks going forward on Easter Sunday.

Alan Dunn 14 25/5/13

4.5. Conversion between LAT and LMT7
For a given location three corrections are made for: (a) the difference between the location
longitude, and the reference longitude, of the time zone, (b) the equation of time, and
(c) daylight saving, . Giving:

( )
4-3

Example
Estimate the time of occurrence in British Summer Time of solar noon at Belfast on 4 August 2007.

Solution
6o 13’ W = -6.217 longitude of Belfast
0o 0’ time zone reference longitude (Greenwich)
31 + 28 + 31 + 30 + 31 + 30 + 31 + 4 + 0.5 = 216.5 days
216.5 (360/365.25) = 213.4o
( ) ( ) hours
1, as summer time is in effect

( ) ( )

5.0 Solar Geometry

Although mathematical models of planetary motion have existed for many years they cannot
account for the complex interplanetary gravitational effects. Each year corrections must be
made based on celestial observations
and these are published in
astronomical almanacs. The almanac
is published each year jointly by the
Her Majesty’s Nautical Almanac
Office at the UK Hydrographic Office
and the Nautical Almanac Office at
the US Naval Observatory. It contains
tables giving the position of the sun,
moon, planets, minor planets as well
as times of eclipses, sunrises and
sunsets. Figure 12 An Orrery - James Ferguson (1710-1778)
Many of the equations used in solar
geometry calculations are empirical i.e. not based on pure mathematics but developed by curve
fitting measured data. Consequently, there often exists slightly different versions of the same

7
Greenwich Mean Time (GMT) is the local mean time on the Greenwich meridian. Although GMT remains the legal basis for civil
time in the UK, all broadcast time signals are based on world standard time, also known as Coordinated Universal Time (UTC), which
is maintained on an atomic clock and periodically adjusted by adding leap seconds, based on astronomical observations.

Alan Dunn 15 25/5/13

 equation, but for practical purposes, the results obtained from each are the same. These notes
follow the approach of (CIBSE, 2002) which closely follows (Scharmer & Greif, European Solar
Radiation Atlas Part 1: Fundamentals and Maps, 2000). Other sources include (ASHRAE, 2009),
and very detailed algorithms and c-code from ( Andreas & Reda, 2008)

5.1. Co-ordinate systems

For building-solar applications the geometry is simpler if we study the apparent motion of the
sun in the sky against our fixed horizon using the Horizontal coordinate system. Here the
position of the sun is defined by two angles: the solar altitude, : the angular elevation of the
centre of the solar disk above the horizontal plane; and the solar azimuth, : the horizontal
angle between the plane containing the centre of the solar disk and the vertical plane running in
a true north-south direction. Solar azimuth is zero at due south and positive to the west,
negative to the east (in the northern hemisphere).

Figure 13 Definition of angles (CIBSE, 2002)

Using the law of cosines for spherical triangles the solar altitude, is given by:

5-1

And the solar azimuth angle is given by:

5-2

5-3

To ensure the correct sign for in all quadrants, apply the following rule.

If then ( ); if then ( )

Alan Dunn 16 25/5/13

 Solar position is uniquely specified by solar altitude, and solar azimuth, ;
both of which are functions of latitude, ; (indirectly) time of year, ; and time of
day, . All of these terms are described below.

5.2. Latitude, and longitude,

Together with altitude, these form the Geographic
Coordinate System that is used specify any location on Earth.

Latitude, is the angle at the centre of the coordinate

system between any point on the earth's surface and the
plane of the equator. Lines joining points of the same
latitude are called parallels.

Longitude, is the angle east or west, at the centre of the

coordinate system, between any point on the earth's surface Figure 14 Geographic
and a north-south line passing between the two geographical Coordinate System
poles and a brass strip on the floor of the Greenwich
Observatory. Lines joining points of the same longitude are called meridians.

5.3. Solar declination,

Solar declination is the angular position of the sun at solar noon with respect to the plane of the
equator, north positive.

Different empirical formula appear in the literature, but the following is widely used in Europe
(CIBSE, 2002) (Scharmer & Grief, European Solar Radiation Atlas Vol. 2: Database and
exploitation software. 4th ed., 2000) and is adopted here.

[ ( )] 5-4

Where is the Day Number (January 1 = 1, February 1 = 32, March 1 = 61), and is the Day
Number expressed in radians:

( )( )

The denominator, 365.258 is the precise number of days in a year, and 360 is the number of
degrees through which the Earth rotates in a year.

8
The tropical year lasts 365.24219879 days, but a year in the Gregorian Calendar (which we use today) is
365 + 97/400 = 365.2425, as there are 97 leap years every 400 years.

Alan Dunn 17 25/5/13

 Figure 15 Solar declination

The equation may be integrated to give a mean monthly value see (CIBSE, 2002) Table 5.2.

5.4. Hour angle,

The angular displacement of the sun east or west of the local meridian: morning negative,
afternoon positive.
( ) 5-5

Where = Local Apparent Time in hours (24 hour clock)

Note: the Earth rotates through 15o in 1 hour.

5.5. Angle of Incidence on an Inclined Surface

We calculate below the angle of incidence ( ) on a surface of tilt and a surface azimuth
angle, (refer to Figure 13).

First calculate the wall-solar azimuth

5-6

If then ; if then

The angle of incidence ( ) may now be found from

( ) ( ) 5-7

A negative value for ( ) indicates that the sun is behind the surface.

5.6. Sunrise and Sunset

Astronomical sunset occurs when the altitude of the centre of the solar disc is zero. Setting
in Equation 5-2 and rearranging gives:

( ) 5-8

Where is the sunset hour angle measured in degrees.

Alan Dunn 18 25/5/13

 At this time half of the sun’s disc visible above the horizon. The average radius of the sun’s disc
(at 1 AU) is 16’ of arc, so theoretically, sunset occurs when . But even this is not true.
Close to the horizon light rays are bent by 34’ due to horizontal atmospheric refraction and the
sun is still visible below the horizon. So in practice sunset, the time when the sun is no longer
visible, occurs when, .

A final correction may be necessary for the observer height above sea level, (metres) and the
following equation accounts for all three factors:

√ 5-9

Even when the sun is no longer visible the sky remains light - twilight. There are various official
definitions of twilight in the UK (Yallop & Hohenkerk, 1996).

Type of Twilight Solar Altitude

Civil -6o
Nautical -12o
Astronomical -18o
Figure 16 Definitions of Twilight

5.6.1. Sunrise and sunset on an inclined surface

The sunrise and sunset times on an inclined surface may be found by setting ( ) in
the Equation 5-7 and solving for . Then use to find .

5.6.2. Horizon File

The above analysis assumes that we have a clear view of the horizon. In the developed world
this situation is increasingly rare. Usually the view of the horizon is obstructed by buildings near
and far, adjacent collectors and the surrounding topology of the land. The effect of these
obstructions is that at certain times beam radiation will be obstructed when solar altitudes are
small; and to permanently reduce the levels of diffuse radiation by reducing the percentage of
the sky vault visible.

In a simulation program these obstructions may be defined in a horizon file as shown in Figure
17. Some software (Meteonorm, 2013) will automatically generate a horizon file from a digital
terrain model (DTM)9

9
See: NASA's Space Shuttle Radar Topography Mission [Link]

Alan Dunn 19 25/5/13

 20o
18o
Horizon Altitude, αH 16o
14o
12o
10o
8o
6o
4o
2o
0o
0o 30o 60o 90o 120o 150o 180o 210o 240o 270o 300o 330o 360o
Solar Azimuth, γS

Figure 17 An example horizon file

A simulation programme will calculate the solar altitude and azimuth and if for a given azimuth
the solar altitude is less than the horizon file altitude then the beam radiation will be set to zero.

Some Geographical Information Systems (GIS) incorporate solar modelling and will automatically
determine the shading by terrain (see GRASS and the plugin r.sun10)

Three conditions must be checked to determine if beam radiation is falling on an inclined surface

a. The solar altitude is less than -50’

b. The angle of incidence on the include surface is > 90o
c. At the given solar azimuth the solar altitude is less than the horizon file altitude;

If anyone (or more) of these conditions is true then beam radiation will not reach the surface.

6.0 Terrestrial Radiation

The total daily irradiation at the Earth’s surface is called the Global Radiation, which may be
divided into two components:

– Direct Beam Radiation from the sun

– Diffuse Radiation from the vault of the sky

6.1. Extraterrestrial radiation on a horizontal plane

Extraterrestrial radiation is often used to normalize data and to calculate clearness indices (See
Section 6.5.1).

By combining Equations 3-1, 3-2 and 5-1 we can show that at a given time the Irradiance (W∙m-2)
on a horizontal plane outside of the Earth’s atmosphere is given by:

6-1
( )

10
See: [Link] [Accessed: 1-9-13]

Alan Dunn 20 25/5/13

 The daily irradiation (kWh) is obtained by integrating the previous equation between the hours
of sunrise and sunset (See Appendix 0 for the derivation).

( ) [( ) ] 6-2

Table 5.4 (CIBSE, 2002) tabulates monthly mean values of daily extraterrestrial irradiation on a
horizontal plane for five UK locations.

6.2. The Earth’s Atmosphere

The standard atmosphere extends 100 km above the
earth’s surface and is divided into four zones with
demarcations occurring at reversals of temperature
gradient. A mean temperature gradient is shown in
Figure 18; the actual gradient varies between
summer and winter and with climatic zone.
Extraterrestrial radiation is greatly attenuated as it
passes through the Earth’s atmosphere by:

• Scattering by air molecules, dust, water

• Absorption by ozone (UV) and water (IR) Figure 18 Atmospheric layers
• Reflection
• Re-radiation (long wave)

Figure 19 shows a comparison between the extraterrestrial solar spectrum and the clear sky
terrestrial solar spectrum for an air mass of 1 showing the extent of atmospheric attenuation.

Figure 19 Terrestrial solar spectrum

6.2.1. Atmospheric scattering

Alan Dunn 21 25/5/13

 Solar radiation is scattered as it passes through the atmosphere by interaction with air
molecules, water and dust. The degree of scattering depends on the number of particles the
radiation encounters and the particle sizes: the
number of particles depends on the path length
which is measured by the air mass, but the particle
sizes vary with time a location.

There are two models of scattering as shown in

Error! Reference source not found.: (a) Rayleigh
Scattering occurs when particles are much smaller
than the wavelength ( ) of the radiation. The
scattering coefficient is found to be proportional to
with shorter wavelengths being scattered more.11

(b) Water and dust can coalesce into particles of a

Figure 20 Atmospheric scattering size equal to or larger than the wavelength of the
radiation, in which case the interaction is best
modelled using Mie Scattering. Mie scattering is less wavelength dependent with a larger
proportion of the radiation scattered in a forward direction.12

In the 1930s Ångström proposed a single attenuation coefficient, given below:

Where is Ångström turbidity13 coefficient and is a measure of the number of aerosols in the
atmosphere in the zenith direction, normally having a value between 0 and 0.5. The wavelength
exponent is a measure of size distribution of aerosol particles, with higher values indicating a
high proportion of small particles. Values range from 0.5 to 2.5.

6.2.2. Atmospheric absorption

Ozone (O3) absorbs ultraviolet radiation with almost complete absorption at wavelengths less
than 290 nm. Infrared radiation is absorbed by both water (H2O) and CO2); as shown in Figure 19
the absorption appears in selective bands. The radiation absorbed will raise the temperature of
the particle which will then emit longwave radiation, some back into space, some towards earth.

6.3. Balance of Atmospheric Energy

11
This is why the sky is blue.
12
This is why mists are white.
13
Turbidity (noun) Muddiness created by stirring up sediment or having foreign particles suspended.

Alan Dunn 22 25/5/13

 Figure 21 Atmospheric energy exchanges

6.4. Clear Sky Radiation

Clear sky radiation represents maximum radiation
that can occur at a given altitude when the sky is SoDa Algorithmic Chain
clear and cloudless. Even under these conditions the
beam radiation is attenuated by scattering and Algorithm 3.2
absorption creating diffuse radiation. CIBSE uses the See §8.0
theoretical clear sky model developed for the
European Solar Radiation Atlas (Rigollier, Bauer, & Wald, 2000) and which is described in section
5.3.2 (CIBSE, 2002). The beam horizontal irradiance for a clear sky, is given by:

( )
6-3
Where
1,367 W∙m-2 Solar constant (see §3.2.1)
- Correction to mean solar distance (see §3.2.2)
Solar altitude
- Linke turbidity factor (for air mass m = 2)
- Relative optical air mass
( ) - Rayleigh optical depth of clear blue dry sky at air mass .

These terms are described below.

Alan Dunn 23 25/5/13

6.4.1. Relative optical air mass, m
Beam radiation travelling vertically down through the atmosphere will encounter a standard
mass of air, . If the path of the beam radiation is not vertical but at a solar altitude, the
path will be longer and it will encounter a greater air mass, usually calculated by

However a more accurate expression is given by (Kasten & Young, 1989)

⁄[ ( ) ]

A further correction for air pressure due to site altitude

( )

For low solar altitudes and therefore a large air mass, a correction may be made to for refraction.

Where

- Air mass at sea level

- Air mass at site altitude
o
Altitude of sun
Pa Atmospheric pressure at site altitude
Pa Atmospheric pressure at sea level

6.4.2. Rayleigh optical depth, δr

The Rayleigh optical depth describes the transmission characteristics of a pure, clean dry
atmosphere without pollution and is a function of the air mass alone.

⁄ ( ) 6-4

For an air mass, use

⁄ ( ) 6-5

6.4.3. Linke Turbidity, TL

The Linke turbidity factor models approximately the atmospheric absorption by water vapour and
scattering by aerosol particle relative to a clean and dry atmosphere. It is the ratio of the observed
optical depth of the atmosphere at solar zenith to the optical depth of a perfectly clean, dry Rayleigh
atmosphere (also estimated at the zenith).

( )
( ) 6-6
( )

Where ( ) is the relative optical thickness relating to Rayleigh scattering by gaseous molecules in
the atmosphere and ozone absorption.

( ) is the relative optical thickness associated with aerosol extinction and gaseous absorption
other than ozone in the stratosphere.

Alan Dunn 24 25/5/13

Worldwide values for monthly mean Linke Turbidity factor are available from the SoDa web service
(ParisTech, 2013).

6.4.4. Clear sky diffuse irradiance on horizontal surfaces

As the clear sky becomes more turbid the beam irradiation decreases and the diffuse irradiation
increases. The clear sky diffuse horizontal irradiance is a function of solar altitude, site elevation
and Linke turbidity factor.

6-7
Where

With a condition on such that

If ( ) then

See the Clear Sky tab on the accompanying spreadsheet

Solar_Resource.xlsm for an example of calculating clear sky
irradiation.

6.4.5. Clear sky global irradiance on horizontal surfaces

The clear sky global irradiance, is the sum of the clear sky beam and diffuse irradiances
calculated above:

6.4.6. Clear sky global irradiation on horizontal surfaces

To obtain daily or monthly values for clear sky irradiation (kWh∙m-2) it is necessary to integrate the
beam and diffuse irradiation between the hours of sunrise and sunset. This may be done numerical
by stepping through the day in hourly or half-hourly time-steps. Equations for beam and diffuse
irradiation may be converted into a form that allows an analytical solution (Rigollier, Bauer, & Wald,
2000) but the equations are complex.

The starting point for the SoDa Algorithmic Chain (Section 8.0) is the calculation of the mean
monthly clear sky global radiation.

Alan Dunn 25 25/5/13

6.4.7. Terrestrial Radiation – Clear Sky Spectral Distribution

Figure 22 Terrestrial spectral distribution (m = 1)

Photovoltaics are particularly sensitive to variations in the solar spectrum. The solar spectrum is
constantly changing with atmospheric conditions and cloud cover but an internationally agreed
solar spectrum may be used to compare the performance of PVs (BS EN 60904-3, 2008). The
spectrum is for global irradiation at an air mass of 1.5 on a surface tilted at 37 o towards the
equator with the sun at a zenith angle of 48.19o.14 The spectral distribution is calculated by the
publicly available SMARTS program (NREL, 2013)

6.5. Total sky radiation

6.5.1. Clearness Index

The effect of all forms of atmospheric attenuation may be combined in the Clearness Index (KT).
KT is the ratio of terrestrial to extraterrestrial irradiation at a given location. There are three
definitions depending on the period of interest.

Hourly KT-value

6-8

Daily KT-value

6-9

Monthly mean daily KT-value

( )
( ) 6-10
( )

14
The curious angle is because ( ) i.e. an air mass of 1.5

Alan Dunn 26 25/5/13

 The European Solar Atlas Project has been mapping Europe by satellite for 10 years. The Atlas
publishes maps of ( ) for Europe on a 10 km grid.

If the clearness index is known then the terrestrial global radiation is predicted. For example if
( ) is known then the monthly mean daily extraterrestrial radiation, ( ) is calculated
from equation XXX and the monthly Mean Daily Global Irradiation, ( ) is given by

( ) ( ) ( )

Generation of synthetic radiation time series

6.5.2. Split of beam and diffuse

As explained later in §9.0, the measurement of diffuse radiation is quite complex and
consequently very few datasets are available. It is usually necessary to predict the split between
beam and diffuse radiation from measured global values. Early regression models attempted to
predict the diffuse radiation as a function of the clearness index, with different models
required for hourly, daily and monthly values.

Beam and diffuse components of monthly radiation

Beam and diffuse components of daily radiation

An example is (Collares-Pereira & Rahl, 1979)

R. Perez, R. Stewart, C. Arbogast, R. Seals, J. Scott, An anisotropic hourly diffuse radiation model
for sloping surfaces: Description, performance validation, site dependency evaluation, Solar
Energy, Volume 36, Issue 6, 1986, Pages 481-497

Beam and diffuse components of hourly radiation

An example is (Erbs, Klein, & Duffie, 1982)

The disaggregation of hourly global radiation into beam and diffuse components is an important
link in the SoDa algorithmic chain (see §8.0) where the correlation of (Perez, Ineichen, Maxwell,
Seals, & Zelenka, 1992) is used.

The Perez model is based on four parameters that define

the sky condition: SoDa Algorithmic Chain
Algorithm 3.2
See §8.0
Alan Dunn 27 25/5/13
 a. Global horizontal radiation, or alternatively, normalized clearness index, , whereby for a
formula independent of the zenith angle is used (Perez et al. 1990b).

b. Zenith angle of the sun.

c. When time series of global radiation are available, a stability index can be calculated
giving the dynamics of the time series. (| | | |) where
points , and refer to the present, previous and subsequent hour.

d. When the dew point temperature is available, it can be used to provide an accurate estimate
of water content (humidity) in the atmosphere which in turn influences the absorption and
the production of aerosols. The humidity is estimated on the basis of dew point temperature
by the method of Wright et al. (1989).

7.0 Total Radiation on a Surface

This is what we have been building towards: all the
geometry and radiation analysis comes together to
SoDa Algorithmic Chain
predict the total radiation on an arbitrarily tilted surface. Algorithm 3.2
The surface could be a solar collector, a photovoltaic See §8.0
module or a passive element in a building. For detailed
analysis hourly values of global radiation are required.

Inclined surface hourly global irradiation has three components and is calculated using the Perez
model (Perez, Stewart, Arbogast, Seals, & Scott, 1986);

• Slope Direct Beam, ( )

• Slope Sky Diffuse, ( )
• Slope Ground Reflected Diffuse, ( )

7.1. Direct Beam Irradiance

The simplest of the three components:

( ) [ ( )] 7-1

Where ( ) is the angle of incidence (see §5.5)

7.2. Sky diffuse irradiance

The simplest models assume an isotropic sky (i.e. that diffuse radiation is uniform across the
vault of the sky:
( ) ( ) 7-2

The term ( ) is a view factor representing the fraction of the sky seen by a tilted
surface. When the tilt angle (the surface is horizontal) then and ( )
.

When the tilt angle (the surface is vertical) then and ( ) ⁄ – it

can only see half of the sky.

Alan Dunn 28 25/5/13

 Of course real skies are not isotropic and modern anisotropic models generally divide the sky
into three sections:

a. A circumsolar region of bright sky surrounding, and moving with the sun. The area is defined
by a solid angle.
b. A region of brighter sky close to the horizon.
c. The remaining sky is treated as isotropic.

CIBSE uses a model developed by Muneer as part of the European Solar Radiation Atlas project,
see §[Link] (CIBSE, 2002); whereas most other key sources (SoDa. PVSyst, Meteonorm,
EnergyPlus) use the Perez Correlation

7.3. Ground reflected radiation

Reflected hourly ground irradiation:

( ) 7-3

( )

If the tilt angle, i.e. the collector is flat and cannot see the ground and the ground view
factor, .

If the tilt angle, i.e. the collector is vertical then the horizon will split the view: half is
the ground and half the sky and, .

is the ground albedo.

Table 5.9 in (CIBSE, 2002) gives typical albedo values for various ground types, and NASA
maintains an online library (NASA, 2013) giving the ground reflectance as a function of
wavelength for more than 2,400 man-made and natural materials.

See Also

Table 5.10 (CIBSE, 2002) Monthly mean daily irradiation on inclined planes for 3 UK locations.
These are compiled from weather data for the period 1981-92.

CIBSE Guide J Appendix A4.3 Algorithm for estimation of monthly mean daily beam diffuse
horizontal irradiation and corresponding hourly irradiances from monthly mean daily global
irradiation on a horizontal surface

7.4. Perez Sky Diffuse Model SoDa Algorithmic Chain

While the sky diffuse model presented up to this
point separated the isotropic, circumsolar, and Algorithm 3.2
horizon components explicitly, Perez developed a See §8.0
more complex model that relies on a set of empirical
coefficients for each term.

Alan Dunn 29 25/5/13

The basic form of the model is:

( )
( ) [( )( ) ( ) ( )] 7-4

Where

( ) kWh∙m-2 Diffuse irradiance falling on a surface with azimuth α and slope

kWh∙m-2 Diffuse irradiance on the horizontal
- Circumsolar coefficient
- Horizon “reduced” brightness coefficient
Incidence-weighted solid angle sustained by the circumsolar region as seen by
steradian
the tilted surface.
Incidence-weighted solid angle sustained by the circumsolar region as seen by
steradian
the horizontal.

[ ( ( ) )] and [ ( ) ( )]

( )  Angle of incidence between the sun and the plane of the array.
Solar zenith angle.
 Surface tilt angle from horizontal.

[ ( )]

The coefficients are defined for specific bins of clearness ( ), which is defined as:

( )⁄
7-5

Where is direct normal irradiance, is a constant equal to 1.041 for zenith angles in radians.

7-6

Where is the absolute air mass, and is extraterrestrial normal incident irradiance.

Perez has published a number of different versions of the coefficients fitted to various data sets [2,
3 , 4]. Table 1 shows the coefficient values published in [3] for irradiance. The bin refers to bins
of clearness, , defined in Table 2.

bin
1 -0.008 0.588 -0.062 -0.06 0.072 -0.022
2 0.13 0.683 -0.151 -0.019 0.066 -0.029
3 0.33 0.487 -0.221 0.055 -0.064 -0.026
4 0.568 0.187 -0.295 0.109 -0.152 -0.014
5 0.873 -0.392 -0.362 0.226 -0.462 0.001
6 1.132 -1.237 -0.412 0.288 -0.823 0.056

Alan Dunn 30 25/5/13

 bin
7 1.06 -1.6 -0.359 0.264 -1.127 0.131
8 0.678 -0.327 -0.25 0.156 -1.377 0.251
Table 1 Perez model coefficients for irradiance

bin Lower Bound Upper Bound

1 Overcast 1.000 1.065
2 1.065 1.230
3 1.230 1.500
4 1.500 1.950
5 1.950 2.800
6 2.800 4.500
7 4.500 6.200
8 Clear 6.200 –
Table 2 Sky clearness bins

8.0 Algorithmic Chains

To simulate the performance of a solar energy system we need a representative annual time series
of hourly values of beam and diffuse radiation on an inclined plane. At one extreme we may actually
have these values, if they have been measured at the actual site; at the other extreme we may only
have monthly mean values for global radiation. The Soda project introduced the concept of
algorithmic chains, so that whatever the starting point, a series of equations or algorithms may be
called to generate the required hourly values. The chains are described in (Remund, Wald, & Page,
2003) and (Remund & Page, 2002) - an extract is shown below.

Alan Dunn 31 25/5/13

 Figure 23 Extract from SoDa algorithmic chain

Figure 23 shows four starting points based on available data and the chain of algorithms to be
employed to generate hourly values of beam and diffuse radiation on an inclined surface. The
algorithms are cross-referenced to sections in these notes.

9.0 Radiation Measurements

9.1. Ground based radiation measurements

Four types of ground based measurement: sunshine recording, pyrheliometers, pyranometers
and pyrgeometers, see section 4.1 of (Scharmer & Greif, 2000)

Ideally for a location we will have measured data for direct beam and diffuse radiation made
with a pyranometer.

Alan Dunn 32 25/5/13

 Figure 24 Pyranometer Figure 25 Shading Band

G. Blaesser, (1997) PV system measurements and monitoring the European experience, Solar
Energy Materials and Solar Cells, Volume 47

Pyranometers

Global Radiation

Diffuse Radiation

9.2. Satellite based radiation measurements

Measured values for global and diffuse radiation are not always available.

Values may be estimated from:

• Sunshine hours
• Hourly cloud data
• Clearness index

10.0 Sources of radiation data

When calculating cooling loads of buildings to size chillers and air conditioning plant, clear sky
data is used. This data is applicable over a wide geographic area and so, for example, CIBSE
provides for XXX UK sites and it is appropriate to use the nearest for the design.

For energy predictions and prediction of renewable technology performance hourly values of
real-sky radiance are required

10.1. Interpolation
Solar daily radiation data should not be used if the measuring station is more than 20 km away,
of 30 km for monthly averaged daily values. (Muneer, 2004). Better results are obtained by
interpolating between a number of stations. Meteonorm uses a 3-D inverse distance
interpolation model; see section 6.2.1 (Meteonorm, 2013). The same method is used to
interpolate temperature and wind data but with corrections for differences in local terrain. ESRA
uses a co-kriging interpolation method.

Alan Dunn 33 25/5/13

 An extensive literature review of meteorological interpolation methods was carried out by the
Royal Netherlands Meteorological Institute and is publicly available, (Sluiter, 2009).

10.2. Generation of synthetic hourly radiation data

Because of the paucity of reliable hourly data, recent approaches have focused on using
stochastic methods to generate synthetic data from known monthly means. The methods used
generate data having the same statistical properties as the measured data i.e. mean, variance
and autocorrelation. Researchers (Gansler, Klein, & Beckman, 1994) have shown that synthetic
data may be used in place of long-term measured data with no loss in accuracy.

This approach is used by both Meteonorm (Meteonorm, 2013) and the European Solar Radiation
Atlas (Scharmer & Greif, 2000)

10.2.1. Markov Transitional Matrices

Daily irradiance values may be treated as a Markov chain which assumes that the probability next value of
radiation depends only on the current value. Markov Transition Matrices can be constructed from large sets
of measured data that have worldwide applicability and may be used to generate sequences of mean daily
global irradiance. The methodology was developed by (Aguiar & Collares-Pereira, 1988) and is
incorporated into the SoDa service (Remund & Page, Advanced parameters WP 5.2b: Chain of algorithms:
short- and longwave radiation with associated temperature prediction resources, 2002).

An introduction to Markov chains and transition matrices is found in Appendix 14.0 and an example
calculation showing how to generate a sequence of daily values is given in Appendix 1.0.

10.2.2. Time series prediction

Attempts have been made to predict hourly values of irradiation using Markov chains, but
correlations were poor. The reason being that the future values of irradiation does not just
depend on the current value, but also on the solar altitude and hence time of day15.

A variety of time series models have been used to predict solar irradiation (Reikard, 2009) but
currently the most widely used method to predict hourly values from daily uses a time-
dependent autoregressive Gaussian model (Aguiar, Collares-Pereira, & Conde, 1992).

Synthetic data does not have to be identical to measured data to produce accurate simulation
results, but it does need to have similar statistical properties. In particular, the mean, standard
deviation, and autocorrelation should be close.

This is a two-part process. First a series of mean hourly values are calculated from the daily
irradiance and then random hourly variations are superimposed on the mean profile.

(a) Mean daily irradiance profile generation

The assumption is that the hourly irradiance values that make up the daily profile will follow
the clear sky irradiance profile.

15
The mathematical term is that the time series is “non-stationary””.

Alan Dunn 34 25/5/13

 Hourly global horizontal irradiance To be found
Daily global horizontal irradiance From Section 10.2.1
Clear sky hourly global irradiance From Section 6.4.6
Daily clear sky global irradiance From Section 6.4.5

and are calculated from the clear sky model and is predicted from the mean monthly
irradiance using Markov Transitional Matrices.

(b) Intermittent hourly variation

The second part simulates the intermittent hourly variations by superimposing an auto-
regressive procedure of the first order (AR-1) (Box & Jenkins, 1994)

( ) ( ) ( )

( ) Predicted hourly value

( ) Mean hourly value calculated in part (a)
( ) First order autoregressive function
Hour

The auto regressive function16 is calculated as follows.

( ) ( )

Where

The autocorrelation function

Normally distributed random variable with expected value 0 and standard deviation

( )

( )

The variable is the standard deviation of the normally distributed random function used to
generate , as is given by:

( )

Finally, is the standard deviation of perturbations and given by:

The function is autoregressive (AR) because its current value ( ) depends on its previous value
16

( ). It does not depend on earlier values, such as ( ) or ( ), so it is a first order

autoregressive model (AR-1).

Alan Dunn 35 25/5/13

 { [ ( )]}

Where
[ ( ) ]
( )

11.0 Data for simulation

There are four formats in which radiation data may be used to simulate renewable systems.

11.1. Measured hourly data

If measured data is available for the site, or has been interpolated from a number of surrounding
locations, this may be used for simulation. However, we cannot just use last year’s data as it may
not be representative of long term weather: last year may have been exceptionally hot, or cold.
Two approaches are possible:

(a) If we have, say, 30 years of data, we could average the hourly values. That is, for all years at
a given hour, we add up the irradiation, and divide by 30 to obtain the mean irradiation
for each hour of the year, ̅ .

̅ (∑ )⁄

This will produce 8760 average hourly values, but the averaging will smooth out the time
series, removing extreme values.

(b) To avoid the smoothing effects of (a) we could instead for each month determine which of
the 30 months was the most average. For example, we might find that the most average
January in the last 30 years was in 1998, and the most average February was in 1985 etc.
These months are then stitched together to create a Typical Weather Year.

11.2. Frequency distributions

A model using hourly data requires a spreadsheet performing calculations 8760 rows (or
columns). The volume of data may be reduced by binning. Figure 26 was obtained by counting
the number of hours each year that radiation fell into each bin. E.g. the irradiation is between
100 and 200 W for 790 hours each year, and between 500 and 600 W for 305 hours. (Hours of
darkness have been omitted so the total hours of irradiation each year is 4,639.)

Alan Dunn 36 25/5/13

 Bin Frequency
0-99 1837
100-199 790
200-299 463
300-399 407
400-499 316
500-599 305
600-699 250
700-799 183
800-899 84
900-999 4
Figure 26 Frequency distribution of irradiation (Kosice)

A frequency distribution may be easily generated using the in-built Excel array function

FREQUENCY(Data_array, Bins_array)

Now only ten calculations (down from 8760) are necessary using the mid-value of each bin. There is
a loss of accuracy when binning which increases as the number of bins decreases. The choice of the
number of bins to be used is a trade-off between ease of calculation and accuracy.

11.3. Typical day of the month

Another approach to reducing the volume of data used in a simulation is to generate hourly
values for a single typical day of each month. As in Section 11.1 two approaches are possible:

(a) For example take the month of January. For each hour sum the irradiation that occurred
during that hour for every day in January for each of the 30 years of data available and then
divide by

̅ ∑∑ ⁄

The Excel VBA function in Appendix 18.0 may be used to generate average hourly values for
each month from annual hourly data.

(b) As before, averaging has the effect of smoothing the data and eliminating extreme values.
An alternative approach would be, for each month (e.g. January) to examine the hourly
values for each of the 990 days and select the most average day to represent the month.

12.0 References
Andreas, A., & Reda, I. (2008). Solar Position Algorithm for Solar Radiation Applications NREL/TP-
560-34302. Golden, Colarado: National Renewable Energy Laboratory.

Aguiar, R. J., Collares-Pereira, M., & Conde, J. P. (1992). TAG : a Time-dependent, Autoregressive,
Gaussian Model for Generating Synthetic Hourly Radiation. Solar Energy, 49(3), 167-174.

Aguiar, R., & Collares-Pereira, M. (1988). A simple procedure for generating sequences of daily
radiation values using a library of markov transition matrices. Solar energy, 40(3) 269-279.

Alan Dunn 37 25/5/13

ASHRAE. (2009). Chapter 14 Climatic Design Information. In ASHRAE Handbook:Fundamentals.
Atlanta: ASHRAE.

Box, G., & Jenkins, G. (1994). Time series analysis: Forecasting and control 3rd Ed. Prentice Hall.

BS EN 60904-3. (2008). Photovoltaic devices. Measurement principles for terrestrial photovoltaic (PV)
solar devices with reference spectral irradiance data. London: BSI.

CIBSE. (2002). Chapter 5 Solar radiation, long-wave radiation and daylight. In C. I. Engineering, CIBSE
Guide J Weather, solar and illuminance data (p. 455). London.

Collares-Pereira, M., & Rahl, A. (1979). The average distributi of solar radiationon - correlation
between diffuse and hemispherical and between daily and hourly insolation values. Solar
Energy, 22(2) 155-164.

DECC. (2011a). Microgeneration Strategy. London: UK Government.

DECC. (2011b). UK Renewable Energy Roadmap. London: UK Government.

Ehnberg, J. S., & Bollen, M. H. (2005). Simulation of global solar radiation based on cloud
observations. Solar Energy, 78(2) 157–162.

Erbs, D. G., Klein, S. A., & Duffie, J. A. (1982). Estimation of the diffuse radiationfraction for hourly,
daily and monthly-average global radiation. Solar Energy, 28(4), 293-302.

Gansler, R. A., Klein, S. A., & Beckman, W. A. (1994). Assessment of the accuracy of generated
meteorological data for use in solar simulation studies. Solar Energy 53((3), 279-287.

IEA. (2012). Key World Energy Statistics. Paris: International Energy Agency.

ISO. (1992). 9845-1: Solar energy -- Reference solar spectral irradiance at the ground at different
receiving conditions -- Part 1: Direct normal and hemispherical solar irradiance for air mass
1,5. International Standards Organisation.

Kasten, F., & Young, A. T. (1989). Revised optical air mass tables and approximation formula. Applied
Optics, 28(22) 4735-4738.

Meteonorm. (2013). Global meteorological database handbook part II: Theory . handbook:
Meteonorm.

Muneer, T. (2004). Solar Radiation and Daylight Models 2nd Ed. Oxford: Elsevier.

NASA. (2013). Home. Retrieved June 28, 2013, from ASTER Spectral Library:
[Link]

NREL. (2013, August 8). SMARTS - Simple Model of the Atmospheric Radiative Transfer of Sunshine.
Retrieved August 7, 2013, from National Renewable Energy Laboratory:
[Link]

ParisTech, A. (2013). SoDa. Retrieved from SoDa: [Link]

Alan Dunn 38 25/5/13

Perez, R. R., Ineichen, P., Maxwell, E. L., Seals, R. D., & Zelenka, A. (1992). Dynamic global-to-direct
irradiance conversion models. ASHRAE Transactions, v98 Pt.2 .

Perez, R., Stewart, R., Arbogast, C., Seals, R., & Scott, J. (1986). An anisotropic hourly diffuse
radiation model for sloping surfaces: Description, performance validation, site dependency
evaluation. Solar Energy 36(6), 481-497.

Reikard, G. (2009). Predicting solar radiation at high resolutions: a comparison of time series
forecasts. Solar Energy, 83(3) 342-349.

Remund, J., & Page, J. (2002). Advanced parameters WP 5.2b: Chain of algorithms: short- and
longwave radiation with associated temperature prediction resources. EU: SoDa.

Remund, J., Wald, L., & Page, J. (2003). CHAIN OF ALGORITHMS TO CALCULATE ADVANCED
RADIATION PARAMETERS. Proceedings of ISES Solar World Congress. Goteborg, Sweden:
International Solar Energy Society.

Rigollier, C., Bauer, O., & Wald, L. (2000). On the clear sky model of the ESRA - European Solar
Radiation Atlas - with respect to the Heliosat method. Solar Energy, 68(1) 33-48.

Scharmer, K., & Greif, J. (2000). European Solar Radiation Atlas Part 1: Fundamentals and Maps.
Paris: Les Presses de l’École des Mines.

Scharmer, K., & Grief, J. (2000). European Solar Radiation Atlas Vol. 2: Database and exploitation
software. 4th ed. Paris: Les Presses de l’Écoles des Mines de Paris.

Sluiter, R. (2009). Interpolation methods for climate data - literature review IR 2009-04. de Bilt:
KNMI.

UK Hydrographic Office. (2011). World Time Zone Map. Retrieved August 21, 2013, from HM
Nautical Almanac Office:
[Link]

Yallop, B. D., & Hohenkerk, C. Y. (1996). ASTRONOMICAL INFORMATION SHEET No. 7 A note on
sunrise, sunset and twilight times and on the illumination conditions during twilight.
Retrieved July 5, 2013, from HM Nautical Almanac Office:
[Link]

International Vocabulary of Basic and general Terms in Metrology (VIM, 1984), ISO, Second edition,
Switzerland, 1993

Guide to the Expression of Uncertainty in Measurement (1995) (GUM, 1995), ISO, Switzerland.

IEA Solar Heating & Cooling programme Task 46 Solar Resource Assessment and Forecasting.

Alan Dunn 39 25/5/13

13.0 Appendix: Derivation of mean monthly daily global extraterrestrial
irradiation
At a given time the Irradiance on a horizontal plane outside of the Earth’s atmosphere is:

( ) 13-1

To find the mean monthly daily global extraterrestrial irradiation we must integrate Equation
13-1 between the hours of sunrise and sunset, using values for the mid-day of the month. (Since
the irradiation is symmetrical around solar noon, we can integrate from midday to sunset and
multiply the answer by 2.) So,

( ) ∫ { }

Where

is the sunset hour angle in radians

Integrating

( ) [ ] [ ]
[ ]

Since radians equals 24 hours

( ) ( ) [ ]

And if we wish to express the sunset angle in degrees rather than radians

( ) ( ) [( ) ]

Alan Dunn 40 25/5/13

14.0 Appendix: Transitional Markov Matrices
This is a technique used to generate hourly weather data from monthly values. The method is
described in general here and then followed be a specific example.

14.1. Markov Chain

A Markov chain is a system made up of a finite set of states. The process starts in one of the
states and moves successively from one state to another. It is a random process and the next
state depends only on the current state. Earlier states do not affect the transition – the system is
memory-less.

Figure 1 represents a system having two states A and E and shows

the transition probabilities. For example if the system is currently
in state E there is a 0.7 probability that after the next time step the
state will be A.

The transition probabilities may be concisely represented in a

transition matrix. If we code state E as 1 and state A as 2, then
( | ) or more concisely, . So for
Figure 1 the transition matrix is:

( ) ( )

Note that each row adds up to 1 as the system must be in either state A or E after the next time
step.

Matrices increase in order as the number of system states increases. For a three state system:

( )

Note that the diagonal elements show the probability of a state remaining the unchanged.

14.2. Multiple Transitions

If the system is currently in state E what is the probability that it will be in state A in two time
steps from now?

There are two transition sequences that could lead to the state changing from E to A in two
steps:

{ } or { }

Overall probability is

Note this calculation is actually the dot product of two vectors. We have dotted the first row of
with the second column of .

Multiple transitions may be clearly represented using a probability tree

Alan Dunn 41 25/5/13

 Figure 27 Probability tree

If is the vector of states at time and is the transition matrix then

Not all transition matrices yield ergodic or steady state distributions. For a transition matrix to
produce ergodic distribution, one of the eigenvalues has to equal 1, while the remaining
eigenvalues have to be below 1. The ergodic distribution shows what the long run distribution is
going to be like if the observed dynamics continue to hold.

14.3. Determining transition probabilities from data

There are a number of ways to determine the transition probabilities for a data set (Refs); the
most intuitive is described below (Ehnberg & Bollen, 2005). For an example we will take hourly
observations of cloud coverage data which is measured in a scale of Oktas, ranging from a
completely clear sky (0 Oktas) to a completely overcast sky (8 Oktas).

To determine the transition probability for, say, moving from 3 Oktas to 5 Oktas, , e analyse
the entire data set and count the number of times that, from one hour to the next, cloud
coverage has changed from 3 to 5 Oktas, . We then count the number of times in the data
set that cloud coverage has changed from 3 Oktas to any other level, including remaining
unchanged, ∑ . The quotient then gives the transition probability. In general each element
of the transition matrix is given by:

( )

( )

Alan Dunn 42 25/5/13

 ( )

( )

( )

Hocaoglu, F.O. (2011) Stochastic approach for daily solar radiation modeling. Solar Energy, 85(2)
278-287.

Alan Dunn 43 25/5/13

15.0 Appendix: Markov transitional matrices
The following Markov Transitional Matrices are used to calculate daily sequences of clear sky daily clearness indices, from knowledge of the
monthly clear sky clearness indices, . Based on the methodology of (Aguiar & Collares-Pereira, 1988) and described in Section 6.6.1 of
(Meteonorm, 2013).

0.0-0.1 0.1-0.2 0.2-0.3 0.3-0.4 0.4-0.5 0.5-0.6 0.6-0.7 0.7-0.8 0.8-0.9 0.9-1.0
0.0-0.1 0.5 0.28 0.15 0.05 0.02 0 0 0 0 0
0.1-0.2 0.2 0.48 0.2 0.1 0.02 0 0 0 0 0
0.2-0.3 0.05 0.2 0.48 0.2 0.05 0.02 0 0 0 0
0.3-0.4 0.02 0.05 0.18 0.5 0.18 0.05 0.02 0 0 0
0.4-0.5 0 0.02 0.05 0.18 0.5 0.18 0.05 0.02 0 0
0.5-0.6 0 0 0.02 0.05 0.18 0.5 0.18 0.05 0.02 0
0.6-0.7 0 0 0 0 0.05 0.2 0.3 0.2 0 0.25
0.7-0.8 0 0 0 0 0.02 0.05 0.2 0.48 0.2 0.05
0.8-0.9 0 0 0 0 0 0 0.05 0.2 0.5 0.25
0.9-1.0 0 0 0 0 0 0 0.2 0.05 0.05 0.7

Table 3 Markov transition matrix for 0.10 < KTm,c ≤ 0.20

Alan Dunn 44 25/5/13

 0.0-0.1 0.1-0.2 0.2-0.3 0.3-0.4 0.4-0.5 0.5-0.6 0.6-0.7 0.7-0.8 0.8-0.9 0.9-1.0
0.0-0.1 0.5 0.28 0.15 0.05 0.02 0 0 0 0 0
0.1-0.2 0.2 0.48 0.2 0.1 0.02 0 0 0 0 0
0.2-0.3 0.1 0.65 0.2 0.05 0 0 0 0 0 0
0.3-0.4 0 0.25 0 0.05 0.3 0.05 0 0 0.05 0.3
0.4-0.5 0 0.4 0.05 0.1 0.4 0.05 0 0 0 0
0.5-0.6 0 0 0 0 0.25 0.5 0.25 0 0 0
0.6-0.7 0 0 0 0 0 0.25 0.5 0.25 0 0
0.7-0.8 0 0 0 0 0 0 0.25 0.5 0.25 0
0.8-0.9 0 0 0 0 0 0 0 0.25 0.5 0.25
0.9-1.0 0 0 0 0 0 0.7 0.05 0 0 0.25

Table 4 Markov transition matrix for 0.20 < KTm,c ≤ 0.30

0.0-0.1 0.1-0.2 0.2-0.3 0.3-0.4 0.4-0.5 0.5-0.6 0.6-0.7 0.7-0.8 0.8-0.9 0.9-1.0
0.0-0.1 0.133 0.319 0.204 0.115 0.074 0.033 0.03 0.044 0.011 0.037
0.1-0.2 0.081 0.303 0.232 0.127 0.088 0.06 0.029 0.031 0.018 0.033
0.2-0.3 0.036 0.195 0.379 0.135 0.087 0.039 0.042 0.027 0.025 0.036
0.3-0.4 0.032 0.19 0.205 0.189 0.119 0.069 0.059 0.038 0.045 0.054
0.4-0.5 0.051 0.175 0.189 0.185 0.14 0.079 0.06 0.04 0.017 0.064
0.5-0.6 0.042 0.213 0.243 0.126 0.117 0.09 0.045 0.036 0.021 0.069
0.6-0.7 0.017 0.166 0.237 0.141 0.1 0.091 0.054 0.062 0.046 0.087
0.7-0.8 0.038 0.171 0.19 0.133 0.095 0.09 0.057 0.062 0.043 0.119
0.8-0.9 0.044 0.093 0.231 0.143 0.115 0.066 0.038 0.06 0.099 0.11
0.9-1.0 0.029 0.131 0.163 0.127 0.062 0.092 0.065 0.072 0.078 0.18

Table 5 Markov transition matrix for 0.30 < KTm,c ≤ 0.40

Alan Dunn 45 25/5/13

 0.0-0.1 0.1-0.2 0.2-0.3 0.3-0.4 0.4-0.5 0.5-0.6 0.6-0.7 0.7-0.8 0.8-0.9 0.9-1.0
0.0-0.1 0.116 0.223 0.196 0.129 0.093 0.077 0.054 0.044 0.032 0.037
0.1-0.2 0.051 0.228 0.199 0.143 0.101 0.083 0.065 0.052 0.035 0.043
0.2-0.3 0.028 0.146 0.244 0.156 0.12 0.092 0.069 0.053 0.04 0.052
0.3-0.4 0.02 0.111 0.175 0.208 0.146 0.104 0.074 0.067 0.044 0.052
0.4-0.5 0.017 0.115 0.161 0.177 0.155 0.102 0.085 0.067 0.054 0.068
0.5-0.6 0.018 0.114 0.147 0.156 0.142 0.123 0.088 0.075 0.06 0.077
0.6-0.7 0.019 0.116 0.152 0.153 0.133 0.1 0.09 0.078 0.061 0.098
0.7-0.8 0.022 0.105 0.145 0.134 0.112 0.109 0.103 0.085 0.077 0.108
0.8-0.9 0.016 0.1 0.119 0.12 0.1 0.105 0.099 0.096 0.12 0.126
0.9-1.0 0.012 0.081 0.109 0.115 0.101 0.082 0.075 0.091 0.107 0.226

Table 6 Markov transition matrix for 0.40 < KTm,c ≤ 0.50

0.0-0.1 0.1-0.2 0.2-0.3 0.3-0.4 0.4-0.5 0.5-0.6 0.6-0.7 0.7-0.8 0.8-0.9 0.9-1.0
0.0-0.1 0.095 0.201 0.14 0.121 0.112 0.076 0.073 0.066 0.055 0.061
0.1-0.2 0.029 0.176 0.158 0.133 0.121 0.096 0.078 0.079 0.067 0.063
0.2-0.3 0.015 0.096 0.171 0.157 0.139 0.121 0.093 0.08 0.066 0.062
0.3-0.4 0.008 0.055 0.103 0.199 0.186 0.13 0.108 0.085 0.063 0.063
0.4-0.5 0.006 0.039 0.077 0.145 0.236 0.167 0.113 0.083 0.064 0.069
0.5-0.6 0.006 0.044 0.08 0.128 0.192 0.166 0.123 0.1 0.081 0.08
0.6-0.7 0.006 0.049 0.082 0.132 0.152 0.139 0.125 0.11 0.095 0.109
0.7-0.8 0.007 0.047 0.086 0.113 0.138 0.125 0.114 0.124 0.112 0.134
0.8-0.9 0.006 0.048 0.079 0.105 0.12 0.108 0.1 0.12 0.138 0.177
0.9-1.0 0.005 0.033 0.062 0.085 0.102 0.086 0.088 0.103 0.144 0.291

Table 7 Markov transition matrix for 0.50 < KTm,c ≤ 0.60

Alan Dunn 46 25/5/13

 0.0-0.1 0.1-0.2 0.2-0.3 0.3-0.4 0.4-0.5 0.5-0.6 0.6-0.7 0.7-0.8 0.8-0.9 0.9-1.0
0.0-0.1 0.061 0.169 0.146 0.095 0.106 0.094 0.108 0.085 0.067 0.07
0.1-0.2 0.023 0.113 0.13 0.114 0.107 0.111 0.102 0.108 0.1 0.092
0.2-0.3 0.007 0.062 0.105 0.132 0.151 0.126 0.113 0.106 0.097 0.1
0.3-0.4 0.004 0.026 0.063 0.15 0.189 0.147 0.118 0.108 0.097 0.099
0.4-0.5 0.002 0.017 0.04 0.098 0.23 0.164 0.13 0.111 0.103 0.106
0.5-0.6 0.002 0.016 0.04 0.084 0.162 0.179 0.149 0.129 0.119 0.12
0.6-0.7 0.003 0.018 0.04 0.079 0.142 0.143 0.153 0.14 0.139 0.144
0.7-0.8 0.002 0.017 0.041 0.079 0.126 0.12 0.135 0.151 0.162 0.167
0.8-0.9 0.002 0.017 0.034 0.069 0.108 0.106 0.114 0.144 0.191 0.215
0.9-1.0 0.001 0.012 0.023 0.05 0.083 0.079 0.088 0.118 0.185 0.362

Table 8 Markov transition matrix for 0.60 < KTm,c ≤ 0.70

0.0-0.1 0.1-0.2 0.2-0.3 0.3-0.4 0.4-0.5 0.5-0.6 0.6-0.7 0.7-0.8 0.8-0.9 0.9-1.0
0.0-0.1 0.049 0.091 0.112 0.07 0.098 0.077 0.105 0.119 0.112 0.168
0.1-0.2 0.019 0.07 0.09 0.105 0.119 0.113 0.103 0.134 0.121 0.125
0.2-0.3 0.005 0.028 0.074 0.114 0.13 0.123 0.113 0.118 0.145 0.151
0.3-0.4 0.001 0.011 0.039 0.102 0.169 0.135 0.123 0.126 0.136 0.156
0.4-0.5 0.001 0.007 0.021 0.062 0.175 0.143 0.132 0.137 0.157 0.167
0.5-0.6 0.001 0.007 0.02 0.049 0.117 0.146 0.15 0.157 0.172 0.182
0.6-0.7 0 0.005 0.015 0.047 0.097 0.122 0.151 0.169 0.197 0.197
0.7-0.8 0.001 0.006 0.016 0.04 0.084 0.098 0.13 0.179 0.224 0.223
0.8-0.9 0.001 0.005 0.011 0.034 0.067 0.079 0.107 0.161 0.262 0.275
0.9-1.0 0 0.003 0.007 0.022 0.045 0.055 0.074 0.112 0.222 0.459

Table 9 Markov transition matrix for 0.70 < KTm,c ≤ 0.80

Alan Dunn 47 25/5/13

 0.0-0.1 0.1-0.2 0.2-0.3 0.3-0.4 0.4-0.5 0.5-0.6 0.6-0.7 0.7-0.8 0.8-0.9 0.9-1.0
0.0-0.1 0 0 0.077 0.077 0.154 0.077 0.154 0.154 0.077 0.231
0.1-0.2 0 0.043 0.061 0.07 0.061 0.087 0.087 0.217 0.148 0.226
0.2-0.3 0 0.017 0.042 0.073 0.095 0.112 0.12 0.137 0.212 0.193
0.3-0.4 0.001 0.003 0.015 0.055 0.106 0.091 0.12 0.139 0.219 0.25
0.4-0.5 0 0.002 0.009 0.035 0.097 0.113 0.123 0.155 0.209 0.258
0.5-0.6 0 0.002 0.007 0.028 0.063 0.089 0.123 0.157 0.235 0.295
0.6-0.7 0 0.002 0.005 0.02 0.054 0.069 0.114 0.17 0.26 0.307
0.7-0.8 0 0.001 0.004 0.015 0.043 0.058 0.097 0.174 0.288 0.32
0.8-0.9 0 0.001 0.002 0.011 0.027 0.039 0.071 0.139 0.319 0.39
0.9-1.0 0 0.001 0.001 0.005 0.015 0.024 0.043 0.086 0.225 0.6

Table 10 Markov transition matrix for 0.80 < KTm,c ≤ 0.90

0.0-0.1 0.1-0.2 0.2-0.3 0.3-0.4 0.4-0.5 0.5-0.6 0.6-0.7 0.7-0.8 0.8-0.9 0.9-1.0
0.0-0.1 0.5 0.25 0.2 0.05 0 0 0 0 0 0
0.1-0.2 0.2 0.5 0.2 0.05 0.05 0 0 0 0 0
0.2-0.3 0 0 0.25 0 0 0 0.25 0.25 0 0.25
0.3-0.4 0 0 0 0 0.048 0 0.143 0.095 0.19 0.524
0.4-0.5 0 0 0.014 0 0.027 0.041 0.041 0.233 0.192 0.452
0.5-0.6 0 0 0 0.008 0.039 0.031 0.078 0.093 0.326 0.425
0.6-0.7 0 0 0 0.006 0.019 0.019 0.067 0.102 0.254 0.533
0.7-0.8 0 0 0 0.005 0.012 0.024 0.041 0.106 0.252 0.56
0.8-0.9 0 0 0 0.001 0.006 0.012 0.031 0.078 0.283 0.589
0.9-1.0 0 0 0 0.001 0.002 0.004 0.012 0.029 0.134 0.818

Table 11 Markov transition matrix for 0.90 < KTd,c ≤ 1.00

Alan Dunn 48 25/5/13

Alan Dunn 49 25/5/13
16.0 Appendix: Generating synthetic daily radiation data
This section gives an example of generating a sequence of synthetic daily clear sky global
radiation values, using the Markov Transition Matrices in Appendix 15.0. The example is
adapted from Section [Link] of (Meteonorm, 2013).

In order to generate synthetic daily data only the Linke Turbidity factor and the Monthly Global
Radiation, needs to be known.

The first step is to calculate the monthly clear sky radiation, from the Linke Turbidity Factor
as explained in Section 6.4. Using the known values of monthly global radiation the monthly,
the clear sky indices, for the location may be calculated from:

For example
Month January February November December
0.424 0.522 0.475 0.389

A simulation for the month of January would proceed as follows:

1. The appropriate MTM is selected using the value of for January (i.e. 0.424).
Since the appropriate matrix is Table 6 in Appendix 15.0. This
table will be used to generate all 31 days in January.

2. As there is no previous daily value to get started assume that the previous day (31
December) has a daily clearness index equal to the December monthly clearness
index i.e. 0.389.

3. This value falls in the interval 0.3-0.4 which is row 4 in Table 6

4. Using a random number generator with a uniform distribution between 0 and 1,

assume a number is found.

5. is used to determine the next state of (on 1 January) by summing the

probabilities in row 4 until the sum is greater than . That is, find so that

6. In this case as

0.0-0.1 0.1-0.2 0.2-0.3 0.3-0.4 0.4-0.5 0.5-0.6 0.6-0.7 0.7-0.8 0.8-0.9 0.9-1.0
0.0-0.1 0.116 0.223 0.196 0.129 0.093 0.077 0.054 0.044 0.032 0.037
0.1-0.2 0.051 0.228 0.199 0.143 0.101 0.083 0.065 0.052 0.035 0.043
0.2-0.3 0.028 0.146 0.244 0.156 0.12 0.092 0.069 0.053 0.04 0.052
0.3-0.4 0.02 0.111 0.175 0.208 0.146 0.104 0.074 0.067 0.044 0.052
0.4-0.5 0.017 0.115 0.161 0.177 0.155 0.102 0.085 0.067 0.054 0.068

Alan Dunn 50 25/5/13

 7. The next state of is given by the heading of column i.e 0.3-0.4. So the
clearness index changes from 0.389 on 31 December to 0.35 on the 1 January

8. To determine the clearness on 2 January the process is repeated from step 2 using
0.35 as the previous day value. This continues until a sequence of 31 daily values
have been generated.

9. Daily global radiation for each day is calculated by multiplying by the clear sky
global radiation, . (See Section 6.4)

10. The average value of the generated sequence may not be equal to the known
monthly mean global radiation. If it deviates by more than a defined tolerance (SoDa
uses 1%), the sequence is discarded and the process repeated until and acceptable
sequence is derived. The process then moves on to February.

In step 7, rather than just using the column heading as the next state, it is possible to
interpolate a more accurate value. This process is described in Section [Link] in
(Meteonorm, 2013).

17.0 Appendix: Generating synthetic hourly radiation data

This model generates synthetic daily sequences of hourly clearness index given the daily
clearness index, .

The following is an example application of the TAG model adapted from the appendix in (Aguiar,
Collares-Pereira, & Conde, 1992). We shall use data calculated from the clear sky model on the
accompanying spreadsheet Solar_Resource.xlsm.

Hour 07:30 08:30 09:30 10:30 11:30 12:30 13:30 14:30 15:30 16:30
Gh 148 393 580 694 749 749 694 580 393 148

For which the horizontal daily clear sky irradiation is 513 W

Procedure:

A. Define the allowed deviation between the observed daily clearness index and the
synthetic daily clearness index (calculated from the synthetic daily sequences of hourly
clearness index generated). A typical value would be .

B. For each day compute:

a. the sunrise/sunset hour angle,
b. The autocorrelation, from

( ) 17-1

Alan Dunn 51 25/5/13

C. For each solar hour, (1 to 24) compute
1. The solar hour angles for the start, centre and end of the hour
( ) ⁄
( ) ⁄
( ) ⁄

And the corresponding solar altitude angles, ( ), ( ) ( )

If or then the sun has either not yet risen, or it has set, so step to next
hour.

2. The hourly clear-sky clearness index, ( )

This may be calculated from the clear sky model presented in Section 6.4 or from the
simplified equation below:

( ) [ ( )⁄ ] 17-2

3. Calculate the average hourly clearness index for the day

( ) ⁄ 17-3
Where

( )

4. Calculate the standard deviation,

( ) ( ) 17-4
Where
( )

( )

5. Generate a random Gaussian number with a null average and a standard deviation

( ) 17-5

This may be done by picking a random number, from a uniform distribution [0, 1] and
applying the transformation:

[ ( ) ]⁄ 17-6

6. Calculate the normalised variable, and the synthetic hourly clearness index,

( ) ( )

Alan Dunn 52 25/5/13

 ( )

( ) is the value of at the current solar hour,

( ) is the value of at the previous solar hour,

( ) ( )
( )
( )

18.0 Appendix: Generation of monthly mean hourly data

The following VBA array function may be used to convert a column of 8760 hourly values into 12
mean monthly hourly profiles. Instructions for use are in the header comments.

Public Function MeanMonthlyHourlyLoad(Data As Variant, Leap As Variant)

' Converts a column of hourly values into 24 mean values for each month
' Select a range of cells 12 columns by 24 rows to display the results
' Insert this User Defined Function
' For Data select the range containing the 8760 hourly values
' Set Leap as 1 for a leap year otherwise 0
' Press <Control><Shift><Return>
'
' Alan Dunn 2013

Dim adResults(1 To 24, 1 To 12) As Double

Dim iMonth As Integer, iDay As Integer, iNdays As Integer
Dim iHour As Integer, iOffSet As Integer
Dim avDaysInMonth As Variant

' The number of days in each month

avDaysInMonth = Array(31, 28, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31)

'Set the starting count for each month

iOffSet = 0

For iMonth = 1 To 12
' Correct the number of days in February for a leap year
If iMonth = 2 Then
iNdays = avDaysInMonth(iMonth) + 1
Else
iNdays = avDaysInMonth(iMonth)
End If

For iDay = 1 To iNdays

For iHour = 1 To 24
adResults(iHour, iMonth) = adResults(iHour, iMonth) + Data(iHour +
(iDay - 1) * 24 + iOffSet) / iNdays
Next
Next
iOffSet = iOffSet + iNdays * 24
Next

MeanMonthlyHourlyLoad = adResults

End Function

Alan Dunn 53 25/5/13

The function works with only one year of data but may easily be adapted to work for multiple
years.

Alan Dunn 54 25/5/13

19.0 Appendix: Statistical Evaluation of Models

19.1. Regression
Many of the models discussed in these notes are based on regression

Observed value
̂ Predicted value
̅ Mean of observed values
Independent value
Coefficient of determination
Coefficient of correlation

Plot the deviation against the independent variable,

19.2. Coefficient of determination,

∑( ̂ ̅)
∑( ̅)

A value of means that 81% of the variation in has been explained by the model,
leaving 9% to be explained by other factors.

19.3. Coefficient of correlation,

19.4. Student’s t-distribution

( ) { ⁄√( )}
Where
Number of data points
( ) Degrees of freedom (d.f.)

Example
A correlation with 12 data points has a coefficient of determination of 0.64. What is the
significance of ?

( ) { ⁄√( )}

Looking up the t-value in a table of percentile values for Student’s t-distribution gives a value of
i.e. 99.8% confidence.

Alan Dunn 55 25/5/13

19.5. Mean bias error, MBE
∑( ̂ )

This gives a measure of the overall trend, i.e. whether the model over or under estimates.

19.6. Root mean square error, RMSE

√( ̂ ) ⁄

RMSE is a measure of absolute deviation in the units of .

19.7. Skewedness, S
Skewedness, is the measure of the asymmetry of a distribution and is given by:

⁄
⁄

Where and are the second and third moments of the distribution.

19.8. Kurtosis, K
Kurtosis, is a measure of the peakedness of a distribution and is given by:

Where and is the fourth moment of the distribution.

19.9. Outlier analysis

An outlier is normally considered to be a datum that lays three or four standard deviations from
the mean. Any outlier should be very carefully examined and only rejected if it can be shown
that they are the result of erroneous observations. They may in fact give insights to special
conditions for which the model does not account.

Alan Dunn 56 25/5/13

19.10. Student’s t-distribution

df α = 0.1 0.05 0.025 0.01 0.005 0.001 0.0005

∞ tα=1.282 1.645 1.960 2.326 2.576 3.091 3.291
1 3.078 6.314 12.706 31.821 63.656 318.289 636.578
2 1.886 2.920 4.303 6.965 9.925 22.328 31.600
3 1.638 2.353 3.182 4.541 5.841 10.214 12.924
4 1.533 2.132 2.776 3.747 4.604 7.173 8.610
5 1.476 2.015 2.571 3.365 4.032 5.894 6.869
6 1.440 1.943 2.447 3.143 3.707 5.208 5.959
7 1.415 1.895 2.365 2.998 3.499 4.785 5.408
8 1.397 1.860 2.306 2.896 3.355 4.501 5.041
9 1.383 1.833 2.262 2.821 3.250 4.297 4.781
10 1.372 1.812 2.228 2.764 3.169 4.144 4.587
11 1.363 1.796 2.201 2.718 3.106 4.025 4.437
12 1.356 1.782 2.179 2.681 3.055 3.930 4.318
13 1.350 1.771 2.160 2.650 3.012 3.852 4.221
14 1.345 1.761 2.145 2.624 2.977 3.787 4.140
15 1.341 1.753 2.131 2.602 2.947 3.733 4.073
16 1.337 1.746 2.120 2.583 2.921 3.686 4.015
17 1.333 1.740 2.110 2.567 2.898 3.646 3.965
18 1.330 1.734 2.101 2.552 2.878 3.610 3.922
19 1.328 1.729 2.093 2.539 2.861 3.579 3.883
20 1.325 1.725 2.086 2.528 2.845 3.552 3.850
21 1.323 1.721 2.080 2.518 2.831 3.527 3.819
22 1.321 1.717 2.074 2.508 2.819 3.505 3.792
23 1.319 1.714 2.069 2.500 2.807 3.485 3.768
24 1.318 1.711 2.064 2.492 2.797 3.467 3.745
25 1.316 1.708 2.060 2.485 2.787 3.450 3.725
26 1.315 1.706 2.056 2.479 2.779 3.435 3.707
27 1.314 1.703 2.052 2.473 2.771 3.421 3.689
28 1.313 1.701 2.048 2.467 2.763 3.408 3.674
29 1.311 1.699 2.045 2.462 2.756 3.396 3.660
30 1.310 1.697 2.042 2.457 2.750 3.385 3.646
60 1.296 1.671 2.000 2.390 2.660 3.232 3.460
120 1.289 1.658 1.980 2.358 2.617 3.160 3.373
∞ 1.282 1.645 1.960 2.326 2.576 3.091 3.291

Alan Dunn 57 25/5/13