Chapter -2
Cold and Heat Wave Indices and Methodology
There are no universal definition of heat and cold waves. Heat and Cold wave
events are anomalous situations with above (below) normal temperatures respectively.
The criteria used to define how much above/below normal could differ. India
Meteorological Department (IMD) uses criteria for heat and cold waves using station
temperature data. Definitions of heat and cold waves based on area average or gridded
temperature data are however different but very similar.
Heat waves have wide range of impacts on different sectors. Because of this, the
definition of heat waves will be different for different applications. Some studies
(Alexander et al. 2006, 2007; Fischer and Schar 2010; Fischer et al. 2011; Avila et al.
2012; Jiang et al. 2012; Perkin and Alexander 2013) used ETCCDI indices for heat wave
analysis and these indices are not good enough to heat stress related studies. The
indices like Predicted Mean Vote (PMV) and the Physiological Equivalent Temperature
(PET) are used for the health related studies (Matzarakis et al. 1999; McGregor et al.
2002; Pantavou et al. 2008). Another index named Heat Index (HI) also known as
apparent temperature (Steadman 1979, Steadman 1984) used for heat stress related
studies in many regions (Zahid and Rasul 2009; Rajib et al. 2011; Rakib 2013; Opitz-
Stapleton et al. 2016; Jaswal et al. 2017) as it combines air temperature and relative
humidity.
There are some similarities across these heat wave matrices, all definitions
include at least one form of temperature (maximum, minimum or average temperature)
(Perkin 2015). A recent study (Perkin et al. 2012) compared the various heat wave
characteristics obtained from 3 different indices (90th percentile for maximum
temperature, the 90th percentile for minimum temperature, and positive extreme heat
factor (EHF)) and it indicates that qualitative measurements across the indices are
17 | P a g e
similar but the quantitative values are different. This implies the importance of selecting
an appropriate index for the impacted region.
People residing in a place for sufficiently long time get more or less acclimatized
to the normal weather conditions of that place. Human body is quite sensitive to any
physiological change which takes place due to significant departure of weather
conditions from normal. These definitions are designed to take the physiological
changes also into account.
2.1 Heat wave Indices.
India Meteorological Department (IMD) considers only maximum temperatures
(Tmax) for defining heat waves. They define heat wave as " if the maximum
temperature of a station reaches at least 40°C or more for Plains, 37°C or more for
coastal station and at least 30°C or more for Hilly region ". The detailed IMD criteria of
heat waves are given below (Table 2.1).
Table 2.1
Criteria used for declaring Heat Wave by India Meteorological Department (IMD)
Nomenclature Departure from Normal Temperature
Criteria for Heat Wave/Severe Heat Wave
a) When normal maximum temperature of station is 40oC or less
Normal -10 C to 1o C
Above normal 2 0C
Appreciably Above Normal 3o C to 4o C
Markedly above normal/Moderate Heat 50 C to 6o C
Wave
Severe Heat Wave 7o C or above
b) When normal maximum temperature of station is more than 40oC (The term
moderate heat wave will not be used)
18 | P a g e
Normal -10 C to 1o C
Above normal 2 0C
Heat Wave 3o C to 4o C
Severe Heat Wave More than 50 C
c) When normal maximum temperature of station is 450C or more for two days
or more
The condition may be declared as Heat Wave
To declare a heat wave, the above criteria should be met at least in 2 stations in
the meteorological sub-division for at least two consecutive days.
For defining heat waves, there are more indices used in scientific literature,
especially with area averaged or gridded data. The first one is the 90th percentile
threshold of maximum temperature (Tmax90) based on a 5-day window (Rohini et al.
2016).
The second index considered is the Excessive Heat Factor (EHF) (Nairn and
Fawcett 2013; Perkins and Alexander 2013, Rohini et al. 2016). This index is based on
two excessive heat indices, namely Excess Heat (EHIsig) and Heat Stress (EHIaccl). The
unit of EHF is oC2.
The Excess heat represents unusually high heat arising from a daytime
temperature that is not sufficiently discharged overnight due to unusually high
overnight temperatures. The Daily Mean Temperature (DMT) averaged over a three-day
period is compared against a climate reference value to characterize this index. The unit
of EHIsig is oC.
The Excess Heat Index is calculated as
(Ti + Ti−1 + Ti−2 )
𝐸𝐻𝐼sig = − T95
3
where T95 is the 95th percentile of DMT (Ti) for the climate reference period of 1961-
19 | P a g e
1990. The daily mean temperature is the average of maximum and minimum
temperatures as defined by
(𝑇𝑚𝑎𝑥 + 𝑇𝑚𝑖𝑛 )
𝑇=
2
The heat stress which arises from a period where the temperature is warmer, on
average than the recent past. Maximum and subsequent minimum temperatures
averaged over a three-day period and the previous 30 days are compared to
characterize the heat stress. This is expressed as a short term (acclimatization)
temperature anomaly. The unit of EHIaccl is oC. The heat stress is defined as
(𝑇𝑖 + 𝑇𝑖−1 + 𝑇𝑖−2 ) (𝑇𝑖−3 +. . . +𝑇𝑖−32 )
𝐸𝐻𝐼𝑎𝑐𝑐𝑙 = −
3 30
where Ti is the DMT on ith day.
The Excess Heat Factor (EHF) is defined as follows:
𝐸𝐻𝐹 = 𝐸𝐻𝐹𝑠𝑖𝑔 × max(1, 𝐸𝐻𝐼𝑎𝑐𝑐𝑙 )
A heat wave is considered when the value of EHF is positive, and the daily
climatological Tmax is greater than 35°C for consecutive three days or more. These
criteria were selected carefully to choose the heat wave events occurring over the
Indian sub-continent. The 35 °C criterion was used to restrict such events occurring over
the hilly regions.
The significant associations between high temperature and mortality and morbidity
have been studied in many parts of the world (Heo et al. 2019). Therefore,
understanding associations between heat waves and various health outcomes is
essential to prevent the current and future burden of heat impacts on human health. In
many countries, the Heat-Health Warning Systems (HHWS) adopted thermal comfort
indices, which quantify a combined effect of a series of meteorological factors (air
temperature, humidity and wind) on perceived temperature, to represent the actual
human thermal situation during heat waves. The effectiveness of a HHWS using a
20 | P a g e
thermal index (including ambient temperature) depends on how properly the heat wave
definition can relate heat wave periods and related health problems, and choice of
which thermal index to use is a critical public health issue for policy (Heo et al. 2019).
There are such indices to define and monitor heat waves in the context of its health
impacts. Some of such indices are Heat Index (HI) and Wet Bulb Global Temperature
(WBGT) (Heo et al. 2019). HI attempts to represent the human perceived temperatures
by the impact of humidity. HI is an index that combines air temperature and relative
humidity to determine a human-perceived apparent temperature to determine how hot
it actually feels. It is a measure of the stress placed on humans by elevated levels of
atmospheric temperature and moisture. It implies that an extended period of unusually
high heat stress may have adverse health consequences for the affected population. The
computation of HI is based upon the formula given by Steadman (1979) where the
author has quantified the physiological effects of high heat and humidity on human
being. The variables used in the formula include heat generation and loss, fabric
resistance, vapour pressure, wind speed, solar radiation, terrestrial radiation,
proportion of body clothed and other factors (Steadman, 1984). The Steadman’s
equation was further modified by Rothfusz (1990) by performing multiple regression
analysis on the data from Steadman’s table. Jaswal et al. (2017) used this index to study
long term trends using IMD station data. More details of the index are available in
Jaswal et al. (2017).
The equation used by the National Weather Service of the National Oceanographic
and Atmospheric Administration (NOAA) is given below.
HI = -42.379 + 2.04901523*T + 10.14333127 *R - 0.22475541 *T*R - 6.83783x10-3*T2
-5.481717x10-2 * R2 + 1.22874 x 10-3 * T2 * R + 8.5282x10-4 * T *R2 - 1.99x10-6 * T2 *
R2
where, T= Ambient dry bulb temperature (0F) and
R= Relative Humidity (Integer percentage).
21 | P a g e
Wet-bulb globe temperature (WBGT) is one of the most commonly
used indices by many organizations since the 1960s. This index was originally invented in
the 1950s in efforts to lower the risk of heat disorders during the training of the US
Army and Marine troops. Since that time, WBGT has been applied in other settings, and
is widely used for the evaluation of occupational heat stress exposure (Budd 2008;
Alfano et al., 2014). WBGT is a weighted average of dry bulb (air) temperature, natural
wet-bulb temperature and black globe temperature. Black globe temperature is a
function of radiant heat, temperature, and wind while natural wet blub temperature
measures the amount of cooling by humidity and wind (Budd, 2008). By incorporating
black globe temperature, WBGT considers radiation effect whereas many other
simplified thermal comfort indices do not. In sunny conditions, the weighting
coefficients are 0.7, 0.2, and 0.1 for natural wet temperature, globe temperature, and
dry-bulb temperature, respectively. At other times, the weighting coefficients are 0.7
and 0.3 for natural wet temperature and dry-bulb temperature, respectively, while the
globe temperature is not considered in the calculation. Until recently, several
approximation formulas of WBGT using readily available meteorological data (e.g.,
temperature, humidity, wind, and radiation) have been suggested for practical use as
the original components for calculating WBGT are not standard meteorological
monitoring data. Further details of WBGT are available in Heo et al. (2019).
The health risk assessment is subject to the designated heat wave periods, and
more extreme heat wave definitions with the higher thresholds generally identify fewer
heat waves of shorter duration and higher intensity. Thus, the definitions with higher
thresholds, which reflect the most intense heat waves but also lead to the shortest heat
wave periods and fewer designated heat waves, may not necessarily result in the
strongest or the most significant health effect estimate.
Effectiveness of a HHWS to prevent health risks from heat exposure is subject to
the precision of weather forecast. Temperature tends to be more accurately forecast
22 | P a g e
than other weather components, which may hinder the use of thermal indices that use
other weather components for heat wave warnings. Currently, historical forecast data
are available through an open data source for everywhere for components including
temperature, humidity and wind speed.
Russo et al. (2014) introduced another index for heat wave activity, namely the
Heat Wave Magnitude Index (HWMI). The computation of HWMI involves multiple
stages that begin with the computation of daily threshold (90 th percentile) with
reference to the baseline period (Das and Umamahesh, 2022). Next, the selection of
heat wave events is carried out according to the minimum three consecutive days Tmax
above the threshold. The selected heat wave is then grouped into sub-heat waves,
where each sub-heat wave is a heat wave of three consecutive days. More details of this
index are available in Russo et al. (2014).
2.2 Cold Wave Indices
Based on station temperature data, India Meteorological Department (IMD) uses
the following definition of cold waves over India, as shown in Table-2.2
Table 2.2
Criteria for Cold Wave/Severe Cold Wave
Nomenclature Departure from Normal Temperature
Criteria for Cold Wave/Severe Cold Wave
a) When normal minimum temperature of station is 10oC or more
Normal -10 C to 1o C
Below normal -20C
Appreciably Below Normal -3o C to -4o C
Markedly Below normal/Moderate Cold -50 C to -6o C
Wave
Severe Cold Wave -7o C or above
b) When normal maximum temperature of station is less than 100C (The term
23 | P a g e
moderate cold wave will not be used)
Normal -10 C to 1o C
Below normal -20C
Cold Wave -3o C to -4o C
Severe Cold Wave -5o C or less
Ratnam et al. (2016 b) considered cold wave events over India when the
normalized area average Tmin anomalies are less than one standard deviation for four
or more days. Some researchers used the 10th percentile criteria for defining cold waves.
2.3 Temperature data
For studying heat and cold waves over India, researchers generally make use of
IMD station data or IMD Gridded daily temperature data set. The IMD gridded
temperature data set was prepared indigenously by IMD researchers in 2009 (Srivastava
et al. 2009). The details of the IMD gridded temperature data set are given below.
IMD maintains around 550 surface observatories in the country, where daily
surface air temperature observations (maximum and minimum) are taken. These
data are compiled, digitized, quality controlled and archived at the National Data
Centre (NDC) of IMD. However, daily data from 1969 onwards only are digitized and
archived at NDC. Later this data set was updated with the daily temperature data
from 1961. From the list of stations for which daily data are available, only those
stations which have minimum 10 years of data, at least for 300 days in a year during
the period 1971–2000, were selected for further analysis. The data were subjected
to basic quality checks like rejecting values, greater than exceeding known extreme
values, minimum temperature greater than maximum temperature, same
temperature values for many consecutive days etc. Unusual high values were flagged
24 | P a g e
by putting a filter which allowed values only in the range mean±(1.76+0.8 N)
standard deviation for further data analysis.
The interpolation was used a modified version of the Shepard’s angular
distance weighting algorithm for the present analysis. In order to avoid biases in the
gridding, daily anomalies were used instead of the absolute values. For this purpose,
climatological normal of maximum and minimum temperatures for the period 1971–
2000 was calculated for each station. The daily anomalies were calculated as the
difference of each daily temperature from its daily normal values. The present
interpolation method requires an under-standing of the spatial correlation structure
of the station data. Therefore, the interstation correlations were calculated to
determine the distances over which observed temperature anomalies are related.
For each month, for each pair of 395 stations lying within 2000 km, correlation was
calculated and then binned according to their separation over intervals of 100 km.
The mean correlation was estimated over each 100-km interval and a two-degree
polynomial function was fitted to these values. For interpolating the station data, we
have to define the radius of influence. This was estimated as the correlation length
scale (CLS), which is defined as the distance at which the mean correlation,
represented by the fitted function, fell below 1/e.
More details of this data set are available in Srivastava et al. (2009). This data
set is being used very extensively by researchers worldwide.
25 | P a g e