Journal of Environmental Management 257 (2020) 110006

Contents lists available at ScienceDirect

Journal of Environmental Management

journal homepage: http://www.elsevier.com/locate/jenvman

Research article

Anthropogenic forcing exacerbating the urban heat islands in India

Sarath Raj a, *, Saikat Kumar Paul b, Arun Chakraborty a, Jayanarayanan Kuttippurath a
a
CORAL, Indian Institute of Technology Kharagpur, 721302, India
b
Department of Architecture and Regional Planning, Indian Institute of Technology Kharagpur, 721302, India

A R T I C L E I N F O A B S T R A C T

Keywords: Urban heat island (UHI) phenomena is among the major consequences of the alteration of earth’s surface due to
Anthropogenic forcing human activities. The relatively warmer temperatures in urban areas compared to suburban areas (i.e. UHI) has
Climate change potential health hazards, such as mortality due to high temperatures and heat waves. In addition, UHI situation
Land surface temperature
demands more energy (e.g. fans and air-conditioners) that would trigger greenhouse gas emissions. Studies on
MODIS
UHI intensity help to assess its impact on urban population, city planning, and urban health planning. This is
particularly important for a country like India, where 32% people (~7% of total world population), live in urban
areas. We conducted a detailed study on surface UHI intensity (SUHII), which is the difference between urban
and surrounding rural land surface temperatures, across all seasons in 44 major cities of India, which shows that
mean daytime SUHII is positive (up to 2 � C) for most cities, as analysed from satellite temperature measurements
for the period 2000–2017, in contrast to previous studies. However, although statistically insignificant, most
cities show a positive trend in SUHII for monsoon and post-monsoon periods, but negative for winter and summer
seasons. The increasing night-time SUHII in all seasons for most cities suggest increasing trend in temperature in
cities due to the impact of the rapid urbanisation, and thus, suggesting the influence of anthropogenic forcing on
SUHII. This is also supported by the analysis of aerosols, night lights, precipitation and vegetation in the study
regions. Therefore, this study shall aid planning and management of urban areas by giving insights about the
effects of nature and intensity of development, land cover and land use mix and the structure of cities on SUHII.

1. Introduction Several studies have been performed to examine SUHII of different

cities around the world (Clinton and Gong, 2013; Imhoff et al., 2010;
More than half of the population in the world lives in urban areas. Jin, 2012; Peng et al., 2012; Zhang et al., 2010). For instance, an analysis
Due to the rapid urbanisation and growth of the population, about 2.5 for more than 400 big cities in the world reported an annual daytime
billion people will be added to the urban areas by 2050, with India, SUHII of 1.5 � C and night-time SUHII of 1.1 � C (Peng et al., 2012). A
China and Nigeria together accounting for about 35% of the urban positive correlation between precipitation and population with annual
population between 2018 and 2050 (UN, 2018). The process of rapid mean day and night UHI, respectively, was found in a study based on 65
urban development leads to elevated temperatures in urban regions cities in the United States of America and Canada (Zhao et al., 2014).
compared to their neighboring suburban or rural areas. This phenome­ The UHI analysis for the 5000 largest cities in Europe showed that SUHII
non is called the urban heat island (UHI) (Oke, 1973). Developments in increases with size of the cities and their fractal dimension, but di­
the field of thermal infrared (TIR) remote sensing have significantly minishes with the logarithm of the anisometry (Zhou et al., 2017). The
improved the study of surface UHI based on the land surface tempera­ seasonal and diurnal variation of SUHII of Asian megacities showed
ture (LST), since these datasets provide better spatial coverage, covering significant positive SUHII, in both magnitude and extent to which the
the entire urban area at the same time, compared to in-situ data which surface UHI was present in 2001 (Tran et al., 2006). Also, a study of 32
are point measurements (Cai et al., 2011; Rigo et al., 2006; Stewart and cities in China showed an annual mean SUHII up to 1.85–1.95 � C (Zhou
Oke, 2012; Voogt and Oke, 2003). Surface urban heat island intensity et al., 2014). A similar study of various cities across the world has
(SUHII) is defined as the difference between the LST of urban and its showed the fact that the urban areas are warmer compared to their
surrounding non-urban area. neighboring rural areas and the intensity of SUHII varies with their

* Corresponding author.
E-mail address: sarath.rajk@iitkgp.ac.in (S. Raj).

https://doi.org/10.1016/j.jenvman.2019.110006
Received 7 August 2019; Received in revised form 29 October 2019; Accepted 16 December 2019
Available online 24 December 2019
0301-4797/© 2019 Elsevier Ltd. All rights reserved.
S. Raj et al. Journal of Environmental Management 257 (2020) 110006

Fig. 1. Major cities of India. Geographical location of the cities selected for this study and digital elevation map of India. The red dots represent the cities classified
as X category, which has a population greater than 5 million and blue dots represent cities classified as Y category that has population less than 5 million. The black
lines represent the Koppen climate regions and the numbers signify the corresponding Koppen climate class. (For interpretation of the references to colour in this
figure legend, the reader is referred to the Web version of this article.)

geographical location, size, population, and vegetation cover (Clinton adjacent rural region for the period 2003–2013 (Kumar et al., 2017;
and Gong, 2013). Doick et al. (2014) and dos Santos et al. (2017) Shastri et al., 2017, 2015). There are also studies of UHI of individual
showed that green cover in urban areas plays a significant role in UHI cities. For example, a study of Delhi for years 2000 and 2010 reported
mitigation. A study of European cities showed that the cities in colder that the increase in urban infrastructure has direct effect on LST and heat
climatic region and cities with higher green cover are more prone to heat fluxes including anthropogenic heat flux (Chakraborty et al., 2015).
waves (Ward et al., 2016). A recent study also investigated the rela­ Average annual SUHII of Chandigarh was shown to be about 4.98–5.43
tionship between surface UHI and land cover types, and concluded that �
C and overall average SUHII has been observed to be 5.2 � C (Mathew
the SUHII was significantly modified by the built-up and vegetation land et al., 2016). The UHI analyses conducted for Jaipur during the period
cover classes (Yang et al., 2017). The intensity and spatial extent of 2003–2015 showed that significant surface UHI exists there with an
surface UHI in China is increasing over the years and this trend will average intensity of 7.86 � C (Mathew et al., 2017). A negative daytime
continue if mitigation strategies are not put in place(Peng et al., 2018; SUHII was observed in Jaipur and a very weak day time SUHII was
Yang et al., 2019; Yao et al., 2019, 2018, 2017). A study of South observed in Ahmedabad, while strong night-time SUHII was observed
American cities concluded the existence of significant positive SUHII over both cities during the period 2003–2015 (Mathew et al., 2018).
both during day and night and also the influence of land cover on SUHII Although there are UHI analyses for single (Chakraborty et al., 2015;
(Wu et al., 2019). Mathew et al., 2018, 2017, 2016) or multiple cities of India together
India has witnessed an increasing trend in urbanisation since its in­ (Kumar et al., 2017; Shastri et al., 2017, 2015), those discussions were
dependence in 1947, and the economic liberalization after the 1990s has mostly done for the connection between SUHII and vegetation or urban
further amplified this trend. According to the 2011 census, 31.8% of the precipitation. However, many factors influence SUHII and they need to
Indian population lives in urban areas, a 4% increase since the year be analysed together for a better understanding of urban climate in the
2001. Furthermore, India is expected to add another 416 million in­ context of global warming induced by anthropogenic forcing (e.g.
habitants to its cities by 2050 and leads the list ahead of China and aerosols, intensity of anthropogenic activities). Therefore, a compre­
Nigeria (UN, 2018). Subsequent infrastructure development is necessary hensive analysis of 44 cities of India with a population more than 1
to meet the needs of growing population in urban areas, which usually million for all seasons is performed. The spatial, diurnal as well as sea­
comes at the expense of conversion of vegetated areas and water bodies sonal variations, trends in SUHII for the 2000–2017 period, and drivers
in the peri-urban areas to urbanised regions having impervious surfaces. of the spatial and temporal variability in SUHII are also discussed.
A recent study on 84 cities in India found negative SUHII during
summer days and was attributed to the low vegetation cover in the

2
S. Raj et al. Journal of Environmental Management 257 (2020) 110006

Fig. 2. Delineation of urban and rural boundaries for SUHII estimates. Annual mean land surface temperature of Delhi during daytime (a) and during night–time
(b) in 2017. Also, the delineation of urban and rural areas is also depicted in the figure. Black line marks the urban boundary and the buffer zone around it considered
as the rural region.

2. Data and methods i) Enhanced vegetation index (EVI), which is a measure of vegeta­
tion, is derived from the MOD13A2 data product of the MODIS
2.1. Datasets sensors aboard Terra satellite (K. Didan, 2015). The data are
available every 16 days at 1 km spatial resolution as a gridded
Forty-four major cities across India are selected for this study. The Level-3 product.
cities are selected such that they fall into X and Y category cities as ii) The MODIS Terra and Aqua combined MCD19A2 Version 6 data
classified by the Government of India. X category cities have population product that provides land aerosol optical depth (AOD) gridded
greater than five million and Y category cities have population between Level-2 product at 1 km pixel resolution is used to study the effect
five lakhs and five million. Furthermore, Y category cities which have of aerosols on SUHII (Lyapustin and AuthorAnonymous, 2018).
population more than one million are only considered in this study. iii) Night-time lights (NL) serves as a good proxy for measuring the
Small towns with population less than a million are excluded from this anthropogenic activities in urban areas (Amaral et al., 2005;
study, since the delineation of urban and rural areas is difficult in such Elvidge et al., 2001). Satellite derived nightlights data are ob­
urban areas, which would corrupt the SUHII analyses for big cities when tained from Defense Meteorological Satellite Program’s Opera­
averaged over the Indian or regional domain; leading to misinterpreta­ tional Linescan System (DMSP/OLS).
tion of the attribution of SUHII. Due to its vast size, the climate in India iv) The Tropical Rainfall Measuring Mission (TRMM) 3B43 monthly
varies largely from region to region. India experience climate from four precipitation data at 0.25� resolution are used to characterise the
major Ko €ppen climate groups and the cities belong to different Ko €ppen effect of rainfall on SUHII (Precipitation Processing System (PPS)
climatic regions (Rubel and Kottek, 2010) (Fig. 1). LST was derived from At NASA GSFC, 2018).
the eight-day composite 1 km LST MODIS Terra (MOD11A2) V6 level-3
land 2017 (Wan et al., 2015). This dataset provides an average eight-day
per-pixel LST. The MODIS bands 31 and 32 are used to derive LST data 2.2. Analysis
by using the generalized split-window algorithm, that corrects both at­
mospheric effects and surface emissivity. The MODIS V6 LST data are SUHII is defined as the difference in mean LST between the urban
more accurate than previous versions and have errors within �1 � C areas and surrounding rural regions (Ward et al., 2016; Zhou et al.,
(Wan, 2014). Each pixel value in the MOD11A2 is an average of all the 2014). The urban areas are identified by applying the city clustering
corresponding daily MOD11A1 LST pixels collected within that algorithm on MODIS MCD12Q1 annual international
eight-day period, which includes both day and night LST and the geosphere-biosphere programme (IGBP) classification scheme dataset
eight-day image composite removes the effects of cloud cover. The Terra that provides yearly global land cover types from 2001 to 2016 (Sul­
satellite acquires data for India during day and night at around 11 a.m. la-Menashe and Friedl, 2015). The rural areas are defined as a buffer
(Indian Standard Time; IST) and 11 p.m. (IST), respectively. Only those area of all non-urban pixels around the city with an area approximately
data were selected in which at least 50% of the pixels were available equal to the area of the city (Peng et al., 2012) (Fig. 2).
without no data points. This will ensure the homogeneity of the analysis Trends are estimated using the annual aggregated time-series
during cloud covered monsoon season. To remove the water pixels from method and slope of the trend is estimated by linear least-squares
the study area, the Terra MODIS Land/Water Mask (MOD44W) Version regression of the annual means, and their significance at 0.05 level is
6 data product, a global land/water mask derived from Terra MODIS is estimated using the Mann-Kendall test (Forkel et al., 2013).
used. MODIS MCD12Q1 dataset that provides global land cover maps at The LST derived from TIR measurements is controlled by radiative
annual time steps is used to delineate urban boundaries. and thermodynamic properties of the surface. The exchange of energy
The dependence of SUHII with its probable drivers are also analysed between surface and atmosphere can be expressed as: Rn ¼ H þ LE þ ΔS,
using satellite derived measurements and following datasets that are where Rn is the net radiation, H is the sensible heat flux, LE is the latent
collected for the period 2000–2017 are used for this purpose: heat flux and ΔS is the surface heat flux (Arnfield, 2003; Voogt and Oke,
2003). With respect to surface UHI, the dominant fluxes in the energy
balance equation during daytime are latent and sensible heat fluxes,
while the storage surface heat fluxes become dominant during night.

3
S. Raj et al. Journal of Environmental Management 257 (2020) 110006

Fig. 3. The urban heat islands of India. The spatial distribution of surface urban heat island intensity (SUHII) for daytime and night-time across different seasons in
44 cities of India, including night-time SUHII of summer, winter, post-monsoon, monsoon and the annual average represented in a, b, c, d, and e respectively and the
daytime counterparts in f, g, h, i, and j respectively.

4
S. Raj et al. Journal of Environmental Management 257 (2020) 110006

Fig. 4. The green cover in Indian cities relative to their rural proximity. The estimated difference in enhanced vegetation index (ΔEVI) of the 44 cities selected
for the study. The bars are stacked and each colour bar represents the value of ΔEVI for the corresponding season. The ΔEVI values of summer, winter, post-monsoon
and monsoon seasons are represented by red, blue, purple and green bars respectively. The grey bar represents the values of annual ΔEVI. (For interpretation of the
references to colour in this figure legend, the reader is referred to the Web version of this article.)

These fluxes are also modified by surface types such as vegetation, the study period will not demonstrate the exact situation of the temporal
impervious surfaces and water. The EVI, which is less influenced by evolution of SUHII and, hence the time series of SUHII in all seasons are
atmospheric conditions, is used to measure the vegetated areas. The also analysed (Fig. S1).
ΔEVI, which is the difference between EVI in the urban and suburban The monsoon and post-monsoon seasons show similar temporal
region is taken as a measure to quantify vegetation. NL data can capture evolution of mean SUHII as most cities exhibit positive SUHII from
the development activities in the urban areas and can be considered as a 0.75 � C to 3.7 � C (Fig. S1). Exceptions to these are the gradual increase
measure of anthropogenic activities in this city/urban area (Peng et al., of SUHII in Jodhpur from 0.62� to 2.12 � C, and the heavy rains and
2012; Zhou et al., 2014). Therefore, ΔNL, which is the difference be­ associated urban cool island in Varanasi in 2010 during monsoon
tween NL data in the urban and suburban region is used as a proxy for ( 2.51 � C). Agra, Delhi, Jaipur and, Jodhpur show negative SUHII for
analysing the effect of anthropogenic activities. post-monsoon days. The night-time SUHII is positive for all seasons,
from 0.33 � C to 2.35 � C, in all years, indicating that the urban surfaces
3. Results emit more radiation than the surrounding rural regions during night.
However, there are few exceptions such as 2016 for Srinagar in summer
3.1. The average SUHII in cities ( 0.7 � C), 2001 for Ahmedabad in monsoon ( 2.46 � C), and 2004 for
Mumbai in post-monsoon seasons ( 0.25 � C).
The average SUHII of 44 Indian big cities are estimated and is The daytime SUHII is highest for Guwahati, Chennai and Ludhiana
depicted in Fig. 1. The cities are selected such that they represent at least (about 2 � C) and lowest for Pune (about 1 � C) for all seasons. This in­
1 million people with significant built up area. In general, the daytime dicates that the SUHII is also related to the local weather/climate of the
SUHII estimated show positive values in all seasons for most cities which region, as the change in SUHII is analogous to the ambient temperature
implies that the urban areas are warmer than its surroundings. Out of 44 of the region (Zhao et al., 2014). However, the highest night-time SUHII
cities 20 cities show negative mean SUHII in summer and 9,8, and 5 (around 2.5 � C) is found for Ahmedabad and Ludhiana, and the lowest
cities in winter, monsoon and post-monsoon seasons, respectively. The for Srinagar and Guwahati (0.3 � C), except during monsoon. The largest
mean SUHII varies between 0.9 � C and 1.5 � C in summer, 1.6 � C and inter-annual variability is found in monsoon and the smallest in
2.6 � C in winter, 0.8 � C and 2.0 � C in monsoon seasons and 0.6 � C and post-monsoon season. The large year-to-year SUHII change during
1.7 � C in post monsoon season. However, six cities show negative SUHII monsoon leave no clear pattern for SUHII, which is directly connected to
in summer and winter, in which Jodhpur and Pune are common for all heavy or intermittent rainfall and evaporation during monsoon season.
seasons. Aurangabad, Gwalior and Jodhpur are the other cities that The results, therefore, show that about 85% of the cities have positive
show mean negative SUHII for all seasons, except in monsoon (Fig. 3). SUHII, with more or less similar values for each year, except for
This is an expected result as the materials that constitute the urban built- monsoon. This implies that the cities expand to their suburban areas and
up gets heated up more than surrounding rural areas. However, it is the temperature also increases there in tune with the changes in the
noteworthy that calculating the mean SUHII has its inherent limitations adjacent cities. The rapid urbanisation of the type-Y (tier 2) cities in
as it can be weighted by values of individual year or an anomalous SUHII India is clearly identified in the SUHII analysis.
occurrence in any year or season. Therefore, the SUHII averaged over The night-time SUHII is positive for all seasons, indicating that the

5
S. Raj et al. Journal of Environmental Management 257 (2020) 110006

Fig. 5. Anthropogenic contributions to the heat islands. a) Dependence of daytime SUHII on ΔEVI and b) night-time SUHII on ΔNL. The left panel depicts the
correlation of the daytime SUHII and ΔEVI and the right panel shows the correlation night-time SUHII and ΔNL, during different seasons. Corresponding r values are
also specified in each panel.

night-time temperature is higher in all cities in all seasons and is in show 0.74 � C/0.87 � C and 1.24 � C/1.76 � C for winter/summer day and
agreement with the previous studies (Kumar et al., 2017; Shastri et al., night, respectively. This difference in SUHII estimates can be due to the
2017). This consistent increase in night-time temperature suggests slow difference in time and datasets used for the study.
warming of urban and rural areas, and indicates the impact of global
warming across the cities. This effect is also visible in daily temperature 4. Discussion
measurements, and is more apparent in big cities (Fig. S2, where
built-up and intensity of development as well as the population density The SUHII exhibit different patterns for cities in the tropical, arid and
is significantly more). temperate climate regions. The EVI, which is less influenced by atmo­
Interestingly, our results are slightly different or even opposite to the spheric conditions, is used to measure the vegetated areas. The ΔEVI,
conclusions drawn from previous studies (Kumar et al., 2017; Shastri which is the difference between EVI in the urban and suburban region is
et al., 2017), where they found negative SUHII in most Indian cities taken as a measure to quantify vegetation.
(mostly daytime SUHII) attributed to continuous decline in vegetation in The large vegetated areas in suburban regions lead to higher
rural areas of the corresponding. These results warranted a more evapotranspiration during day, which in turn leads to larger latent heat
detailed examination of SUHII in big Indian cities, and revealed that the flux. This will cause an enhanced cooling effect on surface in suburban
SUHII average values were largely weighted by those of the small towns areas during day, as found for the higher SUHII in Kolkata during day,
(42 cities in total), where city boundaries are hard to delineate and they where a ΔEVI of 0.08 is observed (Fig. 4). In an arid area, the vege­
house a population between 0.1 and 1 million. Nevertheless, the large tation cover in both suburban and urban areas is smaller compared to a
cities with significant nightlights and distinct rural border still show the tropical city, and thus the effects of sensible heat flux dominate during
positive SUHII (e.g. Kolkata), consistent with the results of this study. It day. Therefore, variability in skin temperatures of suburban and urban
is further attested by the SUHII estimated for the cities in both studies parts of these cities is smaller and that explains the reduced daytime
with population of more than 5 million (e.g. Hyderabad and Coimba­ SUHII. For example, Jaipur, located in a semi-arid area, has an annual
tore). Also, the cities that show contrasting results are also due to the average ΔEVI of 0.03 as the city is in the proximity of the Thar Desert.
difference in data analysis period (e.g. Lucknow) and in population (e.g. Conversely, the night-time temperature is controlled by surface heat
Dhanbad). Therefore, selection of cities and interpretation of SUHII fluxes. The urban built-up store more heat than the sparsely vegetated
must be done carefully for making conclusions on SUHII intensities, as suburban areas of arid regions and thus shows higher SUHII at night.
addition of a year with more than average SUHII would decide the This is also reflected in the correlation between SUHII and EVI, because
long-term average of SUHII in that city. it shows statistically significant correlation for summer and winter
The results, nevertheless, for Jaipur and Ahmedabad are in agree­ SUHII during day (r ¼ 0.60, 0.41, respectively Fig. 5).
ment with the results reported in an earlier study conducted for the A similar result was also found in a study conducted for 32 major
period 2000–2013 using MODIS data, although the values of SUHII are cities of China (Zhou et al., 2014). Correlation between day and
slightly different due to the difference in analysis period (Mathew et al., night-time SUHII is, however, very weak and statistically insignificant
2018). A study conducted for Chandigarh using the MODIS data for the for all seasons. Since the interannual variation of UHI and EVI is very
2000–2013 period show an average SUHII of 5� –6 � C for summer and large during monsoon, there is no correlation between SUHII and EVI.
winter seasons (Mathew et al., 2016) whereas results from this study This implies that EVI alone cannot explain the variability and spatial

6
S. Raj et al. Journal of Environmental Management 257 (2020) 110006

Fig. 6. The nexus between urban heat islands and vegetation, atmo­
spheric pollution, climate and socio-economic factors. Correlation of SUHII
with ΔEVI, ΔNL, ΔAOD, mean precipitation (P), mean air temperature (T),
mean aerosol optical depth (AOD) and population. a) Correlation of annual
SUHII with ΔEVI, ΔNL, ΔAOD, mean precipitation (P), mean air temperature
(T), mean aerosol optical depth (AOD) and population. Red line and blue line
represent daytime and nigh-time SUHII, respectively. b) and c) represent cor­
relation of daytime and night-time SUHII with ΔEVI, ΔNL, ΔAOD, mean pre­
cipitation (P), mean air temperature (T), mean aerosol optical depth (AOD).
The summer, winter, post-monsoon and monsoon seasons are represented by
red, blue, green and orange lines respectively. The symbols y, z, $, &, ¥, and ^
represent statistically significant values during day, night, summer, winter,
post-monsoon and winter seasons, respectively. (For interpretation of the ref­
erences to colour in this figure legend, the reader is referred to the Web version
of this article.)

inhomogeneity of SUHII. Therefore, the analysis of the night lights,

aerosol loading, precipitation in the city and surrounding rural areas has
also been carried out to further examine development of cities in terms
of their built infrastructure and population.
The difference in nightlights between urban and suburban areas
(ΔNL) is used to examine the effect of anthropogenic activities in
nighttime SUHII (Amaral et al., 2005; Elvidge et al., 2001; Letu et al.,
2010). Larger values of ΔNL indicate higher anthropogenic activities in
urban areas, which in turn can lead to higher SUHII. Kolkata having a
ΔNL of 27.7 has higher night-time SUHII during summer and winter
(1.05 � C and 1.5 � C, respectively). However, Dhanbad, with a low ΔNL
of 7.9, has a smaller night-time SUHII in both seasons (0.56 � C and 0.71
�
C for summer and winter, respectively). The results show a statistically
significant positive correlation between ΔNL and SUHII in the summer
and winter seasons (Fig. 6b and c).
Another indicator of the anthropogenic forcing is the aerosol loading
which is analysed using AOD data available from MODIS (Cao et al.,
2016). The cities in the Indo-Gangetic plain exhibit the highest values of
AOD (>0.45) across all seasons with larger AOD during the
post-monsoon season (>0.5). The ΔAOD, which is the difference be­
tween the mean AOD’s of the urban area and rural area is also analysed.
The ΔAOD is negative for most cities (36 out of 44) during summer. In
addition, 16 out of these 36 cities show negative daytime SUHII during
this season and most exhibit smaller SUHII (<1 � C). However, Guwahati,
located in the northeastern region of India, despite showing a high
negative ( 0.054) ΔAOD, experience a significantly higher SUHII (2.32
�
C) during summer seasons. The ΔAOD was positive for most cities (32
out of 44) during the post-monsoon season, implying a higher aerosol
loading over the urban areas during this season. The SUHII was also
positive for most of the cities (35 out of 44) during the same season,
although Guwahati exhibits higher SUHII (2.12 � C) albeit having ΔAOD
of 0.014.
The precipitation data available for the study period was also ana­
lysed. The difference between urban and rural rainfall could not be
estimated since the data available had a coarse resolution to carry out
such an analysis, hence, the monthly rainfall averaged over the rural and
urban areas were analysed. The mean precipitation showed a positive
correlation with summer and winter (r ¼ 0.71 and 0.52, respectively)
SUHII during daytime and negative correlation ( 0.52) with night-time
winter SUHII (Fig. 6 b and c).
(caption on next column)

4.1. Temporal evolution of SUHII

The day and night SUHII trends for all seasons are estimated and the
results are listed in Table 1. In general, the daytime summer and winter
SUHII show negative trends for most cities; about 23 (summer) and 24
(winter) out of 44 cities, in which 10 and 13, respectively, are statisti­
cally significant. This implies that, although the mean SUHII is positive
for most urban areas, the vegetation is decreasing and thereby temper­
ature is increasing in city areas. A similar trend is also observed for

7
S. Raj et al. Journal of Environmental Management 257 (2020) 110006

Table 1
The significance of urban heat islands. Trend (� C/decade) of daytime and night-time SUHII estimated for the period 2000–2017. Trends are shown with 95%
confidence level. The statistically significant values are shown in bold.
No. City Summer Winter No. City Summer Winter

Day Night Day Night Day Night Day Night

1 Agra -0.278 0.054 -0.291 -0.104 23 Kota -0.225 -0.069 -0.258 -0.209
2 Ahmedabad -0.124 0.170 -0.093 -0.096 24 Lucknow 0.007 -0.040 -0.065 -0.264
3 Allahabad -0.114 0.108 -0.014 -0.130 25 Ludhiana -0.376 -0.026 -0.583 0.019
4 Amritsar -0.368 0.019 -0.360 0.042 26 Madhurai -0.210 0.040 -0.513 0.058
5 Aurangabad -0.176 0.188 0.022 0.225 27 Meerut -0.022 0.059 -0.023 -0.045
6 Bangalore -0.239 0.117 -0.130 0.127 28 Mumbai -0.083 0.093 0.124 0.177
7 Baroda 0.261 0.226 0.168 0.024 29 Nagpur -0.202 0.109 -0.221 0.104
8 Bhopal -0.033 0.196 0.103 -0.010 30 Nashik -0.312 0.270 -0.107 0.307
9. Chandigarh -0.343 0.014 -0.241 -0.007 31 Patna 0.300 0.279 0.044 0.099
10 Chennai 0.044 0.065 0.310 0.072 32 Pune 0.025 0.019 0.180 0.016
11 Coimbatore 0.124 0.087 -0.046 -0.013 33 Raipur 0.059 0.116 0.058 0.132
12 Delhi -0.313 0.075 -0.301 -0.082 34 Rajkot -0.214 0.145 0.227 0.088
13 Dhanbad 0.019 0.047 0.062 0.025 35 Ranchi -0.097 0.181 -0.044 0.146
14 Guwahati 0.106 0.245 -0.104 0.254 36 Srinagar 0.345 -0.158 0.458 0.016
15 Gwalior 0.175 0.181 0.146 -0.044 37 Surat 0.008 0.082 -0.173 0.030
16 Hyderabad -0.175 0.238 -0.115 0.242 38 Tata 0.156 0.222 0.195 0.243
17 Indore 0.058 0.191 0.569 0.062 39 Thrissur 0.182 0.101 0.223 0.001
18 Jabalpur 0.372 0.328 0.078 0.039 40 Trichy -0.351 -0.025 -0.311 0.104
19 Jaipur -0.393 0.015 -0.355 -0.37 41 Trivandrum 0.427 0.057 0.434 0.184
20 Jodhpur 0.093 -0.182 0.126 -0.267 42 Varanasi 0.010 0.237 0.028 0.030
21 Kanpur -0.088 0.116 -0.077 -0.192 43 Vijayawada 0.174 0.219 -0.028 0.109
22 Kolkata -0.030 0.165 -0.069 0.270 44 Vizag 0.227 0.053 0.176 0.117

The values in bold font represents statistical significance.

monsoon and post-monsoon seasons during daytime, as the trends of 5. Conclusion

about 10 (monsoon) and 11 (post-monsoon) out of 44 cities show
negative trends ( 0.6� to 0.18 � C/decade), but only 1 city in monsoon This analysis shows that the SUHII was intense during day in cities in
and 2 cities in post-monsoon show (e.g. Amritsar and Ludhiana) statis­ tropical climatic zones (Ko €ppen class A) especially in cities of Chennai
tically significant trends. These are also consistent with the temporal and Kolkata (1.21 � C and 1.33 � C) and during night in arid and
evolution of EVI, which is a measure of vegetation, during these seasons. temperate cities (Ko €ppen class B and C) such as Delhi and Ahmedabad
The cities that show negative trend in SUHII has a negative trend in (2.25 � C and 2.23 � C). The SUHII is weak during night-time in the
vegetation such as Chandigarh and Ahmedabad ( 0.27� and 0.104 � C/ tropical cities such as Trivandrum and Visakhapatnam (0.8 � C and 0.91
decade, respectively). Similarly, a few cities show increasing trend in �
C), and during day in arid and temperate cities of Jaipur and Gwalior
SUHII since the difference in vegetation in urban and suburban areas is ( 0.14 � C and 0.12 � C). The effect of vegetation on determining SUHII
increasing and Pune (0.18 � C/decade) is an example for such an urban was evident as the lower vegetation cover in suburban areas of the arid
agglomeration. and temperate climate regions was the major factor for a weaker day­
The trends in night SUHII show positive values for most cities time SUHII. There is an increasing trend in the night-time SUHII in all
( 0.26� –0.64 � C/decade), although more number of cities show nega­ seasons for most cities (0.007� –0.64 � C per decade). This can be
tive SUHII during winter (15) and monsoon (18), and half of these for attributed to the effect of anthropogenic forcing, that play a major role
other seasons. As shown by the mean SUHII, the night-time SUHII shows in determining the night-time SUHII. The study therefore, demonstrates
positive trends (0.007� –0.64 � C/decade), indicating the continuous the effect of anthropogenic activities on the heat island of Indian cities,
heating of urban surfaces in the night. The time for cooling the urban and the surface UHI effect will be enhanced if the city planning does not
surfaces takes longer, indicating the slow and consistent increase of the account for management of its green cover (e.g. EVI), pollution (e.g.
background temperature of the city, which is a signature of anthropo­ AOD), local weather and climate (e.g. temperature and precipitation)
genic warming or a component of the global warming. In addition, the and socio-economic factors (e.g. population). The findings of this study
night light differences between cities and their suburban areas are can be beneficial for implementing effective strategies to mitigate SUHII
significantly reducing, which suggests the urban sprawl and increase of and its detrimental effects. Future research should be aimed at con­
urban footprint. The conversion of suburban areas to urban agglomer­ ducting in-depth analysis of the effects of SUHII and their drivers within
ations indicates development of the city too. In India, urbanisation is individual cities.
very rapid for many reasons such as migration of people for jobs,
development of the cities to accommodate more built infrastructure for Appendix A. Supplementary data
industry and other amenities. However, the difference in urban and rural
nightlights is increasing in Nagpur and Mumbai, although there are also Supplementary data to this article can be found online at https://doi.
ambiguities in delineating urban and suburban areas in some complex org/10.1016/j.jenvman.2019.110006.
megacities such as Mumbai.
Conservation and expansion of vegetated areas in and around cities References
could be an effective mitigation strategy to counter the effects of high
SUHII. Evidence from this study suggest that more green spaces within Amaral, S., C^amara, G., Monteiro, A.M.V., Quintanilha, J.A., Elvidge, C.D., 2005.
Estimating population and energy consumption in Brazilian Amazonia using DMSP
the city’s boundary could reduce the temperature in the city and sur­ night-time satellite data. Comput. Environ. Urban Syst. 29, 179–195. https://doi.
rounding areas. This comprehensive analysis of SUHII and its relation­ org/10.1016/j.compenvurbsys.2003.09.004.
ships between potential drivers across both space and time could aid in Arnfield, a.J., 2003. Two decades of urban climate research: a review of turbulence,
exchanges of energy and water, and the urban heat island. Int. J. Climatol. 23, 1–26.
implementing accurate mitigation strategies of SUHII in urban areas in https://doi.org/10.1002/joc.859.
future.

8
S. Raj et al. Journal of Environmental Management 257 (2020) 110006

Cai, G., Du, M., Xue, Y., 2011. Monitoring of urban heat island effect in Beijing Precipitation Processing System (PPS) At NASA GSFC, 2018. TRMM (TMPA/3B43)
combining ASTER and TM data. Int. J. Remote Sens. 32, 1213–1232. https://doi. rainfall estimate L3 1 month 0.25 degree x 0.25 degree V7. https://doi.org/10.506
org/10.1080/01431160903469079. 7/trmm/tmpa/month/7.
Cao, C., Lee, X., Liu, S., Schultz, N., Xiao, W., Zhang, M., Zhao, L., 2016. Urban heat Rigo, G., Parlow, E., Oesch, D., 2006. Validation of satellite observed thermal emission
islands in China enhanced by haze pollution. Nat. Commun. 7, 1–7. https://doi.org/ with in-situ measurements over an urban surface. Remote Sens. Environ. 104,
10.1038/ncomms12509. 201–210. https://doi.org/10.1016/j.rse.2006.04.018.
Chakraborty, S.D., Kant, Y., Mitra, D., 2015. Assessment of land surface temperature and Rubel, F., Kottek, M., 2010. Observed and projected climate shifts 1901-2100 depicted
heat fluxes over Delhi using remote sensing data. J. Environ. Manag. 148, 143–152. by world maps of the K€ oppen-Geiger climate classification. Meteorol. Z. https://doi.
https://doi.org/10.1016/j.jenvman.2013.11.034. org/10.1127/0941-2948/2010/0430.
Clinton, N., Gong, P., 2013. MODIS detected surface urban heat islands and sinks: global Shastri, H., Paul, S., Ghosh, S., Karmakar, S., 2015. Impacts of urbanization on Indian
locations and controls. Remote Sens. Environ. 134, 294–304. https://doi.org/ summer monsoon rainfall extremes. J. Geophys. Res. Atmos. 120, 76. https://doi.
10.1016/j.rse.2013.03.008. org/10.1002/2014JD022061 (Received).
Didan, K., 2015. MOD13A2 MODIS/Terra vegetation indices 16-day L3 global 1km SIN Shastri, H., Barik, B., Ghosh, S., Venkataraman, C., Sadavarte, P., 2017. Flip flop of day-
grid V006. https://doi.org/10.5067/modis/mod13a2.006. night and summer-winter surface urban heat island intensity in India. Sci. Rep. 7,
Doick, K.J., Peace, A., Hutchings, T.R., 2014. The role of one large greenspace in 1–8. https://doi.org/10.1038/srep40178.
mitigating London’s nocturnal urban heat island. Sci. Total Environ. 493, 662–671. Stewart, I.D., Oke, T.R., 2012. Local climate zones for urban temperature studies. Bull.
https://doi.org/10.1016/j.scitotenv.2014.06.048. Am. Meteorol. Soc. 93, 1879–1900. https://doi.org/10.1175/BAMS-D-11-00019.1.
dos Santos, A.R., de Oliveira, F.S., da Silva, A.G., Gleriani, J.M., Gonçalves, W., Sulla-Menashe, D., Friedl, M., 2015. MCD12Q1 MODIS/TerraþAqua land cover type
Moreira, G.L., Silva, F.G., Branco, E.R.F., Moura, M.M., da Silva, R.G., Juvanhol, R. yearly L3 global 500m SIN grid V006 [data set]. NASA EOSDIS L. Process. DAAC. htt
S., de Souza, K.B., Ribeiro, C.A.A.S., de Queiroz, V.T., Costa, A.V., Lorenzon, A.S., ps://doi.org/10.5067/MODIS/MCD12Q1.006.
Domingues, G.F., Marcatti, G.E., de Castro, N.L.M., Resende, R.T., Gonzales, D.E., de Tran, H., Uchihama, D., Ochi, S., Yasuoka, Y., 2006. Assessment with satellite data of the
Almeida Telles, L.A., Teixeira, T.R., dos Santos, G.M.A.D.A., Mota, P.H.S., 2017. urban heat island effects in Asian mega cities. Int. J. Appl. Earth Obs. Geoinf. 8,
Spatial and temporal distribution of urban heat islands. Sci. Total Environ. 605–606, 34–48. https://doi.org/10.1016/j.jag.2005.05.003.
946–956. https://doi.org/10.1016/j.scitotenv.2017.05.275. UN, 2018. World Urbanization Prospects: the 2018 Revision. DESA/Population Div.
Elvidge, C.D., Imhoff, M.L., Baugh, K.E., Hobson, V.R., Nelson, I., Safran, J., Dietz, J.B., https://doi.org/10.1016/j.ijgo.2007.03.023.
Tuttle, B.T., 2001. Night-time lights of the world: 1994-1995. ISPRS J. Voogt, J.a., Oke, T.R., 2003. Thermal remote sensing of urban climates. Remote Sens.
Photogrammetry Remote Sens. 56, 81–99. https://doi.org/10.1016/S0924-2716(01) Environ. 86, 370–384. https://doi.org/10.1016/S0034-4257(03)00079-8.
00040-5. Wan, Z., 2014. New refinements and validation of the collection-6 MODIS land-surface
Forkel, M., Carvalhais, N., Verbesselt, J., Mahecha, M.D., Neigh, C.S.R., Reichstein, M., temperature/emissivity product. Remote Sens. Environ. 140, 36–45. https://doi.
2013. Trend Change detection in NDVI time series: effects of inter-annual variability org/10.1016/j.rse.2013.08.027.
and methodology. Remote Sens. 5, 2113–2144. https://doi.org/10.3390/rs5052113. Wan, Z., Hulley, G., Hook, S., n.d. MOD11A2 MODIS/Terra Land Surface Temperature/
Imhoff, M.L., Zhang, P., Wolfe, R.E., Bounoua, L., 2010. Remote sensing of the urban heat Emissivity 8-Day L3 Global 1km SIN Grid V006 [Data set], NASA EOSDIS LP DAAC.
island effect across biomes in the continental USA. Remote Sens. Environ. 114, https://doi.org/10.5067/MODIS/MOD11A2.006.
504–513. https://doi.org/10.1016/j.rse.2009.10.008. Ward, K., Lauf, S., Kleinschmit, B., Endlicher, W., 2016. Science of the Total Environment
Jin, M.S., 2012. Developing an index to measure urban heat island effect using satellite Heat waves and urban heat islands in Europe: a review of relevant drivers. Sci. Total
land skin temperature and land cover observations. J. Clim. 25, 6193–6201. https:// Environ. 569 (570), 527–539. https://doi.org/10.1016/j.scitotenv.2016.06.119.
doi.org/10.1175/JCLI-D-11-00509.1. Wu, X., Wang, G., Yao, R., Wang, L., Yu, D., Gui, X., 2019. Investigating surface urban
Kumar, R., Mishra, V., Buzan, J., Kumar, R., Shindell, D., Huber, M., 2017. Dominant heat islands in South America based on MODIS data from 2003–2016. Remote Sens.
control of agriculture and irrigation on urban heat island in India. Sci. Rep. 7, 1–10. 11, 1212. https://doi.org/10.3390/rs11101212.
https://doi.org/10.1038/s41598-017-14213-2. Yang, Q., Huang, X., Li, J., 2017. Assessing the relationship between surface urban heat
Letu, H., Hara, M., Yagi, H., Naoki, K., Tana, G., Nishio, F., Shuhei, O., 2010. Estimating islands and landscape patterns across climatic zones in China. Sci. Rep. 7, 1–11.
energy consumption from night-time DMPS/OLS imagery after correcting for https://doi.org/10.1038/s41598-017-09628-w.
saturation effects. Int. J. Remote Sens. 31, 4443–4458. https://doi.org/10.1080/ Yang, Q., Huang, X., Tang, Q., 2019. The footprint of urban heat island effect in 302
01431160903277464. Chinese cities: temporal trends and associated factors. Sci. Total Environ. 655,
Lyapustin, A., Wang, Y., 2018. MCD19A2 MODIS/TerraþAqua Land Aerosol Optical 652–662. https://doi.org/10.1016/j.scitotenv.2018.11.171.
Depth Daily L2G Global 1km SIN Grid V006. NASA EOSDIS Land Processes DAAC. Yao, R., Wang, L., Huang, X., Niu, Z., Liu, F., Wang, Q., 2017. Temporal trends of surface
https://doi.org/10.5067/MODIS/MCD19A2.006. urban heat islands and associated determinants in major Chinese cities. Sci. Total
Mathew, A., Khandelwal, S., Kaul, N., 2016. Title: spatial and temporal variations of Environ. 609, 742–754. https://doi.org/10.1016/j.scitotenv.2017.07.217.
urban heat island effect and the effect of percentage impervious surface area and Yao, R., Wang, L., Huang, X., Zhang, W., Li, J., Niu, Z., 2018. Interannual variations in
elevation on land surface temperature: study of Chandigarh city, India. Sustain. surface urban heat island intensity and associated drivers in China. J. Environ.
Cities Soc. 26, 264–277. https://doi.org/10.1016/j.scs.2016.06.018. Manag. 222, 86–94. https://doi.org/10.1016/j.jenvman.2018.05.024.
Mathew, A., Khandelwal, S., Kaul, N., 2017. Investigating spatial and seasonal variations Yao, R., Wang, L., Huang, X., Gong, W., Xia, X., 2019. Greening in rural areas increases
of urban heat island e ff ect over Jaipur city and its relationship with vegetation , the surface urban heat island intensity. Geophys. Res. Lett. 46, 2204–2212. https://
urbanization and elevation parameters. Sustain. Cities Soc. 35, 157–177. https://doi. doi.org/10.1029/2018GL081816.
org/10.1016/j.scs.2017.07.013. Zhang, P., Imhoff, M.L., Wolfe, R.E., Bounoua, L., 2010. Characterizing urban heat
Mathew, A., Khandelwal, S., Kaul, N., 2018. Analysis of diurnal surface temperature islands of global settlements using MODIS and nighttime lights products. Can. J.
variations for the assessment of surface urban heat island effect over Indian cities. Remote Sens. 36, 185–196. https://doi.org/10.5589/m10-039.
Energy Build. 159, 271–295. https://doi.org/10.1016/j.enbuild.2017.10.062. Zhao, L., Lee, X., Smith, R.B., Oleson, K., 2014. Strong contributions of local background
Oke, T.R., 1973. City size and the urban heat island. Atmos. Environ. Pergamon Pres 7, climate to urban heat islands. Nature 511, 216–219. https://doi.org/10.1038/
769–779. https://doi.org/10.1016/0004-6981(73)90140-6. nature13462.
Peng, S., Piao, S., Ciais, P., Friedlingstein, P., Ottle, C., Br�
eon, F.M., Nan, H., Zhou, L., Zhou, D., Zhao, S., Liu, S., Zhang, L., Zhu, C., 2014. Surface urban heat island in China’s
Myneni, R.B., 2012. Surface urban heat island across 419 global big cities. Environ. 32 major cities: spatial patterns and drivers. Remote Sens. Environ. 152, 51–61.
Sci. Technol. 46, 696–703. https://doi.org/10.1021/es2030438. https://doi.org/10.1016/j.rse.2014.05.017.
Peng, J., Ma, J., Liu, Q., Liu, Y., Hu, Y., Li, Y., Yue, Y., 2018. Spatial-temporal change of Zhou, B., Rybski, D., Kropp, J.P., 2017. The role of city size and urban form in the surface
land surface temperature across 285 cities in China: an urban-rural contrast urban heat island. Sci. Rep. 7, 1–9. https://doi.org/10.1038/s41598-017-04242-2.
perspective. Sci. Total Environ. 635, 487–497. https://doi.org/10.1016/j.
scitotenv.2018.04.105.