Article

Enumerating and Modelling the

Seasonal alterations of Surface
Urban Heat and Cool Island: A
Case Study over Indian Cities

Vinayak Bhanage, Sneha Kulkarni, Rajat Sharma, Han Soo Lee and Shirishkumar Gedam

Special Issue
Urban Agenda
Edited by
Prof. Dr. Luis Hernández-Callejo, Prof. Dr. Sergio Nesmachnow, Prof. Dr. Pablo de Frutos Madrazo,
Dr. Tigran Haas, Dr. Yu-Sheng Shen and Prof. Dr. Marcia Eugenio-Gozalbo

https://doi.org/10.3390/urbansci7020038
Article
Enumerating and Modelling the Seasonal alterations of Surface
Urban Heat and Cool Island: A Case Study over Indian Cities
Vinayak Bhanage 1,2 , Sneha Kulkarni 3, *, Rajat Sharma 4 , Han Soo Lee 2,5 and Shirishkumar Gedam 1

1 Centre of Studies in Resources Engineering, IIT Bombay, Powai, Mumbai 400076, India
2 Transdisciplinary Science and Engineering Program, Graduate School of Advanced Science and Engineering,
Hiroshima University, Hiroshima 739-8529, Japan
3 Institute of Engineering Innovation, School of Engineering, The University of Tokyo, Tokyo 113-8656, Japan
4 School of Energy & Environment, Thapar Institute of Engineering & Technology, Patiala 147004, India
5 Centre for Planetary Health and Innovation Science, The IDEC Institute, Hiroshima University,
Hiroshima 739-8529, Japan
* Correspondence: sneha.kulkarni@sogo.t.u-tokyo.ac.jp

Abstract: The present study has been carried out to analyze the seasonal variation of the Urban Heat
and Cool Island over the nine developing cities of India. The magnitude of urban heat/cool island
and vegetation gradient (∆NDVI) were measured from the daytime satellite datasets. Results of this
study show that during the pre-monsoon (March to May) season, the maximum magnitude of the
Surface Urban Heat Island (SUHI) was experienced over Kolhapur city, whereas, in the winter, the
highest intensity of SUHI was noticed over Pune city. Subsequently, outcomes also depict that the
changes in ∆NDVI restrain the pre-monsoon means and the seasonal alterations in SUHI magnitude.
However, during the winter (November to February) season, it is controlled by the temperature–
vegetation conditions of the rural areas. For pre-monsoon and seasonal changes in SUHI, with the
aid of ∆NDVI and the surface temperature of the urban area, regression equations were fitted for
pre-monsoon and seasonal changes in SUHI, which explains nearly 90% of SUHI variation. Similarly,
the variation of SUHI has been modeled for winter, which elucidates up to 84% of SUHI discrepancy.
The study reveals that, on a seasonal scale, a decrement of 0.1 in seasonal ∆NDVI leads to an increase
Citation: Bhanage, V.; Kulkarni, S.;
in the seasonal intensity of SUHI by 1.74 ◦ C, which is quite a significant augmentation.
Sharma, R.; Lee, H.S.; Gedam, S.
Enumerating and Modelling the
Keywords: MODIS; NDVI; seasonal variation; Surface Urban Heat Island; vegetation gradient
Seasonal alterations of Surface Urban
Heat and Cool Island: A Case Study
over Indian Cities. Urban Sci. 2023, 7,
38. https://doi.org/10.3390/
urbansci7020038
1. Introduction
Urbanization is the process where an increasing percentage of a population lives in
Academic Editor: Luis
cities and suburban areas. This process is often linked to industrialization and modern-
Hernández-Callejo
ization, as many people shift from rural to urban areas [1]. These urban areas generate
Received: 1 February 2023 their climatic domain dissimilar from their surrounding rural area as they have different
Revised: 14 March 2023 land surface properties, partitions of surface energy balance, urban geometry, soil mois-
Accepted: 28 March 2023 ture, vehicle density, etc. [2]. This disparity between the urban and rural areas in various
Published: 30 March 2023 physical and biophysical properties leads to a local scale modification of an urban climate.
Though the overall spatial extent of the urban area is minimal, its entanglement is attached
to half of the world’s population living in the urban area [3]. The world’s total urban
area is expanding twice its population growth [4]. Hence, the modifications in the local
Copyright: © 2023 by the authors.
Licensee MDPI, Basel, Switzerland.
climate generated by urban areas are imperative to understand its impact on climate and
This article is an open access article
human health.
distributed under the terms and The phenomenon of the Urban Heat Island (UHI) is one of the best examples of
conditions of the Creative Commons modifying an urban climate mainly caused due to the process of urbanization. It mainly
Attribution (CC BY) license (https:// indicates the elevated temperatures in the urbanized area compared to the surrounding
creativecommons.org/licenses/by/ rural area [5]. Sometimes depending upon the response of local physical and biophysical
4.0/). factors to the existing climatic conditions leads to negative thermal alteration in the rural

Urban Sci. 2023, 7, 38. https://doi.org/10.3390/urbansci7020038 https://www.mdpi.com/journal/urbansci

Urban Sci. 2023, 7, 38 2 of 12

area; this phenomenon is known as the Urban Cool Island (UCI) [6]. These UHI and UCI
reverberations are consequences of the disturbed cycles of the absorption and redistribution
of incoming solar radiation by altered urban and rural surfaces. In the past, several studies
have revealed that UHI/UCI can affect human health [6], economic growth [7], emission
of greenhouse gases [8], stormwater runoff [9], rate of mortality [10], modification of local
climate [11], etc. Thus, it is essential to comprehend the role of various parameters in the
formation of UHI to design mitigation strategies to control the contribution of those factors
in the development of the UHI/UCI.
Several studies found that alterations in vegetation strength, albedo, and partition of
surface energy balance, sky view factor are the major driving factors that regulate the mag-
nitude of the UHI [12–15]. In addition, previous research has shown that these parameters
mentioned above can act as proxy variables to explain the phenomenon of UHI/UCI [16].
Most parameters associated with an urban area, such as the sky view factor and vehicle
density, remain constant during the pre-monsoon and winter. In contrast, parameters, such
as soil moisture, vegetation strength, etc., mainly influence the rural microclimatic condi-
tions and induce alterations in the magnitude of the UHI/UCI concerning the season. Thus,
it is essential to model the behavior of the UHI/UCI and quantify the seasonal changes in
the magnitude of UHI/UCI related to those seasonally altered variables. To date, various
models have been developed and introduced to study the phenomenon of UHI.
There are mainly three types of models viz. Numerical, Physical, and Empirical mod-
els to scale out and quantify the seasonal behavior and changes in the UHI/UCI. Each has
its advantages and disadvantages. As we know, the spatial extent of the cities is much
smaller than the grid size of the global climatic models, which are used as input to the
numerical models. Therefore, it is challenging and computationally expensive to model
other associated parameters to UHI/UCI. Subsequently, these models are developing con-
tinuously; hence, analyzing each parameter and behavior of UHI at stationary conditions is
challenging. In earlier studies, physical models are mainly used for dispersion analysis,
wind flow pattern, surface roughness analysis, etc. [17]. An approach to empirical model-
ing depends upon the statistical algorithm. These models mainly deal with an observed
dataset. The only limitation of an empirical model is that they are restricted to a particular
location. In UHI studies, among all the empirical models, the statistical model is commonly
used to model and analyze the relationship of UHI with other physical and meteorological
parameters. Several researchers have used this method to quantify and model the impact
of UHI [5].
In the case of India, cities are mainly surrounded by agricultural fields, due to which
higher albedo, emissivity, and rate of evapotranspiration have been observed over the
rural area compared to the nearby urban area. This type of vegetation gradients and
evapotranspiration between urban and rural areas leads to lowered surface temperature
in rural areas. Throughout this process of evapotranspiration, vegetation releases water
vapor into the atmosphere, which contributes to reducing the air temperature. On the
contrary, impervious surfaces cover urban areas, such as conventional roofs, roads, and
sidewalks. As a result of urban expansion, more vegetation is removed and replaced by
more impervious surfaces (i.e., buildings and paving), actuating the reduction in the rate
of evapotranspiration that promotes the incremental trends of the temperature inside the
urban areas. The vegetation gradient between the urban and rural areas, which decides the
magnitude of the UHI, is sensitive to seasonal variation and crop cycle. During the pre-
monsoon season, less soil moisture, inactive agricultural practices, and deficient irrigation
make rural land temporarily barren, which leads to having hotter surfaces in the rural
areas, while due to enough irrigated parklands and roadside vegetation inside the cities
makes it cooler than the surrounding rural area which may enhance the magnitude of UCI.
Compared to the conditions mentioned above of the pre-monsoon season, in the winter,
due to enough precipitation and agricultural practices, it shows the ideal the UHI case
where urban areas are warmer than surrounding rural areas. According to [18], nearly
all cities in India act as a UCI during the pre-monsoon season, which promotes a higher
Urban Sci. 2023, 7, 38 3 of 12

mortality rate due to the intense heat wave events in the rural areas near the urban vicinity.
To resist such events, we need to quantify and understand the role of vegetation gradient
between the urban and rural areas on a different spatiotemporal scale to model the behavior
of the UHI. Several studies [19,20] have demonstrated the response of UHI and UCI with
the rurally dominant parameters during the pre-monsoon and winter seasons. However,
little attention has been paid to quantifying the seasonal changes in UHI/UCI associated
with the seasonally altered parameters.
The phenomenon of the UHI is mainly divided into three types, i.e., Surface Urban Heat
Island, atmospheric urban heat island, and sub-Surface Urban Heat Island. In this study, we
mainly focused on the first type of heat island, known as the Surface Urban Heat Island. We
will now indicate the Surface Urban Heat Island and surface cool island as the SUHI and
SUCI, respectively. The SUHI/SUCI mainly deals with the differences between the urban
and rural Land Surface Temperature (LST). In the past, several researchers [21–23] used the
LST data obtained from the MODIS satellite has been used to quantify the phenomenon of
SUHI. For the western part of the Indian region, Pandya et al. [24] evaluated the quality of
LST obtained from the MODIS [24]. He found that LST obtained from the MODIS satellite is
well correlated with the ground observation with the bias ranging from 0.7–7.6 ◦ C with the
correlation index >0.9. The MODIS LST product has been validated using ground-based
measurements and other satellite-based LST products, and the results have shown good
agreement with these independent datasets. Therefore, with the help of LST data obtained
from the MODIS, this study aims to (1) Analyze and model the seasonal variation of the
SUHI/SUCI based on the climatology of SUHI/SUCI; (2) Quantify the seasonal alteration
of the SUHI/SUCI concerning seasonally altered parameters.

2. Study Area
The present study was carried out over nine developing cities in India (Figure 1). The
population of each city is greater than 0.4 million [25]. These cities mainly span the four
states of India, i.e., Maharashtra, Karnataka, Gujarat, and Andhra Pradesh. The physical
and land surface characteristics of selected cities are different from each other. At the same
time, the local climatic conditions of the cities are also different due to the variation in their
topography and geographical location. These cities mainly fall under three climatic zones,
i.e., hot and dry, hot and humid, and composite climate [26]. Due to the strong demand in
the residential and industrial sectors, major cities, such as Hyderabad, Ahmadabad, Pune,
Nashik, and Aurangabad, are seeing rapid urban development growth.

Figure 1. Locations of the selected cities and their elevation variations.

On the other hand, cities, such as Kolhapur, Parbhani, Bellary, and Hubbali, are
surrounded by agricultural land where the urbanization rate is low compared to the larger
cities. This study has been carried out for the seasons of the pre-monsoon (March, April,
and May, MAM) and winter (November, December, January, and February, NDJF). The
Urban Sci. 2023, 7, 38 4 of 12

pre-monsoon and winter seasons have been defined based on the report published by the
Indian Meteorological Department (IMD) [27]. Due to cloud cover, the MODIS datasets
obtained during the monsoon season (June, July, August, and September, JJAS) did not
meet the required quality standards. As a result, we have excluded the monsoon season
from our analysis. The Climatological mean pre-monsoon surface temperature of the cities
ranges between 39.44 ◦ C to 45.68 ◦ C, while during the winter season, it spans between
30.20 ◦ C to 33.42 ◦ C. The detail information about various physical and meteorological
parameters is given in Table 1.

Table 1. Characteristics of selected nine cities of India.

Mean
Mean Winter
Premonsoon
Geographical Elevation Surface
Cities Population Area (km2 ) Surface
Location (m) Temperature (◦ C)
Temperature (◦ C)
(2006–2016)
(2006–2016)
Aurangabad 19.87◦ N 75.34◦ E 1,508,000 568 139 45.68 31.18
Hubbali 15.36◦ N 75.12◦ E 944,000 671 404 42.96 33.32
Hyderabad 17.38◦ N 78.48◦ E 6,810,000 505 650 39.44 30.2
Kolhapur 16.7◦ N 74.24◦ E 549,236 546 120 41.53 32.05
Nashik 19.99◦ N 73.70◦ E 1,486,053 700 268 42.48 31.41
Pune 18.52◦ N 73.85◦ E 3,120,000 560 424 42.26 32.12
Ahmedabad 23.02◦ N 72.57◦ E 5,307,000 53 464 41.99 30.21
Parbhani 19.26◦ N 76.77◦ E 410,445 347 37.8 43.15 32.42
Bellary 15.13◦ N 76.92◦ E 3,070,000 485 89.9 43.46 33.42

3. Data and Methodology

3.1. Dataset Used
For this study, we analyzed data collected from MODIS-Terra, i.e., daytime eight-day
composite data (MOD11A2) of Land Surface Temperature (LST) and monthly fused data
(MOD13A3) of Normalized Difference Vegetation Index (NDVI). These datasets were ob-
tained from the USGS website in a clear sky condition and further processed to maintain
the images’ quality and equal temporal granularity. In addition, MODIS quality control
flags were used to check the quality of the datasets. As a part of pre-processing, the original
projection of the datasets, i.e., sinusoidal, is converted to the geographical projection. We
know that satellite datasets have their qualms depending upon many other factors. In
2002, Trishchenko et al. [28] studied the impact of the sensor’s spectral response function
on NDVI measurements to evaluate that it could result in their qualms between 0.02 to
0.06 in the absolute values for the NDVI estimated from the MODIS dataset. NDVI is
mainly used to quantify vegetation vigor by measuring the difference between vegeta-
tion response within a near-infrared (NIR) and red (RED) channel of the electromagnetic
radiation spectrum. It can be calculated by using the following equation:

NIR − RED
NDVI = (1)
NIR + RED
In this formula, NIR is the reflectance value in the near-infrared band, and Red is the
reflectance value in the red band. The values of NDVI range from −1 to 1. The values
closer to 1 show higher vegetation vigor, while the measurements closer to −1 represent
lower vegetation strength. Several researchers [29,30] conducted their study to compare the
MODIS LST with the in situ and the LST retrieved from the other satellites. Their analysis
found a good confirmation index between LST obtained from the MODIS and other sources
over India and other parts of the world. The secondary data of population, which act as
a proxy to various other parameters, such as anthropogenic heat and built-up area, were
obtained from the 2011 Census Report, Government of India and utilized. The details about
all the datasets and their sources are given in Table 2.
Urban Sci. 2023, 7, 38 5 of 12

Table 2. Descriptions of the dataset used.

Temporal Spatial
Parameter Data Period References
Resolution Resolution
https://lpdaac.
Land Surface
usgs.gov/
Temperature MOD11A2 8 Day 1 km 2006–2016
(accessed on 11
(Daytime)
September 2021).
https://lpdaac.
usgs.gov.modis
NDVI MOD13A3 Monthly 1 km 2006–2016
(accessed on 11
September 2021).
http://www.
censusindia.gov.in
Population Census, 2011 2011
(accessed on 12
August 2021).

3.2. Methodology
LST (8 days) and NDVI (monthly) data were obtained from 2006 to 2016 from MOD11A2
and MOD13A3, respectively. The temporal resolution of both datasets were not equal. Thus,
the eight days of LST data were converted to the monthly mean datasets. In the further part,
the monthly mean LST and NDVI datasets were converted to the seasonal mean datasets
for the pre-monsoon (March to May) and winter (November to February) seasons. The
quality of the MODIS datasets during the monsoon (June to September) season was not up
to the mark due to the cloud cover; hence, we have not integrated the monsoon season for
our analysis. Afterward, all the seasonal data were calculated over the nine developing
cities of India for the 11 years, averaged out, and mean Climatological data were prepared
for the pre-monsoon and winter. During the entire data pre-processing and preparation
procedure, null values in the datasets were excluded. Figure 2 represents a general flow of
data processing and methodology flow.

Figure 2. The General flow of methodology followed for this study.

Urban Sci. 2023, 7, 38 6 of 12

3.2.1. Defining the Boundaries of the Urban and Rural Area

The UHI intensity mainly quantifies the difference in LST between urban and rural
areas, so determining the urban and rural area boundary is essential. Urban area expansion
is not restricted to municipal boundaries in cities, such as Ahmadabad and Pune. To mark
these limits of urban–rural areas accurately, MODIS LULC data are used. Each city’s buffer
distance from the urban boundaries towards the rural region varies. The buffer distance
for the smaller cities lies between 4–6 km, while due to the large urban sprawl, a larger
buffer distance, i.e., 7–8 km, was considered for further evaluation. During this analysis,
the buffer width is kept constant at 1 km.

3.2.2. Estimation of Surface Urban Heat Island (SUHI)

The method of urban–rural buffer analysis has been adopted to investigate the intensity
of the UHI. In this method, the difference in mean LST between the urban and rural areas
has been considered to calculate the SUHI magnitude (Equation (1). This method considers
the averaged LST of rural buffer areas along each direction of the urban areas as its
significant advantage.
∆T = Tu − Tr , (2)
where,
∆T: Surface Urban Heat Island Intensity (SUHII).
Tu: Mean Land Surface Temperature of the Urban Area.
Tr: Mean Land Surface Temperature of the Rural Buffer Area.
With the equation mentioned above (2) and the averaged LST data obtained for 2006–2016,
SUHII for the nine cities has been calculated for both seasons, i.e., pre-monsoon and winter.

3.2.3. Evaluation of Urban-Rural Vegetation Gradient

A recent study documented that the amount of vegetation inside the city or around
the rural area influences the magnitude of the SUHI [31]. However, the association between
vegetation gradient and the UHI modification is still unexplored. Thus, it is essential to
quantify this impact of the vegetation gradient on the SUHI. On a seasonal scale, the vegeta-
tion gradient between urban and rural areas alters due to meteorological changes affecting the
cities’ SUHII. Therefore, at a seasonal scale, it is necessary to elucidate the relationship between
vegetation gradient and SUHII. During this study, the vegetation gradient was calculated
with the help of averaged values of NDVI data obtained for 2006–2016. In this process, the
urban–rural buffer analysis method was adopted to estimate the vegetation gradient over
each city.
∆NDVI = NDVIu − NDVIr, (3)
where,
∆NDVI: vegetation gradient.
Tu: Mean NDVI of the Urban Area.
Tr: Mean NDVI of the Rural Buffer Area.

4. Results and Discussion

4.1. Seasonal Variation of SUHI
For the pre-monsoon and winter seasons, with the aid of 11 years of Terra MODIS LST
datasets, the extensiveness of SUHI over the nine developing cities of India was calculated
(Figure 3). The analysis shows that, during the pre-monsoon season, the maximum intensity
of SUHI (1.47 ◦ C) is observed over Kolhapur, while the minimum magnitude of SUHI
(−0.85 ◦ C) is observed over the city of Nashik. In the pre-monsoon season, out of the
nine cities, four, viz. Parbhani, Hyderabad, Nashik, and Bellary, represent the Urban Cool
Island (SUCI). This SUCI phenomenon mainly arises due to reduced vegetation and rising
temperature in the adjacent rural part of the urban areas. Still, in the case of Hyderabad, no
higher magnitude has been found in the vegetation of the urban area in both seasons. This
Urban Sci. 2023, 7, 38 7 of 12

phenomenon of the SUCI during the pre-monsoon and winter in Hyderabad may be the
effect of a large water body in the middle of the town.

Figure 3. Patterns in the Surface Urban Heat Island (SUHI) over nine Different Cities in India during
pre-monsoon (March to May) and winter season (November to February).

In the winter, the maximum intensity of the SUHI is found over Pune City (0.95 ◦ C),
followed by Ahmadabad (0.61 ◦ C). Some cities, such as Aurangabad, Hubbali, Kolhapur,
Hyderabad, and Bellary, show the negative behavior of the SUHI, i.e., the SUCI, during the
winter season.
Figure 4 depicts that in the winter season, in cities experiencing the phenomenon of the
SUCI, a negligible difference (−0.003 to −0.009) has been observed in ∆NDVI. On the other
hand, cities with a higher magnitude of ∆NDVI represent the typical intensified profile of
the UHI during the winter season. These seasonal alterations in the SUHI intensity may
have occurred due to the seasonal changes in NDVI.

Figure 4. Seasonal Pattern of ∆NDVI over nine Different Cities in India.

4.2. Statistical-Based Modelling of Premonsoon and Winter SUHI

In the further part of the study, a statistical analysis was carried out to explore the
association of the SUHI with other vegetation and climatic parameters on a different spatial
and temporal scale, which may elucidate the spatial and seasonal variations in the SUHI.
Urban Sci. 2023, 7, 38 8 of 12

To model the SUHI regression-based statistical model is developed for the pre-monsoon
and winter seasons. A simple linear regression was performed between the SUHI and the
other variables to identify the suitable forecaster. The other parameters were selected based
on former literature and their physical significance to the phenomenon of the SUHI. To
establish the SUHI model for the pre-monsoon and winter seasons, we mainly considered
the five parameters, i.e., the mean land surface temperature of the urban and rural area,
population, the vigor of vegetation in the rural part, and ∆NDVI. As we know, the total
amount of incoming solar radiation is approximately equal for adjacent rural and urban
areas. Still, the different land surface properties of the urban–rural area led to a difference
in the magnitude of surface temperatures, which gave rise to the SUHI.
Thus, during this study, the LST of the urban and rural areas has been taken as a proxy
variable to the land surface characteristics of the urban and rural areas, such as albedo,
emissivity, heat capacity, etc. Various studies have been conducted, such as [32,33], to
explain the negative relationship between temperature and vegetation. The vegetation in
a particular territory keeps that area cool through evapotranspiration. Hence, the total
amount of vegetation present in the urban and rural areas controls the temperatures of
the respective area. It is considered one of the significant parameters that can pedal the
magnitude of the SUHI. In addition, the disparity of vegetation strength between urban and
rural areas can alter the natural cooling process of the environment, resulting in changes
in the temperature difference between urban and surrounding rural areas. Besides these
climatic and biophysical parameters, the population of the urban area is a significant factor
that plays an essential role in the formation of SUHI. The more population in the urban
area represents the high density of buildings, vehicles, roads, low sky view factor, and
many industries. It creates more anthropogenic heat that gets discharged and trapped
in the urban environment, making the urban area warmer. Hence, the population is also
essential for modeling the SUHI effect. Figure 5 represents the correlation coefficient, i.e., R,
between SUHI, and other predictors for the pre-monsoon and winter seasons.
During the pre-monsoon season, the significant relationship between ∆NDVI and
the SUHI confirms that the difference in the vegetation vigor between urban and rural
areas is one of the major factors responsible for developing the phenomenon of the SUHI
in the pre-monsoon season. In addition, the association between population and surface
temperature of the urban area reveals that an increase in the population of the urban area
makes the city warmer. Compared to the pre-monsoon season, the association between the
population and Tu becomes stronger during the winter, while the SUHI has the strongest
association with Tr. It characterizes Tr as a controller of the SUHI effect for the winter
season. Using the above analysis, we constructed the regression-based model for the SUHI
during winter and pre-monsoon. We have mainly used three predictors to develop a model
for the pre-monsoon season. Tu, population, and ∆NDVI. Likewise, the correlation analysis
and the multivariate regression model (Equation (4)) also denotes the dominance of ∆NDVI
during the pre-monsoon season. This regression model has a coefficient of determination
(R2 ) of 0.97 with p < 0.0001 and the Root Mean Square Error (RMSE) of 0.28 ◦ C. The ∆NDVI
as an alone predictor coefficient of determination (R2 ) was 0.65; other variables were added
to improve the model’s predictability.

SUHI in pre-monsoon = 43.08 − 0.135 (Tu) − 20.72 (∆NDVI) − 0.00000025 (Population) (4)

During the winter, vegetation coverage in the surrounding rural area is a decisive factor
controlling the magnitude of the SUHI. Initially, NDVIr as an alone predictor coefficient of
determination (R2 ) was 0.62. Thus, to perk up the model performance, additional predictors
were added.
SUHI in winter = 107.96 − 0.37 (Tr) + 11.08 (NDVIr) (5)
The model of the SUHI for a winter season has a coefficient of determination (R2 ) of
0.89 with p < 0.0001 and the RMSE of 0.24 ◦ C.
Urban Sci. 2023, 7, 38 9 of 12

Figure 5. Coefficient of Correlation (R) analysis matrix between SUHI and other parameters for
(a) Premonsoon and (b) Winter season.

4.3. The Relationship between ∆Seasonal SUHI and ∆Seasonal Change in Vegetation
The ∆Seasonal SUHI is obtained by calculating the difference between the mean of the
pre-monsoon and wintertime SUHI. At the same time, the ∆Seasonal change in vegetation is
retrieved by calculating the disparity of ∆NDVI between both seasons. Figure 6 elucidates
the relationship between alteration in the seasonal NDVI and the difference between
the seasonal SUHI. The negative affiliation with the correlation coefficient R = −0.646
(p < 0.001) at a 95% confidence level is found between seasonal change in Sthe UHI and
seasonal variation in the difference between urban and rural vegetation. After observing
the association between the ∆Seasonal SUHI and ∆Seasonal change in vegetation, it is
required to find the significant controlling factors which decide the seasonal alterations
in SUHI.
Urban Sci. 2023, 7, 38 10 of 12

Figure 6. Association between seasonal alterations in ∆NDVI and ∆SUHI over 9 Different Cities
in India.

Recently, a few studies noted that agricultural activities in rural areas tend to have
more seasonal variation in the land surface properties than in the urban area [20,34]. It
suggests that the magnitude of perturbations in LST due to land surface properties is greater
over the rural area. Some studies explained that the population data as a proxy for the
built-up area, anthropogenic heat, etc. [35]. Thus, to model the ∆Seasonal SUHI, we have
used a multivariate regression method and considered three parameters viz. ∆Seasonal
change in vegetation, ∆Seasonal rural LST, and population, which quantifies the seasonal
change in SUHI, where

∆Seasonal SUHI = SUHI winter − SUHI premonsoon (6)

∆Seasonal change in vegetation = ∆NDVI winter − ∆NDVI premonsoon (7)

∆Seasonal rural LST = LST of Rural Site (winter) − LST of Rural Site (premonsoon) (8)

The equation, “∆Seasonal SUHI = −8.353 + (0.0142) × ∆Seasonal rural LST—(17.40)

× ∆Seasonal change in vegetation”, explains the changes in the seasonal SUHI with the
coefficient of determination R2 = 0.68 where the p-value < 0.005 for all the attributes of the
equations. It indicates that a decrement in a ∆Seasonal change in vegetation by 0.1 leads to
an increase in the ∆Seasonal SUHI by 1.74 ◦ C, which is quite a significant augmentation if
other variables remain constant. Hence, to reduce the intensity of the SUHI at a seasonal
scale, it is necessary to reduce the magnitude of ∆Seasonal change in vegetation. The
advantage of the above equation is that it can explain the ∆Seasonal SUHI variations of up
to 68% over the nine cities in India using the readily available satellite data of NDVI and
LST. We consider this equation true, at least for cities under study.

5. Conclusions
This study was mainly divided into two parts, i.e., (1) Assessment of the seasonal
variation of SUHI and (2) the response of seasonal alterations in the SUHI to the seasonal
variations in the vegetation strength. This study was performed over nine developing cities
in India using satellite data for 2006–2016. These nine cities were chosen carefully based
on their differences in (a) climatic zones, (b) geographical location, and (c) topographical
characteristics.
Urban Sci. 2023, 7, 38 11 of 12

In this study, during the pre-monsoon (March to May) and winter (November to
February) seasons mean intensity of the SUHI and the ∆NDVI were derived for each
city. A result shows that the SUHI’s magnitude mainly depends upon the ∆NDVI. The
positive values of ∆NDVI represent the phenomenon of the Urban Cool Island, where the
city of Hyderabad was an exceptional case. In both the pre-monsoon and winter seasons,
Hyderabad experiences the SUCI, which may be the effect of a large water body located in
the middle of the city. It needs to be analyzed further to validate and quantify the impact
of the water body on the SUHI.
Further statistical analysis between the SUHI and other studied variables shows that
during the pre-monsoon season, ∆NDVI is the dominant factor which explains 97% of the
SUHI with the RMSE of 0.28 ◦ C, while in the winter season temperature of the rural region
is the most significant parameter, which describes the 89% of the SUHI with the RMSE of
0.24◦ . The performed correlation and regression analysis confirms the sensitivity of each
parameter to the mean seasonal SUHI so that one can concentrate on specific variables
while designing the mitigation strategies for the SUHI. In the second part of the study, we
quantified the response of the SUHI to the seasonal changes in vegetation strength. During
this analysis, we found that an alteration in the ∆Seasonal change in vegetation explains
68% of the variation in the ∆Seasonal SUHI. As we know, the variation in vegetation
strength during the winter and pre-monsoon seasons mainly depend upon the monsoon
disparity. Thus, further exploration of seasonal changes in the magnitude of the SUHI
and its association with the monsoon variation through its impact on seasonal vegetation
strength needs to be evaluated.

Author Contributions: Conceptualization, V.B., S.K., H.S.L. and S.G.; methodology, V.B., S.K., H.S.L.
and S.G.; software, V.B. and S.K; formal analysis, V.B., S.K. and R.S.; investigation, V.B., S.K., H.S.L.,
R.S. and S.G.; writing V.B. and S.K.; supervision, H.S.L. and S.G. All authors have read and agreed to
the published version of the manuscript.
Funding: This work is a part of DST-INSPIRE program (DST/INSPIRE/03/2015/005692 IF160608).
Data Availability Statement: Not applicable.
Acknowledgments: V.B. and S.K. are thankful to the IIT Bombay for providing necessary facilities to
carry out this research. All the authors are grateful to the editor and anonymous reviewers for their
rigorous reviews, which greatly improved the quality of the manuscript. We are also thankful to R.
Latha and B.S. Murthy for their support.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Urbanisation, Demographic Transition, and the Growth of Cities in India. Available online: https://www.theigc.org/sites/
default/files/2016/11/Tumbe-2016-Working-paper.pdf (accessed on 15 February 2022).
2. Erell, E.; Pearlmutter, D.; Williamson, T. Urban Microclimate: Designing the Spaces between Buildings, 1st ed.; Routledge: Milton
Park, UK; Philadelphia, PA, USA, 2011.
3. United Nations. World Urbanization Prospects: The 2018 Revision; United Nations: New York, NY, USA, 2018.
4. Angel, S.; Parent, J.; Civco, D.L.; Blei, A.; Potere, D. The dimensions of global urban expansion: Estimates and projections for all
countries, 2000–2050. Prog. Plann. 2011, 75, 53–107. [CrossRef]
5. Oke, T.R. City size and the urban heat island. Atmos. Environ. Pergamon Press 1973, 7, 769–779. [CrossRef]
6. Heaviside, C.; Vardoulakis, S.; Cai, X.M. Attribution of mortality to the urban heat island during heatwaves in the West Midlands,
UK. Environ. Health 2016, 15. [CrossRef]
7. Santamouris, M.; Cartalis, C.; Synnefa, A.; Kolokotsa, D. On the impact of urban heat island and global warming on the power
demand and electricity consumption of buildings—A review. Energy Build. 2015, 98, 119–124. [CrossRef]
8. Sarrat, C.; Lemonsu, A.; Masson, V.; Guedalia, D. Impact of urban heat island on regional atmospheric pollution. Atmos. Environ.
2006, 40, 1743–1758. [CrossRef]
9. Li, H.; Harvey, J.T.; Holland, T.J.; Kayhanian, M. Erratum: The use of reflective and permeable pavements as a potential practice
for heat island mitigation and stormwater management. Environ. Res. Lett. 2013, 8, 015023. [CrossRef]
10. Gasparrini, A.; Guo, Y.; Hashizume, M. Mortalité attribuable au froid et à la chaleur: Analyse multi-pays. Environ. Risques Et
Sante 2015, 14, 464–465.
Urban Sci. 2023, 7, 38 12 of 12

11. Lin, C.Y.; Chen, W.C.; Chang, P.L.; Sheng, Y.F. Impact of the urban heat island effect on precipitation over a complex geographic
environment in northern Taiwan. J. Appl. Meteorol. Climatol. 2011, 50, 339–353. [CrossRef]
12. Spatio-Temporal Analysis of Urban Heat Island Effect For Different Land Use Land Cover Pattern Over Pune Metropolitan
Region, India. In Proceedings of the AGU Fall Meeting Abstracts. Available online: https://ui.adsabs.harvard.edu/abs/2019
AGUFMGC21I1384B/abstract (accessed on 9 August 2021).
13. Rakoto, P.Y.; Deilami, K.; Hurley, J.; Amati, M.; Sun, Q.C. Revisiting the cooling effects of urban greening: Planning implications
of vegetation types and spatial configuration. Urban. For. Urban. Green. 2021, 64, 127266. [CrossRef]
14. Mariani, L.; Parisi, S.G.; Cola, G.; Lafortezza, R.; Colangelo, G.; Sanesi, G. Climatological analysis of the mitigating effect of
vegetation on the urban heat island of Milan, Italy. Sci. Total Environ. 2016, 569, 762–773. [CrossRef]
15. Vinayak, B.; Lee, H.S.; Gedam, S.; Latha, R. Impacts of future urbanization on urban microclimate and thermal comfort over the
Mumbai metropolitan region, India. Sustain. Cities Soc. 2022, 79. [CrossRef]
16. Bottayan, Z.; Kirsci, A.; Szegedi, S.; Unger, J. The relationship between built-up areas and the spatial development of the mean
maximum urban heat island in Debrecen, Hungary. Int. J. Climatol. 2005, 25, 405–418. [CrossRef]
17. Kanda, M. Progress in the scale modeling of urban climate: Review. Theor. Appl. Climatol. 2006, 84, 23–33. [CrossRef]
18. Shastri, H.; Ghosh, S.; Paul, S.; Shafizadeh-Moghadam, H.; Helbich, M.; Karmakar, S. Future urban rainfall projections considering
the impacts of climate change and urbanization with statistical–dynamical integrated approach. Clim. Dyn. 2019, 52, 6033–6051.
[CrossRef]
19. Shastri, H.; Barik, B.; Ghosh, S.; Venkataraman, C.; Sadavarte, P. Flip flop of Day-night and Summer-Winter Surface Urban Heat
Island Intensity in India. Sci. Rep. 2017, 7, 40178. [CrossRef] [PubMed]
20. Kumar, R.; Mishra, V.; Buzan, J.; Kumar, R.; Shindell, D.; Huber, M. Dominant control of agriculture and irrigation on urban heat
island in India. Sci. Rep. 2017, 7, 14054. [CrossRef] [PubMed]
21. Barat, A.; Kumar, S.; Kumar, P.; Parth Sarthi, P. Characteristics of Surface Urban Heat Island (SUHI) over the Gangetic Plain of
Bihar, India. Asia-Pac. J. Atmos. Sci. 2018, 54, 205–214. [CrossRef]
22. Pramanik, S.; Punia, M. Land use/land cover change and Surface Urban Heat Island intensity: Source–sink landscape-based
study in Delhi, India. Environ. Dev. Sustain. 2020, 22, 7331–7356. [CrossRef]
23. Kumari, P.; Yadav, N.K.; Santra, A.; Upadhayay, U. Identification of urban Heat Islands from multi-temporal Modis land surface
temperature data: A case study of the southern part of West Bengal, India. Int. J. Eng. Sci. Res. Technol. 2019, 135–144.
24. Pandya, M.R.; Shah, D.B.; Trivedi, H.J.; Darji, N.P.; Ramakrishnan, R.; Panigrahy, S.; Parihar, J.S.; Kirankumar, A.S. Retrieval of
land surface temperature from the Kalpana-1 VHRR data using a single-channel algorithm and its validation over western India.
ISPRS J. Photogramm. Remote Sens. 2014, 94, 160–168. [CrossRef]
25. Census of India Census of India, 2011. Econ. Polit. Wkly. 2011, 46, 5.
26. Kottek, M.; Grieser, J.; Beck, C.; Rudolf, B.; Rubel, F. World map of the Köppen-Geiger climate classification updated. Meteorol. Z.
2006, 15, 259–263. [CrossRef]
27. A Technical Note on the Analysis of Modern Meteorological Gridded a Technical Note on the Analysis of Modern Meteorological
Gridded Datasets Using Python. Available online: https://www.imdpune.gov.in/Reports/CRSResearchReports/TECHNICAL_
REPORT_review.pdf (accessed on 22 November 2022).
28. Trishchenko, A.P.; Cihlar, J.; Li, Z. Effects of spectral response function on surface reflectance and NDVI measured with moderate
resolution satellite sensors. Remote Sens. Environ. 2002, 81, 1–18. [CrossRef]
29. Singh, R.; Singh, C.; Ojha, S.P.; Kumar, A.S.; Kishtawal, C.M.; Kumar, A.S.K. Land surface temperature from INSAT-3D imager
data: Retrieval and assimilation in NWP model. J. Geophys. Res. Atmos. 2016, 121, 6909–6926. [CrossRef]
30. Salama, M.S.; Velde, R.v.d.; Zhong, L.; Ma, Y.; Ofwono, M.; Su, B. Decadal variations of land surface temperature anomalies
observed over the Tibetan Plateau by the Special Sensor Microwave Imager (SSM/I) from 1987 to 2008. Clim. Change 2012, 114,
769–781. [CrossRef]
31. Haashemi, S.; Weng, Q.; Darvishi, A.; Alavipanah, S.K. Seasonal variations of the surface urban heat Island in a semi-arid city.
Remote Sens. 2016, 8, 352. [CrossRef]
32. Weng, Q.; Lu, D.; Schubring, J. Estimation of land surface temperature-vegetation abundance relationship for urban heat island
studies. Remote Sens. Environ. 2004, 89, 467–483. [CrossRef]
33. Boegh, E.; Soegaard, H.; Hanan, N.; Kabat, P.; Lesch, L. A remote sensing study of the NDVI-T(s) relationship and the transpiration
from sparse vegetation in the sahel based on high-resolution satellite data. Remote Sens. Environ. 1999, 69, 224–240. [CrossRef]
34. Zhou, D.; Xiao, J.; Bonafoni, S.; Berger, C.; Deilami, K.; Zhou, Y.; Frolking, S.; Yao, R.; Qiao, Z.; Sobrino, J.A. Satellite remote
sensing of Surface Urban Heat Islands: Progress, challenges, and perspectives. Remote Sens. 2019, 11, 48. [CrossRef]
35. Zeug, G.; Eckert, S. Population growth and its expression in spatial built-up patterns: The sana’a, yemen case study. Remote Sens.
2010, 2, 1014–1034. [CrossRef]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.