ISSN: 0973-4929, Vol. 17, No. (2) 2022, Pg.

393-409

Current World Environment

[Link]

Modelling the impact of Urban Heat Island mitigation

strategies on Urban Air Quality
NEHA M. DAVE1*, RUPESH P. VASANI2 and PARMIT CHHASIYA3

1
Environment Engineering, Gujarat Technological University, Ahmedabad, India.
2
Sal Institute of Technology & Engineering Research , Ahmedabad, India.
3
Center for Applied Geomatics, Cept Research and Development Foundation,
Cept University, Ahmedabad, India.

Abstract
Increasing urban sprawl has caused many severe problems like surge in
pollution, rapid climatical variations, and the intensification of temperature in
the urban areas, termed Urban Heat Islands (UHI). Population density has Article History
caused the conversion of most land areas into cities, and cities have expanded Received: 10 March 2021
Accepted: 17 June 2022
vastly. UHI phenomenon has caused temperature rise in the cities. Most of
the metropolitan regions of India are experiencing consequences of UHI and Keywords
the severity of pollution formation, which is a crucial research area. Since the Ahmedabad City;
rising temperature has a direct linkage with urban air pollution, the mitigation Air Quality;
Regression Analysis;
measures for UHI are also linked with urban air pollution mitigation, efficacy of
Urbanization;
mitigation measures of UHI phenomenon in correlation with urban air quality Urban Heat Island.
is being studied extensively, which emphasizes the scientific approach and
planning concerns of implementation agency to consider the same into urban
design and planning aspects. Ahmedabad is one of the growing metropolitan
regions of India. The city has grown economically and physically by expanding
its boundaries in a radial pattern. This study has attempted temporal
assessment of remote sensing data to derive the UHI and the city's growth,
and its changing land uses. Assessment has been performed from 2008 to
2018 from Landsat data for temperature profile at surface level and type of
usage of land of the study area. The spatial profile of Particulate Matter (PM2.5
and PM10) has been generated based on data from the state pollution control
board. Four variables, LST, PM2.5, PM10, and LULC, are taken to establish
the relationship between all variables present in different layers with the help
of regression statistical analysis. A strong positive correlation between PM2.5,
PM10, and LST has been discovered, which was eventually used to assess
the impact of mitigation strategies of UHI, specifically urban greening and
a white roof to particulate matter concentrations.

CONTACT Neha M. Dave nmd_env@[Link] Environment Engineering, Gujarat Technological University, Ahmedabad, India.

© 2022 The Author(s). Published by Enviro Research Publishers.

This is an Open Access article licensed under a Creative Commons license: Attribution 4.0 International (CC-BY).
Doi: [Link]
 DAVE et al., Curr. World Environ., Vol. 17(2) 393-409 (2022) 394

Introduction and radiance value which is further converted into

Urban Heat Islands is the higher temperature in an ground-level reflectance value.
urban area than in surrounding open or rural areas
which has two distinct categories: Atmospheric and UHI mitigation measures such as green area
Surface Heat Island. Atmospheric UHIs are divided development and change of materials which are
into two parts. one is the layer of air to the top able to reduce the adverse effects of the UHI and
of trees or roofs, and another is from the building mitigate future problems.14 Considering variables
roof or tree top level to the upper atmosphere of the urban atmosphere such as evaporation,
of that area. Surface urban heat island depicts h u m i d i t y, t e m p e r a t u r e , t u r b u l e n c e s , e t c . ,
a difference between urban surfaces compared with UHI mitigation strategies result in energy consumption
its surrounding. The surface UHI is highest during the and changes in radiation patterns.15
summer, and its variation is much more than UHIs.11
The spatial characteristics of the UHI are related to Literature Studies
the topography profile of the urban area, which can Anthropogenic activity affects the urban climate,
be differentiated by the urban-rural temperature and its effects are adverse to the environment.
boundary.10 Moreover, urban area expands at a rapid Change in the local natural environment, which
rate from center to fringes. The growth generally is replaced by artificial with concrete and other
happens in areas of city fringes, thus exacerbating radiation absorption materials, affects the
the sprawling level. Changing scenarios of human urban area's heat exchange and micro climate.
settlements, lifestyle changes, and anthropogenic Urbanization changes the LULC characteristics
activities have resulted in two major phenomena, of an area, resulting in the formation of the UHI
increasing heat and deterioration of air quality. effect. Built-up area increase, vegetation decrease,
emission, and pollution are reasons for urban
Urban Heat Islands cause higher use of energy areas hotter than green fields. Urban climate study
for cooling.1 Heat also produces chemical reactions in South Asia shows more than an 8 °C temperature
into atmosphere.2 These secondary air pollutants difference between the highly-dense urban area
harm health of living beings.3 Urban green areas and greenfields.8 Even high-temperature rises in
and white roofs decrease heat in atmosphere and urban areas end up with high energy consumption,
eventually improves air quality.4 which comes from the area covered by buildings
and streets as an impermeable material with
Urban Heat Island intensity, also known as low emissivity and high heat absorption.
UHI magnitude, is the temperature difference
between urban and surrounding areas.5 Climate Remote sensing can map urban areas and
region, local topography, and industry expansion their LULC spatial concentration using several
of a city affect UHI intensity,6 LULC characteristics, classification methods as captured optical data
the neighborhood rural areas,7 and quantity of with spectral information about the area. Land
vegetation in an urban area8 are remarkable for Surface Temperature (LST) is a necessary term
urban heat island intensity, as well as local climatic related to the thermal state of the boundary layer
thermal patterns. It is evaluated that 3-8% higher of the study area. Initially, the land surface temperature
electricity demand in an urban area with a population was coined for surface or air temperature measured
of more than 1 00,000 is used to tackle the heat by a thermometer installed 1.5-3.5 meters above
enhanced by the urban heat island effect, which leads the flat surface. Due to advancements in satellite
to higher emissions enhancing the concentration technology, another type of land surface temperature
of harmful smog, higher air pollution, and human data collection, satellite-based surface temperature
health impacts. collection, i.e., skin temperature, is extremely useful
when the spatial scale is large.10 Through state
Through state-of-the-art satellite technology, land of the art satellite technology, land surface
surface temperature has become a key parameter temperature has become a key parameter to
for estimating surface temperature.9 Satellite sensor, estimate surface temperature, which is not the only
having a thermal band, targets towards surface affecting indicator of climate change but also its
 DAVE et al., Curr. World Environ., Vol. 17(2) 393-409 (2022) 395

relation with radiation, nevertheless with control of of harmful smog, higher air pollution, and human
surface sensibility and heat flux exchange between health impacts.
atmosphere and surface of the earth.5 For instance,
energy exchange in boundary UHI is dominated PM2.5 shows a positive correlation with change
by the disparity between air temperature and of usage pattern of land with a correlational
surface temperature. The air and the surface interact coefficient of more than 0.6. The area with more road
with different time-space scales to external forces, density, sizeable built-up area, and LST are steady
yet they are correlated complexly. with a high focus on PM and SO2, NO2 and CO,
which can be understood by the correlation between
The satellite sensor, having a thermal band, targets air pollutants and land use,9 Furthermore, NO2 is the
the surface, and the radiance value is further most sensitive gas affected by land-use changes,
converted into a ground-level reflectance value. and O3 shows a weak correlation with land [Link]
The reflectance from the ground is converted to relational study between UHI and air quality shows
the top of the atmosphere radiance value. The that increasing UHI increases particulate matter.
Thermal Infrared Sensor (TIR) gets the surface Decrease in Ozone concentration is observed with
radiance radiated towards the sensor until this value usage of high albedo built up.14 Moreover, vegetation
is converted to brightness temperature. To derive of the urban area and high emissivity rooftops could
Land Surface Temperature, it is must require drastically reduce energy consumption, resulting in
to calculate surface emissivity (E). improved air quality[.4

Urban heat island contributes to elevated ozone The arrangement of UHI is, for the most part,
levels because it adds thermal radiation to a chemical supported by the low wind speed and high
reaction from the ground ozone level. Sunlight convergence of PM2.5, with a positive relationship
and enhanced heat levels due to the urban heat between the UHI and PM2.5 focuses. The expanded
island effect can also photochemically cook ozone pressurized canned products in urban climates
to a more hazardous level when air temperature lessen the approaching solar flux and increment
exceeds 90°C; ozone is created when dinitrogen of barometrical long wave radiation in the urban
monoxide and volatile organic compounds areas. The reaction of the material to the difference
photochemically combine with solar radiation. in reflected radiation is solid during the evening
Nevertheless, the ozone concentration can be and powerless during the day. The surface urban
reduced by the lower ambient air temperature. heat island force is improved by 12% in the evening
Under the right chemical conditions, the UHI by the expanded assimilated solarradiation in the
effect can hike up the air temperature by 10 F urban areas.
or more. thus, it acts an additive supportive effect
to solar radiation. Methodology
This study is limited to the boundary of Ahmedabad
Climate region, local topography, and industry Municipal Corporation. Ahmedabad was declared
expansion of a city affect UHI intensity,12 green a Megacity, located betweena latitude of 23.03
roofs have capacity of surface cooling which and a longitude is 72.58 in North-central Gujarat.
is very significant and can help in decreasing The average elevation profile of Ahmedabad city is
temperature in urban areas,16 LULC characteristics, 53 meters. The population of Ahmedabad is more
the neighborhood rural areas, 3 and quantity than 3.5 million as per the census 2001, which is the
of vegetation in urban area 2 are remarkable 7th metropolitan city of India. Ahmedabad is located
for urban heat island intensity, as well as local on the Sabarmati River.
climatic thermal patterns. It is evaluated that 3-8%
higher electricity demand in urban areas with Ahmedabad has a dry and comparatively hot
a population of more than 1,00,000 is used to climate with maximum 41°C and minimum 27°C
tackle the heat enhanced by the urban heat island in summer months and maximum 30°C and minimum
effect, which leads to higher emission of SO2, CO, 15°C in winter months with average rainfall
NO, PM2.5, PM10, enhancing the concentration of 800 mm.
 DAVE et al., Curr. World Environ., Vol. 17(2) 393-409 (2022) 396

In order to begin this study, primary data was process, supervised classification for LULC, Spatial
collected from various sources. Collected data was interpolation for Air Quality data, etc. Furthermore,
in a standard format, so it was needed to select, the study has been divided into scales: Macro level,
acquire, review and process It for further analysis Meso level, and Micro level, which leads to spatial
relevant to the study. Raster Data in terms of Satellite modelling of LST, Air Quality, and LULC. The study
Images were acquired from Landsat ETM & ETM+ is progressed by suggestive mitigation measures
Data (2008 to 2018). Vector Data such as study area for UHI and Air Quality for the study area. Further
boundary, ward, zones of the city, physical features geospatial operations and methods are explained
like rivers and lakes, green areas, etc. were prepared below for this study. Ground truthing has been done
based on secondary data sources of Ahmedabad for LST results and analysis of the data gap between
Municipal corporation, Attribute Data of Air Quality in-situ measurements. Ground truthing dates are
for PM2.5 and PM10 (2008, 2015-2017) selected based upon the satellite temporal resolution
were acquired from State Pollution Control Board. of Landsat-8. Moreover,the study area selection
for ground-truthing has been made based on the
The identified data types can be categorized into typology of the area, for readings, types of features
Raster, Vector, and Non-Spatial (Attribute Data). are identified for temperature measurement.
Raster Data Landsat TM & ETM+ data is used for
LST generation and LULC classification of the study Model Builder in Arc GIS is used for processing
area. Vector data like administrative boundaries the databases. It is an application that automates
of the study area helps perform numerous geospatial geoprocessing workflows, a series of tools chained
operations that bounds research to the study area. together, where the output of one tool is given as
In Nom-Spatial or Attribute data, data of Air Quality input to another tool. After having connected tools
focused on PM2.5 and PM10 of Ahmedabad city in the Model, one can execute and run that model.
has been collected from GPCB. Moreover, Field In simple words, it connects geoprocessing
data collection, which is done by ground-truthing, sequences and gives desired spatial [Link] the
gives ground-based LST and Air Quality data. case of LST derivation, it saves time if there is
The flow chart shows various spatial processes for a model system that contains sequences of raster
study, such as the LST retrieval process, supervised equations. Hence, a model was created to estimate
classification for LULC, Spatial interpolation for LST, containing its intermediate products, e.g.,
Air Quality data, etc. Radiance, Emissivity, and Fractional vegetation
cover. Image classification for land use/land cover
Furthermore, a study has been divided into the is the process of developing interpreted maps
scale of study. Macro level, Meso level, and Micro from remotely sensed images. Consequently,
level, which leads to spatial modeling of LST, classification gives the most important visualization
Air Quality, and LULC. The study is progressed and spatial information for usage pattern of land.
by suggestive mitigation measures for UHI and To derive spatial relation of land use/land cover
Air Quality for the study area. Further geospatial with LST, Air Quality, it is needed to process image
operations and methods are explained below, used classification through satellite imagery for the
for this study. Ground truthing has been done for Ahmedabad region. The Landsat images (2008 to
LST results and analysis of the data gap between 2018) are classified to get the land cover patterns
in-situ measurements. Ground truthing dates are in Ahmedabad. The classes are identified from
selected based upon satellite temporal resolution a range of pixel values, tone variance, visual
of Landsat-8, moreover, the study area for ground- image interpretation of the image, and histogram
truthing has been done based upon the typology peaks. Identified classes for this study are Built-up,
of area, for readings types of features are identified Open Land, Vegetation, and Waterbody.
for temperature measurement.
The three-dimensional interpolation methodis used
Moreover, Field data collection, which is done by to estimate the value of unknown points by known
ground-truthing, gives ground-based LST and Air points, and it is used to predict unknown values by
Quality data. The flow chart shows various spatial existing spatial values. For instance, we have a
processes for study, such as the LST retrieval data variable that has a spatial value at each point
 DAVE et al., Curr. World Environ., Vol. 17(2) 393-409 (2022) 397

(e.g., elevation, temperature, etc.), but the 16 days. The central and West zone is selected for
data variable is not spatially distributed; hence Ground truthing because of building typology, land
it is required to produce a continuous raster surface cover feature, and built-up density. Measurement
that can give the predicted value of other points of temperature for features like the wall, open land,
based upon existing points. road, vegetation, and waterbody has been taken by
IR Gun and PM2.5 and PM10 by Polludrone sensor.
Air Quality, PM2.5, and PM10 data collected from Regression analysisis used here for derivation
GPCB are attribute data brought in point shape files. of co relation of two or more two variables.
Based on available points, it was needed to generate
a surface for the AMC study area so that Air Quality Data Analysis and Interpretation
(PM2.5 and PM10) can also be predicted where Trend Analysis of LST, Air Quality, and LULC was
a collection point is not available. done to get series information with respect to time.
It is a statistical technique that deals with data with
Inverse Distance Weighting is finally used to connect respect to intervals; this interval is a time. In the
all raster and vector database. Ground truthing for case of LST, two seasons, summer and winter of the
temperature and air quality of the study area has year 2008 to 2018, in which 2008, 2015 from 2018
been done for two zones, the Central and West zone are chosen for trend analysis. The calculated LST
of Ahmedabad. The date selectionis made according are divided into ranges for both summer and winter
to the satellite temporal resolution, for Landsat-8, it is seasons to preserve symbology for visualization.

Table 1: Values of Lst In °C for Summer and Winter Seasons

Study year Average Highest Lowest Study year Average Highest Lowest
°C °C °C °C °C °C

2008 40.26 53.85 26.68 2008 25.30 30.43 20.18

2015 30.60 36.98 24.23 2015 21.05 26.78 15.32
2016 35.35 43.95 26.65 2016 24.33 29.66 18.94
2017 41.95 54.41 29.39 2017 25.89 31.24 20.54
2018 35.77 44.16 27.39 2018 22.95 28.06 17.85
For Summer For Winter

Statistics show that there has been a temperature For Summer in 2008 temperature recorded was
rise in the summer and winter seasons in the 40.26 °C; it increased to 41.95°C in 2017, and
year 2017. The mean temperatures for both the 35.77°C temperature was recorded in 2018.
summer and winter seasons of 2008-2018 are 36°C The maps below show temporal changes in LST
and 23°C, respectively. for the Summer season.

Fig. 1
 DAVE et al., Curr. World Environ., Vol. 17(2) 393-409 (2022) 398

Fig. 2

For Winter, In 2008 temperature recorded The maps below show temporal changes in LST
was 25.30°C; it increased to 25.89°C in 2017, for the Winter season.
and 22.95°C temperature was recorded in 2018.
 DAVE et al., Curr. World Environ., Vol. 17(2) 393-409 (2022) 399

Fig. 3

Landcover classes that impact UHI are Built-up, and histogram peaks. Identified classes for this study
Non-Built-up, Vegetation, and Waterbody. Through are Built-up, Open Land, Vegetation, and Waterbody.
the supervised classification, method results are The table below shows the total landcover change
carried out inraster. The classes are identified identified through the generated raster from the
from the range of pixel values, the variance supervised classification method.
of the tone, visual image interpretation of the image,

Table 2: Land Cover Class Area (In [Link])

Landcover class 2008 2015 2016 2017 2018

Urban 163.97 240.19 243.10 248.76 253.96

Vegetation 134.64 120.66 118.55 117.54 116.35
Waterbody 6.71 7.25 8.65 9.65 10.58
Open land 112.13 84.92 82.67 79.65 78.65
 DAVE et al., Curr. World Environ., Vol. 17(2) 393-409 (2022) 400

Fig. 4: The proportion of built up areas wasestimated across the Ahmedabad Municipal
corporation area.

Statistics show that open land has the highest noted in 2008 while it increased in 2018 up to 34.58°C, the
temperature value than other classes, moreover, average temperature of Waterbody was 28.63 °C in
waterbody has the least temperature compared to 2008 while it increased in 2018 up to 31.27°C, the
other land cover class. From 2008 to 2018, it can average temperature of the Urban class was 40.21
be observed that trend of temperature rise has °C in 2008 while it increased in 2018 up to 41.95°C,
increased with time. It can be observed from the the average temperature of open land was 39.78 °C
below table that, for the Winter season, the average in 2008 while it increased in 2018 up to 42.13°C.
temperature of vegetation class was 20.28 °C
in 2008, while it increased in 2018 up to 21.83°C. Raster surface for attribute data of Air Quality (PM2.5
The average temperature of Waterbody was 18.65 and PM10) of Ahmedabad city is derived through
°C in 2008 while it increased in 2018 up to 20.31°C, interpolating variables. For winter and Summer,
the average temperature of the Urban class was a raster surface for PM2.5 and PM 10 is generated
21.01 °C in 2008 while it increased in 2018 up to for both seasons. Data of PM10 was available for
22.32°C, and the average temperature of open land winter and summer from 2008 to 2017, while data
was 22.65 °C in 2008 while it increased in 2018 up of PM2.5 was not available for study for 2008; from
to 22.85°C. For the Summer season, the average 2015 to 2017, data of PM2.5 was taken for study.
temperature of vegetation class was 31.26 °C
 DAVE et al., Curr. World Environ., Vol. 17(2) 393-409 (2022) 401

Table 3: Values Of Lst In °C Per Land Cover Classes

2008 2015 2016 2017 2018 2008 2015 2016 2017 2018

Vegetation 20.28 21.23 22.27 24.62 21.83 Vegetation 31.26 29.75 33.86 38.95 34.58
Waterbody 18.65 19.82 21.06 22.97 20.31 Waterbody 28.63 27.55 30.4 35.85 31.27
Urban 21.01 21.67 22.83 24.59 22.32 Urban 40.21 30.98 35.62 42.73 41.95
Open land 22.65 22.09 23.32 25.31 22.85 Open land 39.78 32.25 37.67 44.71 42.13
Average LST in Landcover Classes for Winter Average LST in Landcover Classes for Summer

Table 4: Pm2.5 and Pm10 Statistics for Winter and Summer Seasons

Study PM10 PM2.5 Study Year PM10 PM2.5

Year concentration concentration concentration concentration
(μg/m3) (μg/m3) (μg/m3) (μg/m3)

2008 108.52 N.A. 2008 102.75 N.A.

2015 84.04 30.50 2015 96.41 37.74
2016 110.41 51.44 2016 102.27 30.45
2017 125.31 64.29 2017 116.15 41.53
For Winter For Summer

From the above table, it can be observed that, for 2008 was 102.75 μg/m3 and raised in 2017 upto
the winter season, PM10 in 2008 was 108.52 μg/ 116.15μg/m3; PM2.5 in 2015 was found at 37.74 μg/
m3 and raised in 2017 upto 125.31μg/m3, PM2.5 m3 and increased to 41.53 μg/m3. It is noticeable
in 2015 was found to 30.50 μg/m3 and increased that the trend of PM10 and PM2.5 from 2008 to
64.29 μg/m3. For the Summer season, PM10 in 2017 is increasing.

Fig. 5: Interpolated maps of PM10 for Winter season (values are in μg/m3)
 DAVE et al., Curr. World Environ., Vol. 17(2) 393-409 (2022) 402

Fig. 6: Interpolated maps of PM2.5 for Winter season (values are in μg/m3)

Fig. 7: Interpolated maps of PM10 for Summer season (values are in μg/m3)
 DAVE et al., Curr. World Environ., Vol. 17(2) 393-409 (2022) 403

Fig. 8: Interpolated maps of PM2.5 for Summer season (values are in μg/m3)

To generate fishnet for the AMC region, spatial cell values of LST, LULC, and Air Quality (PM2.5 and
extend is considered per AMC boundary, and grid PM10) is to analyze the relationship of these layers
size is taken 120*120m. Landsat spatial resolution within each cell. All values falling under the AMC
is 30m; thus, to get the overall average scenario region have been transferred using Zonal statistics
of LST, Air Quality, and LULC in grid level, the to grids of 120*120m. The attribute table of the grid
smallest grid size containing 16 pixels is 120*120m. after using zonal statistics is shown below. Total rows
The total number of cells generated is 47,073. in gird’s attribute table are 47,073.

The simple linear regression model uses

the following equation with independent and
dependant values.

Y= a + bX + ϵ

Where:y – dependent variable, x – independent

variable, a – intercept, b – slope, ϵ – residual/error

In this study, there are four variables, LST, PM2.5,

PM10, and LULC. In order to perform regression
analysis, the Multilinear regression method
has been used. To predict LST, it is taken as
a dependent variable, while LULC classes are taken
Fig. 9
as independent variables.
Zonal statistics is a tool that estimates cell statistics
or a value of a raster layer by a defined zone. Regression Statistics derived were Multiple
To transfer the value of LST, LULC, and Air Quality corelation coefficient 0.809074773, Coefficient
(PM2.5 and PM10). The idea of transferring of determination 0.654601988, Adjusted
 DAVE et al., Curr. World Environ., Vol. 17(2) 393-409 (2022) 404

Coefficient of determination 0.654558728,Standard examined. The simple linear regression model use
Error 9.022893242 for Observations 40773. following equation.

The value of R 2 is 0.65, which is a positive Y= a + bX + ϵ

correlation, which means the independent variable
explains 65.45% of the variance in the dependent Where: y – dependent variable, x – independent
variable. variable, a – intercept, b – slope, ϵ – residual/error

The multiple linear regression is expressed here as, The simple linear regression for the dependent
LST= 32.86 + 0.09 * UrbanP + 0.13 * OpenP + variable PM2.5, PM10 with independent variable
0.038 * VegP LST is examined for the Summer and Winter
seasons. Results are shown below.
W h e r e , L S T = L a n d S u r f a c e Te m p e r a t u r e ,
32.86= Intercept, UrbanP= Percentage For Summer, as shown below in the graph,
of Urban, OpenP= Percentage of Open land, the Dependent variable is LST, and the independent
VegP= Percentage of Vegetation variable is PM2.5. Regression Statistics derived
were Multiple corelation coefficient 0.975516681,
In the equation, it was observed that WaterP Coefficient of determination 0.951632794,
(Percentage of Water), coefficient value coming Adjusted Coefficient of determination 0.95161155.
null, the fact that only 10% of the total land-use
area of AMC having Landcover class of Waterbody.
It was attempted to select only grids with waterbody
as landcover class. Thus, grids with a water area
percentage of more than 50% were selected through
a query, and 679 grids were selected. For these
679 grids, multilinear regression was computed,
and the value of R2 computed at 0.314, which is a
negative correlation.

Furthermore, it was expanded to a multi-linear

regression equation derived before, which has no
significance due to data insufficiency of waterbody Fig. 10
class. Regression Statistics derived were, Multiple
Now for the Dependent variable as shown below
corelation coefficient 0.560449, Coefficient
in graph: LST and Independent variable: PM10,
of determination 0.314103, Adjusted Coefficient
Regression Statistics derived were Multiple
of determination 0.310388, Standard Error 8.065898
corelation coefficient 0.97882, Coefficient
Observations 679. Thus, the Multilinear regression
of determination 0.958088, Adjusted Coefficient
equation with the extracted value of WaterP comes,
of determination 0.958067.
LST= 32.86 + 0.09 * UrbanP + 0.13 * OpenP + 0.038
* VegP -0.035 WaterP

W h e r e , L S T = L a n d S u r f a c e Te m p e r a t u r e ,
32.86= Intercept, UrbanP= Percentage
of Urban, OpenP= Percentage of Open land,
VegP= Percentage of Vegetation, WaterP=Percentage
of Water

In simple linear regression, relationship between

dependent variable and independent variable is
Fig. 11
 DAVE et al., Curr. World Environ., Vol. 17(2) 393-409 (2022) 405

Regression statistics show a strong positive to be considered because it is also correlated with
correlation between PM2.5, PM10 and LST with an Urban Air Quality, and it may affect human health,
R2 value of 0.95. human comfort, and the environment. Thus, some
of these mitigation measures which are being
For the Winter season, the Dependent in common practice can be followed as follows:
variable: LST, Independent variable: PM2.5,
and Regression Statistics derived were Multiple Since vegetation is negatively correlated with
corelation coefficient 0.984914728, Coefficient LST, an increment in vegetation area by an act
of determination 0.970057021, Adjusted Coefficient of tree plantation, gardening, or the concept
of determination 0.970035777. of green building can reduce the intensity of UHI
on the city scale.

The surface albedo of material is negatively

correlated with LST; thus, the practice of white
roofing and material with high solar reflectance
can be used as rooftop pavement since they
have less solar radiation absorption and higher
surface albedo values.

In this study, there are four variables, LST,

PM2.5, PM10, and LULC. In order to perform
regression analysis, the Multilinear regression
Fig. 12
method has been used. To predict LST, it is taken
Now for, Dependent variable: LST, Independent as a dependent variable, while LULC classes
variable: PM10, Regression Statistics derived were are taken as independent variables. Regression
Multiple corelation coefficient 0.993962, Coefficient Statistics derived were Multiple corelation coefficient
of determination 0.98796, Adjusted Coefficient 0.680257033, Coefficient of determination
of determination 0.987938. 0.462749631, Adjusted Coefficient of determination
0.462737156, Standard Error 2.022893242
for observations 209352.

The value of R 2 is 0.46, which is a positive

correlation, which means the independent
variable explains 46.27% of the variance in the
dependent variable.

The multiple linear regression is derived here as,

LST= 34.366 + 0.079 * UrbanP + 0.101 * OpenP
+ 0.062 * VegP
Fig. 13
By analyzing both results, it is seen that in winter Where,LST= Land Surface Temperature, 34.366=
correlation between PM2.5 and PM10 with LST is Intercept, UrbanP= Percentage of Urban OpenP=
stronger than Summer season. The reason could be Percentage of Open land, VegP= Percentage
a suspended particle-like PM2.5 and PM10 having of Vegetation
a higher concentration on the ground in the Winter
season than in summer due to low temperature. The simple linear regression for dependent variable
PM2.5, PM10 with independent variable LST
After primary data analysis, it is found that Urban is examined for 30m grids. Where, Dependent
Heat Island phenomenon is a problem and needs variable: LST and Independent variable: PM2.5;
 DAVE et al., Curr. World Environ., Vol. 17(2) 393-409 (2022) 406

Regression Statistics derived were Multiple of determination 0.000165328, Standard

corelation coefficient 0.013042395, Coefficient Error 9.599812854 for observations 209352.
of determination 0.000170104, Adjusted Coefficient

Fig. 14

Now for the Dependent variable: LST and Independent Coefficient of determination 0.00398107,
variable: PM10; Regression Statistics derived Adjusted Coefficient of determination 0.00397631,
were Multiple corelation coefficient 0.06309568, Standard Error 23.0820537, Observations 209352

Fig. 15

By regression statistics, we have already derived 30m (Landsat 8 resolution). Total 209352 polygon
a model equation to predict LST; The multiple linear rectangular grids are generated with a spatial extent
regression is derived here as, of the central and west zone of Ahmedabad.
LST= 34.366 + 0.079 * UrbanP + 0.101 *
OpenP + 0.062 * VegP As a part of the mitigation strategy, there are two
assumptions in which green cover area is increased
Central and West Zone of Ahmedabad is divided and proposed with 30% and 40% of the total area
into grids of 30m which is the smallest grid size of these zones. The predicted LST with proposed
for study and can have an as smallest pixel size Vegetation area cover is as below:
of the generated raster of LST, and Land use is

Table 5

VegP UrbanP OpenP Existing LST Predicted LST

30% 50% 20% 42.67 41.66

40% 55% 5% 42.67 41.11
 DAVE et al., Curr. World Environ., Vol. 17(2) 393-409 (2022) 407

It can be observed that there has been 1 °C Regression equation for PM10 and PM2.5 has
temperature decrement in both of these scenarios been derived as,
when green cover is made up 30% and 40%. PM10 = 0.5288 * LST + 83.856
PM2.5= -0.0454 * LST + 39.871
Furthermore, PM2.5 and PM10 have been modeled
by regression statistics, so the value of Predicted If the LST value is replaced with predicted LST,
LST can be correlated to Predict PM2.5 and PM10 then predicted PM10 and PM2.5 can be derived
to get the result as a part of a mitigation strategy in to estimate the concentration of PM2.5 and PM10.
which 30% and 40% Vegetation cover is proposed.

Table 6

Predicted PM2.5 Predicted PM 10 Existing PM2.5 Existing PM10 LST (°C)

37.97 105.88 54.81 156.07 41.66

38.00 105.62 54.81 156.07 41.11

The regression model equation estimates the study, 30% and 40% roofs are converted to white
value of PM 2.5 and PM 10. It is noticeable that the roof for Central Zone of AMC as a mitigation strategy.
predicted value of PM10 and PM2.5 is 51 mg/m3 In Central Zone, 30% and 40% area of the total
and 17 mg/m3 less than existing values. rooftops are selected so that the raster value
of emissivity can be changed with the value of white
Central Zone of AMC has a dense typology roof emissivity, which has higher surface reflectivity.
of building with concrete and metal roof which has It leads to a new LST product because the emissivity
low emissivity that absorbs heat, and due to this of the raster has been modified.
phenomenon, the temperature rise at the local
climatic level can be observed. As a white roof It is noticeable that the area converted 40% under
implementation strategy, rooftops with concrete and the white roof has a temperature reduction of 1.73°C,
metal roofs are replaced with white roofs, which while the area converted 40% under the white
affects the local temperature decrement. For this roof has a temperature reduction of 1.31°C.

Table 7

% of White Roof in Area LST Predicted PM10 Predicted PM2.5

30% 38.78 104.36 38.10

40% 38.04 103.63 37.47

Conclusion The land cover maps compared with temperature

To conclude a study with supportive statistical show that where there is less vegetation,
investigations and results, it can be asserted that the temperature was high, and the areas with
Ahmedabad city has been facing both urban heat vegetation patches show less temperature
island phenomena and urban air quality deterioration. than other land use classes.

It is observed that seasonal variation affects the The correlation analysis was done to model LST,
formation of Urban Heat Island by accessing decadal PM2.5, PM10, and LULC. Which shows the positive
data of UHI of the Winter and summermonths. correlation between land use classes and LST.
From 2008 to 2018 gives an idea that there was an Moreover, it has been found that there is a strong
increment in the temperature and air pollution profile. relationship between PM2.5, PM 10, and LST.
 DAVE et al., Curr. World Environ., Vol. 17(2) 393-409 (2022) 408

However, it was observed that in the Winter season, positive impacts on particulate matter concentration
the relationship between PM2.5, PM10, and LST reduction of the study area and recommends both
is stronger than in the summer season. mitigation strategies not only to reduce Urban
Heat Island Impact but to reduce particuate matter
Furthermore, for mitigation strategy, it was concentrations as well. Use of historical data and
considered White Roof mitigation strategy and remote sensing techniques can be incorporated
Green area cover. It is noticeable that both of these into urban design and planning for resolution
mitigation strategies are affecting well to decrease of complex urban climatic issues of air pollution and
temperature rise in Ahmedabad, It is noticeable heat island impacts.
that the area converted 40% under the white roof
has a temperature reduction of 1.73°C, while the Acknowledgement
area converted 30% under the white roof has The author would like to thank, Environment
a temperature reduction of 1.31°[Link] a part of the Engineering, Gujarat Technological University,
green cover increase mitigation strategy, when Ahmedabad, India. for their guidance and support
green cover area is increased and proposed with to complete this article.
30% and 40% of the total zone area there has been
1 °C temperature decrement. Both the strategies Funding
also contributes inreduction in the concentration The author(s) received no financial support
of particulate matter. for the research, authorship, and/or publication
of this article.
The study attempts to establish correlation between
the temperature decrease with use of White Conflict of Interest
roof and green cover increment strategy and their The authors do not have any conflict of interest.

References

1. Devadas, M. D., & Amirtham, L. R. (2014). Urban Engineering Complexity, 3430. https://
Urban Factors and the intensity of Heat Island [Link]/10.1080/15693430412331291698
in the City of Chennai Urban Factors And The 6. Heat, U., & Basics, I. (n.d.). Reducing Urban
Intensity Of Heat Island, (June), 2–6. Heat Islands : Compendium of Strategies
2. Estoque, R. C., Murayama, Y., & Myint, S. Urban Heat Island Basics.
W. (2017). Science of the Total Environment 7. Imhoff, M. L., Zhang, P., Wolfe, R. E.,
Effects of landscape composition and pattern & Bounoua, L. (2010). Remote Sensing
on land surface temperature : An urban heat of Environment Remote sensing of the
island study in the megacities of Southeast urban heat island effect across biomes in
Asia, 577, 349–359. the continental USA, 114, 504–513. https://
3. Fortuniak, K. (n.d.). Global environmental [Link]/10.1016/[Link].2009.10.008
change and urban climate in Central 8. Kotharkar, R., Ramesh, A., Bagade, A.,
European cities. & Asia, S. (2018). Urban Climate Urban Heat
4. Gill, S., Forest, T. M., & Ennos, R. (2007). Island studies in South Asia : A critical review.
Adapting Cities for Climate Change : The Urban Climate, (July 2017), 1–16. [Link]
Role of the Green Infrastructure, (April 2014). org/10.1016/[Link].2017.12.006
[Link] 9. Li, H., Meier, F., Lee, X., Chakraborty, T.,
5. Golden, J. S., & Golden, J. S. (2010). & Liu, J. (2018). Interaction between urban
The Built Environment Induced Urban Heat heat island and urban pollution island
Island Effect in Rapidly Urbanizing Arid during summer in Berlin Science of the
Regions – A Sustainable Urban Engineering Total Environment Interaction between
Complexity Island Effect in Rapidly urban heat island and urban pollution island
Urbanizing Arid Regions – A Sustainable during summer in Berlin, (May). [Link]
 DAVE et al., Curr. World Environ., Vol. 17(2) 393-409 (2022) 409

org/10.1016/[Link].2018.04.254 37–41. [Link]

10. Li, Z., Tang, B., Wu, H., Ren, H., Yan, G., 07.9687275
Wan, Z. Sobrino, J. A. (2013). Remote 13. UN-Habitat. (2007). World Urbanization
Sensing of Environment Satellite-derived Prospects The 2007 Revision Highlights.
land surface temperature : Current status and New York. [Link]
perspectives, 131, 14–37. 14. H. Taha, " Meteorological, air-quality, and
11. Liu, W., You, H., & Dou, J. (2009). emission-equivalence impacts of urban heat
Urban-rural humidity and temperature island control in California," Sustainable
differences in the Beijing area, 201–207. Cities and Society 19 (2015) 207–221, 2015.
[Link] 15. J Fallman, “Urban Heat Island Versus Air
12. Taylor, P., Stathopoulou, M., & Cartalis, C. Quality - a Numerical Modelling Study for
(n.d.). Use of Satellite Remote Sensing a European City”(2015)
in Support of Urban Heat Island Studies 16. Roozbeh Arabi, "Mitigating Urban Heat Island
Use of Satellite Remote Sensing in Support Through Green Roofs"(2015). [Link]
of Urban Heat Island Studies, (January 2015), org/10.12944/[Link]-Issue1.111