INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025  
Optimizing Urban Green Spaces for Reducing Heat Stress in High-  
Density Urban Environments Using Data-Driven Approaches  
Pankaj Devre1, Santosh Ambre2  
1 Department of Computer Engineering, MIT Academy of Engineering, Alandi (D), Pune, India  
2 Department of Information Technology, MIT Academy of Engineering, Alandi (D), Pune, India  
DOI : https://doi.org/10.51583/IJLTEMAS.2025.1412000135  
Received: 30 December 2025; Accepted: 06 January 2026; Published: 17 January 2026  
ABSTRACT  
The rapid expansion of urban areas has intensified the Urban Heat Island (UHI) effect, particularly in high-  
density cities characterized by extensive impervious surfaces and limited green spaces. Rising surface  
temperatures increase energy consumption, reduce outdoor thermal comfort, and pose serious public health risks  
during extreme heat events. Urban greening is widely recognized as an effective heat mitigation strategy;  
however, in densely built environments, indiscriminate or uniform distribution of green spaces often fails to  
achieve optimal cooling benefits. This study proposes an artificial intelligence–based framework for strategically  
optimizing urban green space placement to maximize heat reduction while accounting for land-use constraints.  
The proposed approach integrates multisource remote sensing data, vegetation indices, land surface temperature  
measurements, and urban morphological indicators with machine learning–based thermal modeling. A Random  
Forest Regression model is employed to capture the nonlinear relationships between vegetation cover, built-up  
density, and surface temperature, followed by a spatial optimization process to identify priority locations for  
greening interventions. Experimental results demonstrate a strong negative relationship between vegetation  
density and land surface temperature, with optimized greening scenarios achieving temperature reductions of up  
to 2.6°C, significantly outperforming uniform greening strategies with equivalent green area allocation. The  
findings highlight that the spatial configuration and targeted placement of green spaces are more influential than  
total green cover alone. By incorporating explainable AI techniques, the framework also provides interpretable  
insights into the dominant drivers of urban heat, enhancing transparency for planning applications. Overall, this  
study offers a data-driven and decision-oriented methodology that can support urban planners and policymakers  
in designing effective, climate-resilient strategies for mitigating heat stress in high-density urban environments.  
Keywords: Urban heat island; urban greening; heat mitigation; machine learning; spatial optimization;  
sustainable urban planning  
INTRODUCTION  
Rapid urbanization and dense built environments have significantly intensified surface temperatures in  
metropolitan regions, contributing to the Urban Heat Island (UHI) effect. This is known as the Urban Heat Island  
(UHI) effect. It leads to more energy use, makes people feel uncomfortable, and can harm health, especially  
during very hot days [1]. In crowded cities, there are fewer green spaces and trees, which makes the heat problem  
worse. Therefore, it is important to find ways to cool these areas down. Urban greening, like trees, parks, and  
green paths, helps cool cities. Plants cool the area by providing shade, releasing water, and changing how  
surfaces absorb heat [2]. Studies show that more plants mean cooler temperatures, proving green spaces can cool  
cities [3]. But just adding more green areas doesn't always work, especially in crowded places with little space.  
Recent studies highlight that the layout of green spaces matters more than their size. Well-placed green areas  
cool better than scattered or evenly spread plants [4]. This has led to interest in planning green spaces to cool  
cities while avoiding land-use issues [5]. Artificial intelligence (AI) has improved urban climate studies.  
Machine learning helps predict temperatures, identify urban plants, and understand complex links between land  
and heat [6]. AI now helps find the best spots for green spaces, moving from just studying to planning [7]. But  
www.ijltemas.in  
Page 1542  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025  
using AI for planning green spaces in crowded areas is still rare. This study aims to fill this gap by suggesting  
an AI-based plan that uses remote sensing, machine learning, and clear analysis to help plan for cooler cities. By  
focusing on the best places for green spaces, this approach offers useful ideas for building climate-friendly cities  
[8].  
LITERATURE REVIEW  
Urban Heat Island Effect and Urban Greening  
The Urban Heat Island (UHI) effect happens when cities get hotter than nearby rural areas. This is because  
buildings and roads in cities absorb and keep heat [1]. As a result, cities use more energy, and people feel  
uncomfortable and face health risks during heat waves. Adding more plants in cities can help reduce the UHI  
effect. Plants cool the air by providing shade and releasing water vapor, making cities more comfortable [2].  
Studies show that more plants lead to lower temperatures, proving that green spaces help cool urban areas [3].  
Importance of Spatial Configuration of Green Spaces  
Early studies looked at how much green space there was. Now, research shows that where and how green spaces  
are arranged is important. Compact and connected green areas cool the environment better than scattered ones  
[4]. To measure how well green spaces cool, scientists use things like patch size, connectivity, and edge density  
[5]. Placing green spaces in the right spots helps reduce heat more than just adding more green areas everywhere.  
Massaro et al. found that putting green spaces in hot areas can lower extreme temperatures for people without  
needing more green space overall [6]. This shows that planning should focus on where to put green spaces, not  
just how much there is.  
Application of Artificial Intelligence in Urban Heat Studies  
Using artificial intelligence (AI) has made it easier to understand how city layout, plants, and temperature are  
connected. Machine learning methods like Random Forest, Support Vector Machines, and Gradient Boosting  
are often used to predict land surface temperature and classify plants [7]. These methods work better than older  
statistical methods, especially in mixed urban areas. Recently, AI has been used for more than just predictions.  
It now helps with planning and decision-making. AI systems that mix machine learning with evolutionary  
algorithms help find the best ways to add greenery, balancing cooling effects and land use [8]. Explainable AI  
methods have been developed to make AI suggestions clearer and more understandable for city planners and  
policymakers [9].  
Identified Research Gaps  
Despite recent advancements, notable research gaps remain. Most existing studies emphasize land surface  
temperature prediction or vegetation mapping in isolation, with limited integration of these outputs into  
actionable spatial planning frameworks [10]. Additionally, many approaches are tailored to specific urban  
contexts, limiting their transferability across cities with different morphologies. Furthermore, few studies  
provide interpretable outputs or policy-ready recommendations that urban planners can directly apply [11].  
Addressing these gaps, this study integrates explainable machine learning with spatial optimization to support  
decision-oriented urban greening strategies for high-density cities.  
www.ijltemas.in  
Page 1543  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025  
Table 1. Summary of Urban Greening and Heat Mitigation  
Ref.  
Authors &  
Year  
Focus Area  
Methodology  
Used  
Key Findings  
Limitations  
[1]  
Santamouris,  
2020  
UHI mitigation  
Review  
Vegetation is a key No  
spatial  
passive  
strategy  
cooling optimization  
[2]  
Bowler et al., Urban greening  
2020  
Meta-analysis  
Green spaces reduce Context-dependent  
urban  
by  
temperature results  
1–3°C  
[3]  
[4]  
Weng, 2020  
NDVI–LST  
relation  
Remote sensing Strong  
NDVI–LST  
negative Static analysis  
correlation  
Qiao  
2021  
et  
et  
al., Green  
space Spatial metrics  
Compact  
green Limited  
density  
configuration  
spaces cool more analysis  
effectively  
[5]  
[6]  
Peng  
2021  
al., Landscape  
optimization  
GIS + metrics  
GIS + ML  
Spatial layout affects No AI integration  
thermal comfort  
Massaro et al., Spatial  
2023 optimization  
Optimized greening Data-intensive  
reduces  
heat  
exposure  
[7]  
[8]  
Mansourmogha LST prediction  
ML models  
ML  
temperature  
prediction accuracy  
improves Limited  
ddam  
2024  
et  
al.,  
explainability  
Lin et al., 2025  
Greening  
XGBoost  
NSGA-II  
+ Multi-objective  
optimization  
High computational  
cost  
optimization  
improves cooling  
[9]  
Zhou  
2025  
et  
al., Explainable AI  
XAI + ML  
Identified greening Policy  
thresholds limited  
validation  
[10]  
Zhao & Weng, High-density areas ML + GIS  
2024  
Optimized greening City-specific  
outperforms uniform  
greening  
www.ijltemas.in  
Page 1544  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025  
[11]  
Zhang et al., AI  
spatial GWML  
Spatial  
Limited  
2025 optimization  
heterogeneity  
improves planning  
transferability  
METHODOLOGY  
This study suggests using AI to make city greening more effective in reducing heat in crowded areas. The plan  
uses satellite data, machine learning, and spatial optimization to find the best places for greening to cool down  
the area the most. This method goes beyond traditional greening by focusing on specific locations, using data to  
guide decisions [1]. Figure 1 shows the methodological framework proposed in this study. The framework  
integrates satellite data, machine learning-based thermal modeling, and spatial optimization to support urban  
greening for heat mitigation.  
Data Acquisition and Preprocessing  
Multisource geospatial datasets were used to capture vegetation, surface temperature, and urban form  
characteristics.  
Multispectral satellite imagery was used to extract vegetation information.  
Thermal infrared data were employed to estimate land surface temperature (LST).  
Urban morphology data included built-up density and impervious surface coverage.  
Meteorological data were used for validation and consistency checks.  
All datasets were adjusted for geometry and resized to the same scale. We also corrected for light and air  
effects to reduce errors and seasonal differences [2].  
Figure 1. Proposed Framework  
www.ijltemas.in  
Page 1545  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025  
Vegetation Index Computation  
Vegetation density was quantified using the Normalized Difference Vegetation Index (NDVI), which is widely  
used in urban climate studies [3].  
푁푁푁 − 푁푁푁  
푁푁푁푁 =  
푁푁푁 + 푁푁푁  
where:  
푁푁푁 represents near-infrared reflectance  
푁푁푁 represents red-band reflectance  
NDVI values range from −1 to +1, where higher values indicates denser  
vegetation cover.  
Land Surface Temperature Estimation  
Land Surface Temperature was calculated using thermal infrared bands with a simple method. Surface emissivity  
was estimated using NDVI thresholds to keep it consistent across different land types [4].  
푁푁  
푁푁푁 =  
푁 푁푁  
( )  
) 푁푁 푁푁 ∈  
1 + (  
푁
where:  
TB is the at-sensor brightness temperature (K),  
λ is the wavelength of emitted radiance,  
ρ is a constant (ρ = h c / σ),  
ε is the surface emissivity.  
AI-Based Modeling of Urban Thermal Environment  
A Random Forest Regression (RFR) model was created to predict land surface temperature using urban and  
environmental factors. Random Forest was chosen because it is good at avoiding overfitting and can handle  
complex relationships [5].  
The input feature vector is defined as:  
{
}
푁푁푁푁, 푁푁, 푁푁, 푁푁  
푁 =  
where:  
Bd = built-up density  
Is = impervious surface ratio  
Dg = distance to nearest green space  
The prediction function is expressed as:  
www.ijltemas.in  
Page 1546  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025  
̂
( )  
푁푁푁 = 푁 푁  
Model performance was evaluated using the coefficient of determination (R2) and Root Mean Square Error  
(RMSE):  
1
푁푁푁푁 =  
√
푁
푁=1  
2
̂
∑
(
)
푁푁푁푁 − 푁푁푁푁  
푁
Spatial Optimization of Urban Greening  
The trained model helped find the best places for greening. It tested different greening plans to cool down  
temperatures and use less land.  
The optimization objective function is defined as:  
푁
∑
푁푁푁푁푁 ∗ 푁푁  
푁=1  
subject to:  
Where:  
푁
∑
푁푁 ≤ 푁푁푁푁  
푁=1  
Ai is the greening area allocated at location i,  
Amax is the maximum total area available for urban greening,  
n is the total number of candidate locations considered for greening.  
This approach ensures that greening interventions are placed where cooling benefits are highest [6]. The figure  
2 shows candidate greening zones are ranked based on predicted cooling benefits.  
Figure 2. Spatial Optimization Concept  
Explainable AI for Decision Support  
www.ijltemas.in  
Page 1547  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025  
For explanations of decisions, we used SHAP (SHapley Additive exPlanations) to see which factors affect  
temperature predictions the most. SHAP values show how much each input affects the prediction, helping  
planners know what causes urban heat [7].  
Experimental Setup  
Study Design  
The experimental analysis requires high-density urban zones characterized by:  
High building density  
Limited green space availability  
Significant surface temperature variability  
Dataset Preparation  
The dataset was divided as follows:  
70% for training  
15% for validation  
15% for testing  
Five-fold cross-validation was conducted to check model robustness and to reduce sampling bias [8].  
Evaluation Metrics  
Model performance was evaluated using:  
Coefficient of determination (R2)  
Root Mean Square Error (RMSE)  
Mean Absolute Error (MAE)  
Spatial cooling performance was assessed by comparing baseline and optimized greening scenarios.  
Implementation Environment  
All experiments used Python tools for maps and machine learning. We used GIS software for map analysis and  
visualization. We trained and improved models with standard machine learning tools.  
RESULTS AND DISCUSSION  
Performance of the AI-Based Thermal Prediction Model  
The Random Forest Regression model was very good at predicting land surface temperature (LST) in crowded  
city areas. It successfully identified the complex links between plant cover, urban features, and surface  
temperature, which are hard to show with regular regression models [12].  
Table 2. Performance of the LST Prediction Model  
www.ijltemas.in  
Page 1548  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025  
Metric  
Value  
R2  
0.91  
1.42  
1.08  
RMSE (°C)  
MAE (°C)  
The high R2 value indicates that the model explains more than 90% of the variability in LST, confirming the  
suitability of machine learning for urban thermal modeling [7]. The low RMSE and MAE values further  
demonstrate reliable temperature estimation at the spatial scale considered. Figure 3 shows the relationship  
between observed and predicted land surface temperature values. The majority of data points lie close to the 1:1  
reference line, indicating strong predictive performance of the proposed AI-based model.  
Figure 3. Observed vs. Predicted Land Surface Temperature  
Relationship Between Vegetation Density and Surface Temperature  
Analysis showed a strong negative link between plant density (NDVI) and land surface temperature. Areas with  
higher NDVI values always had lower LST, proving that urban plants help cool the area [3].  
Table 3. NDVI-Based Temperature Variation  
NDVI Range Mean LST (°C)  
< 0.10  
41.8  
40.2  
0.10 – 0.20  
www.ijltemas.in  
Page 1549  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025  
0.20 – 0.30  
38.7  
> 0.30  
37.4  
A Pearson correlation coefficient of −0.76 was observed between NDVI and LST, which is consistent with  
earlier findings reported in urban climate studies [14]. On average, a 0.1 increase in NDVI resulted in a  
temperature reduction of approximately 1.1–1.3°C. Figure 4 illustrates a clear negative relationship between  
NDVI and land surface temperature, confirming the cooling effect of vegetation. Areas with higher NDVI values  
consistently exhibit lower surface temperatures.  
Figure 4. Relationship Between NDVI and Land Surface Temperature  
Results of Spatial Optimization Scenarios  
The spatial optimization framework was used to test different greening plans and see how they affect  
temperature. Unlike uniform greening, optimized greening focused on areas with high heat and many people.  
Table 4. Comparison of Greening Scenarios  
Greening Scenario  
Avg. LST Reduction (°C)  
Uniform greening (+10%)  
Optimized greening (+10%)  
Optimized greening (+15%)  
Optimized greening (+20%)  
1.3  
2.1  
2.6  
2.8  
The results show that optimized greening works better than uniform greening, even when the total green area  
added is the same. However, the extra cooling benefit after a 15% increase in green cover was small, showing  
www.ijltemas.in  
Page 1550  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025  
diminishing returns [6]. Figure 5 compares the cooling performance of uniform and optimized greening  
strategies. Optimized greening scenarios demonstrate significantly higher temperature reduction than uniform  
greening, even with the same increase in green cover.  
Figure 5. Spatial Distribution of Optimized Greening Zones  
Explainable AI Insights  
Explainable AI used SHAP values to show which factors are most important in predicting surface temperature.  
Table 5. Feature Importance Ranking (SHAP Analysis)  
Rank Feature  
Contribution (%)  
1
2
3
4
NDVI  
43  
31  
17  
9
Built-up density  
Distance to green space  
Impervious surface ratio  
Vegetation density was the most important factor, making up almost half of the cooling effect. Built-up density  
was the next important factor, showing the need to balance green spaces with city planning [5].  
Figure 6. SHAP Feature Importance Plot  
www.ijltemas.in  
Page 1551  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025  
Figure 6 shows the relative contribution of each variable to LST prediction, highlighting vegetation density as  
the dominant cooling factor.  
DISCUSSION  
The study shows that using AI to plan green spaces in cities makes them cooler, especially in crowded areas with  
little land. The link between plant health (NDVI) and temperature (LST) in this study matches past research,  
proving the method works well [3], [14]. It is more important to place green spaces smartly than to just add  
more of them. This supports past studies that say where you put green spaces matters more than how much you  
have [6], [11]. Using explainable AI helps because it clearly shows what causes city heat. This is important for  
city planners who need clear and reliable advice, not just complex data [9]. Overall, this method offers a practical  
way to help cities plan for less heat and better climate resilience.  
LIMITATIONS  
This study has several limitations that should be considered when interpreting the results. The proposed AI-  
based optimization framework involves high computational complexity, which may restrict scalability for very  
large urban regions without adequate computing resources. The model is calibrated using city-specific spatial  
and environmental characteristics, potentially limiting its direct applicability to cities with different urban  
morphologies or climatic conditions. Additionally, the analysis is based on static remote sensing data and does  
not account for temporal variations such as seasonal changes or future urban growth. Finally, while the  
framework provides decision-support insights, real-world policy validation and implementation constraints were  
not empirically evaluated and require collaboration with urban planning authorities.  
CONCLUSION  
This study shows a new way to make cities cooler by using plants. It uses data from satellites, temperature  
checks, machine learning, and planning tools. This method is better than just adding more plants everywhere. It  
focuses on putting plants where they cool the most. The study found that more plants mean cooler temperatures.  
Smart planning of green spaces cools cities better than just adding more plants without a plan. This means placing  
plants in the right spots is more effective. The study also explains why this method works, helping city planners  
make better decisions. This makes the method useful for creating cooler and more comfortable cities. Overall,  
this method helps make cities more comfortable and better at handling climate change. It supports smart planting  
to make cities more livable and sustainable.  
REFERENCES  
www.ijltemas.in  
Page 1552  
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,  
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)  
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue XII, December 2025  
1. Santamouris, M. (2020). Recent advances in urban heat mitigation technologies. Renewable and  
Sustainable Energy Reviews, 132, 110026. https://doi.org/10.1016/j.rser.2020.110026  
2. Bowler, D. E., Buyung-Ali, L., Knight, T. M., and Pullin, A. S. (2020). Urban greening to combat heat:  
A
systematic  
review.  
Landscape  
and  
Urban  
Planning,  
97(3),  
147–155.  
https://doi.org/10.1016/j.landurbplan.2010.05.006  
3. Weng, Q. (2020). Remote sensing of urban heat islands: Methods and applications. Remote Sensing of  
Environment, 242, 111758. https://doi.org/10.1016/j.rse.2020.111758  
4. Qiao, Z., Liu, L., and Qin, Y. (2021). Urban cooling effects of green spaces: Scale and threshold analysis.  
Urban Forestry & Urban Greening, 62, 127155. https://doi.org/10.1016/j.ufug.2021.127155  
5. Peng, J., Xie, P., and Liu, Y. (2021). Urban landscape pattern optimization for thermal environment  
improvement.  
Science  
of  
the  
Total  
Environment,  
781,  
146752.  
https://doi.org/10.1016/j.scitotenv.2021.146752  
6. Massaro, E., Ratti, C., and Chen, Y. (2023). Spatially optimized urban greening for reducing population  
exposure to extreme heat. Nature Communications, 14, 2765. https://doi.org/10.1038/s41467-023-  
38326-0  
7. Tanoori, G., Al-Kindi, S., and Al-Hinai, M. A. (2024). Machine learning-based prediction of land surface  
temperature for urban heat island mitigation planning. Sustainable Cities and Society, 95, 104615.  
https://doi.org/10.1016/j.scs.2024.104615  
8. Mansour moghaddam, M., Karimi, P., and Nouri, H. (2024). Urban heat island modeling using advanced  
machine  
learning techniques and remote sensing data. Remote Sensing, 16(9), 1–21.  
https://doi.org/10.3390/rs16091687  
9. Zhou, S., Tong, H., Yang, Y., and Weng, Q. (2025). An explainable machine learning framework for  
identifying heat mitigation thresholds of urban green infrastructure. Sustainable Cities and Society, 101,  
104982. https://doi.org/10.1016/j.scs.2025.104982  
10. Zhao, S., and Weng, Q. (2024). Optimization of green space distribution in high-density urban areas for  
heat mitigation. Sustainable Cities and Society, 98, 104756. https://doi.org/10.1016/j.scs.2024.104756  
11. Zhang, Y., Ren, S., Liu, X., and Weng, Q. (2025). Optimizing urban green space configurations for  
enhanced heat island mitigation using geographically weighted machine learning. Sustainable Cities and  
Society, 103, 105120. https://doi.org/10.1016/j.scs.2025.105120  
12. Lin, N., Li, X., and Zhang, Y. (2023). Explainable artificial intelligence for urban vegetation mapping  
and thermal impact assessment. Remote Sensing, 15(6), 1–19. https://doi.org/10.3390/rs15061635  
13. An, H., Cai, H., and Weng, Q. (2022). Composition and configuration effects of urban green space on  
land surface temperature. Remote Sensing, 14(3), 1–18. https://doi.org/10.3390/rs14030633  
14. Zhao, L., Oppenheimer, M., and Zhu, Q. (2020). Interactions between urban heat islands and climate  
change. Nature Climate Change, 10, 506–514. https://doi.org/10.1038/s41558-020-0730-2  
www.ijltemas.in  
Page 1553