{"title":"Spatiotemporal analysis of thermal islands in a semi-arid city: A case study of Kermanshah, Iran using machine learning and remote sensing","authors":"Peyman Karami , Seyed-Mohsen Mousavi","doi":"10.1016/j.envc.2025.101174","DOIUrl":null,"url":null,"abstract":"<div><div>Studying urban land use/land cover (LULC) and land surface temperature (LST), and assessing their changes, is crucial for understanding and mitigating the environmental and climatic impacts on cities in semi-arid regions, where water scarcity and heat stress exacerbate urban sustainability challenges. In this study, Landsat 8 data from 2013 to 2023 were used to assess changes in Kermanshah city. LST was extracted using a Mono-Window algorithm (MWA) for each year. Following Intensity-Hue-Saturation (IHS) pan-sharpening, LULCs were classified into five categories: built-up areas, vacant land, green spaces, water bodies, and transportation infrastructure, using training samples and machine learning methods. Cold Islands (<em>CIs</em>) and Hot Islands (HIs) were identified for each image using LST and Getis-Ord Gi analysis, and their spatio-temporal changes were evaluated with the Kappa index and landscape metrics. The sample size for assessing the impact of environmental parameters on LST variations was determined using the Cochran formula. Topographic, topoclimate, and biophysical variables, along with machine learning methods and the chi-square test, were employed to model and evaluate LST variation across different LULCs.</div><div>The results demonstrate a significant increase in HIs, particularly in the city's periphery, driven by rapid urbanization, impervious surface expansion, and reduced green cover. Key environmental factors, including elevation, built-up areas, and green space density, critically influenced LST variations. Although green spaces expanded in some areas, they were insufficient to counteract the intensifying HI effect. The analysis revealed a 65 % spatial consistency in HI patterns over the decade, highlighting persistent thermal hotspots.</div><div>The findings underscore the urgent need for sustainable urban planning, prioritizing green infrastructure and resilient design strategies to mitigate HI effects, reduce energy consumption, and improve urban livability. Future studies could focus on the impact of building heights and the expansion of urban green spaces on thermal islands.</div></div>","PeriodicalId":34794,"journal":{"name":"Environmental Challenges","volume":"20 ","pages":"Article 101174"},"PeriodicalIF":0.0000,"publicationDate":"2025-05-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Environmental Challenges","FirstCategoryId":"1085","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S2667010025000939","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"Environmental Science","Score":null,"Total":0}
引用次数: 0
Abstract
Studying urban land use/land cover (LULC) and land surface temperature (LST), and assessing their changes, is crucial for understanding and mitigating the environmental and climatic impacts on cities in semi-arid regions, where water scarcity and heat stress exacerbate urban sustainability challenges. In this study, Landsat 8 data from 2013 to 2023 were used to assess changes in Kermanshah city. LST was extracted using a Mono-Window algorithm (MWA) for each year. Following Intensity-Hue-Saturation (IHS) pan-sharpening, LULCs were classified into five categories: built-up areas, vacant land, green spaces, water bodies, and transportation infrastructure, using training samples and machine learning methods. Cold Islands (CIs) and Hot Islands (HIs) were identified for each image using LST and Getis-Ord Gi analysis, and their spatio-temporal changes were evaluated with the Kappa index and landscape metrics. The sample size for assessing the impact of environmental parameters on LST variations was determined using the Cochran formula. Topographic, topoclimate, and biophysical variables, along with machine learning methods and the chi-square test, were employed to model and evaluate LST variation across different LULCs.
The results demonstrate a significant increase in HIs, particularly in the city's periphery, driven by rapid urbanization, impervious surface expansion, and reduced green cover. Key environmental factors, including elevation, built-up areas, and green space density, critically influenced LST variations. Although green spaces expanded in some areas, they were insufficient to counteract the intensifying HI effect. The analysis revealed a 65 % spatial consistency in HI patterns over the decade, highlighting persistent thermal hotspots.
The findings underscore the urgent need for sustainable urban planning, prioritizing green infrastructure and resilient design strategies to mitigate HI effects, reduce energy consumption, and improve urban livability. Future studies could focus on the impact of building heights and the expansion of urban green spaces on thermal islands.