@article{Diksha2026, 
author = { Diksha and Maya Kumari and Rayudu Venkata Komali Supraja and Varun Narayan Mishra},
title = {Urban heat island assessment using Landsat-derived indices and geographically weighted regression in Sonipat district, India},
year = {2026},
journal = {AIMS Geosciences},
volume = {12},
number = {2},
pages = {302-334},
keywords = {autocorrelation, geographically weighted regression, land surface temperature, normalized difference vegetation Index, ordinary least square, sky view factor, urban heat island, urban thermal field variance index},
url = {https://www.sciopen.com/article/10.3934/geosci.2026012},
doi = {10.3934/geosci.2026012},
abstract = {Urban heat islands (UHI), a confined rise in temperature relative to surrounding areas, are one environmental problem caused by urbanization, despite being essential for societal growth. By analyzing land surface temperature (LST) data from Landsat-7 (L7) and Landsat-8 (L8) Earth observation data and processing them using ArcGIS, SAGA GIS, and Google Earth Engine (GEE) software and cloud platform for 2011 and 2021, this study sought to quantify the UHI in the Sonipat district of Haryana. Results indicated that areas with high Normalized Difference Built-Up Index (NDBI) values correspond to elevated LST, while reduced Normalized Difference Vegetation Index (NDVI) values across all land use types reveal a strong negative spatial correlation with LST, further supported by increasing Moran's index values. Following the mapping and identification of locations exhibiting varying UHI intensities, the Urban Thermal Field Variance Index (UTFVI) indicated that regions with higher UHI intensity were predominantly situated in residential areas, whereas zones with diminished intensity were located in agricultural lands. An ordinary least squares (OLS) analysis demonstrated that Surface albedo (SA), solar radiation (SR), sky view factor (SVF), and NDVI accounted for 37.6% of LST variability in 2011, which considerably increased to 92.8% in 2021, demonstrating a more robust and consistent influence of these components over time. Due to the nonstationarity of LST, a geographically weighted regression (GWR) model was also considered and demonstrated superior performance compared to the OLS model. This study will aid policymakers, stakeholders, and local governments in effectively developing new urban and constructed environments to mitigate significant UHI impacts.}
}