Predictive Modeling of Urban Heat Islands: Integrating Satellite Remote Sensing, Deep Learning, and Local Climate Zoning

Abstract

Urban Heat Islands (UHIs) have emerged as a significant environmental challenge due to rapid urbanization, increasing population density, land-use transformation, and climate change. Urban areas frequently experience higher temperatures than surrounding rural regions because of extensive impervious surfaces, reduced vegetation cover, anthropogenic heat emissions, and changes in surface energy balance. Accurate prediction and spatial assessment of urban heat islands are essential for sustainable urban planning, climate adaptation, public health protection, and environmental management. This study proposes an integrated predictive modeling framework that combines satellite remote sensing, deep learning algorithms, and Local Climate Zone (LCZ) classification to identify, analyze, and forecast urban heat island intensity and spatial distribution. Satellite derived datasets, including Land Surface Temperature (LST), Normalized Difference Vegetation Index (NDVI), Normalized Difference Built-up Index (NDBI), and other environmental indicators, are utilized to characterize urban thermal conditions and surface properties.