Modeling the Subpixel Land-Use Dynamics and Its Influence on Urban Heat Islands: Impacts of Factors and Scale, and Population Exposure Risk

Abstract

Urban heat islands (UHIs) has been proven firmly related to the land use structure. Identifying interactions between UHIs and multiple land use components is a crucial step to obtain human heat welfare information. However, few studies have predicted sub cell scale land use structure dynamics on UHIs due to the lack of subpixel simulation methods. Herein, we present an integrated framework coupling subpixel unmixing and mixed-cell simulation methods to predict fine-scale land-use structural changes. A widely used XGBoost was used to determine the optimal scale for future UHIs prediction. This framework explores how multiscale changes in land use structure will affect future UHIs intensity, taking Wuhan, China as a study area. The results reveal that most influence comes from the scale below the 330-m grid, while the fine-scale land use components of a given position show limited impact on the UHI intensity. Impervious surfaces contribute more than 55% of the importance, while bare soil and water components within the 270-m grid also significantly affect UHIs. We also find that optimizing the structure of land use components can potentially release approximately 599,000 people from high-UHI regions in the study area.