Assessing the severity of urban heat transfer and flow across years: Evidence from thermal environment spatial networks

Analysis of the evolution of urban thermal environment spatial networks can effectively assess the development patterns of the urban heat island effect, which is important for sustainable urban development. We employed a research framework of “heat sources - thermal resistance surfaces - thermal environment spatial network “ to conduct our experiments. Particularly, by focusing on the issue of resistance surfaces within the spatial network of a thermal environment, we developed a method for constructing resistance surfaces that integrates the XGBoost model and SHAP. Using multi-temporal remote sensing data, we established thermal environment spatial networks for different years and reached the following conclusions: the distribution range of high and higher temperatures in 2010 and 2023 is significantly broad; the conversion area from low, lower, middle, and higher temperature zones to high temperature zones has continuously increased, totaling 304.6986 km² from 1992 to 2023; the Normalized Difference Built-up Index (NDBI) is the most significant factor in constructing the resistance surface; the number of thermal corridors was 69 in 1992, 55 in 1999, 81 in 2010, and 65 in 2023; the thermal environment spatial network structure in 2010 was the most complex, with the number of closed loops was the highest, and the network connectivity was at its peak; and the heat transfer and flow in 2023 are more severe. These findings can serve as a valuable reference for urban planning and management, while also enriching the structural framework for studies related to the urban thermal environment.