Enhancing urban thermal resilience in multi-mountainous cities through optimized 3D block morphology: a machine learning framework

Abstract

The accelerated global urbanization process has accelerated the significant transformations in land surfaces, leading to the formation of obvious urban overheating phenomena. Maintaining urban thermal resilience can effectively alleviate the impact of extreme high temperatures on communities during heatwaves. This study takes the built-up area of Guiyang City, a typical multi-mountainous city, as the research area, and establishes an evaluation method for urban thermal resilience (UTR) through the difference between conventional temperature and extreme high temperature. Then, based on multi-source data and the characteristics of the study area, we screened the block scale multi-mountainous city morphology index system. The impact of block scale urban form on UTR was quantified from both global and local perspectives using SHAP interpretation model and geographically weighted random forest model. The research results indicate that: (1) The UTR of Guiyang presents a spatial distribution characteristic of low in the middle and high at the edges; The highest values are mainly distributed in the southwest direction near the edge of the city. (2) The urban morphologies are significantly correlated with UTR and has a threshold effect; The average ground elevation (MGE) and building density (BD) are the main driving factors affecting UTR. (3) The UTR of Guiyang exhibits significant spatial clustering, and the impact of urban morphology on UTR also shows spatial heterogeneity. This study provides a scientific foundation for enhancing the thermal environment and promoting sustainable planning in multi-mountainous cities.