Spatiotemporal heterogeneity of urban thermal environments: Linking urban form and vitality

With the rapid pace of urbanization, urban thermal environmental issues have become increasingly critical. This study takes the central area of Xi'an as a case to investigate how urban morphology and vitality factors influence land surface temperature (LST) across space and time. A multi-scale approach is adopted: Multi-scale Geographically Weighted Regression (MGWR) is used at a 1 km2 scale to examine spatial heterogeneity, while LightGBM combined with SHAP is applied at a finer 1200 m2 resolution to evaluate variable importance. The results reveal that NDVI and MNDWI consistently have strong cooling effects on LST with minimal spatial and temporal variation. Conversely, building height shows significant spatiotemporal heterogeneity in its influence, indicating the complex role of 3D urban form. Population density and nighttime light intensity are positively correlated with LST, especially in extreme seasons like summer and winter. The study also emphasizes that spatial scale significantly affects the observed impact of influencing factors, with some variables (e.g., population density) showing strong effects at local scales but weakening at larger scales. These findings provide insights for urban planning strategies aimed at mitigating the urban heat island effect and enhancing urban thermal comfort.