Temporal and spatial variations of urban surface temperature and correlation study of influencing factors

Current studies on the effect of urban form on land surface temperature (LST) are mostly conducted from the daytime and 2D morphological perspectives, with less attention paid to the vertical structure of urban areas and their seasonal and diurnal variations, which have a significant impact on heat redistribution. In our study, we selected the spatial scale of urban neighbourhoods and calculated six 3D building form factors, and used the Gradient Boosting Machine (GBM) to quantify the effects of 3D building form on LST production captured by Landsat thermal sensors between seasons and Day/Night. The results show that MH, BD, and FAR are seasonal stabilising factors, with MH having the strongest cooling effect on LST, with a four-season average of 2.1°C and a diurnal difference in its effect on LST. There is a strong positive correlation between BD and LST during the daytime, and the strongest heating effect is in autumn, up to 3.5°C. BVD, GFA, and SVF are seasonal variation factors, with GFA and SVF having a cooling effect in all seasons except spring, and BVD having a slight cooling effect in autumn. These results will provide a reference for future urban planning and mitigation of urban heat island effect.