Quantifying 3D building form effects on urban land surface temperature and modeling seasonal correlation patterns

Multiple factors regulate urban land surface temperature (LST), including land cover, climate, and urban form, among which urban form is now receiving more and more attention. Some studies have discussed the planar effects of urban form on LST, whereas less concern has been devoted to the vertical structure of urban areas, which can have a significant effect on heat redistribution. In this paper, we quantify the three-dimensional building form (3DBF) effects on LST captured by Landsat thermal sensors over four seasons by using a random forest (RF) regression method. The five 3DBF factors of building density (BD), building height (BH), sky view factor (SVF), frontal area index (FAI), and building shadow (BS) are calculated within a grid. The seasonal correlation of the 3DBF factors with LST is analyzed, and seven correlation patterns are modeled for the city of Wuhan in China. The results show that: 1) The best grid scale for the building form analysis in a city such as Wuhan is 180 m. 2) The 3DBF factors have a significant effect on urban LST over the four seasons. BD, BH and BS are the season-stable factors, among which BD has a heating effect, which reaches a maximum of 3.6 °C in spring, while BS has a cooling effect, which reaches −3.4 °C in winter. There is also a transition point for BH between heating and cooling at a height of 10 m. 3) SVF and FAI are season-varying factors, in which SVF has cooling effect, except in summer, while FAI also has a cooling effect, but not in winter. These findings will help us to understand how building form affects urban surface temperature, and will provide a reference for urban policy makers and planners in the future.