Spatial heterogeneity in effects of horizontal and vertical environmental features of blocks on land surface temperature: A case study of Shenzhen, China

Abstract

Changes in land surface temperature (LST) can be triggered by both horizontal expansion and vertical growth due to urbanization. A limitation of previous studies is that they have neglected to consider vertical environmental factors and have instead focused on the influence of horizontal features (e.g. land cover composition) on LST. In this regard, this study takes the block space of Shenzhen, a typical high-density city in China, as the basic unit of analysis. A multivariate dataset including panoramic street view images and building footprints is used to extract an indicator system that comprehensively characterizes the horizontal and vertical features of the block. The effects of each feature on LST and its spatial heterogeneity are analyzed using the ordinary least squares (OLS) regression and two local regression models. The results indicate that the two local regression models demonstrate superior performance in explaining LST, particularly the multi-scale geographically weighted regression (MGWR) model (R² = 0.848). In addition to horizontal factors such as percentage cover of forest (PC_{_for}) and plan area building density (${\lambda }_P$), the frontal area index (${\lambda }_F$) and the sky view factor (SVF) in the vertical class of features also had a significant effect on LST. In addition, the majority of horizontal and vertical environmental features demonstrated spatial heterogeneity in their effects on LST. The study provides a scientific basis for differentially regulating block environmental features to mitigate the UHI, particularly in high-density cities.