Understanding the Drivers of Land Surface Temperature Based on Multisource Data: A Spatial Econometric Perspective

Urban thermal condition has seriously affected the quality of residents’ daily life and triggered some environmental issues. Understanding spatial patterns of land surface temperature (LST) and its driving mechanism is important for the sustainable development of cities. Taking Beijing as an example, this study employed spatial econometric models to investigate spatial and temporal heterogeneity of LST from 2014 to 2018 based on multisource remote sensing and statistical data. The global autocorrelation Moran's I index showed the existence of significant positive correlations of LST among regions, indicating the regions with high thermal environments are spatially adjacent. The temperature of a region would increase by more than 0.6% for every 1% increase in LST of surrounding areas based on the spatial Durbin model. In terms of spatial interactions of influencing factors, elevation, normalized difference vegetation index, modified normalized difference water index, nighttime light, and fossil energy consumption of neighbors exhibited significantly positive spatial agglomeration effects on local LST, whereas albedo, GDP, and population density of adjacent areas had negative effects on LST in local areas. Particularly, the indirect effects of drivers were greater than their direct effects, indicating urban thermal condition was an interregional issue and joint control measures should be adopted to mitigate the urban heat island effects as a whole.