Drivers of Land Surface Temperatures From the Perspective of Urban Functional Zones

Urban areas have a high level of spatial heterogeneity. Therefore, examining the degree of drivers in urban functional zones (UFZs) is important. In this study, point-of-interest data were used primarily to identify UFZs. Random forest and geographically weighted regression models were used to analyze the influence of drivers on land surface temperature (LST). Landsat imagery showed that the urban center had a significantly higher LST than the coast and forested mountains in downtown Dalian. Residential zones accounted for most of the uptown area (37%). UFZs in built-up areas generally had high-temperature levels, exceeding 37 °C. The normalized difference vegetation index was the most important in almost all UFZ types. Modified normalized difference water index and albedo showed uncertain positive/negative correlations in different UFZs. There was also a positive correlation between UFZs and road density. The importance and correlation between the gross domestic product and population density were not statistically significant. The patch density (PD) was similar to the landscape shape index in terms of importance; however, the PD was relatively suitable in terms of correlation. The agglomeration index showed a significant negative correlation.