Exploring the driving forces responsible for heterogeneous surface temperatures is important for selecting proper mitigation strategies to improve the urban thermal environment. Many previous studies have examined the relationship between Land Surface Temperature (LST) and landscape composition and configuration of greenspace. However, investigations focused on the influence of artificial surface on LST are limited. This study develops a method for exploring the driving forces of LST heterogeneity at various spatial scales in Beijing, China. We classified LSTs into six sequential temperature classes to which we applied stepwise regression analysis to detect factors controlling LST variation. We found that: (1) high and low temperature patches were unevenly distributed across the city, with high temperature patches often forming high temperature belts that might negate the expected environmental benefits of planned urban greenspaces, (2) the explanatory power of parameters significantly associated with LST variation was noticeably reduced at finer spatial scales, and (3) variations in significant parameters in low temperature areas can serve as reference target for improving the thermal environment of high temperature areas. Our findings provide insights that can be used to mitigate the impact of urbanization on the thermal environments of cities and achieve the sustainable urban development.