Urbanization brings about heat island effect, which affects regional climate change. Land surface temperature can reflect the degree of surface warming, and can directly affect human living comfort. This study retrieved land surface temperature (LST) based on Landsat 8 TIRS thermal infrared remote sensing data and meteorological data, and the local climate zones (LCZ) were divided by GIS spatial analysis and decision tree classification. We quantitatively studied LST differentiation rules of different types of local climate zones from the regional perspective. The results showed that (1) most of the buildings in study area were low- and middle-level, with the intensive buildings accounting for 27.54%, 21.95% and 25.09%, respectively. Urban park contained the main green and water space. (2) In terms of spatial distribution, LST in the city center was higher than that in the suburbs. Forest and river were the main low temperature areas, and the heat island effect was significant. (3) LST in built-up LCZs were generally higher than that of natural LCZs. Among built-up LCZs, LCZ3 (compact low rise) had the highest LST, while LCZ4 (sparse high-rise) had the lowest LST, with a difference of 1.53℃, 2.30℃ and 2.22℃ in Beijing, Tianjin and Shijiazhuang, respectively. Vegetation and water can lower LST, however, the LST of bare soil and paved road were always higher. Therefore, building layout should be fully considered and vegetation and water distribution should be rationally utilized to reduce heat accumulation and improve urban ecological environment.