Abstract: As the rapid urbanization in recent years，it has resulted in serious environmental problems such as urban heat island，air pollution，ecological imbalance and so on. It is of great theoretical and practical significance to study the evolution and influencing factors of urban thermal environment for alleviating the ecological and enviro- nmental problems. In recent fifty years，urbanization of Lanzhou City，Gansu Province，China is in a accelerating period. Compared to the plain city，Lanzhou’s artificial and natural factors bring about the prominent and particular heat island effects. So it is necessary to probe the temporal-spatial characteristics and formation mechanisms of ther- mal environment in Lanzhou. The eighth generation of Landsat series，Landsat Data Continuity Mission（LDCM） sat- ellite，was launched in February 2013 with two sensor payloads，an Operational Land Imager（OLI），and a Thermal Infrared Sensor（TIRS）．One of the biggest advantages of Landsat 8 is the separate TIRS thermal infrared sensor，the thermal infrared spectrum has split in two，set up two thermal infrared bands，which makes the split window algorithm can be used for atmospheric correction and surface temperature inversion of higher precision. Accordingly，with the aim at the trait of Landsat8 data，this paper explored the split-window algorithm suitable for Landsat8 data，especially for the wave feature of Landsat8. This paper selected its inversion parameters and raised the inversion method for earth surface’s real temperature based on Landsat8，with Lanzhou’s central area as an example to extract earth surface’s real temperature and to obtain the distribution pattern of the thermal field on the support of ENVI and GIS. By utilizing the Fractal Net Evolution Approach（FNEA） segmentation element of thermal field，Gi index of spa- tial clustering to extract heat island center and scope，Spectral Mixture Analysis to extract the vegetation coverage and impervious surface information，the paper explored the relative relationship between the earth surface’s real temper- ature and different underlying surface parameters，in order to provide method and case support to Landsat8’s applic- ation enrichment in ecological and environmental inspection research. It found that the split window algorithm based on Landsat 8 ground temperature inversion data is feasible，the distribution  of high temperature district at center regi- on of Lanzhou City is more concentrated，the high temperature areas of Lanzhou City are mainly distributed at the urban core with a dense population，residential and commercial area such as Anning High-tech Development Zone，Qilihe Old Industrial Zone，Chengguan District，Xigu Petrochemical Industry Zone and Yantan District，etc. Urban heat island is mainly concentrated in Chengguan，Qilihe and Xigu District，etc. Based on mixed spectral model，the paper resolved the image into impervious surface，vegetation and other non photosynthetic material to analyze the relationship between temperature and underlying surface. Surface temperature showed a strong negative correlation with vegetation，an insignificant positive correlation with impervious surface，and a positive correlation with other non-photosynthetic material. It will has some helps for further research on the mechanism and laws of the cooling effect of vegetation．

Key words: urban heat island, remote sensing, Landsat8, split window algorithm, impervious surface, lanzhou