Abstract: Split-window algorithm is mainly used for retrieving land surface temperature in thermal infrared remote sensing. Around seventeen algorithms have been published in recent years, but few have been applied to Chinese meteorological satellite data such as FY-3 series. FY-3A, the first of this series is launched from Taiyuan Satellite Launch Centre in China on 27 May, 2008. This new generation satellites series provide three dimensional, quantitative, multi spectrum global remote sensing data under all weather conditions, which will greatly help the operational numerical weather prediction, global change research, climate diagnostics and prediction, and natural disasters monitoring. The visible and infrared scanning radiometer (VIRR), medium resolution spectral imager (MERSI) and microwave imaging instrument (MWRI) aboard this satellite can all be used for retrieving the land surface temperature (LST) and monitoring the process of high temperature weather.Daily LST products are retrieved by the improved Becker algorithm using the FY-3A/VIRR data. And the first process of high temperature weather in 2009 is monitored and the changes of thermal environment in different land types are analyzed. First, a modified Becker's split window retrieval algorithm is developed using VIRR thermal infrared spectral response function for retrieving LST from the FY-3A/VIRR data. A new set of parameters for Becker's LST algorithm is proposed. The algorithm is developed from a surface brightness temperature dataset generated from the MODTRAN program, which uses a range of surface parameters (294.2 K, 294.2 K±5 K, 272.2 K, 272.2 K±5 K, 287.2 K, 287.2 K±5 K, 288.2 K, 288.2 K±5 K) and four kinds of atmospheric model (mid-latitude summer atmosphere, mid-latitude winter atmosphere, sub-polar summer atmosphere and American standard atmosphere 1972) as inputs. The daily brightness temperature data of the Channel 4 and Channel 5 of FY-3A/VIRR (1-km resolution) are used to generate the model parameters of Becker's split window inversion algorithm. Second, as a validation of the algorithm, the retrieved VIRR LST is compared with the instrument measurement data in satellite transit period in Dunhuang radiometric calibration and validation test site and MODIS LST of the same period and area. The results show that the error of LST products is-0.17 K and 1.77 K by two validations in Dunhuang site. The two LST products are found to be consistent, and the root mean square error between FY-3A LST and MODIS LST is 2.64 K. By histogram comparison, the two frequency distributions show no difference. Finally, the retrieved daily FY-3A LST is applied to monitor the first high temperature weather process in 2009, indicating two high-value regions: North China (Shijiazhuang, Zhengzhou, Beijing, etc.) and Northwest Territories (Xi'an, etc.). Land surface temperature in some regions exceeds 320.2 K. There are some spatial distribution differences in different urban land types. The profile data of VIRR LST in main cities verify the existence of the phenomenon of urban heat island. And the heat island of industrial mining land is reflected especially clearly in the LST spatial distribution because the vegetation around the industrial mining land is destroyed seriously.These results show VIRR LST production meet the real-time demand of operation. This also would provide scientific basis data for further study of local climate change, and greatly help the operational numerical weather prediction and global change research.