Abstract: Near-surface atmosphere temperature is an important parameter in the many fields such as plant physiology, hydrology, meteorology and environics. Traditionally,near-surface temperature is obtained from the discrete meteorological observation, and the continuous regional temperatures are obtained by applying interpolation to the discrete station observed results, which is limited in complicating topographical conditions. Satellite observation provides new technology for obtaining atmosphere temperatures with the development of the land surface temperature retrieval approach. A new methodology is suggested to obtain atmosphere temperatures with satellite data and the station-observed data in this paper. The formula describing land surface temperatures from the satellite and station observed near-surface atmosphere temperatures in the light of linear regression method was created by using the datasets from 2340 stations from 1998 to 2007 and the corresponding surface temperature data from the NOAA/AVHRR, with the various factors, including vegetation cover, land cover type, season, wind, air pressure and precipitation. The simulated results from the satellite data by this method are tested with the station observed atmosphere temperature data between 2002 and 2003, which are not employed in the regression model. Additionally, the results are also compared with the interpolated air temperatures with the same resolution in a pixel to pixel way. especially in mountain regions and regions with sparse stations. 