Near-surface temperature is a key parameter for the hydrological modeling. It not only directly affects the process of melting snow/glacier and permafrost,but also controls the exchange of water and heat between the surface and the atmosphere. Base on the observed near-surface temperature data from the six self-established meteorological stations in the typical alpine basin - Tizinafu River Basin over the Kulun Mountains,Xinjiang,China from 2012 to 2016,we analyzed the spatial and temporal distribution characteristics of the near surface temperature in the study area. The results show as follows: (1) The data of the average temperature can be fitted very well by the sine function (R~2 > 87%) . As the altitude of the stations increases, the dispersion degree of the data is lowered,and the temporal variation got delayed,and the temperature fluctuated less. (2) The occuring months of the lowest dailymean temperature and monthly-mean temperature are inconsistent,and the phenomenon appear more frequently at higher attitude. (3) There is a spatial difference about the temperature within a year. The situation of the spatial heterogeneity in temperature had 5 sub-regions: The maximum of difference degree in space is in D1 (Maza-Kudi) in autumn/winter and in D5 (Momoke-Jiangka) in spring; the minimum is in D3 (Kudidaban-Shihshu) in spring, summer and autumn. There is most significant difference in D4 (Shihshu/Kudidaban-Momoke) in summer and lowest difference in winter. (4) The results of correlation analysis show that the degree of spatial variability of temperature is highly correlated with the altitude. This study would be helpful to understand climatic characteristics on high and cold mountains of central Asia and optimize the water recycling modelling between solid state water and liquid water.