Abstract: Global warming has become an indisputable fact，under this background, study on fine spatial-temporal variation of air temperature has important practical significance in Xinjiang. However, the meteorological stations are in sparse and uneven spatial distribution in Xinjiang，it is difficult to get the fine spatial-temporal variation information of air temperature using the traditional technology methods. In recent years, with the expansion of GIS technology applications, the method of regression model by geographic factors （longitude，latitude，altitude，etc.） combining with spatial interpolation was used in grid-based regional climate factors and get good results. In this paper, based on the monthly mean temperature data during 1961-2010 from 93 meteorological stations in Xinjiang，by using the methods of linear regression and Mann-Kendall test, the change trend and mutation characteristics were analysis. With ArcGIS as a platform for direct space interpolation, by longitude, latitude, altitude as independent variables，the multiple regression analysis methods were used to build the macro spatial distribution simulation model of mean seasonal(spring，summer，autumn and winter)and yearly temperature in 1961-2010，before and after their discontinuity ascension in recent 50 years. And the residual terms of 93 meteorological stations about the regression model space interpolation simulation were completed. Using inverse distance weighing （IDW），the residual terms were interpolation simulation. Superimposing the two parts of macro spatial distribution and residual terms, the fine spatial distribution can be gotten of mean seasonal and yearly temperature in recent 50years，before and after their mutations by the grid cells of 200 m×200 m. Then the change range before and after mean temperature mutations can be gotten according to the former and latter differences. The main results are as follows：（1）Various temperature elements are very obviously different in different regions in Xinjiang, generally, seasonal and annual mean temperature in the southern, plain and basin areas are higher than that in northern and mountainous areas of Xinjiang. （2）In the background of global warming, spring, summer, autumn, winter and annual mean temperature shows a significant increasing trend by the rate of 0.24 ℃/10 a、0.21 ℃/10 a、0.39 ℃/10 a、0.49 ℃/10 a and 0.33 ℃/10 a respectively，and their significantly mutations occurred in 2004、1997、1995、1985 and 1994 in recent 50 years, the rising ranges before and after mutations are 1.5 ℃、0.8 ℃、1.2 ℃、1.6 ℃ and 1.0 ℃ respectively. But the rising ranges are very obviously different in different regions, generally, they are higher in northern areas than in southern areas of Xinjiang.

Key words: Xinjiang, air temperature, spatial-temporal change, ArcGIS