Abstract: The basis of meteorological data of annual average precipitation and its trend from 1970 to 2011 in Gansu province was analyzed for spatial interpolation and cross validation, and the spatial temporal variation of precipitation was analyzed further by using methods of IDW, OK, CK, Microtopographic Factors and Analytic Method Based on Multiple Regression and Residues (M AMMRR), improved MAMMRR (I AMMRR). The results showed that I AMMRR, combining with climatic factor, longitude and latitude, altitude and micro terrain factor (slope, aspect) to interpolate based on OK, significantly improved simulation accuracy of precipitation, which was the most suitable interpolation method for arid and semi arid area. Annual average precipitation had a downward trend from southeast to northwest during last 42 years in Gansu province. Accordingly, the distribution of climatic zone varied from humid region , semihumid region to semiarid region and arid region. Meanwhile, the area of arid region was the largest,accounting for 43% of the total area. Apart from Qilian mountain, middleeastern Hexi corridor and north mountains, annual precipitation in Gansu province decreased during last 42 years. Precipitation in summer showed a decreasing trend while increased in winter. Precipitation variation indicated zonal distribution in spring and autumn. The results detailed the distribution of the precipitation in Gansu province, which could provide the basis for adjustment of agricultural structure, the development of characteristic agriculture and precision agriculture.

Key words: Precipitation, Variation trend, I AMMRR interpolation method, Climatic division