Prediction of the spatial distribution pattern of precipitation using geostatistical methods in Urmia region

Prediction of the spatial distribution of rainfall is essential for establishing the water balance and a good estimation of water availability, especially in mountainous areas because of complex rainfall gradient and scarce number of climatological stations. This study was conducted to evaluate the rainfall spatial distribution of Urmia region by interpolation methods. Data related to 38 climatological stations were used and three methods, IDW, Kriging, and Co-kriging were investigated. After normalization of data, variograms were computed. The least RSS and the most powerful spatial structure were considered as criteria for selecting the best model for fitting on experimental variograms. Cross-validation and RMSE were used for selection of best interpolation method. Results showed that Cokriging method, with elevation parameter as an auxiliary variable, has the least error which may be attributed to the significant correlation between the elevation and annual precipitation at the study region. Due to the mountains region and suitable condition for rainfed farming, so using cokriging method and GIS, high- resolution maps of rainfall distribution and determine rainfed farming region was prepared.