Application of Interpolation and Regression Methods in Spatial Estimation of Rainfall (Case Study: Kermanshah Province)

Estimation of rainfall in areas without stations is very important due to the lack of meteorological stations and significance of rainfall in various planning strategies. In this research، the daily rainfall data from 46 rain-gauge and synoptic stations in the Kermanshah province in a 20-year period has been used to estimate the seasonal and annual average rainfall in the region. To do this، univariate methods (deterministic and geostatistical) and multivariate methods (geostatistical and linear regression) have been used. The usual method for spatial estimation of rainfall in previous studies in Iran has been the use of one variable (usually altitude)، and also default settings of the interpolation methods. However، in this study، firstly، in the multivariate methods، in addition to the altitude، other variables such as slope percent، latitude and longitude have been used as covariates (independent variables). Secondly، instead of using the default values of the models، various settings were performed on the 8 parameters in the deterministic methods، and up to 31 parameters in the geostatistical methods depending on the interpolation method used، and then the effect of each method was evaluated in the precipitation estimates considering the error of each method. For example، in the geostatistical methods، optimized semivariogram and covariogram were determined in each case according to the spatial structure of the variable، and then their characteristics were adjusted by taking into account the lowest errors in the estimation of precipitation. The results of cross-validation technique showed that the deterministic methods have more errors than the geostatistical methods in all cases. To estimate the average spring rainfall، linear multivariate regression method، and for average summer and autumn rainfalls، ordinary kriging method، and finally for average winter and annual rainfalls، ordinary cokriging method were selected as the best methods. Based on this، the average annual rainfall in the province was estimated about 479 mm (346 to 848 mm)، with maximum seasonal rainfall of 212 mm in winter (equivalent to 44. 3% of the annual rainfall).