Performance Evaluation of Geostatistics Methods for Rainfall Estimation in Haraz Watershed

Knowing the temporal and spatial distribution of rainfall is one of the most important factors in implementing soil and water conservation, flood control, and drought management projects and one of the most important parameters for water resource planning. Given the significant variation in rainfall on the one hand and the shortage of rain gauge stations to record rainfall, on the other hand, the requirement of correlating the rainfall estimation model to the spatial is inevitable. In this regard, geostatistics is one of the most important methods for estimating the spatial distribution of rainfall. The purpose of this study was to evaluate the geostatistical methods in annual, monthly and maximum 24-h rainfall estimations in the Haraz Watershed in Mazandaran province and to study their spatial variations in the time period of the water year 1976-1977 to 2013-2014. In this research, 40 meteorological stations with 38-year statistical period into and around the mentioned watershed were used to compare the geostatistical methods including Kriging, Co-kriging, inverse distance weighting and radial basis function in the isohyetal map for the Haraz Watershed. For this purpose, using the cross-validation method, by removing the individual stations, their precipitation values were determined and compared with the observed values. The results showed that radial basis function for annual rainfall and simple Kriging for maximum 24-hour rainfall, and simple and ordinary Kriging, Co-kriging, inverse distance weighting and radial basis function for different monthly rainfall is the most appropriate method. There is the spatial rainfall dependency on a monthly scale from 11.6 to 99.2 km, annual scale with 35.9 km and maximum 24 hour rainfall with 21.3 km. These results can be used to design distance among meteorological stations for rainfall measurement.