A New Approach for Interpolation of Precipitation Data by Soft Indicator Kriging and Genetic Algorithm

Interpolation is among the most important areas in geosciences, because there is always a need to obtain reasonable large and accurate information from a finite set of data. Many methods of interpolation have been developed so far. Kriging is one of the most applicable and accurate among them. Also, Kriging has been classified into various sub-types; Such as: Ordinary Kriging, Indicator Kriging and Soft Indicator Kriging. Genetic Algorithm is a common evolutionary algorithm used to obtain an optimum result from a potential answer space. In this paper, Genetic Algorithm has been utilized to train some parameters of SIK relating to an objective function to result in the best answer. In this article, Ordinary Kriging, Indicator Kriging, Soft Indicator Kriging and Soft Indicator Kriging & Genetic Algorithm have been applied to a data set collected from 15 Synoptic stations in Mazandaran province, Iran. The data are the observations of monthly obtained precipitation of Bahman, 1387. By comparing the calculated and observed values in both Cross-Validation and Jackknife approaches in all of the above mentioned methods (4ones) it is proved that the combination of Soft Indicator Kriging and Genetic Algorithm is the most accurate approach. The mean of the absolute differences between the observed and calculated values of each method is the criterion by which the methods have been evaluated. This quantity for the Soft Indicator Kriging & Genetic Algorithm method is 10.3 mm and is the least one compared to the other methods (12.3, 13.3 and 12.8 mm respectively).