Spatial characterization and interpolation of precipitation data

Being able to accurately predict point or a real precipitation data at unsampled locations or areas using measured precipitation data is important in the work of agriculturists, hydrologists, climatologists, engineers, and others. Precipitation phenomenon is a complicated process due to the spatial variability, uncertainty, and complexity of the meteorological processes underlying its formation. There is a need to investigate the use of the most common geostatistical techniques to characterize, interpolate, and analyze precipitation data with the intent to identify the best set of semivariogram and spatial interpolation algorithms for characterizing precipitation data in a region of interest. Linear kriging (ordinary kriging, simple kriging, and universal kriging) and nonlinear kriging (indicator kriging, probability kriging, and disjunctive kriging) algorithms were used in this research project to characterize and interpolate precipitation data. Gaussian, circular, spherical and exponential semivariograms were employed with the six interpolation algorithms to characterize the precipitation data. Statistical measures of correctness (mean prediction error, root-mean-square error, standardized root-mean-square error, average standard error) from cross-validation were used to compare the combination of kriging and semivariogram algorithms. The most accurate results were obtained by using indicator kriging (IK) with a circular semivariogram for spatial characterization and interpolation of the precipitation data. IK and circular variogram algorithms were used to perform multi-scale analysis of the wet and dry months.