Evaluation of the Empirical Bayesian Kriging method in ground water level zoning

Evaluation of the Empirical Bayesian Kriging method in ground water level zoning

Abstract
Background and Objective: Groundwater as one of valuable water resources has always been of interest to researchers. Preparing an interpolated water level zoning map is one of fields that is can be acquired via best interpolation methods among different available methods. The Kriging method, based on semivariogram analysis, is one of traditional geostatistical methods. Interpolation precision is depended on the suitable selection of variogram. Empirical Bayesian Kriging (EBK) method is developed to estimated semivariogram parameters during simulation process. The objective of this study is investigation of EBK to increase the precision of groundwater level interpolation zoning and its results compare to other deterministic interpolation method including inverse distance weighted, radial basis function and local and global polynomial functions.
Material and Methods: This study is based on annual mean groundwater level of 57 deep well in Ghareh-Sou aquifer located in Golestan province during 2005-2016 period. Different deterministic and geostatistical interpolation methods evaluated using cross validation technique to groundwater level zoning. The best semivariogram selected for Kriging and EBK methods and finally the model with minimum error is determined and its map is drawn.
Findings: The results of cross validation in study area showed better results for degree two local polynomial methods among deterministic method even it had less error in comparison with Kriging method. The EBK methods, with simulation of fitness of suitable variogram on groundwater level data, led to decrease in Kriging error (23 to 16 meter) and had close precision to local polynomial method.
Conclusion: Although the error of prepared maps based on EBK and local polynomial methods have not significant differences, but there are considerable discrepancies between these maps. The EBK basis map show smoother spectrum of groundwater level changes and the drawn pattern is proportionate with general slope of study area.
Findings: The results of cross validation in study area showed better results for degree two local polynomial methods among deterministic method even it had less error in comparison with Kriging method. The EBK methods, with simulation of fitness of suitable variogram on groundwater level data, led to decrease in Kriging error (23 to 16 meter) and had close precision to local polynomial method.
Conclusion: Although the error of prepared maps based on EBK and local polynomial methods have not significant differences, but there are considerable discrepancies between these maps. The EBK basis map show smoother spectrum of groundwater level changes and the drawn pattern is proportionate with general slope of study area.