Improvement of water table interpolation and groundwater storage volume using fuzzy computations.

Abstract

The water table is an important piece of data for hydrogeological studies, particularly as input data to groundwater simulation models. Since the accuracy of groundwater simulation models significantly depends on input data, this study highlights the application of fuzzy kriging to improve the accuracy of water table interpolation. The results of the fuzzy kriging approach are compared with common methods in water table interpolation like ordinary kriging, inverse distance weighting (IDW), and Thiessen polygon methods to justify the suitability of the fuzzy kriging. The Gilan and Zanjan plains, located in the northwest of Iran, are used as case study areas. The Gilan Plain is characterized by a dense and regular piezometric network and gentle hydraulic gradient. The longitudinal plain of Zanjan has a sparse and irregular piezometric network and steep hydraulic gradient. Since these plains have different piezometric network configurations, the sensitivity of the interpolation methods to the monitoring point configuration is analyzed. The cross-validation method is employed to validate the accuracy of interpolation methods in water table interpolation. In control points, the average of root-mean-square errors associated with groundwater water table values estimated using fuzzy kriging, ordinary kriging, IDW, and Thiessen polygon methods are obtained to be respectively 1.36, 1.93, 3.49, and 9.10 in the Gilan Plain and 13.60, 22.86, 32.30, and 59.81 in the Zanjan Plain. The results indicate that the fuzzy kriging technique has greater precision in comparison with other methods, especially under the conditions of the sparse piezometric network and steep hydraulic gradient. The results also demonstrate that the used methods generally have higher accuracy in the Gilan Plain with a regular piezometric network than in the Zanjan Plain. Furthermore, Thiessen polygon, IDW, and ordinary kriging methods overestimated water table in comparison with the fuzzy kriging method in our cases. This overestimation may cause large error values in subsequent calculations such as water budget and aquifer storage which play a major role in the appropriate management of water resources.