Comparative assessment of geostatistical and deterministic interpolation methods for seasonal groundwater level mapping in the Arpa River Basin, Central India

Abstract Groundwater level mapping is essential for groundwater resource assessment and sustainable water management. This study examines the performance of deterministic interpolation methodologies and geostatistical methods in mapping groundwater levels by analyzing the groundwater level data from 69 wells belonging to the Central Ground Water Board (CGWB) and covering the period from 2001 to 2025 in the Arpa River Basin in Central India. The seasonal datasets were arranged for the rainy season (June to September), winter season (October to January), and summer season (February to May) and were analyzed through different methods like Inverse Distance Weighting (IDW), Regularized Spline, and Ordinary Kriging with Circular, Spherical, Exponential and Gaussian semivariogram in ArcGIS. The model performance was analyzed through different cross-validation statistics like Mean Error (ME), Root Mean Square Error (RMSE), Mean Standardized Error (MSE), Root Mean Square Standardized Error (RMSSE), and Average Standard Error (ASE). The results indicate that geostatistical methods generally produced more reliable and spatially consistent groundwater level surfaces than deterministic approaches. In spite of the fact that the Exponential semivariogram produced lower prediction errors, the uncertainty-related parameters, in contrast, varied in the Circular, Spherical, and Gaussian models depending on the seasons. Therefore, no single semivariogram model was universally superior across all validation criteria, and model selection should be based on the combined interpretation of prediction accuracy and uncertainty measures. The seasonal groundwater level maps created during the research can serve as the scientific background for more efficient groundwater monitoring, as well as basin and groundwater management of the Arpa River Basin.