Abstract Cokriging is a widely used geostatistical method for modeling the shallow groundwater table, often incorporating digital elevation models (DEMs) as secondary variables. However, existing approaches rarely include robust validation procedures to reduce local uncertainty or iteratively improve spatial predictions. This study presents a novel cokriging algorithm that integrates metaheuristics and iterative residual correction to enhance the estimation of the shallow water table. The method is based on a novel mathematically demonstrated theory based on overlapping kriging maps. Our framework consists of four steps: (a) a genetic algorithm selects the optimal variogram model based on input data; (b) a DEM‐based cokriging routine generates initial estimates; (c) residuals between observed and predicted values are estimated; and (d) residuals are iteratively corrected, and their maps are superimposed to initial estimates until a user‐defined convergence threshold is met. Applied to shallow aquifers in Italy, the algorithm minimized squared residuals between observed and simulated piezometric levels, with the stopping condition based on average annual residuals. Validation against historical data and a global groundwater model demonstrated significant improvements in predictive performance. Within three iterations, the correlation coefficient increased from 2 = 0.85 to 2 = 0.99, an accuracy not achieved with conventional DEM‐based cokriging. Importantly, a multiyear analysis of corrected residuals enabled detection of groundwater level declines linked to drought conditions, offering a novel approach to drought detection and impact assessment. By improving spatial accuracy and supporting long‐term monitoring, this method helps decision‐makers implement informed water conservation strategies, particularly under increasing pressure from climate variability, prolonged droughts, and growing water demand.
Schiavo et al. (Wed,) studied this question.