Ensemble Space Inversion yields efficient and realistic calibration of a multi-layer volcanic aquifer model

Calibration (history matching) of groundwater flow and transport models is an essential step to minimize posterior predicting uncertainty, enabling their use to design interventions such as remediation systems in contaminated sites. To encapsulate all information coming from field observations, especially in complex hydrogeological settings, high-dimensional inverse problems are usually set up, which may be difficult or very lengthy to solve. This study applies the Ensemble Space Inversion (ENSI) method to a multi-layer groundwater flow model developed using MODFLOW-USG to design a hydraulic barrier in an industrial area in Central Italy. ENSI projects the inversion problem into a reduced-dimensional parameter space using super-parameters derived from stochastic realizations. The method yields a fast solution that preserves geological plausibility where standard history-matching tools fail to even reach convergence. Results show excellent agreement between observed and simulated heads and drawdowns, with significant computational savings compared to traditional approaches. ENSI proves to be a robust and practical tool for complex hydrogeological models, especially in data-rich and highly heterogeneous settings.