Robust decision-making for sustainable management of groundwater resources in unconsolidated aquifers using a multi-model ensemble approach incorporating hydrogeological uncertainty

Uncertainty in hydrogeological structures and properties has limited the effectiveness of traditional frameworks for aquifer characterization, groundwater monitoring and modelling in practical applications. How to properly deal with uncertainty is highly relevant for robust decision-making in sustainable management of groundwater resources, which are increasingly stressed between water use and climate change impacts for many drinking water supply sites worldwide that are strongly dependent on groundwater. Previous innovative studies, which rely strongly on the stochastic approach, are predominantly explored in synthetic and scientific cases, creating a gap in presenting how practical and efficient these frameworks can be at regional and local scales. In this work, we developed a holistic approach to better understand and manage uncertainties in hydrogeological structures and properties through groundwater flow modelling. We tested the developed approach for a local drinking water supply site in eastern Germany, which consists of a porous aquifer of glacial-fluvial unconsolidated sediments and is characterized by strong heterogeneity and anisotropy of its hydraulic properties. A large borehole dataset was analyzed to characterize the geological variability and form the basis for a detailed 3D subsurface model. Multiple subsurface structure realizations were generated using ArchPy to represent plausible hydrogeological interpretations of hydraulic conductivity and the groundwater flow dynamics were simulated using MODFLOW 6. The results highlight that hydrogeological uncertainty significantly affects simulated groundwater flow patterns and limits the reliability of deterministic models. The multi-model ensemble approach, incorporating probabilistic assessments, proved to be a robust framework for groundwater management in heterogeneous systems. More specifically, the results highlight the efficiency of the proposed approach to couple ArchPy with MODFLOW via FloPy to incorporate and acknowledge the propagated spatial uncertainty on simulated groundwater dynamics into a robust decision-making process of sustainable groundwater management. Furthermore, this research showed that a highly simplified or highly complex representation of hydraulic conductivity uncertainty almost equally leads to a less reliable and practical groundwater model. The study advances hydrogeological research by providing a practical approach to uncertainty in groundwater modelling that addresses some of the significant implications for sustainable water resource management and provides a framework that is transferable to similar systems facing challenges of aquifer variability and uncertainty.