Quantifying Groundwater Risk Associated with Physical and Anthropogenic Vulnerabilities Across Coastal Aquifer of the Mekong Delta Using Data-driven Approach

The coastal aquifer of the Mekong Delta (MD) is facing groundwater challenges exacerbated by the joint impacts of anthropogenic activities and climate change. Groundwater sustainability in MD needs a clear understanding of both natural and human-related factors that cause risks, a knowledge that is currently lacking in MD. This study introduced a framework to separately quantify physical and anthropogenic groundwater risks in MD’s multi-layered coastal aquifer system. Leveraging a weighted overlay analysis, we combined thematic layers (e.g., distance to sea, distance to drain, aquifer thickness, groundwater level, extraction capacity etc.,) using multiscale geographically weighted regression (MGWR) model and observed groundwater nitrate data of 216 locations. Results were compared to traditional Analytical Hierarchy Process (AHP)-Frequency Ratio (FR) methods. The MGWR model demonstrated superior accuracy in predicting nitrate levels, with R² values ranging from 0.81 to 0.83 on the test dataset. We found that physical vulnerability dominated the groundwater system, with 52% and 57% of the area classified as medium to high risk based on the AHP-FR and MGWR methods, respectively. Anthropogenic vulnerability accounted for a smaller share, with 48% and 37% of the area rated as medium to high risk based on the AHP-FR and MGWR methods, respectively. However, the integration of hazard variables leads to divergent risk category distributions. When combining physical and anthropogenic vulnerabilities, the AHP-FR scheme identified 53% of the area as medium-to-high risk for physical vulnerability, 25% for anthropogenic vulnerability, and 47% for the combined effects. In contrast, the MGWR-based spatial rating scheme reported 10% lower for physical vulnerability, a 17% higher value for anthropogenic vulnerability, and a nearly unchanged estimate for the combined effects. It is worth noting that the MGWR method tends to rate areas as higher risk compared to the AHP-FR method during combined risk assessment (physical + anthropogenic). Validation results of area under the curve (AUC) show slight superiority for MGWR (78%) over AHP-FR (75%). This work underscores the importance of methodological discrimination in groundwater evaluation, assisting decision-makers in executing effective and sustainable groundwater management strategies.