Sensitivity analysis of multiple groundwater vulnerability models for enhancing resource management in the Menzel Bourguiba aquifer system, northern Tunisa

The Menzel Bourguiba aquifer system in northern Tunisia is a major groundwater resource increasingly threatened by climatic changes and human activities. Assessing groundwater vulnerability is essential for sustainable management. This study evaluates vulnerability to contamination using three models: the standard DRASTIC, the modified DRASTIC-LU/LC integrating land use/land cover, and the analytic hierarchy process (AHP). Vulnerability maps were generated using environmental and anthropogenic factors, including aquifer characteristics, soil properties, topography, recharge, proximity to pollution sources, and demographic indicators. All thematic layers were integrated in a GIS to delineate areas at risk. The validation process used nitrate concentration measurements and sensitivity analysis via receiver operating characteristic (ROC) curves, with area under the curve (AUC) values showing differences in discriminatory power. The original DRASTIC model had the lowest discriminatory power (AUC = 0.733) and limited predictive performance owing to the omission of key anthropogenic variables such as land use/land cover, industrial zones, and population density. Performance improved with DRASTIC-LU/LC (AUC = 0.789). The AHP model achieved the highest discriminatory power (AUC = 0.898) and strongest predictive performance, with vulnerability patterns matching nitrate levels, especially in the central sector with dense pollution sources. Bayesian network analysis further confirmed the robustness of the AHP model, showing higher predictive consistency across vulnerability classes, an overall accuracy 0.72, and balanced F1-scores. Conditional probability distributions (CPDs) highlighted the stronger influence of key variables within the AHP framework. Overall, adapting vulnerability models to regional conditions and assigning weights through pair-wise comparisons proved the most effective strategy protecting groundwater in the Menzel Bourguiba aquifer.