The El Merdja agricultural plain of Ain Djasser (northeastern Algeria) is a topographic depression where intensive groundwater abstraction exceeds annual renewal due to agricultural and domestic pressures. Hydrochemical analyses on 40 wells confirm contamination, with nitrate up to 279 mg L-& sup1; and chloride up to 1794 mg L-& sup1; , well above drinking-water guidelines in the central plain. Using a GIS-based DRASTIC model refined by Empirical Bayesian Kriging Regression Prediction (EBKRP) and contamination indicators, we developed a hybrid DRASTIC-EBKRP-Hydrochemistry index. Results show that 84% of the 174 km & sup2; aquifer exhibits moderate vulnerability, while the most vulnerable zones form a continuous north-south axis where shallow groundwater, high recharge, permeable soils, and high hydraulic conductivity converge. The central depression acts as a preferential recharge zone and pollution trap, with the highest nitrate and chloride concentrations coinciding with the highest vulnerability classes (DI > 130) and shallowest water levels. The significant correlation between contaminant levels and low static water levels confirms this axis as the most exposed sector, where vertical pollutant transfer is rapid and attenuation minimal. These findings highlight the urgent need for protective zoning, optimized fertiliser use, improved sanitation, and targeted monitoring to support sustainable groundwater management in semi-arid regions.