Karst aquifers in the Mediterranean region are crucial for water supply and agriculture but are increasingly threatened by climate change and overexploitation. The Jebel Zaghouan aquifer, historically significant for supplying Carthage and Tunis, serves as the focus of this study, which aims to evaluate its availability, quality, and vulnerability to ensure its long-term sustainability. To achieve this, various methods were employed, including APLIS and COP for recharge assessment and vulnerability mapping, SPEI and SGI drought indices, and stable and radioactive isotope analysis. The findings revealed severe groundwater depletion, primarily caused by overexploitation linked to urban expansion. Minimal recharge was observed, even during wet periods. APLIS analysis indicated moderate infiltration rates, consistent with prior reservoir models and the MEDKAM map. Isotopic analysis highlighted recharge from the Atlantic and mixed rainfall, while Tritium and Carbon-14 dating showed a mix of ancient and recent water, emphasizing the aquifer’s complex hydrodynamics. COP mapping classified 80% of the area as moderately vulnerable. Monitoring of nitrate levels indicated fluctuations, with peaks during wet years at Sidi Medien Spring, necessitating control measures to safeguard water quality amid agricultural activities. This study provides valuable insights into the aquifer’s dynamics, guiding sustainable management and preservation efforts.
Karst aquifers are critical yet vulnerable water resources in semi-arid Mediterranean regions, where structural complexity, nonlinearity, and delayed hydrological responses pose significant modeling challenges under increasing climatic and anthropogenic pressures. This study examines the Jebel Zaghouan aquifer in northeastern Tunisia, aiming to simulate its natural discharge dynamics prior to intensive exploitation (1915–1944). Given the fragmented nature of historical datasets, meteorological inputs (rainfall, temperature, and pressure) were reconstructed using a data recovery process combining linear interpolation and statistical distribution fitting. The hyperparameters of the artificial neural network (ANN) model were optimized through a Bayesian search. Three deep learning architectures—Multi-Layer Perceptron (MLP), Convolutional Neural Network (CNN), and Long Short-Term Memory (LSTM)—were trained to model spring discharge. Model performance was evaluated using Kling–Gupta Efficiency (KGE′), Nash–Sutcliffe Efficiency (NSE), and R2 metrics. Hydrodynamic characterization revealed moderate variability and delayed discharge response, while isotopic analyses (δ18O, δ2H, 3H, 14C) confirmed a dual recharge regime from both modern and older waters. LSTM outperformed other models at the weekly scale (KGE′ = 0.62; NSE = 0.48; R2 = 0.68), effectively capturing memory effects. This study demonstrates the value of combining historical data rescue, ANN modeling, and hydrogeological insight to support sustainable groundwater management in data-scarce karst systems.