Groundwater in northwest Tunisia plays a vital role in supporting the domestic, agriculture, industry, and tourism sectors. However, climate change and over-exploitation have led to significant degradation in groundwater quality and quantity. Traditional spatial analysis techniques such as Geographic Information Systems (GIS) and Remote Sensing (RS) are frequently used for assessing groundwater potential and water quality. Yet, these methods are limited by data availability. The integration of Geospatial Artificial Intelligence (Geo-AI) offers improved precision in groundwater potential zone (GWPZ) delineation. This study compares the effectiveness of the Analytical Hierarchy Process (AHP) and advanced Geo-AI techniques using deep learning to map GWPZ in the Majerda transboundary basin, shared between Tunisia and Algeria. By incorporating thematic layers such as rainfall, slope, drainage density, land use/land cover (LU/LC), lithology, and soil, a comprehensive analysis was conducted to assess groundwater recharge potential. The results revealed that both methods effectively delineated GWPZ; however, the Geo-AI approach demonstrated superior accuracy with a classification accuracy rate of approximately 92%, compared to 85% for the AHP method. This indicates that Geo-AI not only enhances the quality of groundwater potential assessments but also offers a reliable alternative to traditional methods. The findings underscore the importance of adopting innovative technologies in groundwater exploration efforts in this critical region, ultimately contributing to more effective and sustainable water resource management strategies.
Read full abstract