In the heart of southeastern Morocco, where the Feija Basin grapples with the dual challenges of shallow aquifers and over-extraction for agriculture, a groundbreaking study offers a beacon of hope for sustainable water management. Led by Abdessamad Elmotawakkil from the Department of Computer Science at the University of Ibn Tofail in Kenitra, Morocco, this research leverages artificial intelligence to predict optimal groundwater recharge zones, a critical need in arid and semi-arid regions.
The study, published in the journal *Frontiers in Remote Sensing* (translated to *Frontiers in Remote Sensing*), introduces a GeoAI-based framework that combines remote sensing, geospatial analysis, and advanced AI models. This innovative approach uses ten conditioning factors, including elevation, slope, topographic wetness index, NDVI, rainfall, soil permeability, and geomorphology, to construct a comprehensive input dataset. The research team trained and optimized five AI models—TabNet, TabTransformer, Multilayer Perceptron (MLP), CatBoost, and AdaBoost—using grid search and particle swarm optimization (PSO). The performance of these models was evaluated using accuracy, AUC-ROC, Cohen’s Kappa, and feature importance, with spatial validation conducted using in-situ borehole data.
Among the tested models, TabNet emerged as the top performer, achieving an impressive accuracy of 97.8% and an AUC of 0.99. TabTransformer followed closely with an accuracy of 97.6%. Both models demonstrated strong generalization and produced spatially coherent recharge maps, identifying optimal zones that corresponded with low-lying, vegetated, and permeable areas. These findings align with known hydrogeological features, providing actionable insights for water resource planners.
“Our study presents a novel application of tabular deep learning models in groundwater science, enhancing the precision and interpretability of recharge zone mapping,” said Elmotawakkil. “The results provide actionable insights for water resource planners, especially in light of recent anomalous hydrological events.”
The implications of this research extend beyond the Feija Basin. In an era where climate change is exacerbating water scarcity, the ability to predict and manage groundwater recharge is crucial. The proposed framework supports the development of rainwater harvesting and artificial recharge systems, ensuring long-term groundwater sustainability in climate-sensitive areas. This is particularly relevant for the energy sector, where water is a vital resource for various processes, including cooling and hydroelectric power generation.
As the world grapples with increasing water stress, the integration of AI and remote sensing technologies offers a promising path forward. This research not only highlights the potential of tabular deep learning models in groundwater science but also sets a precedent for future developments in the field. By providing a data-driven, proactive strategy for water management, it paves the way for more resilient and sustainable water resources in arid regions.
In the words of Elmotawakkil, “This study is a step towards ensuring that we can better manage our water resources, even in the face of climate change and increasing demand.” The future of groundwater management lies in the intersection of technology and sustainability, and this research is a testament to that.