In the arid landscapes of Ethiopia’s Great Rift Valley, where the Abaya and Chamo Lakes shimmer under the equatorial sun, a hidden crisis bubbles beneath the surface. Groundwater—essential for drinking, farming, and industry—is under threat from pollution, yet its vulnerability has never been systematically mapped. That changes with a new study led by Tilahun Wankie Wanjale, a researcher at Arba Minch University’s Faculty of Water Resources and Irrigation Engineering. His team has applied cutting-edge tools to reveal where groundwater is most at risk, and the findings carry implications far beyond Ethiopia’s borders.
Using a hybrid approach combining GIS-based DRASTIC-LU modeling with machine learning, the researchers assessed eight key factors influencing groundwater contamination: depth to water table, recharge rate, aquifer type, soil composition, topography, vadose zone impact, hydraulic conductivity, and most critically—land use. “Land use emerged as a dominant factor,” explains Wanjale. “Areas near agricultural zones and urban settlements showed significantly higher vulnerability, especially where nitrate levels exceeded safe drinking standards.”
The study compared two machine learning models—Random Forest (RF) and Support Vector Machine (SVM)—to evaluate which best predicted pollution risk. The RF model outperformed SVM, achieving an 89% accuracy rate in predicting groundwater vulnerability. Based on this, the team generated high-resolution maps classifying the western catchment into five risk zones, from “very low” to “very high.” Alarmingly, zones adjacent to the lakes registered the highest risk, with nitrate concentrations between 21.41 and 53.60 mg/L—far above the WHO guideline of 50 mg/L.
For the energy sector, these insights are more than academic. Groundwater is a lifeline for cooling systems in geothermal plants, feedstock for biofuel production, and a critical backup during droughts that threaten hydropower generation. In Ethiopia, where geothermal energy is expanding along the Rift Valley, ensuring groundwater quality is not just an environmental issue—it’s an economic one. A contamination event could force costly shifts to surface water or imported water, disrupting operations and raising energy costs.
Wanjale’s work signals a shift toward data-driven groundwater management. “Traditional methods like DRASTIC alone can miss the dynamic influence of land use and human activity,” he notes. “Integrating machine learning allows us to anticipate risks before they become crises.”
Published in *Discover Sustainability*, this study offers a blueprint for similar basins worldwide. As climate change intensifies pressure on water resources, tools that combine geospatial analysis with AI could become standard in safeguarding the unseen reserves that power economies. For Ethiopia and beyond, the message is clear: the future of water—and energy—depends on seeing what lies beneath.