Fresh groundwater is a vital resource for serving domestic, agricultural, and industrial needs, despite its vulnerability to pollution, which poses considerable economic and human health problems. The agriculturally dominated western catchment of the Abaya-Chamo Lakes Basin is no exception. However, there have been no earlier investigations on the sensitivity of groundwater to contamination in this catchment. The aim of this study is to assess the vulnerability of groundwater to pollution in the western catchment of the Abaya-Chamo Lakes Basin, Ethiopian Rift System. Groundwater vulnerability to pollution was evaluated using DRASTIC-LU and machine learning methods, specifically Random Forest (RF) and Support Vector Machine (SVM). Eight parameters were used as inputs for analysis: depth to groundwater table (D), net recharge (R), aquifer media (A), soil media (S), topography (T), impact of vadose zone media (I), hydraulic conductivity (C), and land use (LU). These parameters were integrated using weighted overlay methods within a GIS interface. The AUC-ROC curve was utilized to evaluate the training and validation datasets (measured nitrate) to assess the models’ performance. The evaluation of the models revealed that the RF-based machine learning approach (DRASTIC-LU-RF) yielded better efficacy in predicting groundwater vulnerability (prediction accuracy = 0.89) to pollution. In contrast, the DRASTIC-LU-SVM model showed a prediction accuracy of 0.84. Based on DRASTIC-LU-RF probability prediction index maps, the study area was classified into five groundwater vulnerability to pollution levels ranging from very low to very high. Spatial analysis revealed that areas near the lakes are particularly highly vulnerable to pollution, as verified by high nitrate concentration levels ranging from 21.41 to 53.60 mg/L. This study will help protect and manage groundwater in the future by showing which areas are more at risk. It will be a valuable resource for researchers, water managers, and hydrogeologists focused on the protection and long-term management of groundwater resource sustainability.
Wanjala et al. (Sat,) studied this question.