Assessment of Ensemble Models for Groundwater Potential Modeling and Prediction in a Karst Watershed

Due to numerous droughts in recent years, the amount of surface water arid and semi-arid regions has decreased significantly, so reliance on groundwater meet local regional demands increased. The Kabgian watershed is a karst southwestern Iran that provides significant proportion drinking agriculture supplies area. This study identified areas with potential using combination machine learning statistical models, including entropy-SVM-LN, entropy-SVM-SG, entropy-SVM-RBF. To do this, 384 springs were mapped. Sixteen factors are related from review literature, these compiled for locations randomly separated into two categories training (269 location) validation (115 datasets be used modeling process. ROC curve was evaluate results. models used, general, good at determining location potential. evaluation showed E-SVM-RBF model had an area under 0.92, indicating it most accurate estimator among ensemble models. Evaluation relative importance each 16 revealed land use, vector ruggedness measure, curvature, topography roughness index important explainers presence It also found affecting significantly different non-karst springs.