Groundwater potentiality mapping using machine learning algorithms BouSbaa area, Marrakech, Morocco

Abstract

AbstractGroundwater recharge is crucial for managing freshwater resources. Machine learning algorithms are used to discuss the important aspects of groundwater exploration. For maximum accuracy, Extreme Gradient Boosting and Random Forest have been tested for modeling groundwater potential maps. A database of water point inventories has been prepared, randomly divided into 75% for training and 25% for model validation. A database of flows is used to confirm the feasibility of the model. Groundwater potential maps are generated using various relevant factors (elevation, slope, precipitation, etc.). After validation of the model using ROC-AUC and confirmation of feasibility with flow diagrams, these methods have shown high accuracy and relevant results for groundwater potential models.