An integration of geospatial and machine learning techniques for mapping groundwater potential: a case study of the Shipra river basin, India

Groundwater is an important component of the hydrologic cycle and its significance is quite high due to the lack of surface water and is an important source of fresh water. The amount of surface water alone is not enough to meet the demands of increasing population and increased needs for different purposes due to technological advances. Hence, the need of the hour is to increase groundwater sources and manage them effectively for their sustainable growth. Therefore, the main objective of this study is to map groundwater potential (GWP) zones of the Shipra river basin in India using advanced machine learning and geospatial techniques. Nine factors were used as effective factors such as slope degree, altitude, plan curvature, topographic wetness index (TWI), profile curvature, topographic factor, drainage density, slope aspect, and land use/land cover. The models adopted in this study were classification and regression tree (CART), boosted regression tree (BRT), and random forest (RF). The integrated results of GIS and machine learning techniques were proved to be effective and successful in predicting GWP zones. The area under the curve (AUC) of three models namely BRT, CART, and RF came out to be 0.841, 0.880, and 0.899, respectively. This indicates that all the models are giving a good performance for the GWP zone mapping (>0.80). This study also found that the best technique for prediction is random forest followed by CART and BRT in the case of Shipra river basin. Therefore, this study outcome can prove to be beneficial for effective management, protection, and exploration of groundwater prospects for the different stakeholders.