Assessing groundwater potentialities and replenishment feasibility using machine learning and MCDM models considering hydro-geological aspects and water quality constituents

Abstract

Climate change has significantly impacted rainfall patterns, water availability, and security. Changes in rainfall alter the groundwater table, primarily sourced from rainfall in tropical regions, a crucial source of freshwater on Earth. Assessing its potentiality, quality, and replenishment feasibility continues to pose a challenge. Our study aims to identify potential groundwater zones to define artificial recharge zones by considering hydrogeological aspects and water quality. Additionally, the study aims to propose suitable recharge structures for different lithological groups in Kangsabati Upper Catchment. The present study used the extreme gradient boosting (XGBoost) algorithm and analytical hierarchy process (AHP) model to delineate the groundwater potential zones and suitable zones to replenish the water table. The XGBoost model evaluated the groundwater potential zones with 81% accuracy (SVM > RF > ANN) and identified various levels of potential. The area with very high and high prospects covers 23.36% and 20.14% respectively, while 20.32% and 13.94% of the area is covered by the low and very low prospect zones. On the other hand, according to the AHP approach, the estimated percentage of coverage for the classes is as follows: very good (< 1%), good (21.45%), moderate (57.53%), poor (15.63%), and unsuitable (5.21%). The study unveils that the east-central, east, north and the area within 300 m contour lines are ideal for both groundwater potential and replenishing the water tables. To achieve the objectives of Sustainable Development Goal (SDG) 6, effective strategies for suitable utilization and artificial recharge of water resources may result from implementing Machine Learning-Multiple Criteria Decision Making (ML-MCDM) models with pertinent influencing factors.