Groundwater level dynamics in a subtropical fan delta region and its future prediction using machine learning tools: Sustainable groundwater restoration

Abstract

Study regionThe Damodar Fan Delta, West Bengal, India. Study focusThe depletion of groundwater resources worldwide is escalating due to its profuse demand for drinking, irrigation, domestic, and industrial uses. Overexploitation of groundwater in a subtropical fan delta region with rapid population growth like the Damodar Fan Delta in India is of great concern for sustainable mapping, monitoring, and managing water resources. During 2000–2020, the Damodar Fan Delta portrayed an increase in semi-critical community development blocks, implying a decline in groundwater level. To this end, the present study intends to show the groundwater level dynamics including its future prediction, using machine learning algorithms based on the seasonal groundwater level data from 2013‐14 to 2020‐21 for 30 wells. New hydrological insights for the regionPost-monsoon kharif and rabi depicted a higher fall rate in the groundwater level compared to the pre-monsoon and monsoon periods. Future predictions using the best-fit model indicated an increasing trend in the depth of groundwater levels in the future (2025–26). The extreme gradient boost regressor appeared to be the best model, while the decision tree regressor was the worst performer. The major controlling factors of groundwater level dynamics were decreasing rainfall and increasing groundwater abstraction due to population growth and increased demand for irrigation, domestic, and industrial water.