Seasonal Groundwater Table Depth Prediction Using Fuzzy Logic and Artificial Neural Network in Gangetic Plain, India

Modelling and forecasting of groundwater level fluctuation are essential for sustaining groundwater availability. Therefore, it is necessary to carry out mechanisms and methods that can predict the groundwater level precisely. This study investigates the comparative potential of FL (Fuzzy Logic) and ANN (Artificial Neural Network) to predict the seasonal groundwater table depth (GWTD) between the Ganga and Hindon rivers area located in Uttar Pradesh State, India. The groundwater recharge (GWR), groundwater discharge (GWD), and the antecedent groundwater level data of 21-years (1994–2014) have been utilized to formulate 18 models (nine for pre-monsoon, and nine for post-monsoon) for training and testing of FL and ANN techniques. The outcomes of FL-based models were evaluated against the ANN models based on performance indicators and graphical inspection for the prediction of seasonal GWTD. The comparison of results reveals that the FL-based models performed better than the ANN models during both seasons for GWTD prediction in the study region. The results derived from this study would help the hydrologists and policymakers to formulate a better plan of action for governance under extreme conditions and conservation of groundwater resources in the study region.