Vol. 9, Special Issue 9, Part D (2025)

This study investigates the spatio-temporal variations in groundwater levels across the Cholapur, Harhua, and Chiraigoan blocks of Varanasi district, Uttar Pradesh, using a 20-year dataset (2000-2019). Seasonal groundwater fluctuations were assessed for both pre-and post-monsoon periods within four depth intervals: <4 m, 4-8 m, 8-12 m, and 12-16 m. During the pre-monsoon season, the dominant range was 4-8 m across all periods, with its area proportion varying from 46.34% in 2000-2004 to 43.90% in 2015-2019, while deeper water levels (12-16 m) increased significantly in the latter years. Post-monsoon analysis exhibited a progressive decline in shallow water levels (<4 m) and an increase in deeper zones, indicating gradual depletion. Groundwater development stage reached a maximum of 98.85% in Chiraigoan block (2009) and a minimum of 61.37% in Cholapur block (2019), highlighting spatial variability in resource utilization. For predictive modelling, Artificial Neural Network (ANN)-based Multi-Layer Perceptron (MLP) and Co-Active Neuro-Fuzzy Inference System (CANFIS) models were developed. Model 4 (4-10-1) was optimal for pre-monsoon ANN prediction, whereas Model 1 (2-10-1) performed best for post-monsoon. CANFIS models (Model 4 for pre-monsoon and Model 3 for post-monsoon) demonstrated superior accuracy compared to ANN, indicating strong applicability in groundwater forecasting. The findings emphasize the utility of soft computing techniques for predicting water table depth using minimal hydrogeological inputs such as recharge, discharge, and depth.