Integrating fuzzy C-means clustering and shannon entropy with hydrogeochemistry for irrigation water suitability mapping in semi-arid hard-rock terrains.

Groundwater contamination threatens agricultural sustainability in hard-rock regions. This study presents a novel integration of Fuzzy C-Means (FCM) clustering and Fuzzy Shannon Entropy for evaluating groundwater quality at a micro-watershed scale-an approach not previously applied in the Bandu sub-watershed within Purulia district, India. A total of 64 groundwater samples were collected during the post-monsoon season from dug wells, tube wells, and submersible pumps, and analysed for physicochemical parameters, major ions, and irrigation indices (SAR, MAR, PI, RSC) using standard protocols. Hydrogeochemical assessment showed that 83% of samples fall within the Ca-Mg-HCO3 facies, indicating dominant rock-water interaction. Sodium hazards were low (SAR 0.23-2.81), with most samples classified as C2S1. However, magnesium posed a major constraint, as 60% of samples exceeded the critical MAR limit, and PCA was used to extract and analyze magnesium-salinity processes. FCM clustering delineated two hydrogeochemical zones: Cluster I (64% of samples) with good irrigation suitability and Cluster II (36%) with higher salinity stress (PC = 0.866, ASW = 0.640). Entropy-based prioritization classified the watershed into high (22%), moderate (63%), and low (15%) irrigation potential zones, with a prediction accuracy of 77.3% (AUC = 0.841). The integrated fuzzy-statistical framework offers an effective decision-support tool for micro-watershed management. The findings provide actionable insights for policy formulation, including targeted soil amendment strategies, improved irrigation scheduling, and sustainable agricultural planning in magnesium-affected hard-rock terrains.