Sustainable and informed groundwater management in a critical karstic mining setting of western India: Application of Isotope-enabled Variational Bayesian-End Member Mixing Analysis

Pant, Diksha , Roy, Annadasankar , Mohokar, Hemant , Keesari, Tirumalesh

2026-05-01 GROUNDWATER FOR SUSTAINABLE DEVELOPMENT 2026   33(卷), null(期), (null页)

查看原文

Groundwater in mining regions faces critical challenges due to over-extraction, contamination, and complex recharge pathways, especially in semi-arid settings. Identifying and quantifying recharge sources is essential for managing water resource sustainably, as it clarifies the roles of natural inputs and human activities in replenishing stressed aquifers. The Jhamarkotra phosphate mine in Rajasthan, India, exemplifies these challenges, where intensive mining intersects with karstic dolomitic aquifers and variable lithologies, leading to overlapping isotopic signatures that hinder recharge source apportionment. This study applied environmental isotopes (delta H-2, delta O-18, and H-3) for source characterization and delta O-18 & EC as tracers for Variational Bayesian-End Member Mixing Analysis (VB-EMMA) framework to overcome these limitations and apportion the recharge sources-rainfall, surface water, and irrigation return flows-across shallow/deep aquifers. VB-EMMA improves upon traditional methods by incorporating endmember and observational uncertainties, applying a logistic-normal transformation suited to compositional data, and using Automatic Differentiation Variational Inference (ADVI) for efficient posterior estimation. Results showed irrigation return flow as the dominant fraction (Mean approximate to 0.40, 95% Highest Density Interval 0.20-0.68), followed by rainwater (0.31, 0.09-0.49) and surface water (0.29, 0.03-0.50). Shallow wells (10-50 m bgl) in stromatolitic dolomite showed dominant recharge from irrigation sources (similar to 45%), mid-depth wells (50-120 m bgl) in cherty quartzite and siliceous marble had balanced, slower recharge (similar to 32-37% each), and deep wells (>120 m bgl) in granite gneiss were sustained mainly by irrigation (42%) and rainfall (50%) with minimal surface water contribution. Lithology/land use patterns reinforced these estimates. Overall, VB-EMMA is an efficient tool supporting improved groundwater governance in complex mining and similar karstic aquifer settings that facilitates realising Sustainable Development Goals.