Regime-based interpretation of groundwater pollution using Gaussian mixture modelling and water pollution index in aquifer systems.

Index-based groundwater assessment reduces complex hydrochemical data to a single value, but it typically masks underlying chemical differences. This study applies a regime-based framework to groundwater in the Neogene aquifer of Al-Hassa Oasis using 300 samples. Gaussian Mixture Modelling (GMM) was used to identify seven hydrochemical regimes based on change in Bayesian Information Criterion (ΔBIC) selection, while Principal Component Analysis (PCA) was used to examine structure in reduced space. The results show that Water Pollution Index (WPI) values range from 1.46 to 1.56 across regimes. Although this range is narrow, statistical testing confirms significant differences. Regime-specific analysis reveals distinct associations among iron (Fe), manganese (Mn), and arsenic (As), as well as pH-dependent behaviour of vanadium (V) and molybdenum (Mo). Strontium (Sr) increases consistently with Total Dissolved Solids (TDS), indicating salinity-related enrichment. The findings show that similar WPI values can arise from different chemical compositions. This study provides a framework for linking pollution level with hydrochemical structure and supports more targeted groundwater monitoring in arid aquifers.