Hydrogeochemical Characterization of an Intermontane Aquifer Contaminated with Arsenic and Fluoride via Clustering Analysis

Abstract

The controlling hydrogeochemical processes of an intermontane aquifer in central Mexico were identified through multivariate statistical analysis. Hierarchical cluster (HCA) and k-means clustering analyses were applied to Na+, K+, Ca2+, Mg2+, F−, Cl−, SO42−, NO3−, HCO3−, As, pH and electrical conductivity in 40 groundwater samples collected from shallow and deep wells, where As and F− are contaminants of concern. The effectiveness of each hierarchical and k-means clustering method in explaining solute concentrations within the aquifer and the co-occurrence of arsenic and fluoride was tested by comparing two datasets containing samples from 40 and 36 wells, the former including ionic balance outliers (>10%). When tested without outliers, cluster quality improved by about 5.4% for k-means and 7.3% for HCA, suggesting that HCA is more sensitive to ionic balance outliers. Both algorithms yielded similar clustering solutions in the outlier-free dataset, aligning with the k-means solution for all 40 samples, indicating that k-means was the more robust of the two methods. k-means clustering resolved fluoride and arsenic concentrations into four clusters (K1 to K4) based on variations in Na+, Ca2+, As, and F−. Cluster K2 was a Na-HCO3 water type with high concentrations of As and F. Clusters K1, K3, and K4 exhibited a Ca-HCO3, Na-Ca-HCO3, and Ca-Na-HCO3 water types, respectively, with decreasing As and F concentrations following the order K2 > K3 > K1 > K4. The weathering of evaporites and silicates and Na-Ca ion exchange with clays were the main processes controlling groundwater geochemistry. The dissolution of felsic rocks present in the aquifer fill is a likely source of As and F−, with evaporation acting as an important concentration factor.