Evaluation of natural background levels of high mountain karst aquifers in complex hydrogeological settings. A Gaussian mixture model approach in the Port del Comte (SE, Pyrenees) case study

Sci Total Environ. 2021 Feb 20;756:143864. doi: 10.1016/j.scitotenv.2020.143864. Epub 2020 Nov 28.

Abstract

The hydrogeological processes driving the hydrochemical composition of groundwater in the alpine pristine aquifer system of the Port del Comte Massif (PCM) are characterized through the multivariate statistical techniques Principal Component Analysis (PCA) and Gaussian Mixture Models (GMM) in the framework of Compositional Data (CoDa) analysis. Also, the groundwater Natural Background Levels (NBLs) for NO3 and SO4 and Cl are evaluated, which are specially important for indicating the occurrence of groundwater contamination derived from the anthropic activities conducted in the PCM. The different hydrogeochemical facies found in the aquifer system of the PCM comprises low mineralized Ca-HCO3 water for the main Eocene karst aquifer, and Ca-SO4 and highly mineralized NaCl water types in the minor aquifers discharging from the PCM. The NBL values of SO4, Cl and NO3 obtained for the main karst aquifer are 14.33, 4.06 and 6.55 mg/L, respectively. These values are 35, 3 and 1.2 times lower than the respective official NBLs values that were determined by the water administration to be compared with in the case of conducting a pollution assessment characterization in the main karst aquifer. Official overestimation of NBLs can put important groundwater resources in the PCM at risk.

Keywords: Compositional data; Gaussian mixing model; High-mountain karst system; Model-based clustering; Natural background levels.