A multivariate non-parametric approach for estimating probability of exceeding the local natural background level of arsenic in the aquifers of Calabria region (Southern Italy)

The concept of natural background level (NBL) aims at distinguishing the natural and anthropogenic contributions to concentrations of specific contaminants, as groundwater management and protection tools. This is usually defined as a unique value at a regional scale, even when the hydrogeological and geochemical features of a certain territory are far from homogeneous. The concentration of target contaminants is affected by multiple hydrogeochemical processes. This is the case of arsenic in the Calabria region, where concentrations are definitely variable in groundwater. To overcome the limitation of a traditional approach and to include the intrinsic hydrogeological and geochemical heterogeneity into the definition of the natural contribution to As content in groundwater, an integrated probabilistic approach to the NBL assessment combining aquifer-based preselection criteria and multivariate non-parametric geostatistics was proposed. In detail, different NBL values were selected, based on the aquifer type and/or hydrogeochemical features. Then, these aquifer-based NBL values of arsenic were used in the Probability Kriging method to map the probability of exceedance and to provide contamination risk management tools. This multivariate geostatistical approach that takes advantage of the physico-chemical variables used in the aquifer-based NBL values definition allowed mapping the probability of exceedance of As in a physically-based way. The hydrogeochemical diversity of the study area and all the processes affecting As concentrations in the aquifers have been considered too. As a result, the obtained map was characterized by a short-range and long-range variability due to local hydrogeochemical anomalies and water-rock interaction and/or atmospheric precipitation. By this approach, the NBL exceedance probability maps proved to be less “noisy”, because the local hydrogeochemical conditions were filtered, and more capable of pointing out anthropogenic inputs or very anomalous natural contributions, which need to be investigated more in detail and properly managed.