A multi-method approach for the assessment of natural background levels in groundwater

The assessment of geochemical Natural Background Levels (NBLs) in groundwater, aims at distinguishing the naturally high levels of geogenic compounds from anthropogenic pollution. This is a fundamental issue in groundwater management, in particular when the concentration of inorganic compounds exceeds the threshold values set for the evaluation of the groundwater chemical status, as requested by environmental regulations. In this paper, we describe a new procedure that integrates the pre-selection method and statistical techniques, using the example of two case studies. The pre-selection aims to identify suitable groundwater samples for the NBLs assessment. The nitrate concentration threshold, for the removal of the groundwater samples affected by human activities, is established locally through different graphical and statistical approaches. Then, the statistical distribution of each compound is analyzed and the outliers are identified. Normality tests on the datasets allow one to select the most appropriate value, e.g. one percentile, to be adopted as NBL within the data distribution. In the selected case studies, we have defined the NBLs for As, F, Mn, Fe and SO4. The two sites are part of a volcanic-sedimentary aquifer in central Italy, where the geochemical background is frequently well above the standards for human consumption. The results of the simple and easily reproducible pre-selection method are strengthened by integration with statistical techniques, notably in selecting the appropriate percentile. New criteria are suggested for the choice of the nitrate threshold to be used for the pre-selection of uncontaminated samples. [Display omitted] •Natural background levels are assessed by using pre-selection and statistical methods.•The pre-selection of influenced samples is based on locally derived NO3 thresholds.•Outliers detection allows to identify data deviating from the background population.•The choice of the NBL depends on the normality of the distribution of residual data.•The integration of different approaches strengthens and improves the NBLs evaluation.