The identification of groundwater bodies impacted by inputs of anthropogenic origin is essential in the context of the management (knowledge, protection, remediation) of water resources at the scale of territory. This identification is complex in the case of mining areas impacted by releases of metallic trace elements which are superimposed on natural concentrations. These can be elevated due to natural mineralization. The detection of enrichment relative to the geochemical background noise therefore requires a methodology adapted to this type of context. The objective of this study proposes an approach integrating regional geological and geomorphological factors to improve the detection and quantification of metallic trace element enrichments in groundwater bodies compared to geochemical background noise, in a mining context. From an assumption of normal distribution of the values of the natural geochemical background, a natural component, it is assumed, at an arbitrarily set confidence level, that the disturbances of anthropogenic origin are revealed by superposition. This superposition is revealed on the basis of three types of independent parameters, linked to the study area comprising areas of past or current mining activities: 1 ‐ of geochemical type (median and variance of various subsets of these samples), 2 ‐ geological type (natural predisposition of samples to be enriched by hydrogeochemical processes) and 3 ‐ geomorphological type (hydraulic gradient, distance from sampling points to places known to be the site of anthropogenic activities). The geochemical parameter makes it possible to adjust the global distribution by superimposing two distribution functions: the apparent natural component and the apparent anthropogenic component, and to identify values as “atypical data”. These components are qualified as apparent because they can contain “fuzzy data”: data whose value, alone, does not allow to statistically determine the component of connection (data having undergone an anthropogenic influence, but insufficiently for the characterize as atypical data). The integration of geological and geomorphological parameters makes it possible to filter these fuzzy data to improve the characterization of the natural component. An application to the detection of arsenic and antimony enrichments in the groundwater of the Gardons watershed in the South of France, having undergone intense mining in the 19th and 20th centuries, is presented. The method makes it possible to improve the detection of enrichment in metallic trace elements of anthropogenic origin, at variable spatial scales, less than 10 kilometers, with an integration of local geological, hydrogeological and topographic conditions.