Towards geochemical quality indicators of urban soils. Example of the "Ile de Nantes"(France).

International audienceOptimized management of excavated soil in urban areas involve setting up channels for reuse of unpolluted or slightly polluted soil, to limit landfill, widely practiced at present in France. In this frame, the French guide on the management of excavated materials proposes to ensure the compatibility of these materials with the geochemical background of the soils of the receiving site. But the definition of urban geochemical background and its representation issues are still under consideration. Indeed, if we consider that the geochemical background matches the usually encountered geochemistry in a given area, then its application to highly heterogeneous urban areas raise the question of the definition of "local diffuse pollution".Basing our reflection on a real site, we confront the usual statistical approaches for calculating the geochemical background to spatial approaches. The case study site is a central quarter of the agglomeration of Nantes (France). This former industrial quarter under redevelopment generates a large volume of excavated soil (about 100 000 tones per year are expected between 2015 and 2025). The data used come from pollution diagnostics, which provide a significant amount of analysis on the entire area (about 1850 samples analyzed identified).Initially, conventional methods of calculating the values of geochemical background are compared: upper whisker, 90th and 95th percentile and "Median Absolute Deviation". These calculations do not take into account the geographical position of data. Furthermore, the result of each calculation is reduced to a single value, while a range or a distribution could be preferable in practice, particularly for the guidance of excavated soils. These calculations are applied taking into account the soil typology developed by the BRGM and including three types of made grounds (questionable, various, and natural like) according to their intrinsic potential of pollution. If they show variations in geochemistry consistent with our knowledge of this typology of soil, the results strongly depend on the method of calculation. Furthermore, treatment of high values, at this stage, is simplistic, as values above a predefined threshold are deleted.In a second step, a spatial approach is examined to determine urban geochemical background. This geostatistical mapping by kriging analysis uses land use descriptors as well as geology. The kriging analysis can separate the different scales of spatial variability, which can be used to define geochemical backgrounds.Comparison of the two approaches shows the importance of mapping for the development of geochemical quality indicators of urban soils. These latter can support the management of excavated soil