A general method for creating new datasets for soil geochemical mapping based on multiple pre-existing geochemical datasets

Pereira, Benoît [UCL] Vandeuren, Aubry [UCL] Sonnet, Philippe [UCL]

When undertaking a mapping of geochemical concentrations in the soil, it is often found that several datasets already exist. These datasets originates for instance from geochemical characterization surveys or from the collection of sample analyses requested by environmental protection regulatory bodies. The use of all the available datasets is often a necessity if one has to provide high resolution maps on a large study area. However, analyses from these geochemical datasets were generally carried out by different laboratories at different periods of time. Although these laboratories are required to follow strict protocols and quality control procedures, these datasets often comprise analytical results of questionable quality. In addition, strong biases may exist between datasets. Here we present a general method which addresses these issues thereby making it possible to use data from pre-existing datasets to produce one reliable dataset for soil geochemical mapping. The method consists in two main stages. Stage 1 includes four checks to establish a higher level of guarantee concerning the relevance of the data for soil geochemical mapping. Stage 2 consists in checking for biases between datasets. If any biases are detected, appropriate data leveling is applied at this stage. The method is illustrated with a collection of soil geochemical datasets of georeferenced soil analyses in Wallonia (South Belgium).