Spatial prediction of the variability of Early Pleistocene subsurface sediments in the Netherlands - Part 2: Geochemistry

We started a geochemical mapping campaign in the Early Pleistocene fluviati le Kedichem Formation in the Netherlands in order to meet the demand for mo re information about subsurface sediment compositions. Geochemical data wer e collected during a sampling campaign, and about 600 samples from the Kedi chem Formation were analyzed. By linking the geochemical data with litholog ical classifications from the TNO-NITG borehole database, we established a geochemical prediction model. Elements were divided into classes according to their geochemical behaviour in relation to lithological parameters. For each of the classes, we combin ed lithological groups in to groups with relevant geochemical differences. By calculating for each element the average composition in each of these gr oups, we were able to predict the geochemical composition of subsurface sed iments by 'translating' the spatial lithological data from the TNO-NITG bor ehole database into geochemical data. We visualized this model by calculati ng and interpolating the average composition of horizontal slices of the Ke dichem Formation. The model performance is fairly good, although it has a t endency to underestimate extreme values.