A multi-stage sampling strategy for the delineation of soil pollution in a contaminated brownfield

摘要

A multi-stage sampling strategy, based on sequential Gaussian simulation, was presented to optimize the step-wise selection of a small numbers of additional samples to delineate soil pollution. This strategy was applied to a Belgian brownfield of 5.2 ha polluted with lead (Pb). Starting from an initial number of 240 samples in stage 1, additional samples were added, 25 per stage, and the reduction of the uncertainty in the Pb delineation was monitored. Twenty stages were used. Already in stage 6 a local optimum was found based on the median conditional coefficient of variation. An independent validation confirmed that this index was to be preferred over the median conditional variance. So for the brownfield considered our procedure indicated that 365 selected samples would have been sufficient, representing a gain of 70.7% in sampling effort compared to current practice which resulted in a sampling effort of 1245 samples.