Selecting panels for remediation in contaminated soils via stochastic imaging

Abstract

Selection of panels of contaminated soils for subsequent remediation should account for the panel volume and the underlying spatial uncertainty regarding the contaminant concentration at unsampled locations. For selection, a double threshold circumventing the change of support problem is proposed. Stochastic simulation of the contaminant concentration field allows the spatial uncertainty to be modeled. Selections of soil panels that are declared polluted based on traditional interpolated maps are compared to selections based on p-field simulation via a case study where the true contaminant concentration is known. Uncertainty in the final selection is assessed on a panel-support level and the resulting monetary impact is evaluated using a loss function approach. Additional sampling targets the most uncertain and most critical areas.