Optimizing Sampling for Acceptable Accuracy Levels on Remediation Volume and Cost Estimations

Abstract

The aim of the paper is to present an iterative approach, based on geostatistical methods, to optimize sampling according to financial and environmental criteria. At current stage j of sampling, if the accuracy on remediation volume and cost estimates is not considered as sufficient, we try to anticipate the number of samples that needs to be collected at stage j+1, to reach an acceptable accuracy level. Sampling of various numbers Nj+1 of additional data is modelled, based on one simulation of the pollutant concentrations generated at stage j, conditioned with the available experimental data, in the area where the probabilities of exceeding the remediation cutoff are too high. In the variogram model fitted at stage j, remediation volumes and costs are recalculated with the various Nj+1 additional conditional data. If necessary, the process is repeated. The approach is illustrated on the site of a former smelting works presenting a lead pollution. Since the uncertainties on the remediation volume and cost estimates at the sixth real sampling stage are not satisfactory, a number of additional N7 data is chosen according to volume and cost forecasts calculated for various N7. The choice is non unique since various criteria, objectives, constraints and decision makers preferences can be taken into account. As an example, it is shown which number N7 will be chosen by a risk averse decision maker or by a risk prone decision maker, according to four common environmental and financial objectives.