Uncertainty assessment of radiation inventories in a contaminated site

Abstract

The Canadian nuclear laboratories (CNL) periodically monitor several sites to collect information on surface and subsurface radiation contamination by means of point or small-area measurements. These measurements are used to estimate spatial distribution and inventories of total radiation. The spatial interpolation of the contamination data is commonly performed using the statistical kriging method. Kriging produces predictions at unsampled locations that are further used for estimating the radiation inventory of the contaminated site. However, the kriging output produces a point estimate of the inventory that does not include the uncertainties that may come from different sources.

This study proposes an approach for uncertainty assessment of radiation inventories based on the geostatistical conditional simulation method – a simulation technique that honours the observations at the sampled locations. The histograms of the radiation inventories are obtained by generating several conditional simulations of the prediction map using the fitted kriging model. A practical application of the proposed method is shown by estimating the total beta inventories using groundwater monitoring data at a contaminated site.