A Probabilistic and Conditional Approach for Delineating the Extent of a Contamination Plume within an Aquifer

Abstract

This paper addresses the issue of assessing, in a probabilistic manner, the extent of a potential contamination plume in an aquifer when scant water table and concentration data are available. This initial assessment serves as a basis for locating additional piezometer that may be required to confirm the contamination diagnostic. A novel approach is proposed, whereby an analytical model of transport is used first to generate a series of contaminant concentration maps, each map resulting from a different, yet equiprobable, set of parameters. This suite of concentration images is then summarized into an a priori contamination probability map, describing the probability that, at any location, the contamination concentration exceeds a critical threshold. Finally, this prior probability map is conditioned to available piezometer information using the geostatistical technique of simple indicator kriging with local prior means. This new approach is illustrated by a case study of an aquifer contamination by polycyclic aromatic hydrocarbon (PAH), resulting from former coal-processing activities.