ABSTRACT
The radon-deficit technique is a powerful tool to detect and delineate sub-surface accumulations of organic contaminants. Field measurements of 222Rn in soil air, however, are affected by several confounding factors that can lead to the misinterpretation of results. Among the most influential are: vertical and lateral changes of lithology, fluctuating contaminant saturations with depth, varying water saturation ratios along the soil profile and atmospheric (and, therefore, soil) thermal oscillations. To evaluate and minimize the effect of these confounding factors on the interpretation of the results of the Rn deficit technique, a Matlab® based multi-layer model of 222Rn production-partition-diffusion in unsaturated porous media (1D_RnDPM: One-Dimensional 222Rn Diffusion and Partition Model) has been developed and is freely available as Supplementary Material in this work. A laboratory protocol has also been proposed to obtain site-specific input parameters for the model, i.e., 222Rn equilibrium concentration (as determined by the accumulation chamber method), soil bulk density and soil solid-phase density. The model predictions have been contrasted with field information obtained from successive sampling campaigns in which 222Rn in soil air was measured at a site where the vadose zone, consisting of an anthropogenic backfill underlain by a silt layer, is affected by a complex mixture of benzene, phenol, (poly) chlorobenzenes, (poly) chlorophenols and hexachlorocyclohexane isomers, among other compounds. The model has successfully predicted the vertical profile of 222Rn concentrations in soil air, including the effect of the oscillations of the water table and of ground-level temperature. The results also underline that 222Rn measurements in subsoil air are representative only of local conditions around the sampling point, an expected result given that 222Rn maximum effective diffusion length is very limited. As a consequence, the influence of a highly fluctuating water table at the site goes undetected at the sampling depths used in the field campaigns. MAIN FINDINGS: The combination of a numerical model and a laboratory protocol allows to predict the activity of 222Rn along the soil profile and to assess the influence of site-specific confounding factors.
