PAH mobility in cyontaminated industrial soils: A Markov chain approach to the spatial variability of soil properties and PAH levels

The consideration of spatially variable contaminant sources and sinks is crucial for the quantification of contaminant transport in industrial soils. To assess the seepage of polycyclic aromatic hydrocarbons (PAH) at a former manufactured gas plant, a combined approach was used comprised of a field survey, stochastic representation of site heterogeneity and numerical simulation of contaminant mobility. Based on field and laboratory data, the vertical transition probabilities of soil materials and PAH-contamination classes were derived and a non-stationary Markov chain model of site heterogeneity was developed. The model was used to generate representative soil profiles by stochastic simulation. Eighty profiles covered 61% of the spatial variability of the site in terms of soil forming materials and PAH levels. Positions and thickness of horizons agreed with the field survey. The seepage of different PAH (fluoranthene, pyrene, benzo(b)fluoranthene, phenanthrene, and benzo(a)pyrene) was calculated by numerical simulation using experimentally derived isotherm data. The cumulative output for the individual PAH covered a range of three orders of magnitude, demonstrating the effect of site heterogeneity on contaminant transport. While calculated local maximum concentrations exceeded the critical values for potable water, the weighted average of PAH concentration in seepage water was low. Stochastic generation of soil profiles based on Markov chain theory provides a powerful tool for the consideration of soil variability at contaminated industrial sites. Total profile probability relates to the area fractions represented by each profile. Therefore, contaminant seepage may be estimated without a costly three-dimensional deterministic representation of the field site.