Parameters describing nonequilibrium transport of polycyclic aromatic hydrocarbons through contaminated soil columns: Estimability analysis, correlation, and optimization
Résumé
The soil and groundwater at former industrial sites polluted by polycyclic aromatic hydrocarbons (PAHs) produce a very challenging environmental issue. The description of PAH transport by means of mathematical models is therefore needed for risk assessment and remediation strategies at these sites. Due to the complexity of release kinetics and transport behavior of the PAHs in the aged contaminated soils, their transport is usually evaluated at the laboratory scale. Transport parameters are then estimated from the experimental data via the inverse method. To better assess the uncertainty of optimized parameters, an estimability method was applied to firstly investigate the information content of experimental data and the possible correlations among parameters in the two-site sorption model. These works were based on the concentrations of three PAHs, Acenaphthene (ACE), Fluoranthene (FLA) and Pyrene (PYR), in the leaching solutions of the experiments under saturated and unsaturated flow conditions. The estimability results showed that the experiment under unsaturated flow conditions contained more information content for estimating four transport parameters than under the saturated one. In addition, whatever the experimental conditions for all three PAHs the fraction of sites with instantaneous sorption, f, was highly correlated with the adsorption distribution coefficient, Kd. The very strong correlation between the two parameters f and Kd suggests that they should not be simultaneously calibrated. Transport parameters were optimized using HYDRUS-1D software with different scenarios based on the estimability analysis results. The optimization results were not always reliable, especially in the case of the experiment under saturated flow conditions because of its low information content. In addition, the estimation of transport parameters became very uncertain if two parameters f and Kd were optimized simultaneously. The findings of the current work can suggest some reasons behind the optimization problems and indicate the type of experimental information additionally needed for parameter identification. To overcome the parameterization issues of PAH non-equilibrium transport, the experimental design, timescale, and model refinement need further improvement. The conclusions presented in this paper are not limited necessarily to PAHs, but may also be relevant to other organic contaminants with similar leaching behavior.