Markov Chain Monte Carlo Techniques to Estimate Long Term Contaminant Leaching in Roadways
by
Tara J. Frizzell
Dept. of Mathematics and Statistics, University of New Hampshire, Durham, NH
Coauthors: Defne S. Apul, Ernst Linder

Estimates of long term leaching of contaminants are required to assess the environmental impact of traditional and recycled materials used in roadways. Release and transport of contaminants is dictated in part by the hydrology of the roadway environment. We use HYDRUS2D, a finite element, water movement and solute transport model to determine the hydraulic regimes and contaminant fluxes. We applied Markov Chain Monte Carlo techniques for estimation of hydrological model parameters using data from several different test road sections monitored and maintained by the Minnesota Department of Transportation. We report on several challenges related to correlations and range dependencies between parameters which require careful fine tuning in the Markov chain construction and implementation. This study demonstrates the applicability and usefulness of the Bayesian Monte Carlo approach for estimating environmental impact in the presence of high variability and uncertainty in field leaching conditions.

Date received: September 5, 2003