PROBABILISTIC SCREENING OF GROUNDWATER CONTAMINANTS

The research described in this paper extended a deterministic screening model for ground-water contamination to include specification of variability in water flux and soil and chemical properties. Uncertainty in these parameters was characterized by independent log-normal probability distributions and a Monte Carlo simulation process was used to estimate the expected percentages of chemical mass reaching the water table. These Monte Carlo results were compared with a deterministic analysis that ignored hydrologic and soil spatial variability. For many of the 49 chemicals analyzed in this study, the Monte Carlo results produced substantially higher estimates of groundwater contamination than the deterministic analyses, indicating that neglecting hydrologic and soil spatial uncertainty may underestimate the potential for pollution. However, for those chemicals and site conditions most likely to produce major contamination, the Monte Carlo and deterministic analyses produced comparable results.