Sensitivity study of six public health risk computation cases from the US Department of Energy risk- and cost-estimate process pilot study

Abstract

This report contains a description of the results from the analysis of the sensitivity of estimated public health risks to changes in model parameters relating to the contaminant source releases, contaminant transports, and human exposures contaminants from six waste sites. Estimated public health risks associated with these and other sites at US Department of Energy (DOE) compounds were reported in a pilot study done by the Oak Ridge National Laboratory (ORNL) for the DOE (ORNL 1992). The objective of the sensitivity analysis was to identify the subset of model input parameters whose variations accounted for the majority of the variation in the computed public health risk values. All environmental modeling in this study and the pilot study done by ORNL (1992) was based on the Multimedia Environmental Pollutant Assessment System (Whelan et al. 1992). The results of the sensitivity analysis for the atmospheric case indicate that the most influential variables were emission rate and, to a lesser extent, population size. For groundwater cases, there was no consistent ordering of the influential variables. Depending on the case considered, some influential variables include the following: Equilibrium partition coefficient (K{sub d}), size of population, pore water velocity, constituent inventory, contaminant flux rate from source, and thickness of saturated zone. For the overland transport case, the regression model fit was not adequate for a reliable identification of the influential variables.