| The quantitative characterization of variability and uncertainty in fate, transport, exposure and dose-response assessments of Hazardous Air Pollutants (HAPs) is very important for human health and ecological risk assessments. U.S. EPA's National-Scale Air Toxics Assessment (NATA) outlines a "top-down" multiplicative model that attempts to summarize the uncertainty in air toxic emissions, the transformation from emissions to ambient concentrations, the relationship between ambient concentrations and personal exposure, and the exposure-dose-response relationship for selected HAPs. The objective is to not only characterize quantitatively the uncertainty and variability in risk estimates, but to also identify the major sources of variability and uncertainty, and quantify the relative contribution of each source to the overall variance and range of the model result. This dissertation presents statistical models to address this objective.; A Bayesian hierarchical model is used to estimate fugitive lead emissions from secondary lead smelting facilities, from observed lead ambient concentration data. An aggregate national distribution of fugitive emissions from such facilities is also estimated that characterizes the uncertainty and variability. The National Toxics Inventory (NTI) is shown to significantly underestimate heavy metal emissions, a significant portion of which may be attributed to emissions due to resuspension of contaminated soil particles. The improved estimates can have important implications in industrial ecology and material-flow analysis. Finally, a simple analytical model is developed that characterizes the uncertainty in health risk due to hazardous air pollutants by estimating, characterizing and combining the uncertainties in atmospheric fate-and-transport models, personal exposure estimates, and the dose-response function. The relative contribution of each source to the total uncertainty in risk estimate is quantified, which can have significant policy implications in risk management. |
