EPA published the results of the National-Scale Air Toxics Assessment for 1996 (NSATA) in January 2001. The EPA's Scientific Advisory Board has recommended that EPA include a more detailed analysis of uncertainty and variability in the NSATA. Therefore, EPA is working to design a framework for uncertainty and variability assessment that will address the following major components of the NSATA: 1) the emissions inventory and emission models, 2) transport and dispersion modeling, 3) exposure modeling, 4) risk estimates, and 5) comparison of model results with monitor data.

The National Toxics Inventory (NTI) emission inventory used in the 1996 NSATA includes hundreds of different emission source categories, and millions of individual emission records. An even larger emissions inventory was used to assess secondary formation of HAPs from volatile organic compounds. The framework for assessing uncertainty and variability must address not only the magnitude of emissions in the inventory, but also the emission locations, stack parameters, factors used to estimate diurnal emission variations, and models used to apportion area source emissions from the county level to Census tracts. In addition, the framework must address the uncertainties of the ASPEN atmospheric dispersion model, meteorological inputs, deposition rates, rates of HAP destruction and formation, the impacts of simplifying assumptions such as neglecting terrain effects and limiting the analysis to a 50 kilometer radius from emission sources, population activity and behavior patterns, different exposure scenarios and dose-response factors. This paper discusses the framework developed by EPA for estimating uncertainty and variability in the NSATA, and challenges encountered in the development of this framework.