 |
NARSTO Workshop on Innovative Methods for Emission Inventory Development and Evaluation University of Texas, Austin October 14-17, 2003 |
 |
Quantification of Uncertainty in Emission Factors and Emission Inventories
H. Christopher Frey, Ph.D., Associate Professor
Water Resources and Environmental Engineering Program, Department of Civil Engineering
North Carolina State University, Raleigh, NC 27695-7908
Over the last five years, work has been underway at NCSU to demonstrate methods for quantification of uncertainty in emission factors and emission inventories applied to case studies for point, area, and mobile sources with respect to nitrogen oxides, volatile organic compounds, carbon monoxide, and hazardous air pollutants. Methods include the use of expert judgment and empirical approaches. The latter includes bootstrap simulation to quantify uncertainty based upon either empirical or parametric distributions of inter-unit variability in emission and activity factors. Techniques for dealing with mixtures of distributions, censored (non-detected) data, measurement errors, and dependencies have been demonstrated. Two prototype software tools have been developed. AUVEE demonstrates the quantification of uncertainty in emission and activity factors, and for an inventory, based upon the example of utility NOx emissions, for averaging times of six to twelve months. AuvTool is a stand-alone tool for fitting alternative parametric distributions to data, evaluating goodness-of-fit, and quantifying uncertainty in selected statistics. This talk will briefly review methods supported with illustrative examples based upon data for specific source categories. For example, in recent work, a probabilistic hourly NO emission inventory for 32 units of 9 coal-fired power plants in the Charlotte domain was propagated through the Multiscale Air Quality Simulation Platform. Intra-unit autocorrelation in emissions and inter-unit dependence were accounted for. Work is currently underway with regard to uncertainty in air toxic emissions in the Houston area. The implications of results of this ongoing research program with regard to quantification of uncertainty in emission inventories are addressed.