Assessment of Uncertainty in Urban Air Toxics Simulations: Monte Carlo Study of the Houston Ship

摘要

Because of the interest in more accurately assessing air toxics exposure in urban areas, a Monte Carlo (MC) probabilistic uncertainty study has been conducted by the American Petroleum Institute (API). The API study, located on a 15 km by 15 km receptor domain centered on the Houston Ship Channel, complements a MC uncertainty study that was carried out by the EPA for the entire Houston metropolitan area. The focus of the API study is on uncertainties in model predictions of annual averaged concentrations of benzene and 1,3-butadiene, due to uncertainties in model formulation, emissions and meteorology inputs, and model dispersion parameters. Uncertainties in model formulation were addressed by running both ISCST3 and AERMOD. Typical emission sources were grouped into 21 categories for benzene and 13 categories for 1,3 butadiene. Uncertainly of emissions in each category was estimated based on observed data variability and supplemented by guidance from an API-EPA expert elicitation workshop held on this topic (typical emissions uncertainties are about +/- a factor of three (i.e., covering the 95 % range). Uncertainties in meteorological inputs (such as wind speed, wind direction, mixing height, cloud cover and temperature gradient) were determined from analysis of the field data plus consultation with experts. Uncertainties in the dispersion coefficients calculated by the models were evaluated based on consultation with experts.