Constraining uncertainties about the sources and magnitude of polycyclic aromatic hydrocarbon (PAH)levels in ambient air: the State of Minnesota as a case study

Emissions data are often lacking or uncertain for many airborne contaminants. Chemicals, such as polycyclic aromatic hydrocarbons (PAHs), emitted from combustion sources, fall into this category. Currently available ambient-air emission inventories of PAHs either fail to account for population-based activities (such as residential wood combustion and motor vehicle activity) and/or report 'total PAH' or particulate organic matter emissions instead of individual compounds. We measure the degree of overlap between predicted concentrations from estimated emissions with measured concentrations. Our analysis is, based on probabilistic analysis of measured outdoor air concentrations with those predicted from mass-balance models. . Based on available information, we estimate the relative magnitude of emissions from four major sources ofPAHs to outdoor air- (1) on-road motor vehicles, including light-duty gasoline vehicles and diesel-powered buses and medium and heavy duty trucks; (2) residential wood combustion; and (3) power generation from external combustion boilers. We use the CalTOX regional multimedia mass-balance model to evaluate our emissions estimates in rural and urban regions of the state of Minnesota, USA. We compare model estimatesof outdoor PAH airborne concentrations with those reported by the Minnesota Children's Pesticide Exposure Study (MNCPES). With these measured concentrations we probabilistically evaluate our emissions and interpret the reliability of our emissions estimates for specific PAHs. The median estimates of our predicted outdoor air concentrations agree within an order of magnitude of measured concentrations. For fourrepresentative PAHs, we were able to obtain a reasonable degree of overlap between empirical and predicted distributions of outdoor air concentrations. Our combination of models, emissions estimates, and empirical concentration data estimate exposure in a manner that is more reliable than any of these tools alone. Thereby, we increase our confidence about our plausible ranges of emissions and predicted concentrations