Long-term particulate matter modeling for health effect studies in California – Part 2: Concentrations and sources of ultrafine organic aerosols

摘要

Organic aerosol (OA) is a major constituent of ultrafine particulate matter(PM). Recent epidemiological studies have identified associationsbetween PM OA and premature mortality and low birth weight. In thisstudy, the source-oriented UCD/CIT model was used to simulate theconcentrations and sources of primary organic aerosols (POA) and secondaryorganic aerosols (SOA) in PM in California for a 9-year (2000–2008) modeling period with 4 km horizontal resolution to provide moreinsights about PM OA for health effect studies. As a relatedquality control, predicted monthly average concentrations of fineparticulate matter (PM) total organic carbon at six major urbansites had mean fractional bias of −0.31 to 0.19 and mean fractional errorsof 0.4 to 0.59. The predicted ratio of PM SOA ∕ OA was lower thanestimates derived from chemical mass balance (CMB) calculations by a factorof 2–3, which suggests the potential effects of processessuch as POA volatility, additional SOA formation mechanism, and missingsources. OA in PM, the focus size fraction of this study, isdominated by POA. Wood smoke is found to be the single biggest source ofPM OA in winter in California, while meat cooking, mobile emissions(gasoline and diesel engines), and other anthropogenic sources (mainlysolvent usage and waste disposal) are the most important sources in summer.Biogenic emissions are predicted to be the largest PM SOA source,followed by mobile sources and other anthropogenic sources, but theserankings are sensitive to the SOA model used in the calculation. Airpollution control programs aiming to reduce the PM OA concentrationsshould consider controlling solvent usage, waste disposal, and mobileemissions in California, but these findings should be revisited after thelatest science is incorporated into the SOA exposure calculations. Thespatial distributions of SOA associated with different sources are notsensitive to the choice of SOA model, although the absolute amount of SOAcan change significantly. Therefore, the spatial distributions of PMPOA and SOA over the 9-year study period provide useful information forepidemiological studies to further investigate the associations with healthoutcomes.