SOA Measurements vs. Models: a Status Report

The advent of fast and more detailed organic aerosol (OA) and VOC measurements in the last decade has allowed clearer model-measurement comparisons for OA and secondary OA (SOA). Here we summarize the patterns emerging from studies to date. <li>At least 8 studies have reported a large (x5-10) underestimation of SOA for polluted regions when using traditional models (those developed until ~2006) (Heald GRL05, Volkamer GRL06, Johnson ACP06, Kleinman ACP08, Matsui JGR09, Dzepina ACP09, Hodzic ACP09, Tsimpidi ACP09). This is especially obvious when models are evaluated with the ΔOA/ΔCO ratio. <li>Close to pollution sources, discrepancies of an order-of-magnitude in SOA lead to smaller discrepancies (often x2-3) for total OA due to the presence of primary OA (de Gouw EST09). Such OA discrepancies have been repeatedly observed (e.g. Vutukuru JGR06, McKeen JGR07&09, Heald JGR07, Fast ACP09, Hodzic ACP09). <li>The discrepancy is reduced when recently-updated yields for aromatics (Ng ACP07) and SOA from glyoxal (Volkamer GRL07) are used, and is eliminated when using SOA formation from S/IVOC (Robinson Sci07) although with an overprediction of SOA at long aging times (Dzepina ACP09; Hodzic ACP09b). It is not clear whether the urban discrepancy is removed for the right reasons. <li>4 evaluations of biogenic SOA formed in unpolluted regions find reasonable agreement between SOA from traditional models and field measurements (Tunved Sci06; Hodzic ACP09; Chen GRL09; Slowik ACPD09). One evaluation reports a significant underprediction (Capes ACP09), although the amount of precursor reacted was difficult to ascertain for that case. The difference with the systematic underprediction observed for anthropogenic SOA may be due to the lack of primary S/IVOC in biogenic emissions, or to other reasons (NOx, SO2, POA, etc.). <li>Comparisons for biogenic SOA formed in polluted regions are more complex. Several studies have reported a lack of clear influence of biogenic VOCs in SOA formation in polluted regions (de Gouw JGR05, GRL09; Weber JGR07; Bahreini JGR09), but 14C studies suggest a large fraction of modern C (Weber JGR07). Synergistic effects of pollution and BVOCs appear likely (e.g. Goldstein PNAS09). <li>SOA from biomass burning emissions appears variable in the field (Capes JGR09; Yokelson ACP09), likely due flaming vs. smoldering fraction and biomass identity, and perhaps also to prompt SOA formation triggered by HONO photolysis. Several studies report significant SOA formation, given enough photochemical processing. Models based on traditional precursors appear to underpredict SOA from BB sources (Grieshop ACP09; Hodzic ACP09). <li>The very large SOA source in the free troposphere postulated by Heald (GRL05) has not been reported in later studies. Dunlea (ACP09) did not find evidence of this source across the Pacific near North America, though precipitation removal precludes any strong conclusions. However Carlton (EST09) reported better comparisons after implementing in-cloud SOA formation. Future model evaluations should compare POA, SOA precursors, OVOCs, oxidants, and boundary conditions. Multiple OA measurements (WSOC, OC, AMS, molecular tracers, 14C, etc.) are necessary to overcome the limitations of any one method. Measurements of semivolatile species are critically needed to constrain models.