Evaluating the impact of new observational constraints on P-S/IVOC emissions, multi-generation oxidation, and chamber wall losses on SOA modeling for Los Angeles, CA
- 1Department of Chemistry, Université de Montréal, Montréal, QC, Canada
- 2Center for Atmospheric Particle Studies, Carnegie Mellon University, Pittsburgh, PA, USA
- 3Department of Environmental Science, Policy and Management, University of California, Berkeley, CA, USA
- 4Department of Civil and Environmental Engineering, University of California, Berkeley, CA, USA
- 5Cooperative Institute for Research in the Environmental Sciences and Dept. of Chemistry and Biochemistry, University of Colorado, Boulder, CO, USA
- 6Laboratory of Atmospheric Chemistry, Paul Scherrer Institute, Villigen, Switzerland
- 7Department of Chemistry and Biochemistry & Oeschger Centre for Climate Change, University of Bern, Bern, Switzerland
- anow at: National Physical Laboratory, Hampton Rd, Teddington, Middlesex, UK
- bnow at: Air Pollution Control Division, Colorado Department of Public Health and Environment, Denver, CO, USA
- cnow at: Lucerne University of Applied Sciences and Arts, School of Engineering and Architecture, Bioenergy Research, Technikumstrasse 21, 6048 Horw, Switzerland
Abstract. Secondary organic aerosol (SOA) is an important contributor to fine particulate matter (PM) mass in polluted regions, and its modeling remains poorly constrained. A box model is developed that uses recently published literature parameterizations and data sets to better constrain and evaluate the formation pathways and precursors of urban SOA during the CalNex 2010 campaign in Los Angeles. When using the measurements of intermediate-volatility organic compounds (IVOCs) reported in Zhao et al. (2014) and of semi-volatile organic compounds (SVOCs) reported in Worton et al. (2014) the model is biased high at longer photochemical ages, whereas at shorter photochemical ages it is biased low, if the yields for VOC oxidation are not updated. The parameterizations using an updated version of the yields, which takes into account the effect of gas-phase wall losses in environmental chambers, show model–measurement agreement at longer photochemical ages, even though some low bias at short photochemical ages still remains. Furthermore, the fossil and non-fossil carbon split of urban SOA simulated by the model is consistent with measurements at the Pasadena ground site.
Multi-generation oxidation mechanisms are often employed in SOA models to increase the SOA yields derived from environmental chamber experiments in order to obtain better model–measurement agreement. However, there are many uncertainties associated with these aging mechanisms. Thus, SOA formation in the model is compared to data from an oxidation flow reactor (OFR) in order to constrain SOA formation at longer photochemical ages than observed in urban air. The model predicts similar SOA mass at short to moderate photochemical ages when the aging mechanisms or the updated version of the yields for VOC oxidation are implemented. The latter case has SOA formation rates that are more consistent with observations from the OFR though. Aging mechanisms may still play an important role in SOA chemistry, but the additional mass formed by functionalization reactions during aging would need to be offset by gas-phase fragmentation of SVOCs.
All the model cases evaluated in this work show a large majority of the urban SOA (70–83 %) at Pasadena coming from the oxidation of primary SVOCs (P-SVOCs) and primary IVOCs (P-IVOCs). The importance of these two types of precursors is further supported by analyzing the percentage of SOA formed at long photochemical ages (1.5 days) as a function of the precursor rate constant. The P-SVOCs and P-IVOCs have rate constants that are similar to highly reactive VOCs that have been previously found to strongly correlate with SOA formation potential measured by the OFR.
Finally, the volatility distribution of the total organic mass (gas and particle phase) in the model is compared against measurements. The total SVOC mass simulated is similar to the measurements, but there are important differences in the measured and modeled volatility distributions. A likely reason for the difference is the lack of particle-phase reactions in the model that can oligomerize and/or continue to oxidize organic compounds even after they partition to the particle phase.