Modeling SOA contributions of VOC, IVOC and SVOC emissions and large uncertainties associated with OA aging

Abstract. Secondary organic aerosols (SOA) are an important component of atmospheric fine particulate matter (PM2.5) in China, and elsewhere, with contributions from anthropogenic and biogenic volatile organic compounds (AVOC and BVOC) and semi- (SVOC) and intermediate volatility organic compounds (IVOC). Policy makers need to know which SOA precursors are important but accurate simulation of SOA magnitude and contributions remains uncertain. We reviewed SOA modelling studies in the past decade that have reported the relative contributions of different precursors to SOA concentration and the findings have many inconsistencies due to differing emission inventory methodologies/assumptions, air quality model (AQM) algorithms, and other aspects of study methodologies. We investigated the role of different AQM SOA algorithms by applying two commonly used models, CAMx and CMAQ, with consistent emission inventories to simulate SOA concentrations and contributions for July and November 2018 in China. Both models have a volatility basis set (VBS) SOA algorithm but with different parameters and treatments of SOA photochemical aging. BSOA (SOA produced from BVOC) is found to be more important over southern China whereas SOA generated from anthropogenic precursors is more prevalent in the North China Plain (NCP), Yangtze River Delta (YRD), Sichuan Basin and Central China. Both models indicate negligible SOA formation from SVOC emissions as compared to other precursors. In July when BVOC emissions are abundant, SOA is predominantly contributed by BSOA (except for NCP), followed by IVOC-SOA (i.e. SOA produced from IVOC) and ASOA (i.e. SOA produced from anthropogenic VOC). In contrast in November, IVOC becomes the leading SOA contributor for all selected regions except PRD, illustrating the important contribution of IVOC emissions to SOA formation. Therefore, future control policies should aim at reducing IVOC emissions as well as traditional VOC emissions. While both models generally agree in terms of the spatial distributions and seasonal variations of different SOA components, CMAQ tends to predict higher BSOA while CAMx generates higher ASOA concentrations. As a result, CMAQ results suggest that BSOA concentration is always higher than ASOA in November while CAMx emphasizes the importance of ASOA. Utilizing a conceptual model, we found that different treatment of SOA aging between the two models is a major cause of differences in simulated ASOA concentrations. The step-wise SOA aging scheme implemented in CAMx (based on gas-phase reactions with OH radical and similar to other models) exhibits a strong enhancement effect on simulated ASOA concentrations and this effect increases with the ambient OA concentrations. The CMAQ VBS implements a different SOA aging scheme that represents particle-phase oligomerization and has smaller impacts, or no impact, on total OA. A brief literature survey shows that different structure and/or parameters of the SOA aging schemes are being used in current models, which could greatly affect model simulations of OA in ways that are difficult to anticipate. Our results indicate that large uncertainties still exist in the simulation of SOA in current air quality models due to the aging schemes as well as uncertainties of the emission inventory. More sophisticated measurement data and/or chamber experiments are needed to better characterize SOA aging and constrain model parameterizations.