Evaluation of Secondary Organic Aerosol (SOA) Simulations for Seoul, Korea

Organic aerosols (OA) represent a significant fraction of total submicron particulate matter (PM1) concentrations globally, including densely populated megacities such as Seoul. However, scientific understanding of the atmospheric formation and removal processes of OA, especially for secondary organic aerosols (SOA), is still highly uncertain. In this study, we examine the characteristics of SOA formation in Seoul during spring-summer 2016 and fall-winter 2017/2018, using airborne and ground observations along with a 3-D global chemical transport model, GEOS-Chem. We use four different SOA schemes in the model, including simplified and complex volatility-based frameworks, and evaluate them by comparing the simulations with the observations to examine how our scientific understanding embedded in each SOA scheme affects the observed biases. Our analysis of the model performance of each scheme also provides the most suitable approach in simulating SOA in a typical urban environment. Comparisons of the simulated versus observed OA concentrations show that model biases range from −72% to +118%, with considerable variability among different schemes and seasons. We find that the inclusion of semi/intermediate volatile precursors, in addition to the traditional precursors, and chemical aging (functionalization) are important factors to simulate surface SOA concentrations in Seoul. However, a comparison of observed and simulated SOA/∆CO enhancement ratios suggests that most schemes underpredict SOA aging in upper levels in the boundary layer. We also find that the simplified SOA scheme can reproduce observed OA but often shows overestimation in surface air, indicating that uncertainties exist in bottom-up emissions and precursor parameterization in Seoul. Plain Language Summary We compare four different modeling approaches for simulating secondary organic aerosol (SOA) formation, accounting for a significant fraction of total fine particulate matter concentrations in Seoul, Korea. Using GEOS-Chem, a chemical transport model, we find that current SOA schemes show large variabilities. Including an additional precursor species and further oxidation (i.e., chemical aging) of simulated SOA improves model performance. We also find that a simplified scheme with less computational cost can reproduce observed values but generally shows an overestimation in Seoul, indicating uncertainties in parameterization.