Evaluating volatility basis set approaches for modeling formation and evolution of secondary organic aerosols from Alberta Oil Sands in the Canadian 3-D air quality model GEM-MACH

Many chemical transport models (CTMs) underestimate organic aerosol concentrations, largely due to poorly constrained parameterizations for secondary OA (SOA), which is formed by chemical transformations of gases in the atmosphere. While progress has been made towards closing the model-measurement gap, significant uncertainties remain. This work describes the incorporation of the volatility basis set (VBS) approach for predicting SOA into Environment and Climate Change Canada's 3-D air quality model, Global Environmental Multiscale - Modeling Air quality and CHemistry (GEM-MACH). High resolution results (2.5 x 2.5 km grid) are compared to measurements taken during flights at the Oil Sands in 2013, as well as to ground-based monitoring networks across North America (10 x 10 km grid).

Two modern parameterizations were evaluated for correcting biased-low SOA yields used currently in GEM-MACH and other CTMs. The first parameterization increases low yields by adding multi-generational oxidation reactions to add increased SOA mass at lower volatilities over several timesteps. The second parameterization adjusts experimental SOA yields to correct for losses of vapors to smog chamber walls in experiments, which results in a larger increase in lower volatility organic mass after a single oxidation event. Both parameterizations improve correlations and biases with respect to the observations, with the wall-loss corrected VBS yields performing best both for the ground stations and when compared to aircraft measurements. Preliminary results indicate the wall-loss corrected yields reduce bias in OA in comparisons to the IMPROVE network from -67 to -25%, when replacing the traditional parameterization. Additionally, a reduced bias from -72 to -54% is obtained in the same comparison for airborne measurements near the Oil Sands. Regional SOA around the Oil Sands is underpredicted in all models, indicating limitations in current models when applied to remote northern regions.