Improved Representation of Water-Sensitive Organic Aerosol Concentrations Over North America

Abstract:

Chemical transport models (CTMs) struggle to simulate organic aerosol (OA) concentrations in both urban and rural areas, yet capturing this variability is needed to make policy decisions. The simplification of physicochemical processes governing the formation of OA is a major source of error where model improvement is needed. In this study, we attempted to balance computational efficiency with chemical accuracy by implementing into GEOS-Chem the Binary Activity Thermodynamics (BAT) model, a reduced-complexity model estimating the nonideal mixing in aqueous organic systems and the effects of liquid-liquid phase separation. To treat equilibrium gas–particle partitioning, the BAT model is coupled with a nonideal volatility basis set (VBS). Unlike state-of-art thermodynamic models which require detailed chemical structure information at input, the VBS-BAT framework can be used even when only bulk elemental composition of organics is known, which is usually the case in CTMs. In this project, the standard dry VBS scheme of all OA systems in GEOS-Chem (terpenes, isoprene, light aromatics and intermediate-volatility organic compounds, primary semivolatile organic compounds, and oxidized semivolatile organic compounds) was modified to review the impact of nonideal thermodynamic effects of organic–water systems on OA formation. The VBS and VBS-BAT simulations were compared in terms of surface OA concentration (OAC) across North America during 2013, the same year the SOAS campaign took place over the southeast US. Our tests show that the computational cost of the VBS-BAT scheme in GEOS-Chem is about twice as expensive as that of the default VBS scheme due to the application of deep learning neural networks that reduce the need for numerical iterations. OAC from the updated scheme demonstrates higher variability in space and time. For organics of similar hydrophilicity, the VBS-BAT and VBS calculations agree well in regions of low relative humidity (RH<20%). In locations of higher RH, the enhancement in OAC driven by the VBS-BAT treatment is noticeable, where water uptake can increase OA by up to 40%. Our water-sensitive implementation may allow for a better representation of the seasonal and regional variations of OA formation in GOES-Chem.