Influence of semi- and intermediate-volatile organic compounds (S/IVOC) parameterizations, volatility distributions and aging schemes on organic aerosol modelling in winter conditions

This study presents a high-resolution (5 km) set of new simulations performed with CAMx v6.40 over the Po Valley area (Northern Italy), aimed to enhance organic aerosol (OA) levels prediction and to gain insight into the sensitivity of CAMx to different uncertain features of the input setup. In particular, we mainly investigated the role of (i) volatility distributions of organic emissions, (ii) parametrizations of semi- and intermediate-volatile organic compounds (S/IVOC) emissions and (iii) different aging schemes, by exploiting the latest experimental information available in the recent scientific literature. Model results were validated against two OA-specific datasets, available for both an urban site (Bologna, February 2013) and a rural one (Ispra, March 2013). We found out a remarkable performance enhancement of modelled OA levels when applying revisions in S/IVOC emission parametrizations together with the new volatility distributions, at both the validation sites. This performance enhancement is associated with a very significant improvement in secondary organic aerosol (SOA) prediction, mainly due to revised IVOC emissions. At Bologna urban site, mean fractional bias (MFB) of OA ranged from - 80.1% in the worst run to - 10.1% in the best one and index of agreement (IOA) from 0.52 to 0.75. Notable improvements but overall poorer metrics were found for Ispra site, where MFB ranges from - 84.2% to - 35% and IOA from 0.45 to 0.50. These findings indicate that organic matter in the semi- and intermediate-volatile range are most likely underestimated in official emission inventories for each main source category (i.e. biomass burning, diesel and gasoline vehicles exhaust). Finally, model results did not show a very pronounced sensitivity to aging processes, due to the low photochemical activity typically observed during winter-time. However, we give evidence that enabling aging processes for biomass burning related SOA, which is by default disabled in CAMx v6.40, can help in closing the gap between modelled and observed SOA concentrations.