Investigating the role of chemical and physical processes on organic aerosol modelling with CAMx in the Po Valley during a winter episode

- CAMx performance for OA are worse than for other PM components.
- SOAP scheme shows a better performance than VBS, due to an error compensation.
- VBS allows a better repartition of primary and secondary OA than SOAP scheme.
- POA volatility distribution and SVOC and IVOC emissions need improvement.
- Aqueous phase mechanism is necessary to reconcile OA observations and modelling.

Traditional aerosol mechanisms underestimate the observed organic aerosol concentration, especially due to the lack of information on secondary organic aerosol (SOA) formation and processing. In this study we evaluate the chemical and transport model CAMx during a one-month in winter (February 2013) over a 5 km resolution domain, covering the whole Po valley (Northern Italy). This works aims at investigating the effects of chemical and physical atmospheric processing on modelling results and, in particular, to evaluate the CAMx sensitivity to organic aerosol (OA) modelling schemes: we will compare the recent 1.5D-VBS algorithm (CAMx-VBS) with the traditional Odum 2-product model (CAMx-SOAP). Additionally, the thorough diagnostic analysis of the reproduction of meteorology, precursors and aerosol components was intended to point put strength and weaknesses of the modelling system and address its improvement. Firstly, we evaluate model performance for criteria PM concentration. PM10 concentration was underestimated both by CAMx-SOAP and even more by CAMx-VBS, with the latter showing a bias ranging between −4.7 and −7.1 μg m−3. PM2.5 model performance was to some extent better than PM10, showing a mean bias ranging between −0.5 μg m−3 at rural sites and −5.5 μg m−3 at urban and suburban sites. CAMx performance for OA was clearly worse than for the other PM compounds (negative bias ranging between −40% and −75%). The comparisons of model results with OA sources (identified by PMF analysis) shows that the VBS scheme underestimates freshly emitted organic aerosol while SOAP overestimates. The VBS scheme correctly reproduces biomass burning (BBOA) contributions to primary OA concentrations (POA). In contrast VBS slightly underestimates the contribution from fossil-fuel combustion (HOA), indicating that POA emissions related to road transport are either underestimated or associated to higher volatility classes. The VBS scheme under-predictes the SOA too, but to a lesser extent than CAMx-SOAP. SOA underestimation can be related to corresponding underestimation of either aging processes or precursor emissions. This indicates that improvements in the emission inventories for semi- and intermediate-volatility organic compounds are needed for further progress in this area. Finally, the comparison between modelled and observed SOA sources points out the urgency to include processing of OA in particle water phase into SOA formation mechanisms, to reconcile model results and observations.