PM2.5 source apportionment in Lombardy (Italy): Comparison of receptor and chemistry-transport modelling results

This work discusses the source apportionment results produced by receptor and chemistry-transport modelling for PM2.5 in Northern Italy, selected as case study due to its very critical conditions. Receptor modelling was performed using the Chemical Mass Balance model, while source oriented analysis by means of the CAMx chemistry transport model. Models shared the same source profiles, mostly based on local data. CMB showed a better reconstruction of the mass closure, while CAMx systematically underestimated cold season concentrations. Nevertheless both models provided the same source ranking at several receptors. According to CMB results, the most relevant contributions during the cold season, ranging around 10 mu g m(-3), were yielded by road transport, domestic heating and ammonium nitrate. CAMx provided similar results for the secondary sources, while systematically underestimated road transport and domestic heating. A similar behaviour was observed during the summer season. The main discrepancies between the models were: the questionable results from CMB at some receptors, missing the road transport contribution, clearly pointed out by CAMx/PSAT; the ability of CAMx/PSAT to apportion the contribution of sources sharing similar profiles and strongly correlated such as CI and SI vehicles, altogether recognized as traffic source by CMB; the ability of CAMx/PSAT to identify the contribution of secondary PM deriving from complex chemical transformation, such as anthropogenic and biogenic SOA. Finally, the comparison of specific source contribution pointed out that CAMx missed the reconstruction of the road transport contribution to the OC, mostly as a consequence of deficiencies in the emission inventories concerning the primary OC in the cold season and the concurrent underestimation of emissions and secondary OC formation in the warm season. (C) 2015 Elsevier Ltd. All rights reserved.