Evaluation of European air quality modelled by CAMx including the volatility basis set scheme

摘要

pstrongAbstract./strong Four periods of EMEP (European Monitoring and Evaluation Programme) intensive measurement campaigns (June 2006, January 2007, Septembera??October??2008 and Februarya??March??2009) were modelled using the regional air quality model CAMx with VBS (volatility basis set) approach for the first time in Europe within the framework of the EURODELTA-III model intercomparison exercise. More detailed analysis and sensitivity tests were performed for the period of Februarya??March??2009 and June 2006 to investigate the uncertainties in emissions as well as to improve the modelling of organic aerosol (OA). Model performance for selected gas phase species and PMsub2.5/sub was evaluated using the European air quality database AirBase. Sulfur dioxide (SOsub2/sub) and ozone (Osub3/sub) were found to be overestimated for all the four periods, with Osub3/sub having the largest mean bias during June??2006 and Januarya??February??2007 periods (8.9span class="thinspace"/spanpbb and 12.3span class="thinspace"/spanppb mean biases respectively). In contrast, nitrogen dioxide (NOsub2/sub) and carbon monoxide (CO) were found to be underestimated for all the four periods. CAMx reproduced both total concentrations and monthly variations of PMsub2.5/sub for all the four periods with average biases ranging from a??2.1 to 1.0span class="thinspace"/span?μgspan class="thinspace"/spanmsupa??3/sup. Comparisons with AMS (aerosol mass spectrometer) measurements at different sites in Europe during Februarya??March??2009 showed that in general the model overpredicts the inorganic aerosol fraction and underpredicts the organic one, such that the good agreement for PMsub2.5/sub is partly due to compensation of errors. The effect of the choice of VBS scheme on OA was investigated as well. Two sensitivity tests with volatility distributions based on previous chamber and ambient measurements data were performed. For Februarya??March??2009 the chamber case reduced the total OA concentrations by about 42span class="thinspace"/span% on average. In contrast, a test based on ambient measurement data increased OA concentrations by about 42span class="thinspace"/span% for the same period bringing model and observations into better agreement. Comparison with the AMS data at the rural Swiss site Payerne in June 2006 shows no significant improvement in modelled OA concentration. Further sensitivity tests with increased biogenic and anthropogenic emissions suggest that OA in Payerne was affected by changes in emissions from residential heating during the Februarya??March??2009 whereas it was more sensitive to biogenic precursors in June??2006./p.