To what extent can aerosol water explain the discrepancy between model calculated and gravimetric PM₁₀ and PM_2.5?

Abstract. Inter-comparisons of European air quality models show that regional transport models, including the EMEP (Co-operative Programme for monitoring and evaluation of the long-range transmission of air pollutants in Europe) aerosol model, tend to underestimate the observed concentrations of PM₁₀ and PM_2.5. Obviously, an accurate representation of the individual aerosol constituents is a prerequisite for adequate calculation of PM concentrations. On the other hand, available measurements on the chemical characterization of ambient particles reveal that full chemical PM mass closure is rarely achieved. The fraction unaccounted for by chemical analysis can comprise as much as 30-40% of gravimetric PM₁₀ or PM_2.5 mass. The unaccounted PM mass can partly be due to non-C atoms in organic aerosols and/or due to sampling and measurement artefacts. Moreover, a part of the unaccounted PM mass is likely to consist of water associated with particles. Thus, the gravimetrically measured particle mass does not necessarily represent dry PM₁₀ and PM_2.5 mass. This is thought to be one of the reasons for models under-prediction of observed PM, if calculated dry PM₁₀ and PM_2.5 concentrations are compared with measurements. The EMEP aerosol model has been used to study to what extent particle-bound water can explain the chemically unidentified PM mass in filter-based particle samples. Water content of PM_2.5 and PM₁₀ has been estimated with the model for temperature 20°C and relative humidity 50%, which are conditions required for equilibration of dust-loaded filters according to the Reference method recommended by the European Committee for Standardization (CEN). Model calculations for Europe show that, depending on particle composition, particle-bound water constitutes 20-35% of the annual mean PM₁₀ and PM_2.5 concentrations, which is consistent with existing experimental estimates. At two Austrian sites, in Vienna and Streithofen, where daily measurements of PM_2.5 mass and chemical composition are available, calculated PM_2.5 water content is found to be about 75-80% of the undetermined PM_2.5 mass and there is correlation between them. Furthermore, accounting for aerosol water has improved the agreement between modelled and measured daily PM_2.5 concentrations, whilst model calculated dry PM_2.5 concentrations appear to agree quite well with the total identified PM_2.5 mass. No information on the composition of PM measured at EMEP sites is presently available. Given that PM₁₀ and PM_2.5 concentrations are measured at EMEP stations with gravimetric methods they are likely to contain water. We show that the levels of modelled PM₁₀ and PM_2.5 concentrations with aerosol water included agree with measurements better than dry PM concentrations. As expected, the spatial correlation has not changed significantly, whereas the temporal correlation of daily PM₁₀ and PM_2.5 with monitoring data has slightly improved at most of the EMEP sites. Our results suggest that aerosol water should be accounted for in modelled PM₁₀ and PM_2.5 when compared with filter-based gravimetric measurements.