Search

The role of perturbations of reactive trace gas concentration distributions in turbulent flows in the planetary boundary layer (PBL) is discussed. The paper focuses on disturbances with larger spatial scales. Sequential nesting of a chemical transport model is applied to assess the effect of neglecting subgrid chemical perturbations on the formation and loss of ozone, NO x , peroxyacetyl nitrate (PAN) and HNO3 calculated with a highly complex chemical mechanism. The results point to characteristic differences regarding the process of mixing of chemically reactive species in the PBL and lower troposphere. [ABSTRACT FROM AUTHOR]

Abstract: Air quality within the European Union (EU) is controlled by the Member States'' monitoring networks. In this study, measured data is compared with the EURAD (EURopean Air pollution Dispersion) model system diagnostic output. Simulations for the German state North Rhine-Westphalia (NRW) with a horizontal grid resolution of 5km×5km are analyzed. The comparison is performed for NO2, O3, and PM10, for the 4-year time period from 2002 through 2005. Although the spatial representativity of data of the two systems differs, the analyzed temporal variability of the averages shows good agreement of modeled and observed concentrations for all three parameters. This confirms the applicability of the EURAD model to mesoscale air quality assessment. Discrepancies between the model and observed data occurred at low concentrations close to the detection limits of the analyzers, for high O3 concentrations in summer, and for PM10 before 2004, when earlier versions of the MM5 meteorological module and of the emission inventory were used. Possible causes of O3 overprediction and NO2 underprediction preferably showing up in summer are considered. It is found that modeled Ox =O3 +NO2 leads to better representation of the observations than the individual species themselves. The model performance can probably be increased by further development of the emission inventories and more accurate land use data. These input data seem to be the main reason for deviations between observations and model results. [Copyright &y& Elsevier]