Evaluation of the GEM-AQ model in the context of the AQMEII Phase 1 project

• Published in: Atmospheric chemistry and physics Vol. 15; no. 8; pp. 3971 - 3990
• Main Authors: Struzewska, J, Zdunek, M, Kaminski, J. W, Łobocki, L, Porebska, M, Jefimow, M, Gawuc, L
• Format: Journal Article
• Language: English
• Published: Katlenburg-Lindau Copernicus GmbH 16.04.2015; Copernicus Publications
• Subjects: Aerosols; Air masses; Bias; Boundary conditions; Chemistry; Circulation; Climate models; Context; Correlation coefficient; Correlation coefficients; Emissions; Equator; Errors; Gravitational waves; Mathematical analysis; Mathematical models; Modelling; Ozone; Particulate matter; Performance evaluation; Pollutants; Pollution monitoring; Quality; Regional climates; Resolution; Rural; Simulation; Spatial distribution; Stations; Temporal variability; Temporal variations; Europe; ANE, Europe
• ISSN: 1680-7324; 1680-7316; 1680-7324
• DOI: 10.5194/acp-15-3971-2015

In the scope of the AQMEII Phase 1 project the GEM-AQ model was run over Europe for the year 2006. The modelling domain was defined using a global variable resolution grid with a rotated equator and uniform resolution of 0.2° × 0.2° over the European continent. Spatial distribution and temporal variability of the GEM-AQ model results were analysed for surface ozone and PM10 concentrations. Model results were compared with measurements available in the ENSEMBLE database. Statistical measures were used to evaluate performance of the GEM-AQ model. The mean bias error, the mean absolute gross error and the Pearson correlation coefficient were calculated for the maximum 8 h running average ozone concentrations and daily mean PM10 concentrations. The GEM-AQ model performance was characterized for station types, European climatic regions and seasons. The best performance for ozone was obtained at suburban stations, and the worst performance was obtained for rural stations where the model tends to underestimate. The best results for PM10 were calculated for urban stations, while over most of Europe concentrations at rural sites were too high. Discrepancies between modelled and observed concentrations were discussed in the context of emission data uncertainty as well as the impact of large-scale dynamics and circulation of air masses. Presented analyses suggest that interpretation of modelling results is enhanced when regional climate characteristics are taken into consideration.