Title Effect of horizontal resolution on PM calculations: EMEP model vs. EMEP measurements On behalf of MSC-W presented by Svetlana Tsyro TFMM 10-th meeting, Paris, 15-17 June 2008
Outline Model runs’ setup for different grid resolution: 50x50, 25x25 and 10x10 km2 Calculation results (any difference?) Comparison with regional BG observation Summary and outlook Meteorologisk Institutt met.no
EMEP model runs for 2006 Meteorology: HIRLAM NWP model 0.2x0.2º and 0.1x0.1º spherical rotated coordinates interpolated to EMEP polar-stereographic (PS) Emissions: EMEP (CEIP), distribution as EMEP or based on TNO (ca 7.5 km) Meteorology Emissions 50x50 km2 HIRLAM 0.2x0.2º -> PS 50km EMEP EMEP & TNO distrib 25x25 km2 HIRLAM 0.2x0.2º -> PS 25km 10x10 km2 HIRLAM 0.1x0.1º -> PS 10km Fine and coarse particles are distinguished in order to account for the different dry and wet removal rates. Meteorologisk Institutt met.no
Concentration ratio: 10 / 50 (EMEPem_TNOdis) Meteorologisk Institutt met.no
Concentration ratio: EMEP_TNO_10 / EMEP_50 Meteorologisk Institutt met.no
Do we see impovement in the model performance for finer resolution compared to observations? EMEP50 TNO 50 TNO 25 TNO 10 PM10 Bias - 34 - 41 R 0.46 0.49 0.54 0.56 PM2.5 - 36 - 35 0.66 0.67 0.73 0.71 SIA - 37 (-32) - 37 (-33) - 33 (-28) - 34 (-29) 0.87 (0.49) 0.88 (0.7) 0.87 (0.45) 0.88 (0.51) Na - 4 - 11 0.83 0.88 0.89 Meteorologisk Institutt met.no
Scatter-plots for PM10 for 2006 50x50 10x10 The model reproduce the regional PM gradients somewhat better when using finer grids (and TNO emission distribution) Meteorologisk Institutt met.no
PM10 with TNO emission distribution Bias Correlation Meteorologisk Institutt met.no
PM2.5 with TNO emission distribution Bias Correlation Meteorologisk Institutt met.no
PM10 with EMEP-50 and TNO-25 and 10 Bias Correlation Meteorologisk Institutt met.no
PM2.5 with EMEP-50 and TNO-25 and 10 Bias Correlation Meteorologisk Institutt met.no
Improvement for Payerne Example: Improvement for Payerne Meteorologisk Institutt met.no
with windblown dust w/o windblown dust Meteorologisk Institutt met.no
BIAS for PM10: June 2006 Montelibretti Melpitz Birkenes Meteorologisk Institutt met.no
Correlation: Montelibretti, June 2006 Meteorologisk Institutt met.no
Correlation for PM components: Birkenes, June 2006 Melpitz, June 2006 Meteorologisk Institutt met.no
Main findings Overall Bias - small or no improvement Spatial correlation - certain improvement more for PM10/PM2.5 than for SIA – due to PPM? Larger improvement from 50 to 25 than from 25 to 10 km Individual sites Bias - small changes, both +/- Correlation – variable (btw sites and countries) PM components (3 sites) – no clear pattern BIAS – some improvement, but for diff. reasons CORR – variable btw sites, components, resolutions Meteorologisk Institutt met.no
Outlook Look at the effect of improved resolution on model performance for PM for rural/sub- urbane sites (AirBase) Look more for explanations of the changes in the model performance (due to emissions, meteorology) HIRLAM 20 vs. 10 - performance Meteorologisk Institutt met.no