Regional and intercontinental modelling - Air Pollution Modeling and Its Application - page 212

HOURLY DAILY

NO 2 SO 2 O 3 NO 2 SO 2 PM 10 PM 2.5

TOTAL 205 18 12 60 43 31 25 16

ustria 19 - 2 11 1 2 2 1

Belgium 2 2 - - - - - -

Bulgaria 1 - - 1 - - - -

Czech. Replublic 6 - - 2 2 2 - -

en ark 5 - - 1 1 2 1 -

France 13 2 - 5 2 4 - -

er any 14 - - 4 1 1 6 2

Great Brit ain 16 3 - 10 3 - - -

Hungary 1 - - - 1 - - -

Ireland 3 - - 2 1 - - -

Italy 6 - - - 2 1 2 1

Latvia 5 - - 1 2 2 - -

Lithuania 3 - - 1 1 1 - -

alta 1 - - 1 - - - -

Netherlands 2 1 - - 1 - - -

or ay 2 - - 2 - - - -

Poland 10 - - 3 4 3 - -

Slovakia 5 - - - 4 1 - -

Slovenia 2 - - 2 - - - -

pain 67 10 10 9 10 8 10 10

S eeden 6 - - 3 2 1 - -

S itzerland 15 - - 2 4 3 4 2

Fig. 1. Location (left) and number (right) of EMEP stations with data availability for the year 2004

Turkey 1 - - - 1 - - -

3. Results and Discussions

The temporal series over spring and summer of measured and simulated pollutant
gases (SO 2 , NO 2 and O 3 ) and particulate matter (only PM 10 is shown) at the EMEP
stations are plotted in Fig. 2 . The monthly statistics for gases and particulate
matter are presented in Table 1. The SO 2 concentration tends to be well correlated
throughout the period of study (0.44 < cor < 0.56). Background levels are well
represented (9.6 µg m −3 < RMSE < 28.9 µg m −3 ) but overestimations of peaks are
detected mainly due to those stations directly affected by stack plumes of large
coal power plants. All stations located in regions of very low SO 2 concentrations
present systematic slight overestimations under anticyclonic situations. Background
levels of NO 2 during both spring and summer are underestimated (−55.6% <
MNBE < −49.7%), but high correlation values demonstrate the satisfactory mean
variations (0.41 < cor < 0.63). The best modeled regions for NO 2 were identified
to be the stations in the Iberian Peninsula. The worst simulated levels were found
in Great Britain, most probably due to uncertainties in the description of the
chemical boundary conditions. The general underestimation by the model is most
likely due to country-to-country discrepancies within the EMEP emission
inventory and partly to the “top-down” disaggregation method utilized to build the
EMEP emission scheme with a resolution down to 12 km. The modeled O 3 is well
characterized (13.1% < MNGE < 23.5%, −15.2% < MNBE < 1.5%) and presents
high correlations (0.48 < cor < 0.60) and high daily variability in the warmest
period. However, mean levels of nighttime O 3 are captured with low accuracy in
summer. As regard to the US-EPA guidelines (MNBE ≤ 15% and MNGE ≤ 35%)
and a European Directive 2008/50/EC criterion (Uncertainty ≤ 50%, see European
Community, 2008), the model presents a good behavior for the O 3 simulation.
PM 10 background concentrations tend to be underestimated throughout the year