05.01.2013 Views

Air quality expert group - Fine particulate matter (PM2.5) in ... - Defra

Air quality expert group - Fine particulate matter (PM2.5) in ... - Defra

Air quality expert group - Fine particulate matter (PM2.5) in ... - Defra

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

were for daily <strong>PM2.5</strong> nitrate and for daily PM10 sea salt and sulphate, and some<br />

were for daily PM10 EC/OC (for the last five months of 2008).<br />

13. In general, the model underpredicts hourly <strong>PM2.5</strong> measurements when us<strong>in</strong>g<br />

estimates of modelled <strong>PM2.5</strong> mass without particle bound water (Figure A2.3.1).<br />

A similar conclusion can be drawn for total PM10 concentrations. Despite the<br />

underprediction of PM10, the model replicates the proportion of each species<br />

reasonably well, slightly overpredict<strong>in</strong>g observed NO3 - and SO4 2- fractions and<br />

underpredict<strong>in</strong>g others such as Cl - , EC, SOA and OC (Figure A2.3.2). The model<br />

results for f<strong>in</strong>e mode PM nitrate (Figure A2.3.3) and POC (Figure A2.3.7) are<br />

<strong>in</strong> reasonable agreement with measurements, although some of the peak<br />

concentrations were not well predicted. Some of the coarse mode particles were<br />

underpredicted and require further evaluation, especially <strong>in</strong> the case of nitrate<br />

and sea salt. EC, SO4 2- (Figure A2.3.5) and SOA were also underpredicted.<br />

14. Improv<strong>in</strong>g the predictive capability of the model for some components (EC)<br />

can be achieved <strong>in</strong> a straightforward way by improv<strong>in</strong>g emissions estimates.<br />

Improv<strong>in</strong>g the predictive capability for others (Cl - and coarse mode nitrate) may<br />

prove to be more difficult because there is a comb<strong>in</strong>ation of possible reasons for<br />

the model underprediction.<br />

15. An exam<strong>in</strong>ation of the chemical composition of the <strong>PM2.5</strong> mass <strong>in</strong> 2008 revealed<br />

that 58% was estimated to be particle bound water (PBW) (Figure A2.3.1).<br />

This is a model estimate of ambient PBW and contrasts with measurements,<br />

where PBW is mostly removed by condition<strong>in</strong>g <strong>in</strong> the field or <strong>in</strong> the laboratory<br />

and water content is typically reduced to around 10% (Harrison et al., 2004,<br />

and Green et al., 2009). As a consequence, <strong>in</strong>clud<strong>in</strong>g modelled PBW without<br />

correct<strong>in</strong>g for the measurement conditions can underm<strong>in</strong>e attempts to<br />

evaluate the model for total <strong>PM2.5</strong> or PM10 as well as for hygroscopic particles<br />

such as SO4 2- and possibly NO3 - and SOA. These results demonstrate the<br />

importance of PBW when quot<strong>in</strong>g PM concentrations and it would be worth<br />

consider<strong>in</strong>g quot<strong>in</strong>g the assumptions made alongside the concentrations from<br />

measurements and models.<br />

Hourly predictions of <strong>PM2.5</strong><br />

Annex 2: PM modell<strong>in</strong>g <strong>in</strong> the UK<br />

16. Speciated PM analysis was conducted aga<strong>in</strong>st measurements from the AURN<br />

and London <strong>Air</strong> Quality Network (LAQN) dur<strong>in</strong>g 2008, us<strong>in</strong>g CMAQ (v4.7)<br />

and the WRF meteorological driver. The WRF-CMAQ model was operated<br />

with 23 vertical layers (up to approximately 15 km above ground) and two<br />

nest<strong>in</strong>g levels, downscal<strong>in</strong>g from 81 km grid resolution over Europe to 9 km<br />

grid resolution over the UK. The emissions from EMEP, NAEI and the European<br />

Pollutant Emission Register (EPER) were processed <strong>in</strong>to hourly 3-D gridded<br />

chemical species and used the CB05 chemical scheme which <strong>in</strong>cluded the<br />

aqueous and aerosol (AERO5) extension.<br />

17. In 2008 CMAQ PM10 and <strong>PM2.5</strong> predictions were assessed aga<strong>in</strong>st measurements<br />

from AURN and LAQN, <strong>in</strong>clud<strong>in</strong>g 70 PM10 sites (three rural, 12 suburban and<br />

55 urban background) and 29 <strong>PM2.5</strong> sites (two rural, six suburban and 21 urban<br />

background). In compar<strong>in</strong>g the hourly average concentrations we have removed<br />

the model’s predictions for water. The reason for do<strong>in</strong>g this is that CMAQ water<br />

estimates can be high (58%), as demonstrated by the modelled estimate of<br />

particle bound water (PBW) at ambient conditions <strong>in</strong> Figure A2.3.1a.<br />

153

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!