Evaluation of an urban modelling system against three measurement campaigns in London and Birmingham

The results of three measurement campaigns are presented in this study. The campaigns have been undertaken at an urban roadside site in London, for more than a year and three months in 2003-2004 and for a year in 2008, and at an urban background site in Birmingham, U.K, for about four months in 2002. The concentrations of PM2.5, PM10, NOx and NO2 were predicted using the roadside dispersion model CAR-FMI, combined with a national U.K. emission model, a meteorological pre-processor, and measured values at urban background stations. The agreement of the predicted and measured hourly and daily time-series has been assessed statistically for all of the campaigns and pollutants. For instance, the Indices of Agreement (IA) in all the campaigns ranged from 0.68 to 0.78, 0.87, from 0.70 to 0.80, and from 0.61 to 0.83 for PM2.5, PM10, NOx and NO2, respectively. However, in case of the campaigns in London, both the PM fractions and the nitrogen oxide concentrations were under-predicted. The model performance in terms of atmospheric stability, wind speeds and other factors was analysed, and reasons for the disagreement of predictions and measurements have been discussed. It is useful to consider the model performance statistics for several measurement campaigns simultaneously as some of the results were found to be specific only to one or two campaigns. The spatial concentration distribution of NOx in London for 2008 has also been presented.