The effect of tropospheric ozone as an air pollutant is normally simulated using regional or urban scale computer models, while global models often are used to study the role of ozone as a greenhouse gas. The chemical transformations involved in the ozone chemistry occur on all scales, and are often related nonlinearily to each other. Due to the relatively coarse spatial resolution used in global Chemistry-Transport Models (CTMs), inaccuracies will arise when emissions and chemical processes are averaged in a grid box. Additionally, meteorological small-scale processes (e.g. convection) impact the chemistry, and inaccuracies may increase if parameterizations are implemented on a coarse scale. The problem of neglecting urban scale processes is particularly important in climate studies because of the rapid urbanization that we experience today.
In this thesis we have applied the WRF-Chem (Weather Research and Forecasting with Chemistry) model to study scale interactions in the ozone photochemistry, and to quantify inaccuracies in terms of effective emissions. The model is run for a three week summer period in 2003 over Europe, zooming in on the London metropolitan area using square horizontal grid resolutions of 81 km, 27 km, 9 km, 3 km, and 1 km. We use the RADM2 chemistry scheme, and as input data we apply a 1 km x 1 km anthropogenic emission inventory for the UK (DEFRA, 2007), and a 0.5 x 0.5 degree anthropogenic emission inventory for the rest of Europe (RETRO, 2006), together with assimilated meteorology from ECMWF as initial and boundary conditions. We have focused on column values of ozone and related chemical components in the London area, caused by London emissions, and results from the different resolutions have been compared.
The results show an increase in the average net ozone column caused by London emissions when horizontal grid spacing is reduced from 81 km to 27 km. Most likely, these changes are caused by increased transport of chemical species out of the London area due to better representation of winds and boundary layer height in the latter case. There were only minor changes in the results between the scales 27 km, 9 km, and 3 km, while the 1 km resolution results gave an increase in ozone column values due to London emissions, causing a shift from net negative to net positive ozone in the London area. The changes from 3 km to 1 km probably arise because of higher resolution in emissions, so that the model better accounts for nonlinearities in the ozone chemistry. However, comparisons with measurements of chemical species show that there are uncertainties related to our results, implying that caution should be used when drawing conclusions. The agreement between model results and measurements were relatively good in the first half of the simulation period, but lack of vertical distribution in the emissions data caused large discrepancies during the last ten days.