Improving UK Air Quality Modelling Through Exploitation of Satellite Observations
Pope, Richard1; Chipperfield, Martyn1; Savage, Nick2
1University of Leeds, UNITED KINGDOM; 2Met Office, UNITED KINGDOM

The Met Office's operational regional Air Quality Unified Model (AQUM) contains a description of atmospheric chemistry/aerosols which allows for the short-term forecast of chemical weather (e.g. high concentrations of ozone or nitrogen dioxide, which can trigger warnings of poor air quality). AQUM's performance has so far only been tested against a network of surface monitoring stations. Therefore, with recent improvements in the quality and quantity of satellite measurements, data products (e.g. tropospheric columns, vertical profiles) from several satellite instruments will be used to test the performance of the model. First comparisons between an AQUM simulation for the UK heatwave event of July 2006 and data from OMI, TES (both on AURA) and MODIS (on AQUA) have identified multiple model-satellite biases. The chemical/aerosol species investigated for this simulation include nitrogen dioxide (NO2), ozone (O3), formaldehyde (HCHO), carbon monoxide (CO) and aerosol optical depth (AOD) at 0.55 microns wavelength.

Comparisons between AQUM and OMI July 2006 monthly mean tropospheric columns highlight significant (magnitude of model-satellite bias greater than retrieval error) biases in northern England, Edinburgh and London. The AQUM-OMI positive biases over northern England and Edinburgh have hypothesised links to the AQUM NOx emission datasets. Sensitivity tests have shown that point sources in the emissions datasets have a significant impact on the AQUM NO2 budget. Further sensitivity tests are determining if the vertical extent of the point sources accounts for the AQUM-OMI biases. If so, a Gaussian stack plume model will be introduced to attempt better representation of sources such as Drax Power Station in Yorkshire.

Surface AQUM NO2 fields have been validated against the Automated Urban and Rural Network (AURN). Metrics such as the modified normalised mean bias (MMB) and the fractional gross error (FGE) show that the AQUM generally underestimates monitoring stations in urban regions. These biases/errors tend to be greatest at roadside or kerbside monitoring stations. The working hypothesis for this is the resolution of the AQUM (12 km x 12 km), where the NOx emissions are averaged out over the grid box leading to these negative biases. In background urban sites, the comparisons improve but still with significant negative biases. The rural monitoring stations have much better agreement.