Improvements in Air Quality Modelling by Using Surface Boundary Layer Parameters derived from Satellite Land Cover Data

Many atmospheric dispersion models include only simpletreatment of surface features to estimate the wind profilesand stability parameters. Detailed characterisation of theland cover, particularly in large and complex urbanconurbations, is especially important, as the surfacefeatures can vary significantly over the area. This paperdiscusses the use of satellite land cover data to derivespatially resolved surface boundary layer (SBL) parameters.These parameters have been used in an air quality model,PEARL (Prediction Air Quality in Urban and RegionalLocations) for estimating monthly and annual COconcentrations. Land cover data, derived from LANDSATThematic Mapper Imagery, has been used to estimate SBLparameters (surface roughness length, albeedo, Bowen ratioand anthropogenic heat flux) for a study area of 10000km² encompassing Greater London and the surroundingcounties. The SBL parameters have been assigned according tomajor land cover types for the whole area at a spatialresolution of 1 × 1 km. Predictions from two versions of the PEARL model (one with land cover data and one without)have been compared with each other and with measured data forannual and monthly CO concentrations from seven London airquality monitoring sites. This comparison shows thatdifferences between predicted and observed values can bereduced by up to a factor of three. The use of SBLparameters derived from land cover data also yields moredetailed predicted annual CO spatial patterns especially inand around suburban areas. The performance of both versionsof the model for monthly CO concentrations has been comparedwith a range of statistical measures. This comparisonconfirms that improved agreement is observed betweenmodelled and measured monthly CO concentrations when use ismade of spatially resolved SBL parameters.