AIR-QUALITY PREDICTIONS USING A FIXED LAYER-DEPTH VERTICAL STRUCTURE IN THE URBAN AIRSHED MODEL

A model with a varying layer-depth structure tends to have excessive v ertical diffusion but can accommodate a well-defined mixing height. It also requires a modified vertical wind field to minimize possible loc al accumulation or drainage of species mass. A model with a fixed laye r-depth structure may require many layers to better resolve the mixing height and contains an explicit challenge to define the thickness of the lowest layer that carries most precursor emission sources. This ch allenge, also present implicitly in the Varying layer-depth case, coul d be addressed only if well-resolved mixing information or the eddy di ffusivity profile under realistic physical environments were available , especially at night and in the morning. The impact an air quality pr edictions of using a fixed layer-depth and a varying layer-depth struc ture in the Urban Airshed Model is compared. Under the given input inf ormation and existing model assumptions, the fixed layer-depth approac h yields substantially higher hourly concentrations of NO, CO, and VOC in the lowest layer in isolated areas in the early morning than the v arying layer-depth approach. When different spatial resolutions are us ed in the fixed layer-depth approach, the resulting spatial concentrat ion distributions tend to be more similar between the different spatia l resolutions than between a fixed layer-depth approach and the varyin g layer-depth approach. The fixed, 11-layer model tends to yield lower precursor concentrations than the fixed, 6-layer model in the morning . Insofar as emission control is concerned, because the location of th e predicted peak O-3 concentration tends to shift differently for the different model versions when NOx emissions are reduced, it becomes di fficult to reveal a general pattern rather independent of the vertical layer structure. If we ignore the shift in location and time of the O -3 peak, then it may be fortuitous that the relative changes in the pr edicted peaks appear rather consistent among the different model versi ons, especially for the 8-h O-3 peak.