Air Quality Modeling
7.3 Atmospheric Reaction Modeling
7.3.3 Meteorology for Modeling
In most airshed models, meteorology is specified as an input from measurements. In a few cases, a mesoscale model of the urban boundary layer and the surrounding environment is also included. Such a model can simulate sea-land breezes in coastal cities, as well as other mesoscale circulations associated with the urban heat island and topography. The mesoscale model can provide much more detailed and accurate in-formation on spatially inhomogeneous transport winds in a complex urban area. Available observations from a few sites in and around the city can be used to specify initial and boundary conditions for the mesoscale model(see e.g, Byun,1987;Seaman et al.,1989).
The combination of transport, diffusion, and chemistry submodels in an urban airshed air quality model makes it very much like a regional air quality model, albeit for the smaller domain and smaller grid sizes. It is not surprising that the U.S.EPA’s Urban Airshed Model(UAM)is also frequently used for studies of ozone pollution over regions extending over hundreds of kilometers and including several cities. A brief description of this model is given here as an example of an urban air quality model designed for regulatory applications (Morris ad Myers, 1990; U.S. Environmental Protection Agency, 1996).
The UAM is an urban or small-regional scale, three-dimensional grid type numerical simulation model. It incorporates a condensed photochemical kinetics mechanism for urban atmospheres. In particular, a simplified version of the carbon-bond mechanism (CBM-IV) developed by Gery et al. (1989) is employed. The CBM-IV utilizes an updated simulation of PAN chemistry that includes a peroxy-peroxy radical termination reaction, which is significant when the urban atmosphere is NOx limited. The Mechanism accommodates thirty-four species and eighty-two reactions. The UAM is designed for computing ozone concentrations under short-term, episodic conditions lasting 1 to 2 days resulting from emissions of oxides of nitrogen (NOx), volatile organic compounds (VOCs), and carbon monoxide. The model treats VOC emissions as their carbon-bond surrogates.
Gridded, hourly emissions of CO, NO, NO2,and ten categories of VOCs from low-level sources are required as input. For major elevated point sources, in addition to their hourly emissions, exit velocity, and exit temperature are also required for calculating effective release heights. Initial concentrations of all the CBM-IV species and hourly concentrations of each pollutant at each vertical grid level along the inflow boundaries as well as at the top of the model domain must also be specified.
Meteorological data required as input to the UAM are: hourly, gridded, horizontal wind (and ) components for each vertical grid level; hourly, gridded surface temperatures, water vapor mixing ratios, exposure classes, and mixing heights; hourly vertical potential temperature gradients above and below the mixing height; hourly surface pressures; and gridded roughness lengths. Mean vertical velocity at each vertical grid cell interface is calculated from the continuity equation using the input horizontal wind field. The model has a maximum of eight vertical layers of which five are below the mixing height and three are above the mixing height. In a simplified version of the UAM with five layers, only two are below the mixing height. Thus, the vertical resolution in the model varies with the diurnally varying mixing height.
The horizontal and vertical dispersion are modeled through the mean advection-diffusion equations for the various pollutant species. In these the horizontal-eddy diffusivity is set to a user-specified constant value (nominally 50
). The vertical-eddy diffusivity is calculated using different surface-layer and PBL similarity relations for stable, neutral, and unstable conditions. For the stable boundary layer, the formulation of Businger and Arya (1974) is employed. For the day-time unstable conditions, however, the parameterization of eddy diffusivity in the model is somewhat question-able. A recent sensitivity study of the UAM by Nowacki et al. (1996) suggests that vertical diffusion is considerably overestimated in the model during daytime un-stable conditions. Calculated air pollutant concentrations during several photochemical smog episodes in Atlanta, Georgia, are found to depend strongly on the parameterization of vertical diffusivity. Nowacki et al. (1996) have suggested several alternative parameterization schemes to alleviate the problem of overestimating vertical diffusivity in the model.
The combination of transport, diffusion, and chemistry submodels in an urban airshed air quality model makes it very much like a regional air quality model, albeit for the smaller domain and smaller grid sizes. It is not surprising that the U.S.EPA’s Urban Airshed Model(UAM)is also frequently used for studies of ozone pollution over regions extending over hundreds of kilometers and including several cities. A brief description of this model is given here as an example of an urban air quality model designed for regulatory applications (Morris ad Myers, 1990; U.S. Environmental Protection Agency, 1996).
The UAM is an urban or small-regional scale, three-dimensional grid type numerical simulation model. It incorporates a condensed photochemical kinetics mechanism for urban atmospheres. In particular, a simplified version of the carbon-bond mechanism (CBM-IV) developed by Gery et al. (1989) is employed. The CBM-IV utilizes an updated simulation of PAN chemistry that includes a peroxy-peroxy radical termination reaction, which is significant when the urban atmosphere is NOx limited. The Mechanism accommodates thirty-four species and eighty-two reactions. The UAM is designed for computing ozone concentrations under short-term, episodic conditions lasting 1 to 2 days resulting from emissions of oxides of nitrogen (NOx), volatile organic compounds (VOCs), and carbon monoxide. The model treats VOC emissions as their carbon-bond surrogates.
Gridded, hourly emissions of CO, NO, NO2,and ten categories of VOCs from low-level sources are required as input. For major elevated point sources, in addition to their hourly emissions, exit velocity, and exit temperature are also required for calculating effective release heights. Initial concentrations of all the CBM-IV species and hourly concentrations of each pollutant at each vertical grid level along the inflow boundaries as well as at the top of the model domain must also be specified.
Meteorological data required as input to the UAM are: hourly, gridded, horizontal wind (and ) components for each vertical grid level; hourly, gridded surface temperatures, water vapor mixing ratios, exposure classes, and mixing heights; hourly vertical potential temperature gradients above and below the mixing height; hourly surface pressures; and gridded roughness lengths. Mean vertical velocity at each vertical grid cell interface is calculated from the continuity equation using the input horizontal wind field. The model has a maximum of eight vertical layers of which five are below the mixing height and three are above the mixing height. In a simplified version of the UAM with five layers, only two are below the mixing height. Thus, the vertical resolution in the model varies with the diurnally varying mixing height.
The horizontal and vertical dispersion are modeled through the mean advection-diffusion equations for the various pollutant species. In these the horizontal-eddy diffusivity is set to a user-specified constant value (nominally 50
). The vertical-eddy diffusivity is calculated using different surface-layer and PBL similarity relations for stable, neutral, and unstable conditions. For the stable boundary layer, the formulation of Businger and Arya (1974) is employed. For the day-time unstable conditions, however, the parameterization of eddy diffusivity in the model is somewhat question-able. A recent sensitivity study of the UAM by Nowacki et al. (1996) suggests that vertical diffusion is considerably overestimated in the model during daytime un-stable conditions. Calculated air pollutant concentrations during several photochemical smog episodes in Atlanta, Georgia, are found to depend strongly on the parameterization of vertical diffusivity. Nowacki et al. (1996) have suggested several alternative parameterization schemes to alleviate the problem of overestimating vertical diffusivity in the model.
