correspondingly crude, especially in terms of spatial and temporal resolution.
A well-known box model that was developed initially for regulatory purposes is the EKMA model (.Empirical kinetic Modeling Approach) (Dodge, 1977a, 1977b; Dimitriades and Dodge, 1983). It is based on a box of air containing VOC and NOx at specified initial concentrations. Dilution occurs during the simulation, in which chemical reactions are converting the VOC and NOx to 03 and other secondary pollutants. Injection of pollutants during the day is included, along with a time-varying inversion height and entrainment of pollutants from aloft as the inversion height increases. Diurnal variation of photolysis rates, temperature, and relative humidity are also incorporated. Chemical mechanisms using both types of approach to "lumping" of the organics, i.e., the traditional organic classifications or the carbon bond approach, can be used with this box model. The model output can then be used to generate a series of isopleths for ozone or other secondary species via the ozone isopleth plotting package, or OZIPP. Applications of a research version of this model, called OZIPR (the R stands for research version), are provided with this book (Appendix III).
The "kinetic modeling" nomenclature arises from the incorporation of chemical kinetic submodels in EKMA. The "empirical" term comes from the use of observed 03 peaks in combination with the model-predicted ozone isopleths to develop control strategy options. Thus, the approach historically was to use the model to develop a series of ozone isopleths using conditions specific for that area. The second highest hourly observed 03 concentration and the measured
6-9 a.m. VOC/NOx ratio were located on these iso-pleths. The isopleths were then used to determine how much reduction in VOC or NOx would be required to reduce the peak 03 concentration to the air quality standard.
While the EKMA model has been very useful as a first approach to incorporating the complex chemistry that links primary and secondary pollutants and to including some meteorological variables, it clearly is an oversimplification for large urban areas and for downwind regions. For example, it does not include complex meteorology such as mixing between the surface layer and higher altitudes or the effects of long-range transport, which are known to be important, nor is it useful for simulating multiday episodes, which generally produce the highest pollutant levels.
(2) Lagrangian models The next step in model development was the use of Lagrangian models depicted in Fig. 16.20 (Wayne et al., 1973). These models consider a column of air containing certain initial pollutant concentrations and follow its trajectory as it moves along a trajectory. In effect, it is an expansion of a simple box model to a series of adjacent, interconnected boxes.
With the increase in computing power, both simple box models and Lagrangian models have been largely supplanted with grid-based models described in the following section. The idea of a Lagrangian approach, however, is still useful in field studies where the motion of an air parcel and changes in its chemical composition can be tracked as it moves downwind in a fashion similar to that depicted in Fig. 16.20. Such studies are often referred to as Langrangian experiments.
d. Grid Models: Urban to Regional Scales
Models currently in use for developing control strategy options are grid-based, or Eulerian, models, the
principle of which is illustrated in Fig. 16.21 (Ames et al., 1978). The area to be modeled is divided into grids, or boxes, in both the horizontal and vertical directions. Pollutant concentrations are calculated at fixed geographical locations at specified times based on their initial concentrations, new emissions, transport in and out of the box, dilution, and chemical reactions. By carrying out such calculations for each box, one can develop a 3-D map of pollutant concentrations as a function of time.
While this approach was first directed primarily to urban regions, regional models have been developed subsequently. The need for regional-scale models became apparent in situations such as the northeast of the United States, where long-range transport plays a major role in determining pollutant concentrations at various locations. With the combination of long-range transport, fresh emissions, and complex chemistry occurring along relatively large distances over highly populated centers, application of individual urban-scale models is not appropriate—hence the need for larger, regional models.
There are a number of urban-scale grid-based models. Examples include the UAM (Urban Airshed Model; e.g., UAM-4 is version 4) (e.g., see Reynolds et al. (1973) and Tesche and McNally (1991)), the CIT model (California institute of Technology model) (see McRae et al. (1982b) and Russell et al. (1988)), and the SMOG model (Surface Meteorology and Ozone Generation model) (see Lu et al., 1997a, 1997b). In such models, the horizontal size of each grid is of the order of a few kilometers, e.g., 4 or 5 km square grids with the vertical height split into 5-20 layers of increasing thickness beginning at ground level. For regional-scale models such as RADM (.Regional Acid Deposition Model; see Chang et al., 1987), ADOM (Acid Deposition and Oxidant Model; Venkatram et al., 1988), STEM II (Sulfur Transport Eulerian Model; Carmichael et al., 1991), RTM-III (.Regional Transport Model III; Liu et al., 1984), LIRAQ (Livermore Regional Air Quality Model; MacCracken et al., 1978), CALGRO (Yamartino et al., 1992), the model of McKeen et al. (1991a)), and ROM (Regional Oxidant Model; Lamb, 1983), the scale is of the order of 15-130 km. The number and size of the vertical layers can vary from 6 to 30.
Recently, models that incorporate urban to regional scales have also been developed. These models use the approach of nested grids in which small grids for the urban scale are "nested" within larger grids used for regional scales. [Although this sounds straightforward, it is not—see, for example, Sillman et al. (1990) and Kumar et al. (1994).] Such approaches, combined with
Model region o o o
Elevated emissions J
Chemistry & Elevated emissions
Chemistry & Elevated emissions i Transport
Region top Transport
! Surface deposition
FIGURE 16.21 Schematic illustration of the grid used and treatment of atmospheric processes in one Eulerian airshed model (adapted from Ames et al., 1978).
the development of a modular structure and "user friendly" software, are also under development for direct use in the regulatory arena [see, for example, the Models-3 description in Dennis et al. (1996) and Appleton (1996)]. The widespread application of such models will require a significant increase in computing power, such as the use of massively parallel computers (e.g, see Dabdub and Seinfeld, 1994a, 1994b, 1996).
While such models are becoming increasingly sophisticated and capable of incorporating more detailed emissions, meteorology, and chemistry, there remain a number of important issues and uncertainties in their inputs and, hence, predictions. A few of the important parameters are discussed briefly in the following.
(1) Meteorology Clearly, meteorology plays a determining role in pollutant concentrations at various locations. Simulating complex meteorology and topography is difficult, and in addition, the detailed 3-D data needed to test the meteorology submodels are not available for many regions.
For example, Kumar and Russell (1996) examined the effect on predicted ozone levels in the Los Angeles area of two different approaches now used for incorporating meteorology into a grid-based Eulerian model. The diagnostic approach is based on field measurements of the needed meteorological variables and in terpolation and extrapolation of these data as necessary. The prognostic approach uses models to predict the evolution of the atmospheric system with time by integration of the equations of conservation of mass, motion, heat, and water in space and time. Figure f 6.22 shows the ozone concentrations (dots) observed at three different locations during an air quality study in the Los Angeles area during August 27-29, f987, in which detailed meteorological and chemical measurements were made. Also shown are model predictions using the diagnostic approach (solid line) and the prognostic approach (dashed line), respectively; in general, the prognostic approach leads to lower predicted 03 concentrations for reasons that are discussed in detail by Kumar and Russell (1996).
Another example of the importance of accurately specifying meteorological parameters is described by Sistla et al. (1996). They investigated the effect on model predictions for ozone in the New York Metropolitan area of using either a spatially invariant mixing height or one that varied spatially. The latter was shown to predict 03 concentrations that were in better agreement with the observations. Figure f6.23, for example, shows the observed and modeled 03 concentrations for Westpoint, New York, where the spatially varying mixing height assumption was in reasonable agreement with the observed values but the
Was this article helpful?