Lectures on Air Pollution Modeling

This volume is concerned with the physics and application of air-pollution modeling on scales up to about 50 km. Its eight chapters, comprising the diverse points of view of seven authors, remain substantially in their original, lecture-note form. The result is not a smoothly flowing monograph but instead a richly textured, lively collection of the seasoned thoughts and perspectives of experienced researchers and practitioners. My coauthor, Akula Venkatram, and I hope that the volume conveys some of the excitement of a vital, rapidly evolving field.

Even the most casual observer notices the changes in the wind. Though it has a certain persistence in time, the details of its swirls and eddies seem infinitely variable. Its strength changes day to day as weather systems evolve, and day to night as the sun rises and sets. It is modulated strongly by terrain features and by urban architecture. How can we hope to describe such a complicated phenomenon?

There now exist a multitude of theoretical methods for making diffusion predictions for comparison with results of field and laboratory diffusion experiments. These include gradient transfer, spectral diffusivity, second-order closure, large-eddy simulation, and random perturbation models (Briggs and Binkowski, 1985). These models generally require either detailed meteorological and turbulence measurements or many assumptions about these quantities, and may require considerable computational effort. In this chapter we seek more general theoretical frameworks that allow considerable ordering of diffusion data without requiring complex meteorological information or computational efforts.

Most industrial pollution sources are stacks with discharges of momentum and heat as well as pollutants. The resulting plume rise can be considerable—hundreds of meters—and can substantially aid the dilution of plume constituents before they reach ground level. Thus, plume rise is an important factor to consider in diffusion modeling. For power plants and other moderate-to-large industrial sources, the major contribution to the rise is from the heat flux. For example, a modern power plant typically discharges ~ 100 MW of heat from its stack. Source momentum can be important for smaller sources, such as those typically found in light manufacturing. Although we will address the plume rise due to source momentum, we will give most attention to the effects of source buoyancy.

Dispersion in the planetary boundary layer (PBL) is controlled by the mean and turbulent structure of that layer. In the past decade, significant progress has occurred in our understanding of PBL structure and diffusion, especially for the daytime convective boundary layer (CBL). As discussed in Ch. 1, this has resulted from the collective efforts of numerical modeling, laboratory simulations, and field observations. In this chapter I discuss how this better understanding has improved the theoretical basis and performance of dispersion models for routine applications. I shall focus on point sources of nonbuoyant (i.e., passive) or buoyant material at arbitrary height in the CBL. The CBL is especially important in the case of tall stacks because they generally contribute their maximum ground-level concentrations (GLC) during convection (see Venkatram, 1980; Weil and Brower, 1984).

In this chapter I use current understanding of micrometeorology to describe dispersion of pollutants in the stable boundary layer (SBL). I emphasize how dispersion works rather than particular formulas that can be used to estimate ground-level concentrations. Whenever it is possible, I will make recommendations on the use of available models for dispersion. However, my main objective is to provide the type of understanding of dispersion that will allow you to tackle a problem without reaching for a handbook on air pollution modeling.

The preceding chapters have laid the foundation for the material presented in this chapter. Here, I shall show how the ideas described earlier can be used to develop practical air pollution models. I shall focus on three topics that are considered important in applied dispersion modeling. The first is dispersion in coastal areas. The recent interest in this subject has been stimulated by offshore drilling for oil. Most of the models developed to estimate the impact of these offshore activities use dispersion parameterizations that are strictly applicable to sources over land. In this chapter I shall emphasize the need to relate the dispersion to the physics of the over-water boundary layer, which differs from that over land. A realistic dispersion model must also treat the sharp change in boundary layer structure as air flows from water to land. I discuss methods to incorporate this transition into models and also discuss estimating the height of the coastal boundary layer, one of the critical inputs to a shoreline dispersion model.

One may argue about whether the Navier-Stokes equations of fluid motion are deterministic, or even about their relevance to atmospheric motion, but at a practical level one is forced to accept that it is impossible to make sufficiently detailed measurements to specify the entire velocity field. We therefore treat the flow as a random, or turbulent, field and attempt to make our predictions by means of statistical measures such as fluctuation correlations or probability distributions. The dispersion of a scalar contaminant in this stochastic velocity field is also a random process, and the present recognition of this aspect of dispersion is evinced by the inclusion of this chapter in a monograph on atmospheric dispersion.

The unstably stratified boundary layer is a relatively simple entity in which to try to understand and model dispersion. The depth of vertical mixing (h) in this case is usually clearly defined by the height above which stable stratification commences. The turbulence intensities responsible for the dispersion are rather well known functions of height relative to h (Caughey, 1982) if the convective velocity scale w_* is known. Also, the wind speed and direction can then usually be treated as independent of height above the surface layer and within the boundary layer if the latter is not too baroclinic.