Atmospheric dispersion models differ in several ways. One major difference is whether the model computes concentrations that result from a continuous release of a substance into the air over time (e.g., a smokestack) or from an episodic release (limited in time to a few seconds to a few hours). Models of a continuous release are known as plume models and models of an episodic release are called puff models. Plume models estimate the steady-state airborne concentrations at locations downwind from the source. Puff models estimate the concentration at a given downwind location as function of time since the release. Some models are both puff and plume models. They can estimate downwind concent rations for either continuous or episodic releases.
Atmospheric dispersion models also differ in whether they represent the release as a point, a line, an area, or a volume. Nearly every model can represent a point release. Many models also represent line releases (e.g., a release from an airplane in flight or dispersion of vehicle pollutants from highways), area releases (e.g., a burning field), and volume releases (e.g., dispersion after the source initially explodes). Some models are capable of representing all these geometries of the release as well as plume and puff (continuous or short term) variants for each of them.
Atmospheric dispersion models also differ in the assumptions they make to simplify the problem of computing downwind concentrations. In general, simpler models take less time to compute concentrations. So there is a tradeoff between accuracy of prediction and computational time that is especially relevant to uses of atmospheric dispersion models for prospective surveillance applications. To simplify computation, model developers may make assumptions such as the terrain is flat, the substance does not settle out of the air, the substance behaves as a gas, weather conditions are the same at every location, and weather conditions do not change with time.
Models that attempt to improve accuracy by making fewer simplifying assumptions typically require additional input parameters about weather, the substance itself, and/or terrain.The simplest atmospheric dispersion models require as meteorological input only wind speed, wind direction, and a measurement of atmospheric turbulence such as the stability class of the atmosphere, and do not require input data about terrain or the substance.3
2 The LD50 is the dose that is lethal for half the population.
3 One can compute stability classes from meteorological data that are widely available such as wind speed and cloud cover.
Finally, we note that plume models are useful in biosurveillance even when the release is episodic. We want to know the total quantity of biological agent that individuals would inhale—the inhaled dose—because it determines the probability that an individual will develop an infection. Because puff models compute the airborne concentration of spores at a given instant following the release, they enable computation of the inhaled dose as a function of time. To compute the total inhaled dose over the entire release however, we must sum the quantity inhaled at each instant after the release over all such instants. If we assume a constant breathing rate, we can multiply it by the sum of the concentrations from the puff model over all instants in time to obtain the inhaled dose. Some plume models compute this very sum of concentrations, because they are the time-integrated (or time-summed) form of a puff model: they assume that a continuous release of spores is equivalent to the situation where a puff release occurs at every instant. The concentrations that this type of plume model computes can be interpreted in two ways: (1) steady-state airborne concentrations that result from a continuous release, or (2) the sum over time of concentrations from a puff release.
Was this article helpful?