top of page

Our Recent Posts



Air Dispersion Modeling: Where Is It Open to Challenge?

NSR Law Blog is pleased to have Bill Jones, President of Blue Sky Modeling, as a guest contributor for this edition. Bill has nearly 30 years of modeling experience and shares some of his insights in this article.

Air dispersion modeling is the mathematical simulation of the transport and dispersion of air pollutants in the atmosphere. Modeling is typically conducted using air dispersion models which are computer programs that consider source characteristics, emission rates, meteorology, and topography to describe the behavior of air pollutants in the atmosphere. Models are frequently used in permitting efforts but also are applied in situations that do not involve permitting directly (e.g., litigation scenarios in which a model can help characterize pollutant impacts at a plaintiff location).

It is important to note that air dispersion models are developed primarily as regulatory tools in that they are designed to demonstrate compliance with air quality standards with a reasonable margin of error. To that end there is a certain amount of conservatism inherent in a dispersion model, and it is incumbent upon the user to understand the strengths, weaknesses, and overall tendencies of the model being used.

Because air dispersion modeling is both a science and an art, all decisions on model setup and execution are not always clear-cut. Accordingly, there are some aspects of a modeling analysis that are more open to challenge than others. This blog will briefly review six components of modeling that are common to most analyses and offer some thoughts on how they could be open to challenge.

Model Selection

EPA maintains a list of approved models in 40 CFR 51, Appendix W, and each model has its own strengths. In nearly all cases today AERMOD is the model of choice.

Is it open to challenge? Typically not, especially if the model being used is an approved model listed in 40 CFR 51, Appendix W. There may be instances where one can argue that a model besides AERMOD would be more appropriate (e.g., CALPUFF in very complicated terrain).

Model Switches

These are the settings that govern how the model is executed. Many regulatory agencies (e.g., states, Federal Land Managers, etc.) publish specific modeling guidelines for how to set up the model, and models have default settings that are typically invoked.

Is it open to challenge? Typically not, especially if the default settings are being used. Some models like CALPUFF have many switches/choices and can be complicated to set up—situations like that could be challengeable.

Emissions Inventory

The emissions inventory consists of not only the sources to be included in the modeling analysis but also how those sources are represented in the modeling. Different information is required for different source types (e.g., stack height for a point source, dimensions for an area source, etc.). There are multiple ways to calculate emission rates, especially for 1-hr SO2 and NO2 modeling and startup/shutdown scenarios. The choice of which sources to include is important as well, particularly regarding the inclusion of offsite sources.

Is it open to challenge? Often the emissions inventory is a good opportunity for a challenge to a modeling analysis because there are many different approaches that can be employed to develop that inventory. For example, have the emission rates been calculated properly or is there an equally appropriate approach that would lead to higher or lower emissions? Does the modeling analysis include all the sources that it should, or can one argue that some sources should not be included or perhaps a nearby facility’s sources should be included?

Meteorological Data

Meteorological data are required by nearly all dispersion models. EPA has published detailed guidance on meteorological data for use in dispersion modeling (see Section 8.4 of 40 CFR 51, Appendix W) but the key consideration is that the meteorological data need to be representative of the area being modeled. In the case of AERMOD, meteorological data typically are obtained from a nearby airport, but they also can be developed from an on-site meteorological station or from the use of a prognostic model. Some states provide model-ready meteorological data for applicants.

Is it open to challenge? Typically not, but in some situations the choice of meteorological data to be used can be very challengeable, particularly if the data being used are from an onsite meteorological station. For example, there are well-established QA procedures for gathering meteorological data and if those have not been followed then the data are not eligible for use in air dispersion modeling. Similarly, it is important to ascertain whether or not the data have been processed correctly.


Receptors are user-defined locations that instruct the model where it should predict pollutant concentrations. Pollutant concentrations are predicted only at locations where the public has access, which is defined as “ambient air” (e.g., receptors are not placed inside a facility’s fenceline). Usually receptors are spaced more densely near a facility and less-densely further away.

Is it open to challenge? Typically not, but in some situations the receptors used in a modeling analysis can be very challengeable. For example, sometimes what defines the ambient air boundary can be questioned—it is not always as simple as is there or is there not a fence. Also, it is important to make sure that the controlling predicted impacts are resolved to a fine enough receptor spacing; for example, if the closest receptor to the location of a maximum predicted pollutant concentration is 1 km away, the conclusions of the analysis could be questioned as there might be another location nearby that, if a receptor were placed there, would produce a higher concentration.

Background Concentrations

Required for some but not all modeling analyses, background concentrations of pollutants are added to modeled concentrations to account for impacts from sources not included in the dispersion modeling (e.g., naturally-occurring particulate concentrations). These concentrations are typically derived from air pollution monitoring conducted by EPA or states and there are many different approaches to calculating them depending on the pollutant and averaging period being addressed.

Is it open to challenge? If the total pollutant concentration (modeled plus background) is close to an air quality standard, then the background concentration can be very challengeable. Are the data used for background from a monitor that is representative of the area being modeled? Were those pollutant data gathered correctly? Was the background concentration calculated appropriately?

Concluding Thoughts

When executing an air dispersion modeling analysis there are a multitude of decisions that must be made. Some of those decisions can have a significant impact on the results of the modeling analysis, and therefore should be scrutinized carefully.

Whether advocating for a modeling analysis or debating its merits, there are always aspects of the analysis that are open to interpretation. Understanding the relative influence of those different aspects will help focus one’s scrutiny and identify areas which potentially can be challenged. Ultimately this will lead to a greater confidence in the validity and robustness of an air dispersion modeling analysis.

Bill can be reached at


Single post: Blog_Single_Post_Widget
bottom of page