AERMOD: Alberta's Workhorse Air Dispersion Model

Why is AERMOD the starting point for so much industrial air-quality modelling?

This is my favourite air dispersion model for conventional industrial work in Alberta. That is a professional judgement based on experience, not a claim that one model is perfect for every situation.

AERMOD is versatile, well established, widely accepted for regulatory work and supported by a mature set of meteorological, terrain, building and post-processing tools. For most conventional industrial facilities, it gives us a practical way to turn emissions and weather information into predicted ground-level concentrations.

It is also familiar territory at Calvin Consulting Group.

Most Calvin staff involved in dispersion assessments know how to set up and run AERMOD. Senior Calvin modellers carry out quality assurance and quality control. We routinely check the model inputs against the information supplied by the client, against other available data and against what makes physical sense. 

That last part matters. Getting a model to run is easy compared with deciding whether you have told it the right story about the facility.

AERmod is a user-friendly air quality model

The short answer: why AERMOD?

For conventional industrial projects, four things make AERMOD especially useful.

1. It fits the regulatory landscape - Alberta currently recommends AERMOD-PRIME for refined dispersion assessments. Alberta describes it as a multi-source, steady-state plume model. CALPUFF is also a recommended refined model, with a different role when time- and space-varying meteorology needs to be represented.

AERMOD is also an EPA preferred and recommended regulatory model and remains an actively maintained system.

2. It handles the features found at many industrial sites - AERMOD can represent multiple emission sources, elevated and surface sources, terrain, receptors and building-related effects through the PRIME downwash methods.

3. It has a mature modelling ecosystem - AERMOD is part of a larger workflow that includes tools such as AERMET for meteorology, AERMAP for terrain and BPIPPRIM for building geometry and downwash.

4. It is usually enough - That is an underrated advantage.

A more complicated model can represent more complicated atmospheric behaviour. It also requires more data, more processing and more quality control.

For a conventional industrial site, there is little benefit in adding complexity simply because you can. The simplest defensible model that answers the question is often the best model.

What does AERMOD actually do?

At its simplest, the modelling chain looks like this:

Emissions → Meteorology → Plume behaviour → Dispersion → Receptors → Predicted concentration

Suppose a compressor engine releases nitrogen oxides through a stack. We need to know things such as:

• How much pollutant is being emitted?
• How high is the stack?
• What is the exit velocity?
• How hot is the exhaust?
• What buildings are nearby?
• What is the terrain doing?
• Which way does the wind blow?
• How strongly is the atmosphere mixing?
• Where could someone be exposed?

AERMOD uses that information to estimate concentrations at specified receptor locations.

The result might be a table of maximum concentrations, a map of concentration contours or both. The computer does the mathematics. The modeller decides what the mathematics should represent.

Why does AERMOD fit Alberta so well?

Much of Alberta is remarkably convenient from a dispersion-modelling point of view.

Outside major urban areas and particularly complicated terrain, many industrial facilities are relatively isolated. The major emission sources may be concentrated within a fairly small facility footprint.

That does not mean the atmosphere is simple. Rather, it means the important modelling problem can often be represented effectively with a steady-state approach. This is one reason AERMOD has become such a practical workhorse for Alberta industrial assessments.

At Calvin, the same basic approach has also served projects in Saskatchewan and British Columbia along with work in Ontario, New Brunswick, Quebec and internationally.

The jurisdiction changes. The modelling question often looks surprisingly familiar.

AERMOD is a steady-state model. What does that mean?

Steady-state sounds more intimidating than it is. Imagine a plume leaving a stack while the wind is blowing steadily from west to east.

For each meteorological condition, AERMOD treats the plume as though the relevant atmospheric conditions can be represented by a steady-state approximation as the plume travels downwind.

That works well for many near-field industrial situations. It becomes less attractive when important atmospheric conditions change substantially across the modelling domain.

For example, a mountainous region can have valleys, ridges and changing wind directions that produce a much more complicated flow pattern.

That is where another model such as CALPUFF may deserve consideration. The key lesson is simple:

The model should match the atmospheric problem.

Where does the weather data come from?

This is where things get much more interesting. AERMOD does not simply need a weather forecast. It needs meteorological information that can be processed into quantities relevant to atmospheric dispersion.

A typical meteorological dataset used with AERMOD can contain fields such as:

A real AERMOD meteorological file can therefore look rather cryptic. You might see a row like: 11 01 01 1 01 -36.4 0.370 ... 116.9 0.50 -1.00 1.00 3.39 255.0 10.0 268.9 2.0

That isn't just a list of weather observations. It is a compact description of the atmospheric conditions AERMOD needs to estimate dispersion.

WRF, AERMET and AERMOD

Modern modelling often begins with meteorological information generated by a numerical weather model such as WRF. WRF stands for the Weather Research and Forecasting model.

You do not need to understand WRF's equations to use AERMOD properly. What matters to the modeller is whether the resulting meteorological dataset is appropriate for the site and how it is processed.

The basic chain is: Meteorological data → AERMET → AERMOD

AERMET processes meteorological information into the surface and profile files required by AERMOD. At Calvin, we look beyond the fact that a meteorological file was successfully generated.

We examine the resulting wind behaviour and other characteristics to make sure the dataset makes sense for the modelling location.

A wind rose is particularly useful. It quickly answers: Where does the wind usually come from?

It can also help reveal whether a modelled concentration pattern looks physically plausible.

A real Calvin modelling workflow

A recent Calvin assessment illustrates how much more is involved than simply opening AERMOD. The assignment was to assess maximum predicted ground-level NO₂ concentrations from stationary combustion sources at an Alberta gas plant.

The assessment used five years of meteorological data, terrain information, receptor grids, background concentrations, NOx-to-NO₂ conversion and building downwash.

That sounds like a long list. In practice, it becomes a chain of questions.

What are the emissions?

The team evaluated source information including engine ratings, emission factors, stack heights, diameters, exit velocities, exhaust temperatures and emission rates.

Does the source information make sense?

Client data were checked against other available information. Where information was unavailable, professional experience was used to develop estimates.

For some sources we can also consult Alberta's large inventory of source and emissions information to compare expected values by source type. This is particularly useful when checking whether an unusual stack parameter or emission estimate is plausible.

Is the terrain appropriate?

Terrain data were prepared for the modelling domain. AERMOD's terrain treatment is important because a receptor's elevation can affect the predicted plume interaction and concentration.

Where should the receptors go?

The assessment used several progressively coarser receptor grids extending from the immediate plant area to 10 km, along with fine receptors around the plant perimeter and around predicted maximum concentrations.

The important point is that the receptor grid is designed around the question being asked. More receptors are not automatically better...the right receptors are better.

Building downwash: when the plant changes its own plume

A stack does not release pollution into an empty world. Buildings can disrupt the airflow around a stack and pull a plume downward or otherwise alter its dispersion. That is called building downwash.

Calvin routinely uses BPIP to represent the relevant building geometry in AERMOD assessments.

In the example gas-plant assessment, the model included numerous compressors, heaters, reboilers and other structures with different roof heights and locations.

That information matters. An inexperienced modeller might look only at stack height. An experienced modeller asks: What does the building immediately downstream of this stack do to the plume?

That is one reason AERMOD's building-downwash capability is so useful for industrial facilities.

AERMOD output: don't just admire the pretty map

One of the most recognizable products of an AERMOD assessment is a concentration-isopleth drawing. The black curves on these drawings are isopleths: lines connecting points with the same predicted concentration.

They work much like elevation contours on a topographic map, where  topographic contour joins locations of equal elevation. A concentration isopleth joins locations of equal predicted concentration.

An AERMOD air dispersion isopleth drawing with contours separating high and low concentration points.

Calvin includes these drawings in many of our modelling reports. The useful question is not merely: Where is the high area? 

Ask instead: Why is the maximum there? Is it downwind of a major source? 

Is there building downwash? Does terrain affect the plume? Is the maximum close to the facility or farther away?

Does the shape of the contour make sense given the wind pattern? Does the pattern change when the operating scenario changes?

Those questions turn a colourful map into a quality-control tool.

What I actually check before trusting an AERMOD result

This is the part that rarely appears in a software description.

1. Do the inputs agree with the facility?

Stack parameters, source locations, operating conditions and emissions need to agree with the best available information.

2. Do the numbers look plausible?

An exit velocity that is wildly different from what the equipment should produce deserves investigation. So does an emission rate that is inconsistent with similar equipment.

Here's what AERMOD's output may look like

3. Does the meteorology make sense?

We look at wind direction, wind speed and other meteorological behaviour rather than simply accepting the dataset because AERMET processed it successfully.

4. Does the terrain make sense?

A map of terrain and sources can reveal problems that are difficult to see from a table of numbers.

5. Do the buildings look right?

A misplaced building can produce a misplaced downwash effect.

6. Does the concentration pattern make physical sense?

The model may be mathematically correct while the interpretation is wrong. The modeller still has to ask why the maximum occurs where it does.

7. Does the regulatory comparison use the right statistic?

For a typical Alberta assessment, for example, the reported one-hour NO₂ design concentration is the ninth-highest predicted value after the applicable modelling requirements were applied.

That is very different from simply taking the single largest hourly number from a giant output file.

Background concentrations: the atmosphere was there before your facility

A plant is rarely the only source of a pollutant. That means the concentration predicted from the facility may need to be combined with an appropriate background concentration.

In the recent Calvin assessment, available Alberta monitoring data were evaluated to develop background NO₂ concentrations. The assessment used the monitoring station and processed the available data according to the applicable modelling approach.

The resulting background values were then combined with the modelled facility contribution.

This is an important conceptual distinction: AERMOD predicts the modelled source contribution.

The regulatory assessment may require additional background information to represent the broader air-quality environment. That distinction becomes particularly important when the predicted facility contribution is small compared with background.

What about chemistry?

AERMOD can also handle selected treatments for pollutants that undergo chemical conversion after release. A good example is NOx-to-NO₂ conversion.

In this Calvin assessment, the model used the Plume Volume Molar Ratio Method, or PVMRM, for NO₂ modelling in accordance with the applicable Alberta modelling approach.

That required decisions about the NO₂/NOx equilibrium ratio, in-stack ratio and ozone information.

This illustrates another important point. The modeller does not simply choose NO₂ from a menu and press run. The chemical treatment has to match the pollutant, the available data and the regulatory methodology.

For more specialized atmospheric chemistry problems, another model may be more appropriate.

AERMOD is versatile, and that includes flare modelling

AERMOD is not restricted to ordinary industrial stacks. It is incorporated into Alberta's AERflare tool used for specialized flare evaluations.

Calvin uses AERflare when appropriate for flare assessments and design work. This is particularly useful because flare modelling brings together several unusual source characteristics, including flare geometry, gas composition, heat release and combustion behaviour.

AERflare can also produce the flare-related parameters needed for applications under BC guidance where those specialized parameters are required.

We have also used CALPUFF-based ABflare approaches when the circumstances justified them. An interesting lesson emerged from one regulatory discussion at Calvin: a regulator involved in the development of AERflare questioned why we had used the CALPUFF-based approach in a particular situation (legacy reasons) rather than the AERflare/AERMOD approach.

That was useful feedback. It reinforced something we already believed: If AERMOD can defensibly answer the flare question, adding another modelling system needs a good reason.

Other models are selected at Calvin when there is a strong technical or regulatory case for doing so. 

The easiest model to choose is not necessarily the easiest assessment to defend

AERMOD is now so familiar that it can be tempting to treat model selection as automatic. In many Alberta industrial projects, that is reasonable.

But 'default' does not mean automatic. A good modeller still asks:

• Is the site within the range of situations AERMOD represents well?
• Is the terrain important?
• Are buildings important?
• Are the emissions well characterized?
• Is the meteorology representative?
• Are there unusual temporal or spatial wind patterns?
• Does the regulatory framework call for a different approach?
• Is there chemistry that requires specialized treatment?

Only after those questions have been considered should the model become routine. That is why AERMOD can be both the default and a professional judgement.

Can you run AERMOD yourself?

Yes. Learning to run AERMOD can be an excellent way to understand atmospheric dispersion.

You can learn about plume rise, wind direction, stability, terrain, building downwash and receptor grids by experimenting with a model.

For a regulatory assessment, however, the most difficult decisions tend to happen before the model runs.

• Someone has to decide what the facility actually emits.
• Someone has to decide which scenarios matter.
• Someone has to decide whether the meteorological data are representative.
• Someone has to decide whether the terrain and buildings are properly represented.
• Someone has to decide whether the final concentration pattern makes physical sense.
• And someone has to be responsible for finding the problems before somebody else does.

That is the value of professional review.

A modelling result should be challengeable

There is another reason professional judgement matters. A model does not measure reality directly.

It calculates predicted concentrations from a defined set of inputs, assumptions and mathematical relationships. That means a good assessment should leave the reader able to ask:

• What assumptions produced this result?
• What information could change it?
• What would make us question it?

Calvin reports explicitly recognize this uncertainty.

Our reports identify where information comes from, distinguish client-supplied information from estimates developed through professional experience and state that supplied information is generally assumed to be accurate unless otherwise verified.

That is scientific honesty, not a weakness. It tells the reader what the result actually means.

What AERMOD cannot do

AERMOD cannot tell you whether the source data provided to the modeller are correct.

• It cannot know that a stack height in a spreadsheet is out of date.
• It cannot recognize that a building was demolished last year.
• It cannot know that an emission factor has been applied to the wrong piece of equipment.
• It cannot determine whether a chosen operating scenario is actually representative of the facility.

It will run anyway. That is why professional modelling is much more than operating an AERMOD executable.

The software performs the calculation. The modeller establishes whether the calculation represents the real question.

What makes AERMOD such a useful model?

After years of industrial consulting, my answer has become fairly simple. AERMOD is versatile enough to handle a remarkable range of conventional industrial problems without forcing the modeller into an unnecessarily complicated modelling chain.

It has a strong regulatory foundation. It has a mature technical ecosystem. It can represent multiple sources, terrain, receptors and building effects.

It works with meteorological preprocessing tools designed specifically for dispersion modelling. And most importantly, experienced modellers know how to check it.

That last point may be the most important one. A tool becomes genuinely useful when you know where it works, where it struggles and what its results should look like when everything is behaving sensibly.

The Calvin approach

At Calvin Consulting Group, our default is straightforward:

Start with AERMOD when it fits the problem. Use other models when there is a strong reason to do so.

That reason might be the physical behaviour of the atmosphere, unusual terrain, specialized chemistry, long-range transport, a specific regulatory requirement or a specialized source such as a flare.

AERMOD is often the right answer. Knowing why it is the right answer is part of the professional work.

Senior Calvin modellers review the model setup and results through QA/QC processes. We compare supplied information against other available data and against engineering and physical expectations. When necessary, we draw on Alberta emissions and source information to develop or check source parameters.

The goal is a model that does more than produce numbers and an assessment that can be explained, reviewed and defended.

Need AERMOD modelling?

You can download an AERMOD model. You can learn how to build an input file. You can make a concentration map. The difficult part comes earlier.

• What should you model?
• Which data should you trust?
• Which assumptions matter?
• Which scenarios could control the result?
• Is the meteorology representative?
• Does the maximum concentration make sense?

Those decisions can determine whether a modelling project proceeds smoothly or has to be rebuilt after review.

At Calvin Consulting Group, we can help at any stage — from an early review of the modelling approach through source characterization, meteorology, receptors, building downwash, modelling, QA/QC and regulatory interpretation.

Getting an experienced modeller involved early is often much cheaper than discovering later that the model answered the wrong question.

The computer can run the model. The specialist helps make sure you are running the right model, for the right reason, with the right information.  Please email Barry J. Lough at Calvin Consulting Group Ltd. at this address for help with any questions you might have.

Thank you.

AERMOD in three levels

Level 1: Just tell me what it is

AERMOD predicts how pollutants released from industrial sources are dispersed through the atmosphere and estimates concentrations at locations where people or the environment could be affected.

• It combines emissions, meteorology, terrain, buildings and receptor locations.
• For conventional industrial regulatory work in Alberta, it is usually the starting point.

Level 2: I want to understand it

Think of AERMOD as a mathematical laboratory.

Give it:

• What came out?
• How did it leave the stack?
• What was the atmosphere doing?
• What was the land and terrain like?
• Where do we want to know the concentration?

It calculates the likely concentration at those locations.

The results can then be shown as tables, statistics and concentration isopleths.

Level 3: I have to use it

Now the important questions become:

• Which emissions?
• Which operating scenarios?
• Which meteorological data?
• Which terrain data?
• Which receptors?
• Which buildings?
• Which chemistry?
• Which model options?
• Which regulatory statistics?
• Which background concentrations?
• Which quality-control checks?

That is where the difference between operating software and performing a professional modelling assessment becomes obvious.

Do you have concerns about air pollution in your area??

Perhaps modelling air pollution will provide the answers to your question.

That is what I do on a full-time basis.  Find out if it is necessary for your project.

Have your Say...

on the StuffintheAir         facebook page

Other topics listed in these guides:

The Stuff-in-the-Air Site Map

And, 

See the newsletter chronicle. 

Thank you to my research and writing assistants, and the author remains responsible for the content.

AERMOD checklist for project managers and students

Before accepting an AERMOD assessment, ask:

Sources

• Are all significant sources included?
• Are emission rates defensible?
• Do stack heights, diameters, temperatures and velocities match the equipment?

Meteorology

• Is the meteorological dataset appropriate?
• Does the wind rose make sense?
• Are surface characteristics appropriate?

Terrain

• Does the terrain dataset cover the modelling domain?
• Does the receptor grid capture important terrain and exposure locations?

Buildings

• Have nearby structures been evaluated for downwash?
• Does the BPIP representation match the facility?

Receptors

• Is the grid detailed enough where the maximum could occur?
• Are the plant boundary and other important locations represented?

Chemistry

• Is any conversion or decay treatment appropriate?
• Are background concentrations represented correctly?

Results

• Are the correct regulatory statistics reported?
• Do the concentration contours make sense?
• Can someone explain why the maximum occurs where it does?

Quality assurance

• Has somebody experienced independently reviewed the setup and results?

That last question is often the cheapest one on the list, when you think about it.