How to Determine Air Quality Using Dispersion Models

A practical Western Canadian guide to screening, refined modelling and choosing the right dispersion model

How to determine air quality around an industrial facility is not simply a matter of choosing a model and pressing run.

A dispersion model uses emissions, meteorology, terrain, source characteristics and receptor locations to predict ambient concentrations. The important word is predict. Modelling does not measure air quality; it estimates what is likely to occur under the conditions represented in the model.

That distinction matters because different questions require different levels of modelling.

A quick screening exercise may be enough to determine whether a project is obviously low risk. A detailed regulatory assessment may require hourly meteorology, accurate terrain, building downwash and a carefully designed receptor network. A complicated terrain or regional problem may require a model capable of representing changing three-dimensional winds.

So, when someone asks: 'How do I determine air quality for this project?', the first answer should be: First determine what question the assessment needs to answer.

Three levels of air-quality modelling

A useful way to think about how to determine air quality is to divide modelling into three broad levels.

1. Screening - Could this source create a concern?

A screening model uses simplified or conservative assumptions to estimate potentially important concentrations relatively quickly.

Source data and model recommendations

The current Alberta framework identifies AERSCREEN as its screening model. BC similarly identifies AERSCREEN for Level 1 screening.

2. Refined modelling - What concentrations are predicted under representative meteorological and operating conditions?

Refined modelling uses substantially more detailed source, meteorological, terrain and receptor information.

In Alberta, the current recommended refined models are AERMOD-PRIME and CALPUFF.

3. Specialized or comprehensive modelling - Does the problem involve atmospheric behaviour or chemistry that requires something beyond a conventional dispersion model?

Examples can include:

• complex regional photochemistry;
• secondary PM₂.₅ or ozone;
• specialized deposition assessments;
• unusual source behaviour;
• regional transport;

or other circumstances where an ordinary steady-state plume model does not adequately represent the problem.

This is where models such as CALPUFF, CMAQ, CAMx, ADMS 6 or other specialized approaches may become relevant, depending on the project and regulatory framework.

The important point is that these are different questions, not simply three versions of the same model. 

Screening isn't a bad model

A screening model is designed to answer a different question.

AERSCREEN uses a matrix of meteorological conditions representing a range of possible conditions and calculates worst-case concentrations for a single source. BC's guideline, for example, identifies AERSCREEN for screening maximum 1-, 3-, 8-, 24-hour and annual concentrations from individual sources.

Here's a handy trick to translate one-hour ambient air quality objectives into equivalent 24-hour or annual impact markers. You can calculate an impact parameter and then multiply your one-hour concentration by it to scale down concentrations for longer periods.

Calculate other averages from hourly concentration.

That makes it useful for an early go/no-go question: Is there enough potential impact here that more detailed modelling should be considered?

Screening can therefore save time by identifying projects that don't warrant a major modelling exercise. But screening has limitations.

It is not intended to provide the same detailed picture as a refined assessment using representative hourly meteorology.

Why Calvin normally starts with AERMOD instead

At Calvin Consulting, our normal starting point for a refined industrial dispersion assessment is AERMOD, not AERSCREEN. There is a practical reason.

AERSCREEN can be useful for a quick screening exercise, but we have found that for many projects the effort required to prepare the source information and model setup is close enough to the effort required for AERMOD that the additional usefulness of the refined AERMOD result makes it the better choice. If AERMOD is deemed necessary, it's already underway.

We do still use SCREEN3 occasionally for informal, rough pre-assessments when we want a very quick high-level indication of how an iterative design or scenario is behaving. That is not our approach to a formal regulatory assessment.

The distinction is important: A quick internal screen is a decision-support tool. A regulatory model is an assessment that has to withstand review.

That is one reason it is useful to understand the difference between how to determine air quality informally during project development and how to establish a defensible regulatory assessment.

AERMOD: the practical workhorse

AERMOD is a refined steady-state dispersion model used extensively for industrial air-quality assessments. Its supporting tools include:

1. AERMET for preparing meteorological data.
2. AERMAP for processing terrain and receptor elevations.
3. BPIPPRM for preparing building-related parameters for PRIME downwash treatment.

Together, these tools allow a modeller to represent a surprisingly broad range of industrial situations. AERMOD can handle:

• multiple sources;
• point, area and volume sources;
• variable emissions;
• terrain effects;
• building downwash;
• deposition;
• rural and urban settings; and
• a wide range of averaging periods.

That makes it an excellent catch-all for conventional industrial dispersion modelling. For many projects, there is simply no compelling reason to make the problem more complicated.

When AERMOD is usually a good choice

At Calvin, we commonly stay with AERMOD when the project involves:

• Relatively uncomplicated terrain - Much of our Alberta work falls into this category, particularly isolated facilities in rural areas with gently rolling terrain.
• Conventional industrial sources - Stacks, vents, process sources and other familiar source configurations are generally well suited to AERMOD.
• A local or near-field question - If the important impact occurs relatively close to the facility, a steady-state plume approach may adequately represent the problem.
• No compelling regional chemistry or deposition requirement - If the question is simply where primary pollutants are expected to reach their maximum concentrations, there may be little reason to introduce a regional modelling system.
• A clear regulatory pathway - When the applicable provincial guidance supports AERMOD and the physical characteristics of the project do not require a different approach, AERMOD is often the efficient and defensible choice.

This is one of the most useful lessons in how to determine air quality: Start with the simplest model that can adequately represent the physical problem.

When does CALPUFF become worth considering?

CALPUFF is a non-steady-state puff model capable of representing changing meteorological conditions across space and time.

That additional capability can become valuable when the project involves a situation that a conventional steady-state plume model may not represent adequately.

At Calvin, we would consider CALPUFF when one or more of the following applies.

1. It was used historically, Continuity can matter.

If previous assessments for the same project or region used CALPUFF, changing models can make long-term comparisons more difficult.

Sometimes retaining the established approach is more useful than changing models simply because another model is available.

2. An authority or client recommends it

A regulator, project requirement or client may establish the preferred modelling pathway.

Terrain models help us predict pollution

3. Significant complex terrain is close to the project

Mountains, foothills, valleys and substantial upslope terrain can create atmospheric flows that are not well represented by a simple straight-line assumption.

At Calvin, being within roughly 10 km of mountains or substantial upslope terrain is a practical reason to take a closer look at CALPUFF, particularly when the terrain exceeds minimal complex-terrain conditions.

That is a screening consideration, not a regulatory cutoff.

4. There are many significant sources

A large industrial region may require a broader assessment. However, the number of sources alone is not enough.

AERMOD can model many sources. The real question is whether the sources, terrain, meteorology and assessment objective combine to make a regional modelling approach useful.

5. There is a genuine long-range transport question

If an assessment extends well beyond the near field (perhaps on the order of 30 km or more) we will examine whether a model capable of following changing meteorological fields over that distance provides a worthwhile advantage.

Again, distance is an indicator, not a magic switch.

6. Deposition or secondary pollutants matter

If the assessment requires regional deposition, chemical transformation or secondary pollutants, CALPUFF may offer capabilities that make the additional complexity worthwhile.

7. The client wants the additional analysis and the budget supports it

There can be legitimate reasons to undertake a more sophisticated assessment. But a larger budget does not turn a complicated model into a scientifically necessary one.

8. Terrain and land-use effects need more detailed treatment

Sometimes the reason is not mountains alone. The combination of terrain, land use, water bodies, changing wind fields and source locations may justify a three-dimensional meteorological approach.

These are the sorts of questions that make how to determine air quality a judgement exercise rather than a software-selection exercise.

Model selection guidance at a Glance

Complex terrain does not automatically mean CALPUFF

This is worth emphasizing because it is one of the easiest rules to misunderstand. A project can be in hilly terrain and still be adequately represented by AERMOD for a particular question.

Modeling air quality using simple and complex terrain

Conversely, a project that is not immediately beside a mountain can still justify a more sophisticated model if the assessment concerns regional transport or spatially varying meteorology.

British Columbia's guidance is especially useful here. It recognizes circumstances in which AERMOD can be appropriate even when some more complicated atmospheric processes exist, depending on the area of interest and the objective of the assessment. CALPUFF becomes more attractive when the actual flow field, curved trajectories or spatial distribution of impacts are important.

The practical lesson is: Terrain is a reason to examine the model choice, not a reason to make the choice automatically.

The wrong question: Which model is more advanced? This is probably the most common conceptual mistake.

CALPUFF is more complicated than AERMOD in many respects. That does not make every CALPUFF assessment better. A more complicated model generally requires:

• more meteorological information;
• more processing;
• more input parameters;
• more decisions;
• more quality assurance;
• more interpretation.

A model with more capabilities also gives the modeller more ways to make a poor choice.

So: More sophisticated does not mean more accurate.

The correct model is the one that adequately represents the processes that matter.

The opposite mistake: choosing AERMOD because it is easier

The reverse can happen too. A modeller may already have an AERMOD workflow and simply keep using it even when the assessment has moved beyond the problem AERMOD was intended to solve.

That is equally poor reasoning. If the project requires:

• spatially varying meteorology;
• regional transport;
• important recirculation;
• regional deposition;
• specialized chemistry;
• or another process that materially affects the answer,

...then convenience is not a sufficient justification for staying with AERMOD.

Good how to determine air quality practice means being willing to make either decision.

AERMOD is enough here. 

Or: AERMOD isn't enough here.

The data may determine the model

This is often overlooked. Suppose the project appears to justify CALPUFF.

Do we have adequate information to support it? A CALPUFF assessment may require:

• detailed terrain data;
• land-use information;
• representative meteorological data;
• a suitable CALMET domain;
• emissions from relevant sources;
• appropriate chemical inputs; and
• enough information to interpret the resulting three-dimensional fields.

If that information isn't available, simply selecting CALPUFF does not make the assessment stronger.

Don't choose a model because of what the software can theoretically do. Choose it because of what you can defend with the information you actually have.

That principle applies to almost every how to determine air quality project.

How the Western Canadian provinces approach model selection

The provinces do not all use identical wording or identical assessment frameworks. That is why the current provincial guidance should always be checked before beginning regulatory work.

Alberta currently identifies:

• AERSCREEN for screening.
• AERMOD-PRIME and CALPUFF for refined assessments.

Alberta's regulatory modelling framework continues to be based on the Air Quality Model Guideline 2021, with supplemental guidance for interpreting regulatory modelling assessments effective July 2025.

Consider terrain, buildings and weather

For Alberta projects, the practical question is therefore not whether AERMOD or CALPUFF is approved. Both are recognized refined options.

The question is: Which one is appropriate to this assessment?

British Columbia (BC) organizes modelling by assessment level:

Level 1 — Screening: AERSCREEN

Level 2 — Detailed: AERMOD or CALPUFF

Level 3 — Comprehensive: AERMOD or applicable CALPUFF approaches depending on the situation.

BC's guidance places particular emphasis on the complexity of atmospheric flow, transport distance, the assessment objective and the availability of representative meteorological data.

Its framework is an excellent example of why how to determine air quality cannot be reduced to a single kilometre cutoff.

Saskatchewan currently lists:

• AERSCREEN;
• AERMOD; and
• CALPUFF

as approved standard and specialized models recommended for air-quality studies. The Saskatchewan framework also emphasizes following recommended/default approaches unless there is a reason to deviate and documenting those deviations.

Again, the important question is which model is appropriate to the particular assessment.

Manitoba is a useful reminder that provincial guidance can sometimes lag behind software development.

Manitoba's modelling material still references older guidance that identifies SCREEN3. In a 2024 regulatory submission, the province noted that its guidance had not been updated since 2006, that EPA had delisted SCREEN3 and that most Canadian provinces had moved to AERSCREEN. The same submission used AERSCREEN for its screening assessment.

Manitoba's current public guidance says that dispersion modelling requirements can vary with source complexity, operating conditions and terrain and that special modelling may be required case by case.

This is exactly why a modeller should check current provincial expectations rather than simply copy a model name from an old report.

Why screening can save money but can also waste it

There is a temptation to assume: Let's always start with the cheapest possible model. Sometimes that makes sense. Sometimes it doesn't.

If the project clearly requires refined modelling, spending substantial effort constructing a screening assessment that cannot answer the eventual regulatory question may simply add another step.

This is one reason Calvin commonly goes directly to AERMOD for refined work. The goal isn't necessarily to Do the fewest calculations. We woudl rather do the minimum modelling needed to answer the question confidently and defensibly.

A practical example from Alberta

Consider an isolated industrial facility in rural Alberta. The site has:

• gently rolling terrain;
• conventional point sources;
• no nearby mountains;
• no major regional chemistry issue;
• an assessment focused on local ground-level concentrations.

This is a natural AERMOD problem...Now change only a few things. Suppose the facility is close to the foothills, the assessment needs to consider impacts over a much larger area and the terrain produces spatially varying winds.

Now the problem is different. CALPUFF becomes much more interesting. The important lesson is that the model changes because the atmospheric problem changed.

Another example: the model does not change just because the project is large. A large facility can still be an AERMOD project.

The number of stacks is not, by itself, a reason to use CALPUFF. AERMOD can handle multiple sources.

Likewise, a relatively small facility could present a difficult modelling problem if it is situated in complex terrain or involves an unusual release.

So facility size and model complexity are not the same thing.

Flaring is a particularly useful Alberta example

Flaring demonstrates why model selection should consider not only theoretical capability but also the regulatory tools surrounding the model. Alberta's non-routine flaring framework includes AERflare, which integrates AERMOD-based calculations into the flare assessment workflow.

At Calvin, this is one reason AERMOD remains a particularly practical choice for appropriate flare assessments. We also routinely use AERflare-generated information in our flare modelling work and have experience dealing with weather-dependent flare parameters.

For some applications, an alternative approach may be technically possible. But if an established regulatory workflow already performs the required task well, there is a good reason not to replace it merely for the sake of novelty.

That is another practical lesson in how to determine air quality: Regulatory fit and practical reliability matter alongside theoretical model capability.

Don't confuse modelling with monitoring

There is another important distinction for anyone learning how to determine air quality.

Monitoring - Measures actual concentrations at a particular location and time.

Modelling - Predicts concentrations based on emissions, meteorology, terrain and other inputs.

Monitoring is therefore extremely valuable for understanding what actually happened. Modelling is valuable because it can examine:

• proposed facilities;
• future operating conditions;
• areas without monitors;
• individual sources;
• alternative designs;
• unusual operating scenarios;
• cumulative effects.

Good assessments often benefit from understanding both.

Need help choosing the right modelling approach?

Dispersion modelling can look deceptively accessible from the outside.

• There are free models.
• There are manuals.
• There are spreadsheets.
• There are tutorials.

There are even examples that appear to show exactly what to type into the input file. The difficult part comes when the project doesn't quite match the example.

What happens when the terrain is unusual? What happens when two regulatory documents appear to point in different directions? What happens when the maximum occurs somewhere unexpected?

What happens when a lower operating condition produces a higher concentration? What happens when a source is intermittent, directional or unusually configured?

Those are modelling problems, not software problems.

Calvin Consulting Group has more than 30 years of experience with industrial air-quality dispersion modelling across Western Canada. Our normal refined approach is AERMOD, but we use CALPUFF and other approaches when the project provides a compelling technical, regulatory or practical reason to do so.

We also know when not to make a modelling problem more complicated than it needs to be.

Instead of the one with the most features, the best model is the one that can answer the question you actually need answered (and withstand the next question from the regulator).

If your project is still at the stage of deciding how to determine air quality, getting that decision right before building the model can save substantial time, cost and rework.

Contact Calvin Consulting Group to discuss the modelling approach before you commit to the wrong model, the wrong assumptions or more complexity than the project requires.

Calvin Consulting can help you master the air quality aspect of your project, and we tailor our services to suit your project's environmental and economic needs. 

Clean air is our Passion...Regulatory Compliance is our Business.

Common mistakes when deciding how to model air quality

AERSCREEN is cheaper, so we should always use it. - Not necessarily. A refined assessment may make more sense from the beginning.

AERMOD is only for flat terrain. - Incorrect. The issue is whether AERMOD adequately represents the particular terrain and assessment objective.

Any span over 50 km requires CALPUFF. - Too simplistic.

Anything near mountains requires CALPUFF. - Also too simplistic.

CALPUFF is more advanced, therefore it is more accurate. - Not necessarily.

We used AERMOD on the last project. - That doesn't make it appropriate for the next project.

We have CALPUFF available, so we might as well use it. - More modelling is not automatically better modelling.

The model ran successfully, so the assessment must be sound. - Software success is not the same thing as physical or regulatory validity.

A practical checklist: how to determine air quality for a new project

Before beginning the detailed modelling, ask:

1. What is the question? - What decision will the modelling support?
2. What pollutants matter? - What is emitted directly and are secondary pollutants or deposition relevant?
3. What is the geographic scale? - Local? Regional? Long range?
4. What does the terrain look like? - Could it alter atmospheric flow?
5. What are the sources like? - What are their heights, emissions, temperatures, velocities, configurations and operating schedules?
6. What model does the current provincial guidance recommend? - Check the current version.
7. Does the recommended model adequately represent the physical problem? - If yes, use it. If no, investigate the alternatives.
8. Are the data adequate? - A sophisticated model with weak inputs is not a sophisticated assessment.
9. Is there previous modelling? - Can continuity provide useful information?
10. Does the final approach make sense? - Can you explain why the model is appropriate to someone reviewing the assessment?

If those questions have good answers, the modelling approach is usually becoming clear. The real skill isn't running the model.

Software is remarkably good at performing calculations. The difficult part is deciding: Which calculations need to be performed in the first place.

That is the heart of how to determine air quality through dispersion modelling.

AERSCREEN can provide a useful screen. AERMOD can provide a powerful refined assessment. CALPUFF can represent atmospheric conditions that call for a different approach. CMAQ, CAMx, ADMS 6 and other specialized tools can address problems that go beyond conventional dispersion modelling.

But no software package can decide what the project actually needs. That requires understanding:

• the source,
• the atmosphere,
• the terrain,
• the regulatory framework,

and the question the assessment is supposed to answer.

What an experienced modeller checks

Before selecting a model, an experienced modeller is likely to ask:

What exactly does the regulator want demonstrated?

• A maximum concentration?
• A spatial concentration pattern?
• Cumulative impacts?
• Deposition?
• Regional transport?

Where does the plume need to be followed?

• Near the source?
• Across a valley?
• Across an airshed?

What is the terrain really like?

• Gentle rolling terrain?
• Foothills?
• Mountains?
• Coastal terrain?

How complicated is the meteorology?

• Can a steady-state representation adequately describe it?

What are the sources like?

• Tall and buoyant?
• Low level?
• Directional?
• Intermittent?
• Multiple major sources?

What data do we actually have?

• Is there enough information to support the chosen model?
• Has this project been modelled before?
• Would consistency with earlier work be valuable?
• What does the provincial guidance say today? (Not what did it say ten years ago.)
• Will the more complicated model change the answer?

This may be the most important question.

If the answer is no, why add the complexity?

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.

A simple model-selection decision tree

Start here:
What are you trying to determine?

A quick conservative indication of whether a source could be important?

→ Screening

A detailed regulatory concentration assessment?

→ Refined modelling

Regional transport, complex terrain, deposition, chemistry or another specialized problem?

→ Consider a more specialized approach

If the assessment is refined:

Is the atmospheric flow reasonably represented by a steady-state plume?

Yes

→ AERMOD is a strong starting point

No / significant spatial variation

→ Consider CALPUFF

Then ask:

Does the assessment require deposition, regional transport or chemical transformation?

Yes

→ Consider CALPUFF or another appropriate specialized model

No

→ AERMOD may be sufficient

Finally:

Can the available data support the selected model?

No

→ Address the data limitation before committing to the modelling approach.

Yes

→ Proceed with the model that best answers the regulatory question.