Air Quality Dispersion Modelling: How It Works, When You Need It and What the Results Mean

Air quality dispersion modelling is used to predict how emissions from industrial facilities will behave after they enter the atmosphere. The basic idea sounds simple:

Emissions → Meteorology → Dispersion → Ground-level concentrations → Regulatory decision

In practice, the difficult part is rarely getting AERMOD or CALPUFF to run. The difficult part is deciding what should be modelled, how it should be represented and whether the result makes physical and regulatory sense.

Safeguarding community health

I've spent decades working with industrial air-quality assessments in Alberta and elsewhere in Canada. One thing that experience has taught me is that a technically correct model can still give the wrong answer if the facility, operating scenario, emissions, meteorology, buildings, terrain or regulatory question has been represented incorrectly. And sometimes the right answer is not to do more modelling at all.

Before commissioning another modelling study, it is worth asking what question the assessment actually needs to answer. What is air quality dispersion modelling?

An air-quality dispersion model estimates how pollutants released from a source are transported and diluted through the atmosphere. Depending on the project, the model may consider:

• emission rates and source characteristics,
• wind speed and direction,
• temperature and atmospheric stability,
• terrain and land use,
• building downwash,
• neighbouring emission sources,
• chemical transformation,
• deposition,
• operating scenarios and
• the location of people or environmental receptors.

The model then calculates predicted concentrations at selected locations.

Those predictions are compared with applicable air-quality objectives, guidelines, standards or health-based criteria. The result is more than a measurement of what will happen. It is a technical prediction based on assumptions and data.

That distinction matters. A good assessment therefore asks not simply: Did the model run?

...but: Does the model represent what can actually happen at the facility?

Where does air quality dispersion modelling fit into a project?

Modelling is usually part of a larger regulatory and engineering process.

A simplified project sequence looks like this:

1. Facility design
2. Emission inventory
3. Operating scenarios
4. Meteorological and terrain data
5. Model selection and setup
6. Air quality dispersion modelling
7. Results and interpretation
8. Comparison with regulatory criteria
9. Engineering or operating changes, if needed
10. Regulatory submission or decision

The important point is that modelling is more than an isolated exercise.

A change in the facility may change the emissions. A change in emissions may change the plume. A building may change the plume again. A different operating scenario may produce a more important result than maximum production.

And a regulator may ask a different question from the one the original assessment was designed to answer.

Before opening the model: What question are we trying to answer?

This is one of the most important steps in an assessment. A project may appear to need a new model when the real issue is much narrower. For example:

• Has production increased?
• Has a stack changed?
• Has a flare been added or removed?
• Has the fuel changed?
• Has an operating condition changed?
• Is a new source close enough to matter?
• Is existing meteorology still representative?
• Did the regulator ask for something that the previous model did not address?
• Does an existing assessment already answer the question?

At Calvin Consulting, we occasionally recommend not doing additional modelling when the existing information is sufficient. That may mean:

• using existing modelling,
• revising an emission rate,
• clarifying an operating assumption,
• reviewing the previous model inputs,
• comparing a proposed change against an existing assessment or
• confirming expectations with the regulator before doing additional work.

More modelling is not automatically better modelling.

The objective is to provide the evidence needed to answer the regulatory question.

The building blocks of a modelling assessment

A defensible assessment generally depends on a number of interconnected inputs.

1. Emissions - Everything starts with understanding what is actually being released. Emission rates may come from:

• engineering calculations,
• equipment specifications,
• emission factors,
• stack testing,
• continuous emissions monitoring,
• process simulations,
• operating records or
• regulatory inventories.

An industrial facility rarely has a single emission rate. Different emissions can occur during:

• normal operation,
• maximum production,
• reduced operating rates,
• startup,
• shutdown,
• maintenance,
• flaring,
• blowdown,
• upset conditions and
• emergency operation.

The modeller's job is to determine which of those operating states matter.

See:

1. What Different Types of Air Pollution Do You Actually Need to Model?
2. Method of Variation of Parameters — Air Quality Dispersion Modelling Guidelines: Sources
3. What Is Gas Flaring?

2. Meteorology - Pollution does not simply travel downwind in a straight line. Wind speed, wind direction, temperature, atmospheric stability, turbulence, mixing height, precipitation and other atmospheric conditions all affect dispersion.

Depending on the assessment, meteorological data may come from:

• site-specific observations,
• representative nearby stations,
• regulatory datasets or
• numerical weather prediction models.

For many refined assessments in Western Canada, five years of hourly meteorological data are used. Selecting a dataset is not just a matter of finding the nearest weather station. The important questions include:

• Is the dataset representative of the site?
• Does it adequately represent the terrain and surface characteristics?
• Are there gaps or quality-control issues?
• Are the meteorological conditions appropriate for the model being used?

See:

1. How Numerical Weather Prediction NWP Models Help You Assess Air Quality
2. Air Quality in Canada — Refined Modelling and Meteorological Data

3. Terrain and land use - Terrain can change both wind flow and plume behaviour. A model may therefore require detailed elevation data covering the modelling domain.

This becomes particularly important when a source is located near:

• hills,
• valleys,
• ridges,
• coastlines,
• large lakes,
• complex industrial terrain or
• rapidly changing land-use conditions.

See: 

1. Digital Terrain Elevation Data - Terrain is closely connected to meteorology, which is one reason modelling in complex terrain can become considerably more complicated than modelling on a flat site.
2. Recommended default land surface characteristics to use in your air quality dispersion modelling.

4. Buildings and downwash - Buildings can dramatically change airflow around an industrial source.

A sketch of airborne emissions

Wind flowing around a building creates turbulence and a wake region. A plume from a nearby stack can be drawn into that wake and brought closer to the ground. This is known as building downwash. The effect can be especially important when:

• stacks are relatively short,
• sources are located on or near buildings,
• several structures surround a source or
• the plume interacts with complex building geometry.

Programs such as BPIP are used to characterize the building dimensions and geometry required by models such as AERMOD.

See: BPIP, How Building Downwash and the Plume Are Connected

This is one of the areas where experience matters. Engineering drawings, plot plans, 3-D models and actual site conditions do not always agree perfectly.

5. Receptors - A model only evaluates concentrations where receptors have been placed. Receptors may represent:

• the facility boundary,
• nearby residences,
• communities,
• roadways,
• sensitive environmental locations,
• specified elevations above ground or
• a carefully designed grid surrounding the source.

Choosing the receptor grid is therefore part of defining the question being answered. A coarse grid can miss a localized maximum.

See: What Is a Receptor in Air Dispersion Modelling?

6. Background and neighbouring sources - A facility rarely operates in isolation. Depending on the assessment, existing air quality and nearby industrial sources may need to be considered.

Background concentrations can come from representative monitoring data, while neighbouring emission sources may need to be incorporated into the model.

See: What Is Normal Air Quality?

This is also where professional judgement becomes important. A monitoring station may be geographically close but still be a poor representation of the project's surroundings.

What operating scenarios should be modelled?

One of the most important decisions in an assessment is identifying the operating conditions that could produce the important impacts. Typical scenarios may include:

An air dispersion assessment is part of an EIA

A critical point is that maximum production does not necessarily produce the maximum ground-level concentration.

For example, lower exhaust temperature can reduce plume rise. A different flow rate can change momentum. Building downwash can dominate the result. Meteorological conditions can turn an otherwise modest release into a localized maximum.

That is why scenario selection is often more important than simply increasing the number of model runs.

See: 

1. What Different Types of Air Pollution Do You Actually Need to Model?
2. Air Quality Assurance

What pollutants are typically modelled?

The answer depends on the facility and regulatory question. Common contaminants include:

• nitrogen oxides (NOx and NO₂),
• sulphur dioxide (SO₂),
• hydrogen sulphide (H₂S),
• particulate matter,
• volatile organic compounds (VOCs),
• carbon monoxide (CO),
• ammonia (NH₃),
• metals,
• odour compounds and
• other site-specific contaminants.

Some assessments also require consideration of contaminants formed after release. For example, nitrogen oxide can be converted to nitrogen dioxide in the atmosphere.

See: How Nitrogen Dioxide Is Formed

Other projects may require modelling of:

• ozone interactions,
• secondary particulate matter,
• acid deposition,
• chemical transformation or
• deposition to environmental receptors.

See:

1. Secondary Pollutants of Air Pollution
2. What Is Acid Deposition?

Choosing the right model

There is no single model that is best for every project. The appropriate choice depends on the:

• distance involved,
• terrain,
• meteorology,
• source configuration,
• operating conditions,
• contaminants,
• chemistry,
• deposition requirements and
• regulatory framework.

AERMOD is widely used for regulatory dispersion modelling and is particularly useful for many local and regional industrial assessments. It works with hourly meteorological data processed through AERMET and can account for effects such as:

• terrain,
• building downwash,
• multiple source types,
• different operating scenarios and
• selected NO₂ conversion methods.

See:

1. How to Determine Air Quality - understanding Key Models
2. Examples of Models in Math Magic - Making Sense of the World

CALPUFF is a non-steady-state modelling system that can be useful where meteorological conditions vary substantially in space and time or where more complex transport processes need to be represented. The CALPUFF system includes:

• CALMET for meteorological fields,
• CALPUFF for dispersion and deposition and
• CALPOST for processing and analysis.

Applications may include:

• complex terrain,
• longer-range transport,
• regional modelling,
• deposition,
• time-varying emissions and
• specialized meteorological conditions.

See:

1. Long-Range Transport — Modelling Domains and Cumulative Effects
2. What Is Acid Deposition?

The more sophisticated model is not automatically the better model. The appropriate model is the one that adequately represents the problem and is appropriate for the regulatory application.

The Western Canadian regulatory picture

The basic scientific problem is similar across Western Canada, but the regulatory approaches are not identical.

Alberta - Alberta's Air Quality Model Guideline provides a framework for selecting models, preparing inputs, evaluating emissions, treating terrain and buildings, selecting meteorological data, assessing results and documenting the assessment.

The guideline also addresses specialized subjects such as:

• flaring,
• NO₂ conversion,
• regional modelling,
• deposition,
• secondary pollutants,
• modelling qualifications and
• assessment content.

See:

1. Air Quality Assessment Reports
2. Air Quality Assurance
3. Method of Variation of Parameters
4. What Is a Gas Flare?
5. How Nitrogen Dioxide Is Formed
6. What Is Normal Air Quality?

British Columbia - BC provides particularly detailed guidance for modelling in complex and unusual circumstances. The provincial framework places considerable emphasis on:

• assessment level,
• model selection,
• meteorological data,
• source characterization,
• terrain,
• building effects,
• NO₂ conversion,
• quality assurance and
• professional judgement.

See:

1. Numerical Weather Prediction NWP Models
2. Digital Terrain Elevation Data
3. Building Downwash
4. How Nitrogen Dioxide Is Formed

Air quality dispersion modelling process

Saskatchewan - Saskatchewan provides modelling guidance for assessing the effects of new and modified emission sources and identifies approved approaches and models for different applications. Particular attention may be required for:

• variable operating conditions,
• intermittent sources,
• flaring,
• dust,
• complex terrain,
• cumulative effects and
• meteorological data.

See:

1. Air Quality Particulate Matter
2. What Different Types of Air Pollution Do You Actually Need to Model?

Assessing new developments' air quality

Manitoba - Manitoba's modelling guidance similarly emphasizes the quality of the source information, meteorology, terrain, receptors, background concentrations and modelling methodology.

Project-specific consultation may be important where a standard air quality dispersion modelling approach does not adequately represent the facility.

What does "good" modelling look like?

A good assessment should be:

Representative - The inputs describe the facility and its operating conditions realistically.

Defensible - The assumptions and methods can be explained and supported.

Regulator-ready - The results answer the questions the regulator needs answered.

Physically plausible - The spatial and temporal behaviour of the predicted concentrations makes sense.

Efficient - The assessment does enough work to answer the question, without creating unnecessary analysis.

That last point deserves emphasis.

A modelling project can become expensive very quickly. Running another five scenarios does not necessarily improve the answer.

Sometimes a single corrected input does...in other cases, an engineering change does.  But sometimes the existing model already answers the question.

A final thought: more modelling isn't always the answer

Air quality modelling is most useful when it reduces uncertainty. That does not necessarily mean running the biggest, most complicated model available.

Modeling tools that are brilliant

Sometimes the best approach is a refined model but it may be a screening calculation. Occasionally  an engineering change solves the problem more efficiently than another round of air quality dispersion modelling.

And sometimes, after reviewing the existing information, the most defensible recommendation is: Don't model it. The information you already have is sufficient to answer the question.

It is understanding what the work it seems like we're avoiding is supposed to accomplish.

How Calvin Consulting approaches air quality dispersion modelling

Calvin Consulting Group Ltd. provides air quality dispersion modelling and related meteorological services for industrial projects across Canada. Our work includes:

• new facility assessments,
• approval renewals and amendments,
• facility expansions,
• flare and incinerator modelling,
• emergency and upset scenarios,
• NO₂ modelling,
• cumulative assessments,
• regional modelling,
• air-quality assessment reports,
• meteorological data evaluation and
• review of existing modelling.

We use tools such as AERMOD, AERflare and CALPUFF, together with site-specific meteorological data, terrain information, engineering data and regulatory guidance. Our approach is deliberately practical.

Science-based public health protection

We don't assume that more modelling is automatically better. We first determine what changed, what needs to be demonstrated, what information is already available and what modelling is actually necessary.

That can mean a detailed new assessment; it can mean revising an existing model. Or it can mean telling a client that another modelling exercise is not justified.

With more than 30 years of experience in meteorology and dispersion modelling, our team has worked on complex industrial assessments and has provided training to personnel from organizations including Alberta Environment and Protected Areas, the Alberta Energy Regulator and Environment and Climate Change Canada.

The objective is straightforward: Clear science. Defensible modelling. Practical answers.

Contact Barry at Calvin Consulting Group Ltd. to discuss an air-quality modelling question, a proposed project or an existing assessment you would like reviewed.

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

A practical starting checklist

Before beginning or revising an air quality dispersion modelling assessment, confirm:

□ Facility configuration
□ Emission sources
□ Maximum and typical emission rates
□ Operating hours
□ Startup and shutdown conditions
□ Flare or blowdown rates and durations
□ Gas composition where applicable
□ Stack parameters
□ Building dimensions
□ Terrain data
□ Land use
□ Meteorological data
□ Receptor locations
□ Background concentrations
□ Neighbouring sources
□ Applicable air-quality objectives
□ Required operating scenarios
□ Existing modelling
□ Regulatory expectations

For a more detailed version, see: Types of Data Quality Checks That Matter

A library of air quality modelling topics

This page is intended as a starting point. The subjects below go considerably deeper into individual parts of an assessment.

Sources and emissions

Method of Variation of Parameters: How modelling guidelines address emission sources and source characteristics.

What Different Types of Air Pollution Do You Actually Need to Model? How normal, maximum, startup, shutdown, flare, upset and other operating conditions affect modelling.

What Is Gas Flaring? An introduction to flaring, its environmental role and how it is assessed.

Models and modelling methods

How to Determine Air Quality: A guide to choosing and understanding dispersion models.

Examples of Models in Math Magic A more accessible introduction to the mathematical thinking behind environmental modelling.

Air Quality in Canada An overview of modelling approaches and refined assessments.

Meteorology

How Numerical Weather Prediction NWP Models Help You Assess Air Quality: How NWP data can supplement or support meteorological datasets used for air quality dispersion modelling.

Buildings and terrain

BPIP How building geometry and downwash affect plume behaviour.

Digital Terrain Elevation Data: How terrain information is prepared and used in dispersion models.

Receptors and background air

What Is a Receptor in Air Dispersion Modelling? Why receptor selection matters and how provincial guidance approaches it.

What Is Normal Air Quality? How background and baseline concentrations are established.

Chemistry and regional effects

How Nitrogen Dioxide Is Formed: NOx-to-NO₂ conversion methods and regulatory approaches.

Secondary Pollutants of Air Pollution: Ozone, secondary particulate matter and pollutants formed after release.

What Is Acid Deposition? Regional modelling, deposition and the role of CALPUFF.

Long-Range Transport: Modelling domains and cumulative effects of nearby and distant sources.

Regulation and professional practice

Air Quality Assessment Reports: How modelling results are presented and interpreted in an assessment report.

Air Quality Assurance: How emission scenarios and modelling choices affect confidence in an assessment.

Certified Regulatory and Compliance Professional: 
The qualifications, technical knowledge and professional competencies involved in dispersion modelling.

What an experienced modeller actually checks

A model can run perfectly and still be wrong.

Before I accept a result, I want to know:

• What is actually being emitted?
• Does the emission rate correspond to the operating scenario?
• Where is it being emitted?
• Does the source location match the engineering information and actual facility layout?
• Under what conditions?
• Can the sources actually operate simultaneously in the way they have been modelled?
• Does the meteorology represent the site?
• A technically complete dataset isn't necessarily a representative dataset.
• Are buildings represented correctly?
• Could downwash be controlling the result?
• Could terrain change the answer?
• Are there hills, valleys, shorelines or other terrain features that matter?
• Are the receptors appropriate?
• Could the grid miss a localized maximum?
• Are background and neighbouring sources reasonable?
• Is the assessment accounting for the air that is already there?
• Does the result make physical sense?
• If the result changes dramatically after a small change in an input, there may be a physical reason—or there may be a modelling problem.
• Does it answer the regulatory question?
• This is ultimately the most important check.

A model can be mathematically correct and still answer the wrong question.

Check out Types of Data Quality Checks That Matter for more.

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.

What can go wrong?

Some of the most important modelling problems are surprisingly mundane.

• An engineering drawing may show one stack height while the modelling spreadsheet contains another.
• A process engineer may provide a normal emission rate when the regulatory assessment requires a maximum credible rate.
• A plot plan may not accurately represent the current facility configuration.
• A meteorological dataset may be valid but poorly representative.
• A nearby source may have been omitted because it belongs to another company.
• A rooftop source may be treated as an ordinary point source even though the surrounding buildings dominate its plume behaviour.
• A flare may have different emissions at different flow rates.
• A temporary operating scenario may be modelled using assumptions that don't match the actual authorization.

None of these problems requires a software malfunction.

They require someone to notice them.