Types of data quality checks that matter

Accurate air quality modelling depends less on whether you use AERMOD or CALPUFF and more on the quality of the inputs and the types of data quality checks used. A sophisticated model cannot compensate for poor emissions data, unrealistic assumptions or missing sources.

When I complete a 30-second review of an assessment, it starts with three questions:

What is being emitted? Where is it being emitted? Under what conditions is it being emitted? Everything else follows from those answers.

A Modeller's Method for Determining if a Dispersion Model is Trustworthy.

The first and most important of our types of data quality checks is the emissions inventory. 

• Emission factors,
• stack tests,
• process calculations,
• operating hours,
• flare events,
• upset conditions,
• chemical composition, and
• the distinction between typical and maximum emissions

...all affect the results. If the emission rate is wrong, the entire assessment is wrong.

Next, I review the source parameters. Stack height, diameter, flow rate, exit velocity, temperature, orientation and location determine how a plume behaves. Small errors can significantly change predicted concentrations and compliance outcomes.

Buildings and downwash are another common source of mistakes. Building dimensions, rooftop sources, parapets, nearby structures and release geometry must be represented correctly using tools such as BPIP. A mathematically correct model can still be misleading if building effects are not captured properly.

Meteorology is equally critical. Representative modelling requires quality hourly weather data, typically five years, including wind speed, wind direction, temperature, atmospheric stability and turbulence. The question is more than whether meteorological data exist and includes whether they accurately represent conditions at the site.

Terrain, land use and surface characteristics also influence dispersion. Elevation changes, complex terrain and local land-use conditions can affect plume transport and resulting concentrations.

Receptor placement is often overlooked but essential. Property boundaries, residential areas, sensitive receptors, facility fencelines, receptor spacing and terrain-following receptors all determine whether the assessment answers the right question.

A model must also account for background and neighbouring sources. An assessment can be technically flawless yet still be wrong if important external emission sources or cumulative effects are ignored.

Model configuration requires careful review as well. This includes AERMOD, CALPUFF, AERMET, AERMAP, BPIP, ADMS6, averaging periods, deposition, chemical transformation methods, NOx-to-NO₂ conversion approaches and other key settings.

Finally, my most important types of data quality checks: Do the results make physical sense?

The model running successfully is not enough. The plume behaviour, concentration patterns, exceedances and scenario comparisons should all be consistent with engineering principles and site conditions. Results that appear unusual or highly sensitive to minor assumptions deserve closer scrutiny.

The last step is ensuring the model addresses the actual regulatory question. A technically correct model can still be the wrong model if it evaluates the wrong scenario, whether normal operations, maximum operations, flaring, upset conditions, temporary authorizations or cumulative effects.

Air quality predictions based on good data.

Air Quality Impact Assessments and Dispersion Modelling: This page explains how air quality impacts are assessed using atmospheric dispersion modelling. Based primarily on Alberta Air Quality Modelling Guidelines, the process evaluates how emissions move through the atmosphere and whether a facility can comply with air quality requirements.

A typical assessment begins with a description of the facility, project, purpose, location, contact information and assessment level. It then identifies the dispersion model being used, any modifications to the model and the outputs to be produced, such as concentration isopleth maps, deposition estimates and compliance summaries.

A key part of the process is Data Quality Management (i.e., types of data quality checks), which verifies that model inputs, assumptions and outputs are reasonable, representative and suitable for regulatory decision-making.

In BC for instance, regulators review the modelling plan and may request revisions before the assessment proceeds. Requirements become increasingly rigorous from Level 1 to Level 3 assessments, with higher levels requiring more detailed emission inventories, stack information, receptor grids, modelling methods and quality assurance procedures.

Ultimately, a defensible air quality assessment combines reliable emissions data, representative environmental inputs, appropriate modelling methods and thorough quality control to demonstrate potential environmental impacts and regulatory compliance.

Notes from Official Provincial Guidelines

While British Columbia, Alberta, Saskatchewan and Manitoba each have their own regulatory requirements, they all rely on the same core principle: a dispersion model is only as good as the emissions, meteorological, terrain and operational data used to build it. The goal is not simply to run AERMOD or another model, but to produce technically defensible results that support regulatory decisions.

Every province requires a clear description of the facility, project purpose, emission sources, stack parameters, meteorological inputs, terrain, receptors, modelling methods, assumptions and quality assurance procedures. Regulators expect enough documentation to understand, reproduce and verify the assessment.

What All Western Provinces Have in Common: Regardless of jurisdiction, a credible air quality assessment requires:

• Accurate emissions inventories
• Realistic stack and source parameters
• Representative meteorological data
• Appropriate terrain and land-use characterization
• Building downwash evaluation
• Proper receptor placement
• Baseline or background concentrations
• Sound model configuration and quality assurance
• Clear, reproducible reporting

The common objective across all four provinces is straightforward: to determine whether a facility can operate without causing unacceptable impacts on ambient air quality and to provide regulators with defensible evidence for decision-making.

Types of Data Quality Checks for Each Province

BC - Structured Assessment Levels: British Columbia places strong emphasis on planning and assessment scope. Modelling plans identify the facility, project, assessment level, modelling approach, expected outputs and quality-control procedures before modelling begins.

Quality data for environmental protection

Requirements become more rigorous from Level 1 screening assessments to Level 2 and 3 detailed assessments, which may require site plans, elevation data, meteorological analyses, cumulative-effects assessments, isopleth mapping and electronic model files. The focus is on demonstrating that emissions, background concentrations, terrain, building downwash, meteorology and chemical transformations have all been appropriately represented.

AB - Detailed Environmental Characterization: Alberta follows a similar framework but places particular emphasis on environmental characterization and cumulative effects. Screening assessments provide a rapid estimate of impacts, while advanced assessments require detailed emissions inventories, operating scenarios, nearby industrial sources, climatology, terrain and representative hourly meteorological data.

Turbulence and air quality predictions.

Advanced studies may also evaluate acid deposition, particulate deposition, plume chemistry, visibility impacts, uncontrolled releases and other specialized issues. Alberta's approach is especially focused on demonstrating whether facilities can meet ambient air quality objectives under realistic operating conditions.

SK - Clear Documentation and Compliance Focus: Saskatchewan emphasizes clear, well-documented modelling reports that demonstrate a new or modified source will not cause unacceptable impacts on ambient air quality.

Reports typically include project descriptions, facility processes, site plans, emission calculations, modelling methods, terrain and meteorological analyses and assessments of flares, odours or deposition where relevant. Results are compared against applicable standards and presented through maps, tables and figures. Supporting model files are commonly submitted electronically to ensure transparency and reproducibility.

Accurate modelling helps keep the air clean.

MB - Verification and Traceability: Manitoba places particular emphasis on documentation, verification and traceability. Beyond presenting modelling results, reports must clearly justify model selection and fully document emissions sources, receptors, meteorological data, land use, terrain, background concentrations and any Good Engineering Practice (GEP) stack height analyses.

A critical quality-control step is ensuring that modelling inputs match engineering drawings and site information. Small discrepancies in stack heights, building dimensions, source locations or release characteristics can significantly affect results, particularly where building downwash is important. If exceedances are predicted, the assessment should identify mitigation measures and explain how regulatory requirements can still be met.

What types of data quality checks do you need?

We Solve Air Quality Challenges with Expert Dispersion Modeling

Does the Model Answer the Right Question? When reviewing a dispersion model, don't just ask whether AERMOD ran successfully. Ask whether the model accurately represents the facility, its emissions, the meteorology and the regulatory question being asked. A model can be technically correct yet still provide the wrong answer if its assumptions, inputs or scenarios do not reflect reality.

That's the approach we take at Calvin Consulting Group Ltd. We help businesses across Canada navigate complex air quality requirements with practical assessments. From selecting representative meteorological data to evaluating every emission source, we tailor each assessment to the facility, project and regulatory framework.

Our focus is not simply on running additional modelling. Before recommending further work, we first determine whether regulators actually require it. That practical, regulator-focused approach is one reason our work has been relied upon by organizations including Alberta Environment and Environment Canada.

We provide fast, reliable and technically rigorous air quality assessments that help clients demonstrate compliance, understand environmental risks and move projects forward with confidence. We handle the technical details so you can focus on running your business.

Contact Barry at Calvin Consulting Group Ltd. to discuss your air quality assessment or dispersion modelling project.

Before recommending further work, we first determine whether regulators actually require it.  A technically competent modeller should be able to tell the difference between a genuine modelling gap and a problem that can be resolved by clarifying an assumption, correcting an input or confirming the regulator's expectations.

If you're unsure whether an existing assessment is adequate, whether additional modelling is actually required or whether the inputs going into a model can be trusted, that's the kind of problem we solve.

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

How an experienced modeller actually decides whether the information going into the model can be trusted. The 10 checks I make

• Emissions
• Stack parameters
• Buildings/ downwash
• Meteorology
• Terrain/ land use
• Receptors
• Background/ neighbours
• Model configuration
• Results/ physical plausibility
• Regulatory question

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 in modelling?

Emission rate

A process engineer may provide a normal operating rate when the regulatory assessment requires a maximum credible rate.

Stack parameters

A plot plan and engineering drawing may show different stack or building heights.

Buildings

A rooftop source may be modelled without properly accounting for nearby structures.

Meteorology

A dataset may be technically valid but poorly representative of the site.

Receptors

A coarse receptor grid can miss a localized maximum.

Neighbouring sources

A project can appear compliant when important existing sources haven't been accounted for.

I've encountered projects where the engineering drawing, plot plan and modelling input file contained different stack or building dimensions. The model will still run perfectly. The problem is that it may be modelling a facility that doesn't actually exist. Thus, the most dangerous modelling error is sometimes not a modelling error at all. It's modelling the wrong facility, source, emission rate or operating scenario.