Land Use and Land Cover: How Surface Conditions Affect Air Quality Dispersion Modelling

Land use and land cover are easy to overlook in air quality dispersion modelling. They shouldn't be.

AERMOD, CALMET, AERSCREEN and related modelling systems do not see the landscape the way we do. They see numerical descriptions of the surface with information that helps characterize how the atmosphere interacts with forests, fields, water, urban areas, industrial development and other land-cover types.

Those surface characteristics can affect wind, turbulence, atmospheric heating and ultimately the way an emission plume disperses. The basic chain is:

Real landscape → land-use/land-cover data → surface characteristics → meteorology → dispersion → predicted concentrations

For most projects, the established procedures for preparing these data work well. The important qualification is that the government dataset is not necessarily the current landscape. Where there has been recent urban or industrial development, I have occasionally found it necessary to update or adjust the land-use information before using it for modelling.

The model needs the landscape that exists (or will exist for the assessment) not necessarily the landscape shown in an older database.

Why land use and land cover matter

A dispersion model needs to represent the surface surrounding the emission source because the surface influences the atmosphere above it.

Different surfaces interact with sunlight, heat, moisture and wind in different ways. A forest is not an agricultural field. A lake is different from a parking lot. A city is not open grassland.

Those differences are represented in a number of parameters used by dispersion and meteorological models. For AERMOD and AERMET, important surface characteristics include:

• Surface roughness length, how strongly the surface disrupts airflow.
• Albedo, how much incoming solar radiation the surface reflects.
• Bowen ratio, how available surface energy is divided between sensible and latent heating.

For CALMET and CALPUFF, Land use and land cover information is represented spatially so that surface characteristics can vary across the modelling domain.

The important point is simple: Land use and land cover are more than just map labels. They are part of the physical description of the atmosphere used by the model.

The three surface parameters to understand

You don't need to be a meteorologist to understand the basic physics.

Consider the difference between an open agricultural field and a forest.

The forest has a much rougher surface. It also interacts differently with solar energy and moisture. Those differences affect the lower atmosphere and therefore the conditions under which a plume disperses.

This is why land-cover classification is ultimately a meteorological input, not simply a GIS exercise.

AERMOD and CALPUFF do not use land information in exactly the same way

This distinction is worth understanding.

AERMOD uses surface characteristics in conjunction with AERMET-processed meteorological data. Depending on the application, the modeller must establish appropriate values for:

• surface roughness,
• albedo,
• Bowen ratio,
• land-use classification and
• seasonal conditions.

Impact of land use and land cover on the environment

Tools such as AERSURFACE can be used to help determine appropriate surface characteristics. The result is a set of surface parameters used in preparing the meteorological input to AERMOD.

CALPUFF and CALMETuse spatially distributed information.

Land-use data are incorporated into the meteorological modelling domain through gridded information such as the GEO.DAT file. Surface characteristics can therefore vary from one part of the domain to another.

This becomes particularly useful when the modelling area contains significant changes in:

• terrain,
• vegetation,
• urban development,
• water bodies or
• other surface characteristics.

The choice is therefore not simply "AERMOD versus CALPUFF." The modeller also has to understand how the selected modelling system represents the surface.

Where does land use and land cover data come from?

The basic information may come from government or publicly available datasets. Depending on the province and application, sources can include:

• federal land-cover datasets,
• Canadian Digital Elevation Data and related geospatial products,
• GeoBC,
• provincial government datasets,
• regulatory modelling datasets,
• satellite imagery,
• maps and aerial imagery and
• site-specific information.

Software may then be used to classify, translate, grid and process the information into the format needed by the modelling system.

For example, Canadian land-cover information may need to be translated into the coding system expected by a particular model or preprocessing program. The important distinction is between:

• data that are available and
• data that are appropriate for the assessment.

The important field check: does the dataset still match the landscape?

This is one of the simplest and most useful checks a modeller can make. Government datasets are generally reliable and provide a trustworthy starting point. But they may be several years old.

That can matter when a facility is located in a rapidly changing area. A previously open field may now contain:

• warehouses,
• industrial buildings,
• roads,
• parking lots,
• residential development or
• other substantial changes in surface characteristics.

What I actually check

When reviewing land use and land cover information, I ask:

1. Does the dataset look like the site today?
2. Has there been significant development since the dataset was produced?
3. Are there large forests, water bodies or urban areas that need to be represented correctly?
4. Does the surface classification make sense around the meteorological station or source?
5. Would I be comfortable explaining the classification to a regulator?

For most projects, the established data-processing procedures are sufficient. Occasionally, however, a map or database is simply out of date.

In those cases, I have found it appropriate to adjust the land-use information to reflect recent urban or industrial development. That doesn't mean the government data are poor.

It means the landscape changed faster than the database.

Urban, rural, forest and water are not interchangeable

One of the reasons land-use classification can become interesting is that regulatory models don't necessarily classify the landscape in the same intuitive way a person would.

Calculating surface roughness

Urban areas can contain large amounts of concrete, asphalt, buildings and other surfaces that alter roughness and heat exchange.

Rural areas generally have lower surface roughness and different thermal characteristics, although "rural" can include a wide variety of actual land covers.

Forests can have substantial aerodynamic roughness and distinctive energy and moisture characteristics.

Water has fundamentally different surface and thermal behaviour from land.

Large water bodies can also create meteorological effects such as lake breezes and other local circulations that may become important to dispersion modelling. This is one reason a simple statement such as "the site is rural" isn't always sufficient.

Alberta: Surface roughness, albedo and Bowen ratio

Alberta's modelling guidance provides recommended surface characteristics for use in AERMOD assessments. The guidance also allows proponents to develop more site-specific or realistic values where appropriate, provided the basis for using different values is explained.

Shaping the air with surface features

One of the important practical considerations is the classification of land within the area surrounding the source.

The Alberta approach includes specific guidance for determining whether the setting should be treated as urban or rural, including the proportion of applicable urban land uses within the area considered. Forested areas also receive particular treatment.

The AERflare modelling system incorporates surface-characteristic information for flare applications. For a project that follows the standard regulatory procedure, these established methods provide a solid starting point.

The experienced modeller's responsibility is to determine whether the standard treatment still represents the site.

British Columbia: detailed surface treatment

British Columbia provides particularly detailed guidance on determining surface characteristics for dispersion modelling. For AERMOD applications, land-use information can be used to determine parameters such as:

• surface roughness,
• albedo and
• Bowen ratio.

Tools such as AERSURFACE can assist with the process. For CALMET/CALPUFF applications, spatially distributed land-use information is incorporated into the meteorological field.

The BC approach also emphasizes the importance of considering:

• seasonal conditions,
• geophysical similarity,
• data quality,
• land-use classification and
• the representativeness of the meteorological data.

See the British Columbia Dispersion Modelling Guideline for the detailed procedures.

Eastern Prairies

Saskatchewan: AERSURFACE and MMIF

Saskatchewan's modelling guidance similarly recognizes the importance of surface characteristics in dispersion modelling. The guidance discusses procedures involving tools such as:

• AERSURFACE for AERMOD and
• MMIF for preparing information for CALMET-related applications.

The assessment of land use includes consideration of:

• urban versus rural conditions,
• surface roughness,
• albedo,
• Bowen ratio and
• anthropogenic heat effects where applicable.

Again, the underlying principle is the same: The model's surface representation should be appropriate for the site being assessed.

Manitoba: Urban or rural?

Seasonal shifts alter pollution spread in Canada

Manitoba also uses land-use classification in determining appropriate dispersion characteristics.

The province's guidance uses the Auer land-use classification system and considers land use within the area surrounding the source when determining whether urban or rural dispersion characteristics are appropriate.

This is a good example of why provincial guidance matters. The general physics of dispersion does not change at the provincial boundary, but the regulatory procedure for representing those physical conditions can.

Calvin Consulting: Getting the Surface Representation Right

Calvin Consulting Group Ltd. uses established regulatory procedures to prepare land-use and surface-characteristic information for air-quality dispersion modelling. Our modelling work incorporates:

• land use and land cover information,
• terrain data,
• meteorological data,
• engineering and facility information,
• building geometry,
• emission sources and
• applicable provincial modelling requirements.

Our normal procedures are designed to produce reliable and defensible model inputs. At the same time, we don't treat a government dataset as automatically correct simply because it came from a government source.

Where recent development has materially changed an area's characteristics, we review the available information and can adjust the land-use representation when necessary.

That is a relatively small step in the modelling process, but it illustrates an important principle: The objective is to represent the site accurately rather than just process the dataset correctly.

With more than 30 years of experience in meteorology and dispersion modelling, Calvin Consulting has worked on industrial projects throughout Western Canada involving AERMOD, CALPUFF, AERMET, CALMET and related modelling systems.

We can help determine whether an existing modelling assessment remains representative, whether the surface data need to be updated, and what level of modelling is actually justified.

Sometimes the answer is a new assessment or an existing model is adequate. Or just one input needs to be corrected.

Good modelling begins with good information—and with somebody willing to look closely at it. Contact Barry at Calvin Consulting Group Ltd. 

...to discuss a new modelling assessment or review an existing one.

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

What an experienced modeller notices

This is where professional judgment matters most.

I don't ask only: Did the software process the land-use data?

I ask: Does this representation of the landscape make physical sense for this site?

A few things can trigger a closer review.

Recent development - Aerial imagery or local knowledge shows substantial urban or industrial development that isn't present in the government dataset.

A mixed landscape - The site is near the transition between agriculture, forest, industrial development and urban areas.

A large water body - A lake or other substantial water surface is close enough that land-water contrasts may influence meteorology.

Complex terrain - Terrain and land cover change together, making the surface representation more important.

Unexpected model behaviour - The resulting meteorology or dispersion pattern behaves differently than expected.

At that point, the question becomes: Is the unusual result caused by the facility, the atmosphere, the terrain, or is it the way the surface has been represented? That's a bit of a dig.

Land use is only one part of the picture

Surface characteristics interact with several other components of an air-quality assessment. The most useful way to explore them is to follow the modelling workflow:

• Start with the surface - Land Use and Land Cover (right here)
• Then look at the terrain - Digital Terrain Elevation Data
• Then the atmosphere - How Numerical Weather Prediction NWP Models Help You Assess Air Quality
• Then the source - Method of Variation of Parameters
• Then buildings - BPIP: How Building Downwash and the Plume Are Connected
• Then receptors - What Is a Receptor in Air Dispersion Modelling?
• Then the chemistry - e.g., How Nitrogen Dioxide Is Formed
• And finally, the result - How to Read an Air Quality Assessment Report

Those pieces fit together because an air-quality model is ultimately trying to answer one question: Given this facility, this landscape and these atmospheric conditions, where could the emissions go and what concentrations could result?

What can go wrong?

Most land-use problems are not dramatic. They are simply discrepancies that quietly make their way into a model.

An outdated dataset - Recent industrial or residential development may not appear in the government database.

An incorrect classification - An important forest, water body or urban area may be assigned the wrong surface category.

Incorrect model coding - Land-cover information may be correct but translated incorrectly into the coding system required by the model.

Wrong spatial scale - A surface feature may be represented at an inappropriate resolution for the assessment.

Seasonal mismatch - Vegetation and other surface characteristics can change considerably between seasons.

Ignoring the surrounding area - The surface immediately around a source may not represent the larger area influencing the meteorology.

These problems do not necessarily cause an obvious model error as the software may run perfectly.

That's why land-use data need scientific review, not simply a successful preprocessing run.

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 practical land-use and land-cover checklist

Before finalizing the surface representation, check:

□ Current land-use information is available

□ Recent development has been considered

□ Major urban areas are identified

□ Forested areas are identified

□ Significant water bodies are identified

□ Appropriate seasonal characteristics are used

□ Surface roughness is reasonable

□ Albedo is reasonable

□ Bowen ratio is reasonable

□ Urban/rural classification is justified

□ The selected data source is documented

□ Data have been converted into the correct model format

□ AERMET/CALMET processing is appropriate

□ The resulting meteorology has been reviewed for reasonableness

For a broader modelling QA checklist, see:

Types of Data Quality Checks That Matter