Canada’s Air Pollutant Emissions Inventory Report 2025: chapter 3

The Air Pollutant Emissions Inventory (APEI) is a comprehensive and detailed inventory of air pollutant emissions in Canada, developed using two types of information:

• facility-reported data, consisting of emissions from relatively large industrial, commercial and institutional facilities
• in-house estimates, including diffuse sources and other sources that are too numerous to be accounted for individually, such as road and non-road vehicles, agricultural activities, construction, and solvent use

The APEI is developed using many sources of information, procedures and emission estimation models. Emissions data reported by individual facilities to Environment and Climate Change Canada’s (ECCC) National Pollutant Release Inventory (NPRI)^{Footnote 1}  are supplemented with documented, science-based estimation tools to quantify total emissions. Together, these data sources provide a comprehensive overview of pollutant emissions across Canada. A framework has been developed that makes use of the best available data, while ensuring no double counting or omissions. This chapter presents information about the inventory development process.

3.1 Overview of Inventory Development

The process of developing comprehensive emission estimates for the APEI is presented in Figure 3–1. It consists of categorizing facility-reported data (section 3.2), calculating in-house estimates (section 3.3), and reconciling the facility-reported data and the in-house estimates in a database, where necessary (section 3.4), followed by compiling and reporting the results (section 3.5). Quality control (section 3.6) is performed throughout the inventory development process. Every year, the whole time series (from 1990 to the latest year) is estimated and continuous improvement often results in revisions to previously published estimates, called recalculations (section 3.7).

Facility-Reported Emissions

As a first step, 17 pollutants reported in the APEI are extracted from the NPRI verified database, which contains facility-reported data. New facilities are identified in the extracted data and classified within the APEI according to the nature of their activities. This step results in a compiled database containing most of the facility-reported emissions used in the air pollutant emissions inventory report.

More information on facility-reported emissions is presented in section 3.2.

In-House Emission Estimates

In-house estimates are based on documented estimation methodologies which are periodically reviewed and updated through literature searches, the collection and analysis of recent emission factors and activity data, and comparisons with alternative sources of information. Updated estimates are calculated using new or updated activity data. Where possible, inventory estimates calculated in-house use the most rigorous (highest-tier) methods. However, owing to practical limitations, the exhaustive development of all emissions categories is not possible. In these cases, estimates are generally calculated using activity data and emission factors following relatively basic (lower-tier) methodologies. Calculations are performed in spreadsheets (Excel), relational databases (MS Access and SQL server), using computational scripts (R and Python), and may include spatial data quantified using geographic information systems software (GIS-ArcGIS and QGIS).

More information on in-house estimates can be found in section 3.3.

Reconciliation

The next step in the compilation process is eliminating any double counting of emissions between the in-house estimates and the facility-reported data through a process of reconciliation. Table 3–1 illustrates the origin of the emissions for each sector and subsector: facility-reported data, in-house calculated data or a combination of both, for the latest available year. The origin of the emissions can change depending on the year. Reconciliation of in-house estimates with facility-reported data is required for sectors or subsectors where both in-house and facility-reported estimates exist. For 2023, reconciliation was performed for 31 sectors.

More information on reconciliation is available in section 3.4.

Compilation and Reporting

The final steps in the development process involve compiling all reconciled data within a final database and generating the results. The final database houses all APEI data and is the source of data for all APEI-related products, including:

• Canada’s Air Pollutant Emissions Inventory Report ^{Footnote 2} 
• open data emissions tables ^{Footnote 3} 
• online Search Tool ^{Footnote 4} 
• emissions maps of certain air pollutants for selected categories
• input to other products, such as Canada's greenhouse gas and air pollutant emissions projections,^{Footnote 5}  air quality modeling, Canadian Environmental Sustainability Indicators, ^{Footnote 6}  and reports under the Canada-U.S. Air Quality Agreement ^{Footnote 7} 
• Canada’s submission to the United Nations Economic Commission for Europe (UNECE) under the Convention on Long-range Transboundary Air Pollution (Annex 4)

More information on compilation and reporting is available in section 3.5.

Figure 3–1: Overview of the Annual Air Pollutant Emissions Inventory Compilation Process

Table 3–1: Origin of 2023 Air Pollutant Emission Estimates by Inventory Category

Notes:

1. Based on the most recent facility-reported data from NPRI.
2. Estimated by ECCC
3. Facility-reported data consists of facilities located in Atlantic Canada. For other provinces, it consists of in-house estimates.
4. Facility-reported data consists of facilities located in Atlantic Canada and SO₂ emissions from Alberta's natural gas processing facilities.
5. In-house estimates for Wood Products were estimated by the Forestry Products group of the Environmental Stewardship Branch at ECCC. All other in-house estimates were estimated by PIRD.
6. Emissions reported under Other (Miscellaneous) are from breakage, transport and recycling of mercury-containing products using the Hg in Products methodology. Products include: automotive mercury switches, batteries, dental amalgams, fluorescent lamps, fungicides, measurement and control devices, non-fluorescent lamps, switches and relays, thermometers, thermostats and tire balancers.
7. Hg in Products estimates for Residential Waste Burning are not estimated after 2008 as a result of the updates for the Hg in Products models.
8. Emissions from Hg-containing products were calculated as a separate inventory. Emissions are reported under many sectors such as Iron and Steel Industry, Municipal Incineration, Human, Other (Miscellaneous) and Landfills. All in-house estimates for Hg in Products emissions continue to be estimated and reported under these sectors.

3.2 Facility-Reported Emissions Data

Facility-reported emissions data generally refer to any stationary sources that emit pollutants through stacks or other equipment at specific locations. The major source of facility-reported data is the NPRI, Canada’s legislated, publicly accessible inventory of pollutant releases (to air, water and land), disposals and transfers for recycling. The NPRI has provided facility-reported data on all the 17 pollutants included in the APEI for industrial and commercial facilities since 2002. For some pollutants, the data collection began earlier. It started as early as 1993 for the three heavy metals (Pd, Cd and Hg), in 1995 for ammonia and in 2000 for polycyclic aromatic hydrocarbons (PAHs), dioxins and furans and hexachlorobenzene (HCB). Prior to 2002, facility-level emissions for the criteria air contaminants (CACs) were collected and compiled by provincial, territorial and regional environmental authorities across Canada and provided to ECCC for inclusion in the APEI.

Facility-reported data from the NPRI are used in the APEI without modification, except when 1) data quality issues are detected and not addressed during the quality control exercise, or 2) adjustments to particulate matter (PM) emissions are necessary to respect their size fraction. The NPRI reporting requirements and thresholds vary by pollutant and, in some cases, by industry. Details on these reporting requirements and thresholds are available on the NPRI ECCC’s website in the substance list by threshold section.^{Footnote 8} 

A distinction has been made between reporting facilities and non-reporting facilities. Reporting facilities meet the threshold required to report to the NPRI, while non-reporting facilities do not meet the threshold owing to their size or emission levels. Some facilities may be required to report emissions of certain pollutants only. Therefore, emissions from the non-reporting facilities or of non-reported pollutants must be estimated in-house to ensure complete coverage.

Historically (e.g., for the years 1985, 1990, 1995 and 2000), facility-reported data were primarily provided by provinces and territories. In some cases, additional information was calculated to fill in intervening years or to update the original submissions. Trends for the intervening years were interpolated. The compilation of emissions for 2001 to 2005 occurred during a transition to the use of emissions data reported to the NPRI as the major source of industrial emissions. In general, facility-reported data from the NPRI and data provided by provinces and territories were used for years 2002, 2004 and 2005, and interpolation was used for 2001 and 2003.

Since 2005, information on facility-reported data has originated mainly from the NPRI, with limited data obtained from some provincial governments on selected sources that are not reported to the NPRI.

The NPRI groups substances into five parts, as listed below. Each part has its own reporting thresholds or triggers for mandatory reporting.

• Part 1A – Core substances, and Part 1B – Alternate threshold substances
• Part 2 – PAHs
• Part 3 – Dioxins, furans and HCB
• Part 4 – CACs
• Part 5 – Speciated volatile organic compounds (VOCs)

Table 3–2 shows the 17 air pollutants reported in the APEI and their NPRI reporting thresholds. No VOC data collected under Part 5 are used in the APEI.

Table 3–2: National Pollutant Release Inventory Air Pollutant Reporting Thresholds

Notes:

MPO = Manufactured, processed or otherwise used

NA = Not applicable

In 2023, 6682 facilities reported releases to air of one or more APEI pollutants to the NPRI. Since 1993, more than 21 000 facilities have reported at least one of the seventeen APEI pollutants. Over the years, many facilities have fallen below the emissions reporting threshold or have ended operations and no longer report to the NPRI program. There may also be times, for example, for oil and gas facilities, that facilities have changed ownership. The new owner would not necessarily reuse the same number used to identify the facility in the NPRI reporting system by the previous owner. The end result would look like the opening of a new facility and the closing and an old facility. Therefore, the total true number of facilities over time could be somewhat less than 20 000. Using the 2023 NPRI database, with data available as of October 10, 2024, facility information and air emissions data for pollutants listed in Table 3–2 were extracted for each province and territory. The quality control process described in section 3.6 was applied to the NPRI data to identify outliers or missing substance reports.

For facilities reporting to the NPRI for the first time, the North American Industry Classification System (NAICS) codes (Statistics Canada, 2017), reported by the facilities, are used to assign preliminary APEI sector and subsector classifications. Additional research and verification on facilities' operations are then performed to confirm or correct the classification into the appropriate APEI sector or subsector. The assigned classification is used for subsequent reporting years, as long as the facility does not change operations.

NPRI reporting facilities may not report all three PM size fractions: TPM, PM₁₀ and PM_2.5. For cases where only one or two of the three PM size fractions are reported to the NPRI, a distribution procedure is applied to estimate a complete set of PM emissions for facilities. The procedure is based on sector-specific PM distribution profiles developed using 2006–2016 facility-reported PM emissions data for most sectors, 2002–2017 facility-reported emissions data or detailed studies for other sectors, or derived from NPRI toolbox guidance (e.g., unpaved roads). Where ratios were calculated using facility-reported data, the ratio for each facility is calculated and then averaged by sector. The resulting PM distribution ratios are available online on the Government of Canada Open Data Portal.^{Footnote 9} 

The TPM, PM₁₀ and PM_2.5 emissions calculated using the distribution procedure are added to the list of facility-reported data and flagged as an ECCC estimate within the compiled APEI final database.

3.3 In-House Emission Estimates

The reporting of substances by facilities to the NPRI remains the primary source of industrial air pollution data in Canada. Sectors with significant sources of facility-reported data (e.g., petroleum refineries, smelters) are well represented by emissions data from the NPRI.

The completeness of the APEI is assessed by the level of inclusion of all known quantifiable sources of pollutant emissions in the provincial, territorial and national totals that are attributed to anthropogenic activities. Where NPRI facility-reported data do not provide for complete sector coverage, additional estimates are developed in-house by ECCC. An overall estimation of completeness in this case is related to the availability and reliability of the activity data and methodologies used for the in-house estimates.

The development of complementary in-house estimates is not required in sectors where NPRI facility data provide complete coverage of air pollutant emissions (e.g., pulp and paper). To produce a complete inventory of emissions, complementary in-house estimates are necessary for subsectors that have limited coverage in the NPRI because many facilities do not meet the reporting thresholds (e.g., Natural Gas Production and Processing, Light/Medium Crude Oil Production, Sawmills, Ferrous Foundries, etc.).

Other sources of air pollutants, such as Residential Fuel Combustion, Transportation and Mobile Equipment or Fires, are not subject to reporting to the NPRI, and coverage is assured solely through the calculation of in-house emission estimates for these sources.

Although all major sources of air pollutant emissions are included in the APEI, several sources are not, such as the burning of agricultural wastes and demolition activities in the construction industry.

In-house estimates are calculated using information such as production data and activity data, using various estimation methodologies, emission models and emission factors.^{Footnote 10}  Depending on the source, there are three methodological tiers that represent varying levels of complexity: Tier 1 is the simplest; Tier 2, the intermediate; and Tier 3, the most demanding in terms of complexity and data requirements. Tier 2 and 3 methods are referred to as higher tier methods and are considered more accurate. Tier 1 methods typically apply a simple linear relation between activity data and emission factors. The default Tier 1 emission factors are chosen such that they represent typical process conditions, and they tend to be technology independent. UNECE provides Tier 1 methods for all sources and substances that countries that have ratified the protocols of the Convention on Long-range Transboundary Air Pollution are required to report. Tier 2 methods use the same or similar activity data as Tier 1 methods, but apply country-specific emission factors, which need to be developed using country-specific information. Tier 3 methods go beyond the previous two methods and may include using facility-level data, specific information on the types of technologies being used at facilities, pollution abatement equipment, and/or sophisticated models. It is a good practice to use higher tier methods for categories that are large contributors to total emissions.

Calculations of in-house estimates are based on the latest data available at the time of inventory development. When possible, the data are updated each year. These emission estimates are calculated at the provincial, territorial and national level. Table 3–1 illustrates the complete list of sectors and subsectors of the APEI for which emissions are based on in-house estimates and provides the activity data year on which the 2023 in-house estimate is based.

A summary of in-house estimation methodologies is presented in Annex 2. For more detailed information on methodologies, please refer to Canada’s Air Pollutant Emissions Inventory Methodology Document on the Government of Canada Open Data Portal.^{Footnote 11}

3.4 Reconciliation

In several sectors, as noted in Table 3-1, estimation of total emissions involves both estimates provided by facilities and estimates developed in-house by ECCC. To prevent double counting of emissions and to confirm that the APEI includes all emissions, a comparison and reconciliation of emission estimates from various sources is performed for each pollutant, industry sector and geographical region, as appropriate.

3.4.1 General Procedures

The approach for reconciling facility-reported data and in-house estimates for a specific pollutant is as follows:

• Case 1: If the total of the in-house estimates is greater than or equal to the total facility-reported data, the reconciled in-house estimate is equal to the total of the in-house estimates minus the total of the facility-report data, as outlined in Equation 3–4.

Equation 3–4

• Case 2: If the total in-house estimate quantity is less than or equal to the total of the facility-reported data for the source, the reconciled in-house estimate is equal to 0, as outlined in Equation 3–5; i.e., facility-reported data are considered to reflect all the sector emitting sources.

Equation 3–5

In addition, there are sectors where a variation of the general reconciliation approach is used. Sections 3.4.2, 3.4.3 and 3.4.4 provide information on reconciliation approaches that are unique in nature.

3.4.2 Wood Products

Particulate matter emissions (TPM, PM₁₀ and PM_2.5) from the Sawmills and Panel Board Mills subsectors (Wood Products sector) were not reconciled using the procedure described in section 3.4.1. Rather, NPRI facility-reported data from these subsectors were used to characterize the entire industry. The facility-reported data, together with several production indicators, were used to estimate the PM emissions from facilities that are not required to report to the NPRI. The sum of the resulting emission estimates represents the total emissions for these subsectors. All other pollutants were reconciled at the subsector and provincial and territorial levels, according to the standard procedure and equations outlined in section 3.4.1.

3.4.3 Dry Cleaning, General Solvent Use, Printing and Surface Coatings

The in-house estimates in the Dry Cleaning, General Solvent Use, Printing, and Surface Coatings sectors (Paints and Solvents source category) include a total of 92 different kinds of solvents and applications. The in-house emissions estimates for these categories are reconciled with facility-reported data, which include a variety of sources (solvent use as well as processes, fuel combustion, road dust, etc.) grouped under the same NAICS. Given the complexity of the solvent sectors, reconciliation of in-house estimates with facility-reported data from the NPRI requires that the following steps be performed (Cheminfo Services, 2020):

1. allocation of the solvent use in-house estimates to the 4-digit NAICS level from the NPRI
2. allocation of the NPRI VOC inventory totals at the 4-digit NAICS level to “Process” and “Solvent” type emissions
3. subtraction of the “Solvent” type NPRI emissions from the solvent in-house emission estimates

If subtraction of the facility-reported data from the in-house estimates for a certain solvent use yields a small negative value, the emission estimate for that in-house estimate is set to zero.

3.4.4 Mercury in Products

Mercury (Hg) can be released to air throughout the life-cycle of mercury-containing products, including during manufacture, distribution, use, disposal, transportation and final disposition, as well as through waste streams. Releases can also result from breakage and processing. Reconciliation of Hg air emissions from mercury in products with NPRI involves a review and characterization of the source of the Hg air emissions included in the facility-reported estimate. This is to ensure that the Hg emissions estimated through the life-cycle approach are not duplicated in the facility-reported data. In situations where overlap exists, either the in-house estimate emissions from mercury in products are removed from reporting in the APEI or a proportion method is applied. If there is no overlap, the facility-reported and in-house estimate emissions are simply added together. The proportion method changes the mercury in product emissions, while the point-source emissions remain unchanged (Equation 3–6 and Equation 3–7):

Equation 3–6

Equation 3–7

This is done at the provincial and territorial level by year.

3.5 Compilation and Reporting

The time interval between the receipt of emissions data from industries and submission of the emissions and report to UNECE is relatively short. Tools used to compile emissions, populate the UNECE Nomenclature for Reporting (NFR) tables, perform quality control tests and generate the different tables and figures for this report are automated as much as possible to allow quick compilation, ensure efficient corrections and reduce the possibility of errors.

In addition, as each file prepared by an expert is submitted for compilation, it must first pass all the control tests of a verification tool before being integrated into the compiled database. More than 25 tests are carried out. This step enables errors to be detected as early as possible in the compilation process. More details on the tests performed are given in section 3.6.3.

3.5.1 Gridded Maps of Air Pollutant Emissions

While the APEI provides valuable information on emissions within Canada, it does not distinguish localized sources of emissions within the provincial and territorial level aggregations. In this respect, work is ongoing to attribute emissions closer to where they originate geographically. Models have been used to spatially distribute the air pollutant emissions. As the project moves forward, the models will be refined, and distribution of emissions will become more precise.

For this edition of the inventory, 14 static maps have been produced. For the following sources of air pollutant emissions, the most significant pollutants have been selected and mapped on a 1 km grid (by ecodistrict for Agriculture):

• Ore and Mineral Industries
• Oil and Gas Industry
• Transportation and Mobile Equipment
• Agriculture
• Commercial/Residential/Institutional

The maps produced represent the total emissions over the year as opposed to an atmospheric concentration at a certain point in time. The maps are published on the Government of Canada Open Data Portal. Please contact ec.dirp.donnees-data.pird.ec@ec.gc.ca if additional details regarding the maps are needed.

3.6 Data Quality Control

Quality control for the inventory takes place at each step of the process, in three main phases. In phase 1, quality control is performed on the most recently submitted NPRI facility-reported data, prior to inclusion of the data in the estimates. Phase 2 of the quality control occurs on the in-house estimates at a subsector level, while phase 3 is performed on the final database of reconciled and compiled emissions, including NFR tables. See Figure 3–2 for a visual representation of the different quality control check points.

Figure 3–2 : Quality Control Check Points

3.6.1 Phase 1: Facility-Reported Emissions Data

The quality control process involves a system of documented activities and procedures performed to identify data outliers, inconsistencies, missing data, inaccuracies and errors. This phase is divided into two parts.

First, high-level completeness tests are completed on NPRI data before sharing facility data with sectoral experts. This step involves comparison with the previous year’s dataset and identify any significant changes. High-level checks on the number of facilities reporting, number of records included in the database, number of new facilities and total emissions by pollutants by year are performed to ensure sufficient completeness before proceeding to more detailed analysis and quality control (refer to 3.6.1.a in Figure 3–2).

Once the initial checks are satisfied, the facility-reported dataset is prepared and shared with sectoral experts for more specific and in-depth quality control. The quality control process is adapted where necessary such that category-specific or sector-specific quality control procedures are applied, as appropriate (refer to 3.6.1.b in Figure 3–2). A key part of the quality control process is identifying missing NPRI facility reporters and assessing new reporters to ensure that the correct data are captured and allocated to the appropriate sectors and subsectors.

Identifying outliers (i.e., reports that significantly depart from comparable NPRI facility-reported data) is critically important to ensuring the usability of the NPRI facility-reported data.

Potential outliers are defined as any NPRI facility report that:

• has a large year-over-year change; and/or
• contributes an unrealistically high proportion of the total reported quantity of an air pollutant in the current or previous reporting year.

In addition to identifying missing NPRI facilities and outliers, the quality control review includes analysis of:

• the impact of first-year reporting;
• substances that are no longer reported, inconsistently reported, or that have never been reported by a facility that conducts activities that are expected to be emissive;
• substance reports with identical reported quantities of an air pollutant within a five-year period;
• substance reports with significant variation over a five-year period; and
• facilities assigned to incorrect subsectors.

Finally, quality control checks are also performed on facility information. These checks include facility identification numbers and geographical information (i.e., city, province and territory, address and latitude and longitude).

Once the review of the facility data is complete, facilities are contacted to resolve identified issues with significant impacts. Identification, facility follow-up and resolution of such issues are conducted at the earliest stage of the quality control review. Where unresolved issues persist, any updates to the data will be reflected in the next inventory edition (refer to 3.6.1.c in Figure 3–2).

3.6.2 Phase 2: In-House Emission Estimates

The objective of Phase 2 of the quality control process is to identify and verify inconsistencies in the APEI at the subsector level. A series of verifications and quality control checks are undertaken on the in-house emission estimates to ensure quality, accuracy and consistency. The following are verified:

• activity data
• emission factors
• unit conversions
• emission calculations

Activity data (refer to 3.6.2.a in Figure 3–2) and emission estimates are reviewed by multiple sector experts to identify outliers, similar to the review of facility reported data. Potential outliers are defined as sector-level activity data and emissions that:

• have large year-over-year changes;
• have changed significantly since the previous reporting year.

Emission estimates (refer to 3.6.2.b in Figure 3–2) are further compared against other metrics for the sectors, such as: heating degree days, electricity generation, population, or gross domestic product. These comparisons are used to confirm general trends identified. Additional information is gathered from industry associations or news releases related to temporary shutdowns, plant closures, and retooling of facilities which is used to confirm trends.

Best-available emission factors are chosen by sector experts to reflect Canadian conditions for the various sectors. For example, emission factors for residential firewood burning are currently taken from the U.S. EPA as their technology is also used in Canada.

Prior to implementation, in-house models are rigorously tested to ensure activity data and emission factors are correctly applied, unit conversions are consistent throughout and resulting emission estimates are in the appropriate sector (refer to 3.6.2.c in Figure 3–2).

3.6.3 Phase 3: Compiled Air Pollutant Emissions Inventory

Phase 3 includes all tests performed immediately prior to compiling the estimates as well as the analysis of the results and different products once they have been compiled in a final database. Before integrating the emission estimates from all sources, automated quality control tests are done on each individual part. The tests performed include checking for duplicates, ensuring that all sources are considered, and all mandatory fields are filled according to the standards, and verifying units. The purpose of the tests is to ensure the quality of the compiled data (refer to 3.6.3.a in Figure 3–2).

Once all estimates have been compiled, trend analysis and recalculation graphics are produced to analyze the consistency of the estimates. Data visualization tools, such as Microsoft Power BI, are also used to perform trend and recalculation analysis and to identify any abnormal gaps. Data are analyzed at different levels. They are analyzed by pollutant, by source, sector or subsector. Gaps can be identified either from their impact on the overall contribution to the national trend or from their impact on the category itself. Trends are also analyzed by province and territory. Any significant changes from year to year and any recalculated emissions are identified and explained (refer to 3.6.3.b in Figure 3–2).

Quality control is also performed on all other APEI products, including the data tables presented in this report as well as data published online (refer to 3.6.3.c in Figure 3–2). At this stage, the quality control tests mainly consist of verifying that totals (for different layers of disaggregation, different years and different pollutants) match the compiled estimates. The various end products are also compared against each other as an additional quality control step.

As a last step, quality control tests are made on the NFR tables (refer to 3.6.3.d in Figure 3–2). Some tests are automated and are run on the compiled tables that will be submitted to the UNECE. They include verifying totals for each pollutant and each year and comparing those values with what is reported in this report. A completeness test is also run to make sure every cell has a value, either a numerical value or a notation key. Other quality control checks are also made by sectoral experts and are, in some cases, cross-referenced with the sector-level estimates. For more information on the NFR tables, refer to Annex 4.

3.7 Recalculations

Emissions recalculation is an essential practice in the maintenance of an up-to-date air pollutant emissions inventory. The APEI is continuously updated with improved estimation methodologies, statistics and more recent and appropriate emission factors. As new information and data become available, previous estimates are updated and recalculated from the base year (1990) to ensure a consistent and comparable trend in emissions. Recalculations of previously reported emission estimates are common for in-house estimates and can also occur with facility-reported emissions data. More information on recalculations is provided in Annex 3.

References

Cheminfo Services. 2020. Update Solvents Volatile Organic Compound (VOC) Emissions from 2015 to 2018, Final report. Unpublished report. Markham (ON): Cheminfo. Prepared for Environment and Climate Change Canada.

Statistics Canada. 2017. North American Industry Classification System (NAICS) Canada 2017 Version 3.0. Ottawa (ON): Statistics Canada.