The model is only as good as the assumptions and the data that are used to create it. This section lists known issues, limitations and suggested improvements. Would you like to contribute work or funding to improve the model? Please contact us.

Disclaimer: The CIRAIG and other contributors to the OpenIO-Canada model and tool cannot be held responsible for damages or losses resulting from the use of the OpenIO-Canada model, the OpenIO-Canada tool or any of its associated content including spreadsheets, code, manuals and websites.

(If you feel the list is incomplete, please do not hesitate to contact us)

Validation

Issue: The model has been built by LCA researchers with little or no background in economics. Validation by experts in the field would increase the robustness of the model.

Proposed improvement: The approach, assumptions and calculations can be audited and validated by experts.

Missing elements in the original Input and Output tables

Issue: For confidentiality reasons, Statistics Canada wilfully excluded some elements from the Supply and Use tables. Some of these elements were estimated by CIRAIG in version 1.0 of OpenIO-Canada. Missing elements ultimately lead to an underestimation of impacts and potential errors in contribution analyses.

Proposed improvements: These estimates made by CIRAIG can be validated or improved. Other missing elements can be filled based on expert estimates.

Imports, exports, investments, etc.

Issue: Imports, exports and any other issue not directly covered in the Supply and Use table were ignored in the model.

Proposed improvements: A method to deal with these issues in the model can be proposed and/or implemented.

(If you feel the list is incomplete, please do not hesitate to contact us)

Granularity

Issue: OpenIO-Canada uses economic tables (Supply, Use) from Statistics Canada that are at the "L61" level, consisting of 112 industries, 246 commodities and 143 final demand categories. More granular tables exist: the Detailed level tables include 234 industries and 470 commodities. However, these more granular data have more missing data, necessary for confidentiality reasons.

Proposed improvement: If you have the expertise to use the public Detailed tables and to make assumptions to fill in the blanks, please consider contributing new tables to the project.

Limited to the industry-technology assumption

Issue: Different “technology assumption” can be used to deal with coproducts, i.e. when commodities are produced by multiple industries. See methodology report for more information. For now, only the industry technology assumption, which is akin to using allocation in LCA, is implemented.

Proposed improvements: Commodity-technology assumption (akin to system expansion in LCA) and hybrid approaches (that mix the two assumptions) could be implemented, making OpenIO-Canada relevant for more types of applications and allowing sensitivity analyses.

Rectangular matrices

Issue: The Supply and Use tables used in the model are rectangular (more commodities than industries). In order to implement other technology models (e.g. the commodity technology assumption), the tables would have to be converted to square tables (number of commodities = number of industries).

Proposed improvement: The tables can manually be made square by manually disaggregating industries (112 x 122 model), which would also entail allocating industry emissions to commodities, or by allocating commodities to specific industries (246 x 246 model).

Dimension

Issue: The direct requirement and total requirement matrices in OpenIO-Canada are at the industry-by-industry level.

Proposed improvements: A commodity-by-commodity model can be built.

Adding a waste component

Issue: As cited here, “standard IOA has the weakness that it does not consider the physical flows of waste and the activity of waste management”. This is the case for OpenIO-Canada.

Proposed improvements: Creation and integration of a waste IO component to the model.

Endogenizing final demand

Issue: Final demand was treated as exogenous to the model and was completely ignored, limiting the types of analyses that can be carried out by the model.

Proposed improvements: A method to deal with final demand in the model can be proposed and/or implemented.

No physical flows

Issue: The model is currently strictly composed of monetary flows. Expressing some flows in physical terms would improve the robustness of the model.

Proposed improvements: A method to create a mixed unit hybrid model can be proposed and/or implemented.

Regionalization – Canadian MRIO model

Issue: The model is currently Canadian in scope and does not consider regional specificity.

Proposed improvements: Data and methodology to create a multi-regional IO (MRIO) model that accounts for all provinces and their trade OR that singles out a specific province (e.g. a Québec + Rest of Canada model) can be provided and/or implemented.

Linking with other IO models

Issue: The model is currently Canadian in scope and does not consider trade with other countries.

Proposed improvements: Data and methodology to create a multi-regional IO (MRIO) model that accounts for Canada and other countries (in priority: USA) can be provided and/or implemented.

(If you feel the list is incomplete, please do not hesitate to contact us)

Greenhouse gases coverage

Issues: The data from Statistics Canada on GHG emissions used in this model only covers three gases: CO2, CH4 and N2O. What is more, the emissions data is only made available as an aggregate Carbon dioxide equivalent (CO2e). Only emissions from the following sources are covered: “combustion of fossil fuels; non-combustion uses of fossil fuels; industrial processes; agricultural soils; livestock manure and enteric fermentation”.

Proposed improvements: Split the total GHG into the three GHG (CO2, CH4 and N2O). Include other GHG. Include other sources of emissions.

Incomplete data on emissions due to excluded facilities in NPRI

Issue: Emission data for non-GHG substances comes from the NPRI, a pollutant release inventory managed by Environment Canada. It collects data from Canadian industries on their emissions of over 300 substances or grouped substances. However, not all facilities are required to report to the NPRI. For example, facilities where less than 20 000 employee-hours are worked in a given year are not, under certain conditions, required to report. See here for all requirements. This necessarily leads to an underestimation of emissions.

Proposed improvements: Missing data could be estimated.

VOC emissions

Issue: VOC emission data comes from the NPRI. The sum of speciated VOC emissions was subtracted from total VOC emissions to avoid double counting. This resulted in negative emissions for 4 industries (GS91300-Other municipal government services, BS541D0-Computer systems design and other professional, scientific and technical services, BS31B00-Clothing and leather and allied product manufacturing and BS31110-Animal food manufacturing). The cause for these negative values is unknown. The sum of unspeciated VOC emissions from these sectors was set to 0.

Proposed improvement: We are open to any suggestion.

Total reduced sulphur

Issue: Total reduced sulphur emission data comes from the NPRI. As explained on their website, emissions of Total reduced sulphur (TRS) are actually the sum of six emissions, three of which (hydrogen sulphide [H2S], carbon disulphide [CS2] and carbonyl sulfide [COS]) are also reported separately. In order to avoid double counting, the sum emission of these three substances, expressed in H2S equivalents, was removed from the reported Total reduced sulphur emission. This was done separately for Air, Water and Soil emissions. This resulted in negative TRS emissions for some industries. The cause for these negative values is unknown. The TRS values for these industries was set to 0. This happened in the following cases:

  • Air: BS21220, BS21300, BS22110, BS327A0, BS33100 and BS56200.
  • Water: BS221A0
  • Soil: BS21100, BS221A0

Proposed improvement: We are open to any suggestion.

Particulate matter emissions

Issue: Particulate matter emission data comes from the NPRI. The particulate matter emissions, reported as Total PM, PM10 and PM2.5, are converted to the elementary flow names (and corresponding values) used in ecoinvent and most LCIA methods, i.e. “Particulates, > 10 um” (PM-PM10), “Particulates, > 2.5 um, and < 10um” (PM10-PM2.5) and Particulates, < 2.5 um (PM2.5). This resulted in negative emissions of “Particulates, > 10 um” in 31 cases and negative emissions of “Particulates, > 2.5 um, and < 10um” in one case: these were set to 0. Contrary to TRS and VOC emissions, the cause of these negative calculated emissions is known and is unavoidable with the current data.

Proposed improvement: We are open to any suggestion.

Compounds reported as elements

Issue: In the NPRI, some substances are reported as elements and their compounds/salts. This necessarily overestimates the weight of the actual elements, and hence leads to an overestimation of their impacts once characterized. The substances are: Acrylic acid (and its salts); Aniline (and its salts); Antimony (and its compounds); Arsenic (and its compounds); Cadmium (and its compounds); Chromium (and its compounds); Cobalt (and its compounds); Copper (and its compounds); Hexavalent chromium (and its compounds); Lead (and its compounds); Manganese (and its compounds); Mercury (and its compounds); Nickel (and its compounds); Nonylphenol and its ethoxylates; Selenium (and its compounds); Silver (and its compounds); Zinc (and its compounds).

Proposed improvement: We are open to any suggestion.

Grouped emissions: isomers

Issue: In the NPRI, some substances are reported as “molecules and their isomers”. Isomers will likely not have the same characterization factors, and hence this leads to a misrepresentation of impacts (uncertainty). The molecules are "HCFC-123 and all isomers" and "Xylene (all isomers)".

Proposed improvement: We are open to any suggestion.

Polycyclic aromatic hydrocarbons (PAH)

Issue: The NPRI provides data for specific PAH. Some of these do not have corresponding characterization factors in the Impact2002+ model. For these, a characterization factor for the generic "PAH, polycyclic aromatic hydrocarbons" substance was used. This affects the following emissions:

  • 5-methyl Chrysene
  • Benzo(e)pyrene
  • Benzo(j)fluoranthene
  • Cresol
  • Dibenzo(a,h)pyrene
  • Dibenzo(a,i)pyrene
  • Dibenzo(a,l)pyrene
  • Perylene

Note that this only affects the calculation of life cycle impact scores with the Impact2002+ method and not the IO model itself.

Metals to water compartment

Issue: The NPRI provides data for metal ane metalloid emissions to water. The characterization factors in Impact2002+ method contains characterization factors for these same metals/metalloids in ionic form. These characterization factors are used, which will overestimate the contribution of metal/metalloid emissions to ecotoxicity. This affects the following metals and metalloids:

  • Arsenic
  • Cadmium
  • Copper
  • Nickel

Note that this only affects the calculation of life cycle impact scores with the Impact2002+ method and not the IO model itself.

(If you feel the list is incomplete, please do not hesitate to contact us)

Validation

Issue: Validation of data used and of its treatment by a third party could increase the robustness of the model.

Proposed improvement: The approach, assumptions and calculations can be audited and validated by experts.

Subcompartments for air emissions

Issue: All emissions to air are presently assigned to the “unspecified” subcompartment.

Proposed improvements: (1) Emissions that are based on NPRI are associated with a specific stack height. This could allow assigning part of the emissions to the “non-urban air or from high stacks” (for stacks > 100m). (2) The fraction of releases to air in urban areas (density greater than 400 people per km2) could be broken out of the total in order to use the proper sub-compartment (“urban air close to ground”), using geographic information contained in the “NPRI_Geo” NPRI field.

Uncertainty information for flows based on NPRI data

Issue: Currently, the OpenIO-Canada model does not consider uncertainty.

Proposed improvement: In the eventuality that uncertainty information is added to the model, the uncertainty of emissions could be based on information contained in the “xxxxxx_E” fields of the NPRI “SubsRele” table.

New flows from the NPRI (dust, waste, transfers, tailings)

Issue: Only emissions of pollutants to air, water and soil are presently included in the model.

Proposed improvements: The relevance of adding other types of flows reported in NPRI , such as dust emissions (as PM), waste flows, transfers and tailings, could be evaluated.

Primary energy not included in the model

Issue: Because the energy data on consumption per industry provided in Statistics Canada in their Environmental Accounts is an aggregate figure that apparently includes electricity, some energy would be double-counted (energy required to produce electricity in the Electric power generation, transmission and distribution sector + the electricity consumed by other industries). Primary energy was therefore temporarily left out.

Proposed improvements: Energy should be included, either by finding an alternate source of data (e.g. the Canadian Industrial Energy End-Use Data and Analysis Centre database, CIIEDAC) or by finding more disaggregated data from Statistics Canada.

Water use not included in the model

Issue: Water data is available from Statistics Canada. However, it is not reported using the same industry classification system as that of the OpenIO-Canada model, and has hence been temporarily excluded from the model.

Proposed improvements: Water flows can be included if data were manually reassigned to the industries as described by the IOIC L61 classification used in this model or if an alternate source of data were found.

Resource extraction not included in the model

Issue: Resource extraction is currently not included in the model.

Proposed improvements: A data source (for example from NRCan) could be used to integrate resource extraction in the model.