Methodology

2. Boundaries, gases and period

Organisational boundary

You choose the consolidation approach when a report is generated: operational control, financial control or equity share, as the Corporate Standard defines them. The approach is recorded on the report run and on the frozen base year, so a filed figure always carries the boundary it was built under. Under equity share the share composes down the organisation tree: a 60% holding in a subsidiary that wholly owns a site gives a 60% share of that site, not 100%. Under the two control approaches a site counts at 100% where the control test is met and 0% where it is not, and the report lists the sites excluded by that test. The organisation hierarchy is effective-dated, so a report for a past period consolidates the organisation that existed in that period.

Operational boundary

Scope 1, Scope 2 by both methods, and 6 of the 15 Scope 3 categories. The catalogue of measures you can record is the operational boundary written down: every measure names its scope, its method and the data-quality tier it produces (section 9 lists them). Anything not in the catalogue is not calculated, and the 9 Scope 3 categories we do not model are printed on the report as stated omissions rather than left blank.

Gases

Carbon dioxide, methane, nitrous oxide, hydrofluorocarbons, perfluorocarbons, sulphur hexafluoride and nitrogen trifluoride, as the GHG Protocol’s 2013 amendment requires. Each gas is stored as a mass in kilograms wherever the publisher provides a split, and reported both as mass and as CO₂-equivalent. Biogenic CO₂ from biomass and biogas combustion is separated at calculation time and held outside the scope totals as a memo item; the methane and nitrous oxide from the same combustion stay inside Scope 1. The biogenic memo line is not yet printed on the filed report; that is a gap in the report, not a modelling choice, and it is listed in section 12.

Reporting period and exclusions

A report covers a stated period and consolidates the activity records dated inside it. Before a report can be filed the completeness register checks every site’s declared sources against the records supplied; what is missing is listed, and a requirement that no loaded factor source can meet is marked as our gap, counted separately from anything you did not supply.

4. Factor sources, versions and licences

Every publisher in the registry, with its licence and whether we serve its factors. Almost nobody publishes the licence column. We do, because a factor you cannot legally cite is a factor your auditor cannot accept.

Three entries are not served, and the reasons matter more than the fact. The AIB European residual mix is not licensed to us, so market-based Scope 2 in Europe falls to a disclosed grid fallback rather than quietly using a figure we have no right to serve. EXIOBASE is under CC BY-SA, and until counsel has ruled on whether the ShareAlike terms permit serving derived factors, Category 1 for a European organisation stays on the disclosed United States proxy. The GHG Protocol calculation tools say “free for public use” and nothing about redistribution, which is exactly what serving a factor into your report is; the Protocol remains the method we follow, and its default factors are already covered by DESNZ and EPA under licences that do say so.

Datasets as loaded on 2026-10-17

One row per dataset the parsers load, with the version, what it covers, the validity window stamped on its rows, how the publisher states per-gas values, and the GWP set the publisher used. This table is checked against the parser manifest at each update of this document; the date above is the date of that check.

Deliberately absent

• IEA Emission Factors. Commercial licence whose terms prohibit exposing the values to users.
• ecoinvent. Commercial subscription whose terms prohibit sublicensing the data.
• Climatiq. A paid API. Its public data explorer is used only to cross-check our own values.

Updates

National publishers release annually: DESNZ in June, EPA and eGRID in January, AIB in May, CEA in late autumn, DCCEEW in August, MfE in spring. A new edition is loaded as a new dataset with its own validity window; the previous edition stays in the library and stays pinned to every figure that used it. Loading a dataset never changes a filed report (section 13). The newest edition of an annual series is open-ended, and the temporal rule in section 10 widens the stated uncertainty as it ages rather than refusing to resolve.

Which countries can be calculated today

Six with a national publisher: Australia, France, Germany, New Zealand, United Kingdom, United States. A country opens only when its publisher can supply grid electricity, natural gas, diesel, which is the floor for finishing Scope 1 and 2. India is the instructive exclusion: its national publisher gives grid electricity and no fuels, so an Indian company could finish Scope 2 and never finish Scope 1. Ember adds grid factors for roughly ~190 countries where no national publisher is loaded, and a grid factor alone does not open a country.

6. Global warming potentials

100-year values, selectable between IPCC AR4, AR5 and AR6; AR6 by default. The basis applied is recorded on every figure. The operative source is the GHG Protocol’s own table (version 2.0, August 2024), which is what an assurer checks a corporate inventory against and which resolves the choices the IPCC chapters leave open.

The methane rule

Methane has two published values and the split is not fossil fuel versus biofuel. The fossil value applies only to fugitive methane of fossil origin, such as a gas-system leak, because it includes the CO₂ formed when that methane oxidises. Methane from burning any fuel, fossil or not, uses the non-fossil value, because the CO₂ from that combustion is already counted in the same source’s CO₂ figure and the higher value would count it twice. This is the GHG Protocol’s instruction, quoted in the engine at the one place the choice is made.

Refrigerant blends

Blends are not carried as values. They are priced from their components by mass fraction, so a change of basis re-prices every blend consistently:

• R-410A: HFC-32 50% · HFC-125 50%
• R-404A: HFC-125 44% · HFC-143a 52% · HFC-134a 4%
• R-407C: HFC-32 23% · HFC-125 25% · HFC-134a 52%
• R-507A: HFC-125 50% · HFC-143a 50%

SF₆

Under AR6 the engine applies 24,300, the figure in the GHG Protocol’s table, where IPCC AR6 Table 7.15 prints 25,200. The engine once carried 25,200 and so put one gas on a different basis from the other thirteen; the correction is recorded in the engine and is referred to the external methodology review.

One basis per inventory, and the exception

The GHG Protocol asks for a single assessment report per inventory where possible and for the source to be disclosed. Every figure built from gas masses is on the basis you chose. Figures built from total-only factors (ADEME, CEA, Ember, USEEIO) keep the basis the publisher used, shown in the dataset table, because a total cannot be re-based. The filed report derives its GWP line from the rows: the basis and its source, and, where more than one assessment report is present, which publishers’ totals sit on another basis, with the off-basis share of the total stated beside the totals. Each such figure records its basis in the ledger.

How the table is checked

Two independent tests run against the engine’s table on every change: one re-transcribes all 41 values from the GHG Protocol PDF in a separate pass and asserts each; the other computes four published blend GWPs from the components and their compositions and matches them exactly. The subset printed above is checked a third time by this site’s own test.

8. Scope 2, location-based and market-based

Both methods are calculated for every electricity record, independently, as the Scope 2 Guidance requires where contractual instruments exist. The two halves can come from different publishers and even different per-gas bases; each is converted and priced on its own.

Location-based

kWh × the grid average factor for the site’s region: subregion where the publisher gives one (eGRID subregions, Australian states), otherwise national. The Guidance asks for combustion-only factors and forbids marginal factors; the CEA weighted average is used for India, with the operating, build and combined margins carried for reference only. District heating, purchased steam and purchased cooling use the published factor for that supply. Where no meter exists, a floor-area intensity may be used and is recorded as an estimate.

Ember, and its boundary.Ember’s intensity, used for countries without a national publisher, is computed by Ember with IPCC AR5 Annex III life-cycle generation factors (coal 820, gas 490, solar 48, wind 11 gCO₂e/kWh). It is therefore a life-cycle figure, higher than a combustion-only value by the upstream share of generation. The Guidance says location-based factors should be combustion-only and permits otherwise with disclosure, so each Ember row is filed with a life-cycle boundary and the filed report names the source and the boundary under its emission factor sources. Until 3 October 2026 the rows were labelled combustion-only; the changelog records the correction.

Market-based

The highest tier available, and the tier is recorded on the figure and printed on the report:

• Contractual instrument.A certificate, a power purchase agreement or a supplier-specific factor that you enter and that names the factor it conveys. It is honoured even when its vintage is older than the period, and the vintage is disclosed. We do not yet check the Guidance’s Scope 2 Quality Criteria for an instrument automatically; you attest them when you enter it.
• Residual mix.The published residual mix for the site’s country, no more than one year older than the reporting year, because residual mixes are published in arrears and a stricter window would push every timely European report to the next tier. In Europe this tier is currently empty: the AIB residual mix is not licensed to us (section 4). No residual mix exists for India, the United Kingdom or the United States at national level.
• Grid fallback.The location-based grid average, disclosed as a fallback. As the Guidance requires where no residual mix is available, the report notes that an adjusted factor was not available to account for others’ voluntary purchases and that this may result in double counting.

CEA and AIB publish CO₂ only; where a grid factor excludes methane and nitrous oxide the dataset table says so, as the Guidance and ESRS E1 ask to be disclosed. Energy consumption in kWh is kept alongside the emissions so it can be reported separately.

10. Data quality and uncertainty

Tiers

• primary. Measured by you: a meter reading, a fuel-card litre total, a refrigerant mass balance.
• supplier-reported. The supplier’s own figure, disclosed to us rather than modelled by us.
• activity-based. A measured physical quantity × a published factor for that thing.
• spend-based. Money × an input-output average for a whole industry sector.
• estimated. A proxy where no measurement exists: a leak rate, a floor-area intensity.

Uncertainty

A figure carries a stated uncertainty when the emission factor publishes one, combined in quadrature with any activity-data uncertainty, as the IPCC Guidelines describe. Reporting a tight activity-data uncertainty against a factor of unknown quality would overstate confidence, so a figure with no published factor uncertainty states none, with two exceptions where the engine knows exactly how far it travelled: a factor borrowed from a wider region, and a factor older than the period. Both add a term from the pedigree matrix’s geographical and temporal dimensions:

What is not scored, stated. The pedigree matrix has five dimensions. Only the geographical and temporal ones reach a stored result today. Reliability, completeness and technological correlation would need provenance the import pipeline does not yet capture (a meter identifier, an invoice reference), and writing a score nobody measured would assert a measurement nobody made. The engine says this in its own comments and so does this page.

12. What the filed report prints

A report is a snapshot taken when it is filed; no renderer reads the ledger afterwards, so a re-rendered report cannot disagree with the filing it re-renders. The methodology block on every filed report states: the consolidation approach; the reporting boundary; the base year, its emissions, version, approach and basis, or the statement that none is established; the GWP basis chosen; every emission factor publisher used, with its licence and attribution in the publisher’s required wording; the restatement position; the Scope 2 tier behind each market-based figure; and the stated omissions; and, for any source whose factors carry a boundary other than combustion-only, the source, the boundary and what it includes. GRI 305-1, 305-2 and 305-3 and the ESRS E1-6 datapoints are produced from the same snapshot; every E1 datapoint is either reported or omitted with its reason on the face of the report.

Known gaps in the report, as of this version

• Biogenic CO₂ is held in the ledger but not yet printed as a memo line.
• The percentage of each Scope 3 category calculated with supplier data is derivable from the tiers but not yet printed.
• Scope 2 Quality Criteria for contractual instruments are attested by you, not checked by the product.

Each is in the risk register with an owner and a date, and will be removed from this list by a changelog entry, not silently.

15. How we checked our own numbers, and what review would add

Against the publisher, not against ourselves. Each of these has actually run and each reading is the one it produced. All of them ran on our own hardware against a local database on the dates recorded; none is a production service level.

Also checked

• A local restore drill that brought back 43 forced tables, 59 policies, 394 grants and the deliberate REVOKE UPDATE on filed reports. The managed-database half, and a measured recovery time, are still owed.
• A one-million-row, 100 MB import read in 19 minutes 15 seconds on a developer machine: 950,000 rows imported and 50,000 quarantined by design, with every cap holding and announcing itself.
• Zero open dependency advisories and zero suppressions: the allowlist is empty.
• A route guard that found 34 unreachable API paths the first time it was run by hand, and 7 on its first run inside CI.

External review

No external review of this methodology has been performed. It is budgeted and briefed but not yet commissioned: no practitioner is engaged and no date is set. Until one has reviewed it, we describe the output as ready rather than audited, and we would rather lose a deal on that sentence than win one without it.

What such a review is, precisely: ISO 14064-3 verifies GHG statements, so no software can be certified under it. What a vendor can obtain is an independent limited- or reasonable-assurance statement over a named methodology version and engine version, against named criteria, by an accredited body. This document, the engine’s versioning, the pinned factor library, the resolver record on every figure and the rounding rule are the object such a statement would cover. When one is commissioned, its scope, the body and the date will be published on the trust page before any result is.