EDATSLATE

Data Sources

Where every number comes from

Slate doesn't make up emission factors. It imports them, unmodified, from six public publishers, and every calculated figure keeps a reference back to the exact release and row it came from. This section describes each one: what it covers, which releases Slate holds, how they're licensed, and where to download the originals.

6

publishers

40

factor releases imported

13

IPCC method parameter sets

6

GWP sets, AR2 to AR6

The publishers

United KingdomUK DEFRA / DESNZFuels, UK electricity, heat and steam, travel, freight, materials and waste2 releasesOGL v3.0United StatesUS EPAFuels, US grid electricity by subregion, steam, travel, freight and waste12 releasesPublic domainUnited StatesEPA WARMWaste management scenarios: 61 materials across disposal and recovery pathways1 releasePublic domainInternationalIPCCCombustion defaults, fugitive coal, oil and gas, industrial processes, agriculture, land use and waste methods14 releases© IPCC, attributedCanadaCanada ECCCGrid electricity consumption intensity for 13 provinces and territories4 releasesOGL – CanadaAfrica (55 countries)EmberGrid electricity emissions intensity for every African country7 releasesCC BY 4.0ReferenceGWP setsThe global warming potentials Slate uses to turn methane and nitrous oxide into CO₂e.

Who covers what

No single publisher covers everything, and they don't overlap neatly. The table shows which publisher can answer which kind of activity. Where a cell is empty, that publisher simply doesn't publish it.

You're reportingDEFRAEPAIPCCECCCEmberWARM
Fuel burned on site●●●———
Vehicle fuel (litres, kg)●●●———
Vehicle distancenot yet●————
Grid electricityUKUSby design, noneCanadaAfrica—
Market-based residual mixUK (DESNZ)—————
Purchased heat and steam●●————
Upstream (well-to-tank)●——inside figureinside figure—
Business travel and freight●●————
Waste you send away●●————
Waste scenarios—————●
Industrial processes——●———
Livestock, soils and land——●———
A landfill you operate——●———
Refrigerant leaks need no publisher: they are leaked mass × the gas's GWP.

How a source is chosen

Each organisation lists the publishers it wants in order of preference. For every record, Slate asks the first one. If that publisher doesn't cover this kind of activity at all, Slate moves to the next. If it should cover it but something is missing, the record stops with an error instead of quietly borrowing another publisher's figure.

  1. 01

    Ask the first source

    e.g. UK DEFRA for a UK site's diesel.

  2. 02

    Not published there?

    A documented boundary, like DEFRA and overseas electricity: try the next source, and label the row as a fallback.

  3. 03

    Should be there, but isn't?

    A data gap: stop and show the reason on the record, never a silent substitute.

Why this matters

A figure that came from a different publisher than the one you chose is exactly what's hard to spot in a report. Slate marks every such row as a fallback. The full reasoning is in Factor Resolution.

Publishing calendar

PublisherWhen it publishesWhat Slate holds now
UK DEFRA / DESNZAnnually, each June. The 2027 set is expected June 2027.UK DEFRA 2026, revised v1.2 file
US EPAHub annually each January, with eGRID alongside. Check for a later eGRID revision before every import.Hub 2025 + eGRID2023 rev 2
EPA WARMIrregular. v16 was published December 2023.WARM v16
IPCCIrregular. The 2019 Refinement is the latest revision; it changes some chapters and leaves others untouched.2006 Guidelines (2019 Refinement imported, not active)
Canada ECCCAnnually, after each April National Inventory Report. Version 4.0 was published 9 September 2026.v4.0, Tables 5.1–5.4 (2023–2027)
EmberUpdated monthly. Yearly figures firm up the following year.Snapshot of 15 Sep 2026 (data to 2025)

How the data is kept honest

Originals, unmodified

Each publisher's file is stored as downloaded, with its SHA-256 checksum. A correction has to come from the publisher as a new file.

Blank is never zero

An empty cell stays empty. A zero factor would remove real emissions while looking like an answer.

Checked against the source twice

Reconciliation tests read the publisher's file by a separate route, or a second hand transcription, and compare.

Import is not activation

A new release is loaded, compared with the last one, and only then switched on, as a deliberate step.

Years stay reproducible

The first calculation in a reporting year pins the release it used, so a published year never moves when a new release arrives.

Inconsistencies kept as published

Where a publisher contradicts itself, Slate stores what was printed and says so, instead of quietly correcting it.