reference-data-pack-plain-guide-v2.md, sha256 f6115f8aa77c). Portfolio edition: Kith Climate cohort case, fictional client. Small grey tags such as J-081 point to entries in the project record (journal, findings, decisions), so every statement can be traced.The reference data pack in plain English
Version 2, 4 October 2026 (replaces v1; adds the traceability pass, in which each statement that rests on a fact, number or rule carries a tag such as [G-07] pointing to its row in the claims register, and a separate check re-tests every row against the record). This is a short, non-technical guide to the reference data pack. The fuller readme beside it (reference-data-pack-readme-v2.md) has the detail; this one answers four questions: what is it, why does it matter, how do I use it, and what should I not trust blindly?
What it is
Northwood's carbon inventory does not invent its own numbers for things like "how much carbon dioxide does a gallon of diesel produce". It takes those numbers from published sources: the US Environmental Protection Agency (EPA), the Intergovernmental Panel on Climate Change (IPCC), the Federal Highway Administration, the California Air Resources Board, and a few others.
The reference data pack is the shopping list and receipt for those outside numbers. For each source it records:
- what it is and who published it,
- which edition (the "vintage"),
- where it came from and when we downloaded it,
- a fingerprint of the file, so anyone can tell whether it has been changed,
- what the inventory uses it for, and
- how to refresh it.
Why it matters
Northwood asked for a system that keeps working after Kith leaves. If a number on the dashboard looks wrong, someone needs to answer "where did that come from, and is it still current?" without phoning us. The pack answers that for every outside number. It also tells you exactly which files to replace when a new edition comes out.
What is in it
Three files you will use, plus the source files themselves in the _reference folder:
| File | In plain terms |
|---|---|
reference-data-registry-v1.csv | The master list: 14 sources, one per row [G-02] |
reference-values-v1.csv | A list of the 16 outside numbers the calculation actually uses, each pointing to its source [G-03] |
verify-reference-pack-v1.py | A checker that confirms every file is present and unchanged |
The 14 sources fall into seven groups: [G-04]
- Emission factors (EPA): how much carbon each fuel and each unit of electricity produces.
- Electricity regions (EPA): maps that tell us which part of the US power grid each site sits in, because grids differ.
- Refrigerant leakage (EPA): typical yearly leak rates for air-conditioning and cooling equipment, used when a company has no service record.
- Vehicle fuel use (federal highway data): the average fuel a vehicle burns, used only to spot fleet figures that look far too high or too low.
- Carbon prices (California, the EU auction market, the European Central Bank): three scenario prices used to show what emissions could cost.
- Warming strength of gases (IPCC, GHG Protocol): how much more a tonne of methane or a refrigerant warms the planet than a tonne of carbon dioxide.
- Sector background (Kith research): context for the materiality work. It does not feed any figure.
How to use it
Check nothing has changed. Open the reference-data folder, and run:
python3 verify-reference-pack-v1.py
You should see 14 lines saying ok. If any line says CHANGED or MISSING, stop and find out why before trusting the numbers. [G-05]
Refresh it each year. Roughly once a year, or before each report:
- Look for a newer edition of each source (the readme says where).
- Download the new file next to the old one. Never replace the old one.
- Add or update its row in
reference-sources-v1.csv. - Rebuild the registry (the re-run note gives the exact command), which stamps fresh fingerprints and dates.
- For carbon prices, also update the three price numbers in the settings file.
The re-run note walks through the whole update, including these steps.
What we checked, and what we did not
Checked on 4 October 2026 against the original source: [G-01]
- The warming strength of the old refrigerant R-22 (1,760) matches the IPCC table and the GHG Protocol list. [G-06]
- The warming strengths for the four refrigerant gases behind the two common blends match, and the blends follow from them.
- All 11 leak-rate rows match EPA's table. [G-07]
- Every emission factor matched the EPA workbook when the calculation was built. [G-13]
Known gaps. Please do not paper over these:
- Electricity region maps. We did not record when or from which exact web address they were downloaded. The registry says "not recorded" and leaves it that way. [G-12]
- Methane. The inventory uses one warming strength for all methane (28). The GHG Protocol gives a higher figure (30) for methane from fossil fuels. Methane is a small part of the total, so the effect is small. [G-08, G-09]
- Newer gas values. The GHG Protocol recommends its newest set of gas values (AR6). We stay on the previous set (AR5) to match the EPA, and we add a newer-set check for any company where refrigerant leakage passes 5 percent of its emissions. That did not happen for any company. [G-10]
- Carbon prices are scenarios, not forecasts. And the EU price is a benchmark for how a European buyer might think; no Northwood company pays it.
- The leak-rate PDF's file name says 2020-12, but the document itself is dated December 2023. We kept the old name so its fingerprint stays valid. [G-11]
Where to go next
- For the full detail on each source:
reference-data-pack-readme-v2.md. - For how the numbers are used in calculations: the methodology note.
- For how to repeat the whole process: the re-run note.