Inzinion/cbam-sector-facility-registry

CBAM-Sector Global Facility Registry

Open screening registry of cement, iron & steel and aluminium facilities worldwide in three CBAM Annex I good categories, with modelled CO2 (Climate TRACE) and regulator-reported CO2 (EU ETS EUTL / US EPA GHGRP) kept in separate columns, each reported figure carrying its match evidence.

Canonical record: doi.org/10.5281/zenodo.22172573 · Publisher: Inzonex

Load

from datasets import load_dataset

ds = load_dataset("Inzinion/cbam-sector-facility-registry", split="train")
print(ds)

Or straight into pandas:

import pandas as pd

df = pd.read_csv("https://huggingface.co/datasets/Inzinion/cbam-sector-facility-registry/resolve/main/cbam_sector_facilities.csv")

3,421 facilities in three CBAM Annex I good categories — cement (2,160), iron & steel (953), aluminium (308) — across 145 countries · 3,102 outside the EU-27 · CC BY 4.0

An open screening layer for research around the EU Carbon Border Adjustment Mechanism, whose definitive phase started on 1 January 2026. Each facility carries: name, country/ISO3/continent, coordinates, owner (95% coverage), capacity where known, modelled annual CO₂ (Climate TRACE estimate, 4,318.8 Mt total across the registry), and — for 360 matched facilities — regulator-reported CO₂ (EU ETS EUTL or US EPA GHGRP) with the full match evidence of that join: confidence class (292 high / 68 medium), match distance in km, name-similarity score and the source pipeline's high-quality flag, so every reported figure can be audited. An eu27 flag separates domestic EU producers from the non-EU producers CBAM actually concerns.

Files

File	Contents
data/cbam_sector_facilities.csv	3,421 facilities × 22 columns
data/cbam_sector_facilities.json	Same records, JSON
data/cbam_country_category_summary.csv	Country × category aggregates (counts, estimated and reported CO₂ totals)
data_dictionary.csv	Field-level documentation

What this is — and is not

This registry answers "who are the producers in CBAM good categories, where, owned by whom, and what is the best available public figure for their CO₂". It does not answer "what will an importer owe":

1. Facility totals are not embedded emissions. CBAM liability is calculated per tonne of a specific CN-code good (specific embedded emissions), not from a site's total annual CO₂. This registry has no production volumes or CN-code splits.
2. Climate TRACE values are modelled estimates, not measured or verified figures — Climate TRACE itself labels them emissions estimates. Where a regulatory figure exists it is given separately in co2_tonnes_reported, never merged.
3. The EU operates its own CBAM registry of declarants and third-country installations (data exchanged between operators and importers). It is not publicly downloadable; this open layer is a complement for researchers and analysts, not a substitute for compliance data.
4. Scope: 3 of 6 CBAM categories. Fertilisers, hydrogen and electricity are also CBAM goods but are not separable in the underlying facility data and are excluded here (lime is likewise excluded — CN 2522 is not an Annex I good). Coverage inherits Climate TRACE's facility detection; it is not a census of every plant.

Research context

Recent work joins facility-level emissions estimates with proprietary trade data to study CBAM effects (e.g. Nature Climate Change, 2026, on early CBAM effects in EU–India steel trade) — but proprietary trade layers are closed-license, so the underlying joined data cannot be shared. This dataset publishes the open half of that equation. The natural extension — production volumes and confirmed EU trade links, starting with iron & steel in a single exporting country using the Global Energy Monitor Iron & Steel Tracker — is planned as a follow-up.

Sources and attribution

Climate TRACE (facility detection and CO₂ estimates) · EU ETS Union Registry / EUTL (reported CO₂, EU) · US EPA GHGRP (reported CO₂, US) · Regulation (EU) 2023/956, Annex I (category scope). Derived and redistributed under the sources' open licenses; see datapackage.json.

Versions

v1.1.0 (2026-08-30): negative numeric values (longitudes, latitudes) are now written as plain numbers — v1.0.0 escaped them with a leading apostrophe as an over-broad spreadsheet-formula guard, which broke numeric parsing; three match-evidence columns added (match_distance_km, name_similarity, match_high_quality) so reported-CO₂ joins are auditable; croissant.json upgraded to conformant MLCommons Croissant 1.0.

Cite

See CITATION.cff. Published by Inzonex — engineering data for high-temperature plant equipment.

Citation

Canonical DOI (always resolves to the newest version):

Aheiev, D. (Inzonex) (2026). CBAM-Sector Global Facility Registry [Data set]. Zenodo. https://doi.org/10.5281/zenodo.22172573

@dataset{inzonex_cbam_sector_facility_registry,
  author    = {Aheiev, Dmytro},
  title     = {CBAM-Sector Global Facility Registry},
  year      = {2026},
  publisher = {Zenodo},
  doi       = {10.5281/zenodo.22172573},
  url       = {https://doi.org/10.5281/zenodo.22172573}
}