Seeking pilot sites, engineering partners and regulatory collaborators across Europe.

Research · methodology & data

How the figures are madeModels · scope · conversions · corrections

Every number on this site traces to evidence — this page is the trace. It states what the models are, what the dataset does and does not cover, how units convert, what is excluded, and how to dispute any of it.

Dataset · JSON · v1.0 Six-mine table · CSV Interactive study

Status: venture research artefact · not peer-reviewed · corrections invited · reviewed 2026-08-08

Scope and definitions

What the study inventory does and does not cover
Geographic scope	European Union member states.
Mine-type scope	Underground hard-coal mines are modelled; surface mines are counted (17) but not modelled.
Closure-date scope	Operations ceased 2015–2023, as recorded in the GEM Global Coal Mine Tracker abandoned-mine methane dataset, June 2024 release. Earlier closures are outside the study inventory — which is why national inventories, covering older and different populations, report different totals.
Closed vs abandoned	Regulation (EU) 2024/1787 Arts. 2(53)–(54): a mine is “abandoned” when no responsible operator, owner or licensee can be identified, or closure was unregulated; otherwise it is “closed”. Responsibility differs accordingly (Art. 25(7)).
Flooded-status unknowns	34 of 53 underground mines have no public flooded-or-dry record. Aggregates are therefore presented as a range across bounding scenarios, never as a single point.
Asset-level vs national accounting	This study models individual assets; national GHG inventories aggregate by different methods and populations. The two disagree today — flagged as Conflicting wherever cited.

The decline model

Per-mine emissions follow the adapted Kholod et al. (2020) M2CM decline family, anchored to each mine’s closure year and GEM’s modelled closure-time emission rate.

Dry (non-flooded) path — hyperbolic decline:
q(t) = q₀ · (1 + b·D·t)^(−1/b) with b = 2.017, D = 0.302 yr⁻¹

Flooded path — exponential decline:
q(t) = q₀ · e^(−0.672·t) — effectively zero within about a decade of flooding

Dry-path structural envelope: EPA Appendix B Table B5 — nine venting-decline curves of form (1 + A·t)^(−E) (three permeability classes × three mine sizes); at each year the band spans the fastest and slowest of the nine.

“Dry path” means the non-flooded mine scenario. It does not refer to dry-gas volume conditions.
Seal quality is a separate unknown from flooding: a surveyed seal changes which fraction of generated gas reaches the atmosphere, not how much is generated.
Flooding is not a binary in reality — partial flooding, changing water levels, compartments above water and leakage pathways all occur. The model’s dry/flooded toggle is a simplified bounding scenario, not a prediction of actual mine condition.

Units and conversions

MCM = million cubic metres of methane per year (volume).
Mass conversion: 1 MCM CH₄ ≈ 0.67 kt CH₄ (density 0.67 kg/m³, IPCC 2006 convention, 0 °C and 1 atm).
CO₂-equivalence: fossil-origin CH₄ at GWP₂₀ = 82.5 (±25.8) and GWP₁₀₀ = 29.8 (±11), IPCC AR6 WG1 Table 7.15. Both lenses are always shown together — neither is “the” number.

Gross, addressable and expected

Every figure in the mitigation ladder is gross modelled source emissions. Technically addressable emissions (reachable with existing equipment at real sites), economically addressable emissions (viable under plausible project economics), and expected destruction (after capture efficiency and uptime) are each strictly smaller numbers that this study does not yet quantify. Treating the global rung as an addressable market would be a category error.

Known exclusions

Surface-mine emissions (counted, not modelled).
Mines abandoned before 2015 (outside the GEM window).
Site-specific gas composition, pressure behaviour and migration pathways — the model has no instrument data to carry them.
Reopened, merged or renamed assets are carried as GEM records them; no independent reconciliation was performed.

Versions and corrections

Version history
Version	Date	Change
1.0.0	2026-08-02	First public release: dataset, methodology page, corrected evidence taxonomy (Reported and Reference labels added).

To dispute a figure, a label, or a modelling choice: report a correction. Corrections are logged in the version table above; a figure shown to be wrong is changed, not defended. Review status: this is a venture research artefact — it has not been peer-reviewed, and technical criticism is actively invited.