Methane doesn’t read the label

When coal is mined, methane escapes from the seam. The same methane is unevenly measured and, for metallurgical coal, barely regulated. The exemption rested on two claims – that methane should be counted at the mine (Scope 1), not the mill (Scope 3), and that mine-level data was unavailable. But the situation is changing on both fronts.

Japan is better placed than any other importer to act on that shift. Through the CLEAN initiative, Japanese buyers already ask their LNG suppliers to disclose upstream methane. This shows that an importing country can create transparency through purchase relationships. Coal moves through the same trading houses, and the data can be produced, but the coal side has no operational mechanism to request it.

Currently, no institution holds that function. The Ministry of Economy, Trade and Industry (METI) treats coal as energy security, the Ministry of the Environment (MoE) inventory stops at Japan’s border, as UNFCCC rules require, and steel-sector accounting excludes upstream methane by default. The emissions fall into the gap between mining and steel.

Coal mine methane adds roughly 27% to steel’s 20-year climate impact, yet that cost stays outside coal markets, and outside the plans to decarbonise steel.

The lever for change is accountability. Once coal mine methane is counted in steelmakers’ Scope 3, investors can price it, procurement standards can reference it, and buyers can ask their suppliers for the data.

A coal-side framework would combine three elements Japan already operates for LNG. These include disclosure requests from buyers, standardised methane data from suppliers, and independent aggregation. CLEAN provides the template, and a first coal-methane disclosure cycle modelled on it would establish Japan’s own standard while the international rules are still forming.

Japan already governs methane in gas imports. The Coalition for LNG Emissions Abatement toward Net-zero (CLEAN) has shown that buyer-side disclosure works in LNG, lifting the share of projects that disclose methane data from 35% to 64% in two reporting cycles and producing the first project-level baseline of its kind.

The same traders, steelmakers, and Japan Organisation for Metals and Energy Security (JOGMEC) divisions sit at the centre of a coal supply chain where no equivalent system exists, and where upstream methane is comparable in scale and less well measured.

1.1 CLEAN: a buyer-led disclosure initiative

Governance through purchase terms, not regulation

CLEAN was announced in July 2023 at the LNG Producer-Consumer Conference in Tokyo, co-hosted by METI and the International Energy Agency (IEA), under a joint statement pledging support for methane abatement across the LNG value chain. The founding buyers, JERA and KOGAS, send disclosure questionnaires to their upstream suppliers, while JOGMEC, with METI’s backing, operates the platform that aggregates responses. The premise is that buyers hold the leverage, and the responsibility, to govern methane in their supply chains.

Unlike the European Union (EU)’s producer-focused approach, Japan embeds methane in purchase relationships. Disclosure questionnaires go directly to projects holding active sales and purchase agreements (SPAs) with participating buyers. This makes disclosure a part of the commercial relationship rather than a regulatory requirement.

But CLEAN’s leverage stops short in two ways. It stops at only gas. Further, it attaches no exclusion or pricing penalty. As a result, disclosure never becomes a true condition of market access.

Measurable results within two reporting cycles

By October 2025, CLEAN had expanded to 27 participating LNG buyers across Japan and South Korea, covering roughly 25% of global LNG consumers and 40% of producers, though participation remains voluntary and is short of full coverage. Between 2024 and 2025 reporting cycles, questionnaire response rates by LNG sellers rose from 80% to 95%, and project-level data provision rose from 35% to 64%. It generated the first buyer-driven, project-level methane disclosure baseline in LNG markets.

End-use agnostic, and only as a start

CLEAN does not distinguish LNG by end use, making its buyer-side architecture transferable to sectors where methane remains ungoverned, including coal. The UN Secretary-General’s 2026 methane call cites CLEAN as a notable example of buyer-side LNG governance while naming metallurgical coal methane an overlooked issue in steel decarbonisation. Disclosure alone is insufficient to drive methane performance, but Japan has already built the architecture. Therefore, it can extend that architecture to coal while the rules are still being written or adapt later to standards set elsewhere.

1.2 Awareness without a mechanism inside JOGMEC

JOGMEC’s LNG and Methane Management Division hosts CLEAN and partners with UNEP’s International Methane Emissions Observatory (IMEO). Further, in partnership with Pertamina and JGC, it ran a two-year methane measurement and abatement project at onshore gas production facilities, aiming at OGMP 2.0 Level 5 in Indonesia. Its Coal Development Department finances coal resource development globally and has commissioned coal mine methane research on governance, regulation, and measurement technologies. Yet no buyer-disclosure mechanism comparable to CLEAN has emerged on the coal side.

For example, in its 2025 H1 coal market update, JOGMEC reported the Queensland coking coal mine accidents of March-April 2025, including the Moranbah North methane fire, as a supply and price disruption, with no reference to methane or its governance implications.

Furthermore, JOGMEC’s own coal-side research has described each element of the gap in its own research. Its FY2023 report surveys the GHG Protocol value-chain framework and separately names Australia and Indonesia, Japan’s two largest metallurgical coal suppliers, yet does not apply the framework to those imports. Its FY2024 report introduces IMEO’s Steel Methane Programme as an emerging OGMP 2.0-style framework and observes that under IFRS S2 industry guidance, coal mine methane is Scope 1 for coal miners but not captured under Scope 3 for steelmakers. Therefore, it falls between the two. The same report examines mine gas monitoring across the US, Australia, Poland and Japan as a safety question with occupational thresholds of 1-2%, rather than the parts-per-million resolution OGMP 2.0 Level 4 and 5 required for climate reporting.

JOGMEC appears to have the capacity to act on coal. Where it has been asked to act on methane, it has acted thoroughly, taking a two-year measurement programme to Indonesia. Its coal work shows the same awareness of methane, disclosure frameworks and supply-chain emissions, but reads methane as a price signal or a safety threshold. Nothing published on that side yet treats it as an emission to be governed.

The same limitation also appears in policy. Japan’s Global Methane Pledge submission confines fossil fuel measures to LNG, via CLEAN. Its 11% methane target and 26.7 Mt CO2e FY2030 ceiling are set against the territorial inventory, which counts the residual methane from the Kushiro mine but not the 699.9 kilotonnes (kt) of methane (CH4) embedded in Japan’s 2024 coal import, 57% of which came from coking coal.

1.3 Locked into coal with unverified methane

Import dependence and blast-furnace lock-in

Japan imports 99.7% of its coal. Despite declining thermal coal use, 80.6 million tonnes (Mt) of blast furnace capacity, operated by Nippon Steel, JFE Steel, and Kobe Steel, locks in continued coal dependence. Together, they account for 86% of Japan’s total crude steel production (Nippon Steel 50%, JFE 29%, Kobe 7%, 2023 share). Output fell from 96 Mt in 2021 to 84 Mt in 2024, but this reflects cyclical demand, rather than a structural transition. Even under decarbonisation pathways such as Nippon Steel’s COURSE50, which injects hydrogen into the blast furnace rather than replacing it, blast-furnace steelmaking, and therefore coking coal demand, is likely to persist into the 2030s.

Indonesia’s rising share is measured with low confidence

Japan’s annual coking coal imports declined from 64 Mt in 2021 to 55 Mt in 2025. Across those five years, Australia’s share slipped from 54% to 50%, while Indonesia’s rose from 20% to 26%, making Indonesia the second-largest supplier of the coal Japan classifies as coking, and the one whose methane is hardest to verify.

Ember rates Indonesia’s national estimate as low confidence, because its UNFCCC reporting is outdated and uses Tier 1 methods – global emission factors applied to aggregated national production, rather than measured mines. Indonesia’s own inventory puts national coal mine methane at around 0.2 Mt; GEM and IEA assessments put it 6 and 14 times higher, respectively. Nothing requires anyone to find out which is closer, so switching origin would only shift the uncertainty elsewhere.

The methane released from mining the coal Japan imports is now measurable, and large enough to matter for steel’s emissions, yet it goes uncounted and ungoverned. The gap travels through a label. Finance and reporting frameworks that exclude thermal coal leave coal declared “metallurgical” untouched, yet no two customs or commercial systems draw that line the same way, so the methane stays off the books.

2.1 The scale of the problem

Coal mining emitted around 35 Mt of methane in 2025, comparable to the gas sector. Metallurgical coal alone contributes an estimated 12 Mt CH4 a year, about 990 Mt CO2e on a 20-year basis, adding around 27% to steel’s short-term climate impact once upstream methane is counted. For the gassiest mines, methane alone can double blast-furnace steel’s emission intensity.

Japan’s eleven blast-furnace mills took in 54.5 Mt of coal in 2024, carrying an estimated 308.1 kt CH4 of upstream methane, 4.6 times Japan’s entire reported domestic energy-sector inventory of 67 kt CH4. That is 9.2 Mt CO2e on the 100-year basis used in corporate reporting, or 25.4 Mt CO2e weighted for methane’s near-term warming (GWP20).

Three companies own those mills. Nippon Steel alone received 33.13 Mt at its seven mills: the 24.87 Mt of coking coal it discloses, plus about 8 Mt of PCI and captive-power thermal coal, outside the coking boundary, but inside the mill gate. Ember estimated the imported methane emissions using IEA methane intensities: 181.7 kt CH4 for Nippon Steel, 100.5 kt for JFE Steel and 25.9 kt for Kobe Steel. The gap follows intake more than grade, though JFE’s coal carries more methane per tonne.

Coal’s upstream emissions belong in Scope 3 Categories 1 and 3. Nippon Steel’s 181.7 kt CH4 is 5.4 Mt CO2e on the 100-year basis or 46% of the 11.7 Mt CO2e that the company reports across those two categories. On a 20-year basis, this is 15.0 Mt CO2e. Whether on a GWP-20 or GWP-100 basis, the coal mine methane emissions are substantial to a steelmaker’s Scope 3 emissions.

How much methane arrives depends on where the coal was mined, not just how much of it is mined. Depending on origin, fugitive methane can add 0 to 140% to steel’s short-term climate impact, a spread no purchasing decision currently reflects.

Australia’s reported intensity has fallen 0.1 tCH4/kt as surface mining grew, but satellite evidence suggests open-cut emissions are underestimated. Therefore, it is likely the decline is partly an artefact of measurement rather than a real reduction.

And Australia is among the better-measured origins. Ember scores 59 coal-producing countries on the quality of their reported coal mine methane: five reach good confidence, but none supply Japan. Australia rates medium; Indonesia, low, but the gap in the top band is measurement, which a buyer can ask its suppliers to close.

Data visibility arrived without usability

Coal mine methane is increasingly measurable, but measurement is not governance. Only 10% of global hard coal production has ever had a methane plume attributed to it, largely because of where institutions choose to look.

Satellites (Carbon Mapper’s Tanager-1, IMEO’s MARS, Japan’s GOSAT-GW) are one layer. Ground sensors, aircraft surveys, and operator measurements add to what satellites cannot monitor, particularly at underground mines and in cloud-covered or mountainous terrain.

Japan’s three principal coal suppliers show the full spread:

• Australia: observable but undercounted. 92% of mines sit under conditions suitable for satellite detection, and emissions are reported mine-by-mine under the Safeguard Mechanism. Yet, independent estimates put fugitive methane 40% to 80% above UNFCCC inventories – at Hail Creek, satellite estimates repeatedly exceeded reported emissions by multiples even as reported emissions fell and credits were issued.
• Indonesia: effectively unmeasured. 84% of production sits in hard-to-monitor regions, underground mines are excluded from inventories, and the energy ministry’s 168 kt CH4 for 2024 compares with Ember’s estimate of 722 kt.
• Canada: a counter-case. Only 16% of hard coal mines are classed as favourable for satellite observation, yet plumes have been detected across 31% of production because Carbon Mapper and institutional attention were directed there.

Where coal mine methane is measured tracks where measurement is required, not where the coal is. Poland, though not a major supplier to Japan, shows the same under the EU Methane Regulation, where that attention is required rather than chosen. Japan’s own supply chain spans that range, from the United States, rated medium for methane visibility, to Mozambique, a growing source that has yet to report coal mine methane emissions to the UNFCCC.

The data can be produced. Yet for coal, no channel exists to receive, process, or act on it – the function JOGMEC already performs for gas.

2.2 Measured, but in no one’s remit

Coal mine methane sits outside most steel reporting by default. Among major accounting and disclosure frameworks, only ResponsibleSteel V2.1.1 and worldsteel’s Life Cycle Inventory methodology explicitly require net coal mine methane within the cradle-to-gate emissions of purchased coal.

The others include upstream coal within Scope 3 in principle but impose no methane-specific or mine-level requirement. GHG Protocol Scope 3 and the Steel Science-Based Target-Setting Guidance place purchased-coal emissions in Scope 3 (Category 1 or the mandatory Category 3 fuel-and-energy target) without isolating mine methane. The IFRS S2 industry guidance provides no steel-specific Scope 3 metric. Mine-gate methane is therefore captured, if at all, only through generic default factors.

The policy-level gap reinforces the reporting gap. No Japanese ministry treats methane from imported coal as part of its climate remit. METI treats coal primarily as an economic resource. Its 2025 Strategic Energy Plan values coal for its “relatively low cost per unit of energy” while describing its climate impact only as the “highest greenhouse gas emissions among all fossil fuels.” The MoE inventory stops at Japan’s border, as UNFCCC rules require, while steel disclosure follows global frameworks centred on mill-gate CO2. With negligible domestic mining, agriculture and waste dominate Japan’s territorial methane profile, leaving coal mine methane largely outside the climate conversation.

Even the new GX is blind to imported coal mine methane emissions

Green Transformation (GX) is Japan’s national decarbonisation programme, legislated in the 2023 GX Promotion Act. It works through two pricing instruments:

• GX-ETS. An emissions trading scheme mandatory for large emitters since April 2026, covering participating firms’ direct CO2. Allowances are allocated free until 2033, and there is no absolute cap.
• GX Surcharge. Formerly the Fossil Fuel Surcharge, a levy on fossil fuel importers from FY2028, charged on the fuel’s carbon content.

Both price carbon where it is imported or burned, but neither price the methane released to extract it. GX’s blast-furnace benchmark runs from raw material intake to crude steel, so coal mine methane, released before the coal reaches the mill, falls entirely outside it, as does methane intensity, the quantity the EU Methane Regulation governs.

The ¥20 trillion (roughly USD 130 billion) transition finance programme and over 60% of Green Innovation Fund steel R&D stay weighted toward blast-furnace technologies, CCS and hydrogen injection, rather than electric arc conversion. This is at a point where 61% of Japan’s steel capacity faces reinvestment, conversion or closure by 2035.

IEEFA’s June 2026 assessment finds that blast-furnace pathways do not address upstream methane from extraction. Nor would displacing them. Displacement reduces how much coal flows through the supply chain over time, while methane governance addresses what is already in it.

GX’s price signals are weak even for the carbon they do cover. There is no emissions cap, free allowances until 2033, and prices too low to drive fuel switching, which makes the methane omission more visible as Japan leans harder on carbon pricing. The two regimes that do regulate coal sit at opposite ends of the chain. Australia’s Safeguard Mechanism is at the producer end, and Japan’s GX-ETS at the buyer end, and neither differentiates coal by verified methane performance.

Certification carries the blind spot too. Japan’s GX Steel programme certifies reductions on a mass-balance basis, allocating emission cuts across products rather than measuring their full footprint, an approach Japan’s own Carbon Footprint guidelines question for product-level claims. Its boundary is the steelmaker’s Scope 1 and 2, so a product can qualify as “green steel” while the coal behind it carries upstream methane the label never counts.

Under METI’s Green Steel for GX push, JISF set out product-specific carbon-footprint rules for steel in October 2025, and the building lifecycle-carbon regime starting FY2028 will require cradle-to-gate emission intensities for steel – drawn from that same certification. Both inherit the blind spot on coal mine methane.

2.3 Coal mine methane emissions for steelmakers are hiding behind labels

Coal mine methane goes unaccounted for, whatever the coal is called. Yet the metallurgical label can change how the coal is treated by financial institutions and regulations. Most banks’ coal policies, and the SBTi standard for financial institutions, exclude thermal coal and leave metallurgical coal alone. The EU Methane Regulation caps venting at thermal coal mines from 2027 and defers the equivalent rule for coking coal mines to a delegated act due by August 2027; both apply to mines inside the EU. Imports run on a separate track that, as written, does not distinguish thermal and coking coal: importers must show MRV equivalence from 2027, report methane intensity from 2028, and meet a Commission-set maximum intensity value from 2030.

The report counts every tonne of coal arriving at the mill gate, not the share labelled metallurgical. Metallurgical coal is a commercial and end-use category, not a property of the coal, and not a synonym for coking coal. The grades inside it are real – steelmakers blend hard coking, semi-soft coking and PCI-grade coal to specification – but the boundary is not fixed. PCI counts as a metallurgical grade, yet it is essentially high-quality thermal coal that can be sold into the thermal market. The coal does not change as it crosses from one system to the next, but the category applied to it does, and none of these labels tracks the methane released at the mine.

The clearest evidence is a single trade flow read three ways. In 2024, the coal shipped from Indonesia to Japan was 39% “coking” by Indonesia’s export customs, 52% “for coking” by Japan’s import customs, and only around 2% “metallurgical” by Kpler’s commercial-grade tracking – the broadest of the three labels producing the smallest number. The two customs agencies roughly agreed on the total tonnage, then disagreed on how much of it was coking. The gap is not a single-year artefact – it has run between 2.6 and 4.1 Mt in every year since 2021.

The inconsistency is built into the categories themselves. The Harmonized System distinguishes coal only at the six-digit level, on measurable thresholds. Below that, each country defines “coking coal” categories. Japan has no single criterion. Above 8% ash, the schedule distinguishes “other coking coal” from plain “other” – a distinction only a declaration can make. At or below 8% ash, it offers no thermal line at all, so any low-ash bituminous cargo that is not heavy coking coal is entered under a heading that says nothing about coke. Japan’s statistics then merge the two into Coal for Coking N.E.S. (その他のコークス用炭). These two lines carry more than nine-tenths of everything recorded as “for coking” from Indonesia. Every coal line is free of customs duty, so nothing at the border depends on where the boundary falls.

Both descriptions can be tested against where the coal was unloaded. In 2024, every tonne of hard coking coal landing in Japan, from every origin, Indonesia included, was discharged at a port serving a coke oven. The residual line behaves differently. According to Kpler’s mill-gate records, the three steelmakers received only a small fraction of the 14.9 Mt recorded as “for coking” from Indonesia.

The problem is not that any of these labels is wrong. Each does legitimate work in the system that created it – but none was built to say anything about methane. Refining the definitions cannot change that. Methane is released when the seam is cut, before the coal is washed or sorted into grades at all, and a single mine’s output is divided between coking, PCI and thermal products by the market rather than by the geology. The category can keep doing its commercial work, but climate meaning has to come from the mine, where methane can be measured, reported and verified, rather than inferred from a tariff line.

What rests on “metallurgical”

Urgewald’s October 2025 Still Burning finds that 146 of 318 major global financial institutions restrict coal financing, but only 14 extend those restrictions to metallurgical coal. In Japan, five top commercial banks provided roughly $1.18 billion to metallurgical coal developers through lending and underwriting, between 2022 and 2024, with no methane conditions attached.

SBTi’s Net-Zero Criteria for Financial Institutions require the in-scope coal value chain to cover all thermal coal grades but only recommend metallurgical. Japan’s Petroleum and Coal Tax (石油石炭税), an excise levied on use, exempts coal certified for steelmaking from the ¥1,370/t (roughly USD 9/t) levy that other imported coal pays, and the Japan Iron and Steel Federation (JISF)’s September 2025 submission asked for that exemption to be made permanent.

The exception is the Climate Bonds Initiative, whose 2024 Steel Criteria already require coal suppliers to demonstrate methane intensity below 5 kg CH4 per tonne of coal, with MRV and reporting to the steelmaker at OGMP 2.0 Level 5 “adapted for coal or equivalent practice”. The machinery to govern coking-coal methane exists, but it lacks wider adoption.

The exemption’s foundations are wearing thin

For policymakers and financial institutions, the case for exempting metallurgical coal is weaker than it was. For steelmakers, methane has to be accounted for from all coal consumed, not just the share classified as “metallurgical.”

The metallurgical coal exemption rested on two claims. The first was that methane should be counted at the mine (Scope 1), not the mill (Scope 3), and the second was that mine-level data was unavailable. Both have been overtaken. On the responsibility side, IFRS S2 establishes the reporting responsibility even where it sets no steel-specific metric, and the EU’s Corporate Sustainability Reporting Directive (CSRD) does the same. On the data side, Australia’s Safeguard Mechanism, IMEO’s Steel Methane Programme, and satellite verification.

The shift is visible on the demand side too. In June 2026, investors representing around US$9.85 trillion in assets declared methane in the steel value chain a material climate, regulatory, and operational risk, calling for site-level measurement and abatement by producers and steelmakers.

The relevant boundary is the steelmakers’ procurement, not customs classification: responsibility follows the buyer and its supply chain, as Scope 3 requires. The evidence and investor expectations behind the carve-out have changed. The politics has not – and neither has the gap a buyer-side mechanism could close.

Japan’s steelmakers and trading houses buy a large enough share of the seaborne coal trade that their purchasing terms move the market – and they already use those terms to get methane data from LNG suppliers. Extending that to coal needs no new institution or law. It needs a buyer to ask.

3.1 Commercial relationships already exist

CLEAN was built on concentrated, long-term relationships between a small number of buyers and producers. Japan’s coal supply chain has the same architecture, long-term contracts, repeated counterparties, and upstream equity stakes, and in several cases the same participants.

Mitsubishi holds a 50% stake in BHP Mitsubishi Alliance (BMA), operator of five Bowen Basin coal mines. Mitsui, Itochu, and Marubeni, all LNG buyers within CLEAN, between them hold equity positions and long-term offtake agreements across Australian and Indonesian coal.

Japan’s integrated steelmakers are increasingly embedded. Nippon Steel’s recent investments – 20% in Canada’s Elk Valley Resources (2024) and 20% of Queensland’s Blackwater coal mine (2025) – sit on top of minority stakes across a string of Australian coking coal mines held since the 1990s, lifting its coal self-sufficiency to 35%. JFE added 10% of the Blackwater coal mine in 2025, on top of long-held coking coal stakes in Byerwen and Sonoma coal mines. Kobe, by contrast, has no publicly disclosed equity in any coking coal mine.

That reach extends to the least-measured part of the supply. Badan Pusat Statistik Indonesia (BPS) data show around 42% of Indonesia’s reported coking coal went to Japan over 2021-2025, so a Japanese disclosure request would travel to most of the stream nobody currently measures.

Shared actors do not, by themselves, establish that CLEAN transfers. Coal contracts, counterparties, disclosure rights and confidentiality arrangements differ from LNG, and a coal disclosure cycle would have to work within those differences. What the overlap does establish is narrower but still material – the commercial relationships through which a disclosure request would travel already exist.

An import methane performance standard has already been modelled for Japan, though only for gas and oil – coal was left out. Operational risk is already evident. The 2024 Grosvenor and 2025 Moranbah methane incidents disrupted supply and investment, contributing to a 7.2% year-on-year fall in exports from Queensland’s four main coal ports in H1 2025, and leading Peabody Energy to abandon a USD 3.8 billion acquisition of Anglo American’s Queensland assets.

3.2 Three things a first cycle would ask for

CLEAN’s core insight is transferable: methane can be governed through procurement. A coal analogue needs no new institutions, only the same logic – a convening body requesting contract-linked disclosure, backed by independent verification. The governance layer that surfaces and prices methane can be built today, even while the abatement stack remains uneven.

JOGMEC already operates each component through CLEAN. The same trading houses hold coal equity stakes, the same satellite infrastructure detects emissions, and the same buyer concentration exists in coal markets. What has not yet happened is the first request.

• Coordinated disclosure requests. A convening body would run a disclosure cycle that requests mine-level methane data from suppliers through buyers’ existing commercial relationships (long-term offtake agreements and equity partnerships). As under CLEAN, questionnaires would be issued and linked to active sales and purchase agreements.
• Standardised mine-level data inputs. Suppliers provide mine-level methane data from national reporting systems, satellite detection, and emerging frameworks (IMEO’s Steel Methane Programme), with country-level intensities as interim input where mine-level data are not yet collected.
• Independent aggregation and verification. A coordinating body aggregates and standardises the data for comparability across suppliers and alignment with international monitoring, reporting and verification (MRV) practice.

The mechanism does not require a new policy architecture. Methane can be added to instruments Japan already operates – methane targets in voluntary industry commitments, eligibility conditions on investment incentives, thresholds in green public procurement – and, where GX’s boundary is widened to reach upstream emissions, MRV requirements in the allowance system itself.

3.3 What changes with data availability

Mine-level methane intensity (tCH4/kt coal) is the starting point. Underground mines, over half of coal mine methane, concentrate emissions in ventilation shafts and drainage systems that can be metered; surface mines are where monitoring is currently weaker. Because low reported intensity may simply reflect poor measurement, the question is not which mines emit the least, but which improve MRV, reduce emissions, or should be prioritised for closure.

At the import level, intensity becomes decision-useful. Weighting Japan’s coal imports by mine-level intensity would yield supplier-specific Scope 3 exposure, directly relevant to GX compliance, procurement, and climate accounting. The relevant unit is total coal imports, as methane is a property of the mine. Across Japan’s full 2024 import of 165.8 Mt, the country- and grade-based estimate comes to roughly 700 kt CH4, of which this report quantifies the 308 kt attributable to the three integrated steelmakers. What such a signal measures should stay constant regardless of what it is later used for – continuous, verified emissions rather than a binary pass/fail label.

Buyer concentration is what makes this tractable. Japan and South Korea together import 83.6 Mt of coal classified as coking, roughly 28% of the 303 Mt global seaborne metallurgical coal trade. Adding the EU’s import-methane rules, phasing in from 2027 would bring the covered share of seaborne metallurgical coal to roughly 41% – and Climate Analytics models a wider four-importer coalition cutting 4.4 Mt CH4 a year, though Japan does not need the coalition to exist before it acts.

An import-weighted methane signal changes three things visible inside existing decision frameworks.

• Methane intensity becomes part of mill-gate Scope 3 reporting, where it currently disappears at the mine gate.
• The metallurgical coal exemption weakens further, resting as it does on a label no two systems apply the same way.
• GX’s carbon pricing gains the upstream signal it currently lacks: the same project-level data CLEAN already collects for LNG.

IEA estimates about half of coal mine methane can already be cut with existing technology; the mechanism’s job is to make that visible. Disclosure attaches a questionnaire to contracts buyers already hold, as CLEAN shows for LNG. Verification sits outside it – CLEAN works with OGMP 2.0 under UNEP’s IMEO review, and IMEO’s Steel Methane Programme could offer the same for coal, though no Japanese steelmaker has joined. Abatement costs sit at the mine and can be shared through buyer-producer negotiation.

3.4 Importance of the next eleven months

The premise that coal methane sits outside active governance is expiring. The EU Methane Regulation, in force since 2024, will set coking coal venting requirements by 5 August 2027. In 2025, coking coal mines accounted for 90 of 109 satellite-detected super-emitter events in Poland while making up only 25% of national production; and discussion is underway on pricing methane in imported fuels, potentially through the emissions trading system or an extension of CBAM.

China, the world’s largest coal producer and steelmaker at around 63% of global metallurgical coal capacity, has begun governing coal mine methane at home, through venting limits, utilisation targets, and carbon-credit eligibility. Australia, the anchor of Japan’s coking coal supply, exports the overwhelming majority of its black coal production, making methane governance primarily a question for importing countries. Without equivalent action, Japan risks becoming the destination for higher-methane coal displaced from Europe.

The window is also closing from the inside, fixing coking coal demand into the 2040s. ACCR’s 2025 analysis identifies six relining decisions for Nippon Steel, JFE and Kobe falling in 2026-2030, and three more in 2031-2035, each committing a furnace to a further 20-year campaign. Some calls are already in: Nippon Steel has shut the No. 3 blast furnace at Kashima, and JFE will replace a Kurashiki blast furnace with a 2 Mtpa electric arc furnace by 2028, but the rest – including Nippon Steel’s Kimitsu and Kobe’s remaining Kakogawa furnace – will land on either side of the EU’s 2027 decision.

As frameworks consolidate elsewhere, Japan’s position shifts from rule-setter to rule-taker: to govern methane through these cycles, the mechanism must be operating before the steelmakers sign, putting a first coal-disclosure cycle inside the eleven months between now and the August 2027 EU decision.

Coal mine methane is now measurable and increasingly comparable across jurisdictions. Japan already asks its gas suppliers to report it, but nobody has asked its coal suppliers.

Japan is uniquely positioned to build that mechanism. Through CLEAN, it built disclosure, standardised aggregation, and coordinated supplier engagement for LNG, demonstrating that an importing country can use buyer relationships to create upstream methane transparency. Coal should not remain outside that logic.

Ember’s recommendations track the UN Secretary-General’s latest Methane Call, which urges governments to treat coal mine methane as a “fix-it-first” opportunity by mandating mine-level measurement and public reporting, time-bound mitigation plans for high-emitting sites, and enforceable post-closure requirements. On the buyer side, it proposes a common methane intensity benchmark, against which importing countries could require fossil fuels to fall below a threshold and prioritise low-intensity suppliers. Its framing names metallurgical coal as the overlooked case, but the recommendation is written for coal production as a whole.

For Japan, Ember recommends:

• Buyers. Steelmakers and trading houses should account for the methane in the coal they buy and part-own: disclose it, fold it into Scope 3 reporting and climate targets, and set out how they will cut it. Those with internal carbon prices should extend them to coal mine methane, a sizable emission they currently carry unpriced – and one that makes each blast-furnace relining decision look cheaper than it is.
• Government. A first CLEAN-style coal methane disclosure cycle needs a convening body, and Japan already has one in JOGMEC, which coordinates CLEAN’s public-private platform. It should start with the steelmakers and trading houses that already hold offtake and equity ties — Japan would then be setting the standard rather than adopting one later.
• Finance. The metallurgical coal exemption – only 14 of the 146 institutions that restrict coal financing extend those restrictions to it – rests on a label that does not track what the coal is or where it goes. Financial institutions should ground coal finance restrictions in verified measurement rather than coal classification, so a self-declared label can no longer decide what it financed.
• Standards and advocacy. Standards bodies should bring coal mine methane inside responsible steel assessment boundaries. For civil society, the relining decisions now underway commit mills to coking coal into the 2040s, and on that timeline, abating upstream methane and campaigning for phase-out are complementary rather than competing.

The metallurgical coal exemption was built on a world where coal mine methane was invisible, and that world is passing. A category that no two jurisdictions apply consistently is a weak boundary for an exemption to rest on.

This is not to propose a better way to classify coal. Rather, Ember proposes decoupling climate governance from coal classification altogether. Responsibility follows the supply chain; climate obligations follow measured emissions.

Even the industry’s remaining defence, that coking coal cannot yet be replaced, has become an argument for governing its methane rather than exempting it. The question is no longer whether methane will be governed, but whether Japan helps shape those rules or simply adapts to them.

Supporting materials

Downloads

Methodology

Japan methane by source

Values are methane mass, kt CH4. Domestic sectors are from the National Greenhouse Gas Inventory Document of Japan, 2024 (FY2022), back-converted from CO2eq at the AR5 GWP100 of 28 that the inventory itself uses; imported coal and LNG upstream methane are Ember calculations, as embedded import methane falls outside every territorial inventory.

Agriculture (874 kt CH4): NID 2024 categories 3.A Enteric Fermentation (8,661 kt CO2eq) + 3.B Manure Management, CH4 (2,709) + 3.C Rice Cultivation (13,068) + 3.F Field Burning (30) = 24,468 kt CO2eq, divided by 28.

Domestic energy (67 kt CH4): 1.A Fuel Combustion (38.4 kt CH4) + 1.B Fugitive Emissions (29.2 kt, of which 18.2 kt from coal mining, post-mining, and abandoned mines) = 67.5.

Waste (124 kt CH4): 5.A Solid Waste Disposal (1,637 kt CO2eq) + 5.B Biological Treatment (86 kt CO2eq) + 5.C Incineration and Open Burning (9 kt CO2eq) + 5.D Wastewater Treatment (1,733 kt CO2eq) = 3,465 kt CO2eq, divided by 28. Industrial processes (1.5 kt) and LULUCF (2.9 kt) are excluded.

Imported coal (700 kt CH4): Japan’s entire 2024 coal import (“Attribution to Japan’s total coal import“), of which the three integrated steelmakers’ mill-gate intake is 308 kt (“Attribution to three steelmakers“) and other buyers 392 kt.

Imported LNG (314 kt CH4): Japan’s 2024 LNG imports by origin (World Integrated Trade Solution, HS 271111; 65.89 Mt) converted at the IPCC default net calorific value for natural gas, 48.0 TJ/Gg (1.15 toe/t at 41.868 GJ/toe, IEA) and weighted by upstream gas methane intensities from the IEA Global Methane Tracker 2024 (2023 estimates) as converted to Gg CH4/Mtoe by Rystad Energy for Clean Air Task Force (2024). For each supplier:

Methane (kt CH4) = LNG volume (Mt) × 1.15 (toe/t LNG) × Intensity (Gg CH4/Mtoe)

Intensities applied: Australia 2.0, Malaysia 4.8, United States 5.3, Russia 6.1, Papua New Guinea 7.5, Indonesia 7.0, Brunei 6.8, Oman 3.6, Qatar 3.2, other origins 6.0; weighted average 4.15. The NCV’s 95% range (46.5–50.4 TJ/Gg) implies 304–329 kt.

Blast-furnace mill coverage

The analysis covers all operating integrated blast-furnace steelworks in Japan owned by the three major producers.

Coal import allocation

Kpler cargo records are used for mill-level attribution, while Japanese Ministry of Finance statistics are used for national import totals. The two datasets differ by approximately 3-4% in 2024 but do not materially affect the attribution. Discharge ports were matched to coke-oven sites using Kpler destination records.

Grade splits are Kpler commercial classifications. Company disclosures use different reporting boundaries and are therefore not directly comparable: (a) coking coal, FY2024, Nippon Steel Sustainability Report; (b) total coal with no grade split, FY2024, JFE ESG Data Book; (c) combined coking coal and coke, FY2024, Kobe ESG Data Book. All broadly corroborate the aggregate coal intake estimated here, but none discloses the grade split on which the methane attribution depends.

Customs classification and source data

Coking coal is identified differently in each national customs schedule. Indonesia identifies it in a single line (2701.12.10); Japan splits it by grade into Heavy Coking (2701.12-011, -091) and Coal for Coking N.E.S. (2701.12-019, -092); South Korea into Hard and Other Coking. Figures are drawn from official customs statistics: Indonesia (BPS-Statistics), Japan (Ministry of Finance) and South Korea (Korea Customs Service).

Country methane intensity calculation

Country-specific methane intensities (t CH4/kt coal) were derived by combining IEA Global Methane Tracker 2024 estimates of coal mine methane emissions by producing country, which cover operating mines only, with World Mining Data 2026 coal production statistics for the corresponding coal categories. For each country and grade:

Methane Intensity = Coal Mine Methane Emissions (t CH4) / Coal Production (kt)

Separate intensities were calculated for metallurgical (coking) and thermal (steam) coal where emissions and production data permitted.

Country intensities are conservative estimates in two respects: they exclude abandoned mine methane entirely, and they are likely understated for Australia and Indonesia, where independent evidence puts fugitive methane above reported inventories. The steelmakers’ total is not sensitive to the Indonesian figure: substituting Indonesia’s own national inventory for the IEA intensity lowers the 308 kt by under 10%, because most Indonesian coal arriving in Japan goes to power stations rather than blast furnaces.

Default intensity assumptions

Where country- or grade-specific data were unavailable, these defaults were applied:

Country intensities used in this report:

Attribution to Japan’s total coal import

This national estimate (700 kt) and the steelmaker estimate (308 kt) are not nested. The former allocates all customs tonnage by customs category; the latter classifies mill-gate receipts by commercial grade. The 398 kt of customs-classified coking coal and the 308 kt at the mill gate therefore cannot be reconciled line by line.

Each source country’s 2024 customs tonnage (Japan Ministry of Finance) is split into coking coal (Heavy Coking, 2701.12-011 and -091, plus Coal for Coking N.E.S., 2701.12-019 and -092) and thermal and other non-coking coal (anthracite, 2701.11, plus Coal N.E.S. 一般炭, 2701.12-099 and 2701.19) and multiplied by the matching grade intensity from the Country Methane Intensities table.

CH4 = ∑ (Import Tonnage_country,grade × Methane Intensity_country,grade)

Because customs data classify every cargo, no unclassified residual arises. Methane is converted to CO2e at AR6 GWP20 = 82.5 (default, reflecting near-term warming) and AR6 GWP100 = 29.8 (where compared with corporate reporting):

CO2e = (CH4 × GWP)

Japan’s 165.8 Mt of 2024 coal imports embed an estimated 699.9 kt CH4 (57.7 Mt CO2e, GWP20). Coking coal is about 35% of the tonnage but 57% of the methane (397.8 kt), as coking seams are gassier per tonne; thermal and other non-coking coal carry 302.2 kt. Australia, Indonesia, and the United States account for around 90%.

Attribution to three steelmakers

Company-level methane is attributed on the same grade-intensity basis as the national estimate above, applying Kpler cargo records in place of customs tonnage. Each seaborne cargo in the 2024 Kpler dataset was assigned to its destination steelworks, aggregated by source country and coal grade (metallurgical, thermal or unclassified, following Kpler Product Type), then rolled up to the company level. Domestic transhipments and vessels with unidentified consignees were excluded.

Unclassified cargoes – 0.58 Mt across the three companies, about 1% of intake – are assigned the metallurgical intensity of their origin. The SBTi Financial Institutions Net-Zero Standard defaults unknown coal to thermal because its purpose is thermal coal exclusion; here the purpose is embedded methane, and metallurgical is the appropriate default: it is the grade steelmakers buy, coking seams are gassier, and it stops a cargo being metallurgical for an exemption but thermal for its methane. Results:

Implied intensity (final column) reflects each company’s import origin mix, not mine-level performance: Nippon Steel’s is lowered by Canadian coal (2.69 tCH4/kt), JFE’s raised by Chinese coal (10.5 tCH4/kt).

The 46% figure sets Nippon Steel’s estimated coal mine methane against the Scope 3 it reports in Categories 1 and 3, which do not include it. Ember estimates 181.7 kt CH4 across the full 33.13 Mt mill-gate intake, not only the 24.87 Mt disclosed as coking coal, since methane is released in mining every tonne delivered, whatever grade it is recorded under. At AR6 GWP100 (29.8), that is 5.41 Mt CO2e, 46% of the 11.70 Mt reported; at GWP20 (82.5), it is 14.99 Mt CO2e, which would raise the reported total by a factor of 2.3.

Whether the steelmakers’ own disclosures already capture any of this cannot be verified: Nippon Steel and Kobe Steel report CO2 rather than CO2e, and JFE’s CO2e footnote does not indicate that upstream coal mine methane is included. This report treats inclusion as unverifiable rather than assuming exclusion. Cargo classifications are a proxy for grade, not a record of mill consumption or captive power use; firmer figures would need disclosure of procurement volumes by source and grade.

Indonesian classification reconciliation

Indonesia’s coking coal exports (HS 2701.12.10) are modest to begin with, totalling 119.19 Mt over 2021 to 2025, with Japan the largest single destination in every year:

The same Indonesia-to-Japan coal flow is classified differently at each end of the trade, while Japan remains a stark outlier among importers of Indonesian coal. Indonesia records comparatively little of the flow as coking on export; Japan records far more on import, almost all in a residual category defined by declared end-use:

Between 2021 and 2025, Japan classified 67.73 Mt from Indonesia as “for coking”, against 49.82 Mt that Indonesia exported to Japan under its own coking line, a cumulative gap of 17.91 Mt. Roughly 90% of Japan’s “for coking” total from Indonesia sat in the N.E.S. residual each year. Because N.E.S. is defined by declared end-use, not measured coal quality, it is the structural source of the 13.62 Mt grey zone identified in section 2.3.

Import-standard coverage of seaborne metallurgical coal

Japan (58.67 Mt, Ministry of Finance, “Coal of Coking”) and South Korea (24.96 Mt, Korea Customs Service, hard plus other coking coal) imported 83.6 Mt of coking coal in 2024.

Europe (39.5 Mt, Argus definition, wider than the EU-27) and the global seaborne total (303.0 Mt, 2024 actuals) are from Argus, Seaborne Coal Outlook, Issue 25-6, 13 June 2025. Japan and South Korea are therefore 27.6% of the seaborne metallurgical coal market, rising to 40.6% once Europe is included. The denominator is seaborne trade only, excluding overland flows.

Yen-dollar conversion

Yen figures are converted at the 2025 average of ¥149.5 per US dollar (Federal Reserve H.10); the ¥20 trillion GX programme is approximately USD 134 billion, rounded to USD 130 billion in the text.

Acknowledgements

Contributors

Ember: Tito Das, Dody Setiawan, Bex Horner, Adomas Liepa, Ardhi Arsala Rahmani, Taiki Asato, Sabina Assan, Nishant Bhardwaj, Sougol Aghdasi, Chelsea Bruce-Lockhart.

Reviewers

We thank our external reviewers, Sue Lyn Stubbs (Fidelity International), Kenta Kubokawa, Sala Tsuzuki (Transition Asia), Jasmine Fowler-Morrow (ACCR), Lia Wagner (Urgewald), Roger Smith, Anand Pathanjali (SteelWatch), Hongyou Lu, Bo Shen (UC Berkeley).

Cover image

A yellow heavy-duty mining truck stands on ground level inside a quarry. In the background, tiered layers of rock strata rise up toward a partly cloudy sky.

Credit: Gansham Ramchandani / Pexels