Two methods can study emissions from the same region and arrive at significantly different results. This is one of the findings of the review of AMPAC-Net, a network dedicated to Arctic methane and permafrost, published by the European Space Agency on 20 October 2026. The gap does not mean we can choose the figure that is most reassuring, or the one that is more alarming. It points to a measurement problem that needs to be resolved.
In the summary reported by ESA, an estimate built from processes and emission sources comes to around 39 Tg CH₄-C per year, compared with around 15 Tg CH₄-C per year for an approach based on atmospheric observations. The uncertainty ranges do not overlap. These two figures therefore do not describe a fall in emissions between two years: they result from different methods.
What the unit used means
The symbol Tg stands for a teragram, or one million tonnes. But the notation CH₄-C specifies that the mass being counted is the carbon contained in methane, not the mass of the entire molecule. Confusing the two would change the reported figure without a single emission having changed on the ground.
An annual flux must also be distinguished from a stock. Carbon accumulated in frozen soils represents a reserve; emissions indicate what enters the atmosphere over a given period. A large reserve does not mean that its contents escape in full, immediately, or in the form of just one gas.
Finally, the figures cited by ESA concern the balance examined in the review. They are neither a global methane inventory nor a personal measurement of what every patch of permafrost would release. Keeping to this scope avoids turning a scientific finding into a universal figure.
Starting from the ground or working back from the atmosphere
The so-called ‘bottom-up’ approach starts from processes and emission sources. The ‘top-down’ approach uses atmospheric observations and inverse modelling to infer emissions from them. In the comparison presented, this second group comprises fifteen atmospheric inversions.
The two approaches therefore pose the question from different starting points. Measuring a concentration in the air is not the same as directly measuring the amount released from a particular patch of soil. Conversely, describing the mechanisms of an environment is not enough to establish, without uncertainty, their contribution to an entire region.
The ESA review identifies a systematic difference that calls for further investigation. It does not designate one method as definitively correct. Nor would taking the average of the two figures resolve the disagreement: it would produce a third number without explaining the source of the gap.
Why thawing can produce different gases
Permafrost is ground that remains at a temperature of zero degrees or below for at least two consecutive years. The National Snow and Ice Data Center notes that it can preserve organic remains for a long time, with decomposition slowed by the cold.
When conditions become favourable for microbial activity, this material can decompose. The carbon then enters the atmosphere, notably in the form of carbon dioxide. Very wet, oxygen-poor conditions also favour methane production. So simply saying ‘the ground is thawing’ does not specify which gas is emitted, or in what proportion.
Vegetation also plays a role: as plants grow, they absorb CO₂. An ecosystem’s balance must take this absorption and emissions into account. The climate risk associated with thawing remains real; understanding its scale requires tracking these exchanges rather than looking at just one point in the year.
A field example that shows why the whole year matters
The National Park Service describes monitoring carried out near Eight Mile Lake, on the edge of Denali National Park in Alaska. Sensors measure soil conditions, and a tower tracks CO₂ exchanges between the tundra and the atmosphere. The site is sufficiently well drained that methane makes little contribution to its carbon balance.
The example is valuable for understanding one limitation: a permafrost site is not automatically a major methane emitter. Local conditions matter. At this site, carbon absorption during summer growth did not offset emissions over the year in five of the six years studied between 2008 and 2013.
This is an older, local finding, presented here to explain the method, not to provide a snapshot of the whole Arctic in 2026. It shows why favourable summer measurements are not enough to conclude that the annual balance is favourable. Comparing several sites and several seasons is essential.
What new observations need to clarify
Satellites extend coverage to regions that are difficult to access, but their observations need to be brought together with field measurements and models. The useful work lies in making these datasets comparable: what area is covered, over what period, in what unit and with what uncertainties?
AMPAC-Net’s contribution is precisely to bring together the teams working on these questions. The key point for readers is simple: disagreement over the quantity does not alter the physical mechanisms of thawing. It shows where knowledge still needs to become more precise, without justifying either a date for catastrophe or a promise that there is no risk.
Illustrative photograph: an ice lens exposed by erosion on the bank of Birch Creek, Alaska. It does not depict the Eight Mile Lake site. Craig McCaa / BLM Alaska, Wikimedia Commons, CC BY 2.0.
