Siberian Methane Emissions Are Doubling Each Decade: What It Means for Carbon Markets

Methane is the second-largest contributor to human-caused warming after carbon dioxide, and it is short-lived and potent — which makes it both a serious near-term problem and one of the fastest levers available for slowing warming. That combination is why atmospheric methane trends get scrutinised so closely, and why a paper on Siberian emissions is worth the attention of people who work in carbon markets rather than atmospheric science.

The study — Zhu et al., Science 393, 615 (2026) — is a careful piece of measurement work. This post covers what it found, how it was done, and the three things it actually changes for the rest of us.

What the study found

Against a regional mean of roughly 22.9 Tg of methane per year, a trend of about 10.4 Tg per year per decade is the "doubling" in the paper's title. Three details give it shape:

• It is seasonal. More than 67% of the annual Siberian methane budget is emitted during the growing season, and the statistically significant upward trend appears only in that window. Outside the growing season there is no significant trend.
• It is regional. Of the three major high-latitude regions the authors examined, only Siberia shows a clear upward trend. Alaska and boreal Canada do not.
• It has two different causes. Western and eastern Siberia are behaving differently, and the authors treat them separately for good reason.

The two mechanisms

Western Siberia: wetland methanogenesis

Methane in wetlands is produced by microbes breaking down organic matter in waterlogged, oxygen-free soil — a process called methanogenesis. Two things control the rate: how much of the ground is saturated, and how warm the soil is.

The paper finds western Siberian emissions of about 12.96 Tg per year, rising at 0.36 ± 0.10 Tg per year each year. Three lines of evidence support the wetland interpretation: the increase is concentrated in the thaw season, it is spatially concentrated over inundation-prone terrain, and the largest trends occur in exactly those wet grid cells that also experienced the strongest preceding-winter warming.

That last point is the most interesting mechanism in the paper. Anomalously warm winters appear to set up high-emission summers — a warm February changes surface albedo and the surface energy balance, and those changes carry through to the warm season. Blocking high-pressure systems and elevated incoming shortwave radiation show up alongside the highest-emission years.

Eastern Siberia: wildfire

Eastern Siberia is a different story. There, the surge tracks fire. The authors estimate about 9.91 Tg of methane per year from the region, and check it three ways: the year-to-year variation follows independent fire-derived CO₂ estimates; during the intense July–August 2021 fire peak, methane and carbon monoxide signals rise together over the burning region; and across the full 2010–2023 record, grid cells with more burned area show both higher methane fluxes and more strongly rising trends.

What it is not

Given where these emissions occur, the obvious hypothesis is leaking Russian gas infrastructure. The authors address it directly and reject it. Oil and gas emissions were included in the prior inventory, but published Russian oil and gas methane trends are relatively small and are not specific to western Siberia, which makes them a poor match for the pattern actually observed. The signal is seasonal, thaw-linked and concentrated over wetlands — a signature that industrial leakage does not produce.

This matters for how the result should be read. It is a climate-feedback finding, not an oil-and-gas-accountability finding.

The methodological finding carbon markets should care about most

Set the Arctic aside for a moment. The most transferable result in this paper is a gap between two ways of counting.

There are two broad approaches to estimating emissions. Bottom-up inventories multiply activity data by emission factors — burn this much of that fuel, apply the published factor, get a number. Nearly all corporate and national greenhouse gas reporting works this way. Top-down estimates work backwards from what is actually in the atmosphere, using satellite and surface measurements plus a transport model to infer what must have been emitted to produce the observed concentrations.

When the authors compared their top-down eastern Siberian fire estimate against the standard global biomass-burning inventories — GFED, FINN, GFAS and QFED, the datasets that underpin a great deal of fire emissions reporting — their estimate was 6.7 to 41 times higher.

They attribute the gap to three things: emission factors derived from very few field or laboratory studies, uncertainty in satellite-derived burned area and fire radiative power, and sources the inventories may simply omit — such as methane released from permafrost thaw triggered by the fire itself.

We cover what this means for corporate inventories in more detail in satellite MRV and methane monitoring and why your emission factors may be wrong.

How the estimate was produced

Briefly, because the method is the reason to take the result seriously. The authors combined column methane retrievals from Japan's GOSAT satellite with NOAA's global network of surface measurements, and used an Ensemble Kalman Filter to adjust a prior emissions inventory until the GEOS-Chem atmospheric transport model reproduced what the instruments actually observed.

They then validated the result against data that was not used in the inversion: ground-based spectrometers from the TCCON network, flask samples collected by Japan Airlines aircraft, the TROPOMI satellite, and eddy-covariance flux towers. Agreement with independent observations improved substantially after the inversion — for Siberian surface stations, the mean correlation between model and observations rose from 0.23 to 0.72, and model bias fell by 70%.

Three things this changes

1. The carbon budget has a smaller margin than a purely anthropogenic accounting implies. The paper's projections are explicitly conditional sensitivity estimates rather than forecasts — they assume the observed temperature–emission relationship continues to hold — but they are still sobering. Under SSP3-7.0, eastern Siberian emissions are projected to reach 97.5 ± 32.0 Tg per year by 2081–2100; under the higher-warming SSP5-8.5, 169.2 ± 54.1 Tg per year. Natural sources that grow with warming consume budget that would otherwise be available to emitters, which is one more reason to expect compliance stringency to ratchet in one direction only. That logic is worth understanding if you are planning around India's CCTS.

2. Fire is a growing risk to land-based carbon projects. If fire-driven emissions in a boreal region are large and rising, the same physical risk applies to forestry and agroforestry projects everywhere, where a fire is a reversal event that can undo years of sequestration. Buffer pools and permanence risk ratings exist precisely for this, and they deserve more attention than they usually get at the feasibility stage.

3. Measurement is catching up with reporting. The instruments used here — GOSAT, TROPOMI, and a growing fleet of successors — are the same ones now being pointed at landfills, coal mines, oil and gas facilities and agricultural regions. The era in which a reported methane number could only be checked against the spreadsheet that produced it is ending.

What not to take from it

Two cautions, both because the subject invites overstatement.

First, this is a measurement and attribution study covering 2010–2023. It does not claim a tipping point, and its future numbers are conditional projections under specified scenarios, not predictions. The authors are explicit about that.

Second, it says nothing about the integrity of any particular carbon credit or project type. Natural methane emissions from Siberian wetlands and the quality of a soil-carbon project in Maharashtra are unrelated questions.

The practical takeaway

For most businesses, the useful residue of this paper is not the Arctic. It is the reminder that greenhouse gas numbers are increasingly checkable — and that the gap between what is reported and what is measurable can be very large. Companies whose footprint includes significant methane, whether from rice cultivation, livestock, landfill or process sources, are the ones for whom that gap is most likely to matter.

Carbon Credit Consulting builds defensible greenhouse gas inventories and designs MRV systems that hold up to third-party and satellite scrutiny. Explore our GHG accounting and carbon offset project development services, or talk to us.