Climate feedbacks could add another 0.2–0.4°C of warming this century

Climate feedbacks could add another 0.2–0.4°C of warming this century

A warming planet does not merely receive heat. Some natural systems respond by releasing more greenhouse gases, which can produce still more warming.

A new 2026 analysis by Sam Abernethy, Danielle Monteverde and colleagues estimates that these warming-induced emissions could add roughly 0.2–0.4°C of global warming by 2100, amplifying human-caused warming by around 20–30% across the scenarios they examined.

The estimate combines projected greenhouse-gas releases from wetlands, permafrost, freshwaters and wildfires. Methane from wetlands, thawing permafrost and inland waters, together with carbon dioxide from permafrost and fires, is fed into the reduced-complexity MAGICC climate model to estimate how much additional warming those emissions could produce.

The result is consequential, but its most useful interpretation is narrower than the headline suggests.

Climate science has not simply forgotten that warming can trigger further emissions. The IPCC’s Sixth Assessment Report explicitly assessed carbon-cycle and non-CO2 biogeochemical feedbacks, including emissions from thawing permafrost and natural methane sources. It also adjusted remaining carbon-budget estimates for some feedbacks that are incompletely represented in standard Earth-system models.

What remains incomplete is the modelling of several important processes, and the confidence with which their combined effect can be quantified.

The IPCC concluded with high confidence that thawing permafrost will release carbon, but with low confidence in the timing, magnitude and balance between carbon dioxide and methane. Most CMIP6 Earth-system models did not represent permafrost carbon processes comprehensively, while abrupt thaw, wildfire-permafrost interactions and some natural methane feedbacks were still missing or weakly constrained.

That gap is what the new work attempts to quantify more systematically.

In an abstract presented at the 2026 European Geosciences Union General Assembly, Monteverde and colleagues estimated that warming-induced methane emissions could reach about 70 ± 40 million tonnes a year under a low-emissions scenario and 200 ± 70 million tonnes under a higher-emissions scenario by 2100. Their preliminary calculation produced approximately 0.2°C of extra warming in the low-emissions case and about 0.5°C in the higher-emissions case.

A related April preprint by the broader research team reports a somewhat tighter central range of 0.2–0.4°C, corresponding to roughly 20–30% amplification of anthropogenic warming. The authors estimate large uncertainty around both quantities.

That uncertainty matters. The calculation synthesises temperature-emission relationships from existing process studies and then applies them through MAGICC rather than demonstrating the full feedback behaviour in a new generation of coupled Earth-system models. Some of the relationships are treated approximately as linear with temperature, while the underlying processes can vary strongly by region, hydrology, fire regime, vegetation and timescale.

Observations are another constraint. A 2026 assessment of global methane monitoring found that uncertainties for natural methane sources remain substantially larger than for anthropogenic sources, particularly in remote wetland, Arctic and tropical regions. Detecting whether natural methane emissions are already accelerating because of warming is therefore harder than estimating global atmospheric methane growth itself.

There is also a useful reversal in the policy implication.

If natural feedbacks amplify warming, that does not make emissions cuts less effective. It makes them more valuable. Lower human-caused warming means less thaw, less warming-driven methane release and generally less activation of the feedbacks themselves.

The EGU analysis illustrates this directly: its estimated warming-induced methane emissions are much smaller under the low-emissions SSP1-2.6 pathway than under the higher-emissions SSP4-6.0 pathway.

So the new estimate should not yet be read as “climate models are underestimating warming by 30%.” Existing assessments already incorporate several feedbacks, and the new calculation remains preliminary. The stronger claim is that some natural greenhouse-gas feedbacks may still be incompletely represented, and their combined effect could be large enough to matter for carbon budgets and temperature targets.

That proposition is now testable rather than merely cautionary. The Warming-Induced Emissions Model Intercomparison Project is bringing multiple modelling groups together to quantify these processes more consistently, with the aim of informing the IPCC’s Seventh Assessment Report.

If the higher estimates survive that process, climate projections will need adjustment. If they shrink, the uncertainty itself will have been reduced. Either outcome is more useful than treating poorly constrained feedbacks as either negligible or catastrophic.

Sources

• Abernethy et al., Projected amplification of global warming from warming-induced greenhouse gas emissions (2026 preprint)
• Monteverde et al., Future warming-induced emissions are substantial and poorly constrained, EGU General Assembly 2026
• IPCC AR6 Working Group I, Chapter 5: Global Carbon and Other Biogeochemical Cycles and Feedbacks
• IPCC AR6 Working Group I, Technical Summary
• Ciais et al., A Global Methane Observation System to Reduce Uncertainty for Anthropogenic and Natural Sources and Sinks for Detecting and Attributing Climate Feedbacks
• Warming-Induced Emissions Model Intercomparison Project
• Phys.org, Unaccounted climate feedback loops may amplify warming by 20–30% this century, but cutting emissions will reduce risks