A Missing Piece in Climate Models: Nature’s Own Emissions

“If you’re not including all the emissions going into the atmosphere, you’re hamstrung from the get-go,” says Brian Buma, a climate scientist at the Environmental Defense Fund. “People are recognizing that the longer we go without taking these emissions into account, there’s just going to be a bigger gap.” 

“Four decades ago, the scientific community was saying we think there are going to be these surprises in the Earth system as the planet warms,” says Benjamin Poulter, the lead scientist at Spark Climate Solutions, a San Francisco-based nonprofit that aims to identify climate “blind spots,” include them in policy frameworks, and assess the best way to deal with them. “Now, we’re starting to see these feedbacks become the reality.”

The biggest sources of warming-induced emissions are wildfires, wetlands, and permafrost — all of which have shown recent indications of rapid change. Since 2001, global carbon emissions from wildfires have increased by 60 percent. In 2020, researchers reported an alarming spike in the concentration of methane in the atmosphere, attributed partly to wetter conditions expanding wetlands in Africa and Asia and partly to warming temperatures, which accelerate the rate at which plants decompose in water. 

Colder regions are seeing some of the fastest shifts: The 2024 Arctic report card found that wildfires and permafrost melt have caused the northern tundra to become a source of emissions, after acting as a sink for millennia. Hotter, drier conditions fuel Arctic wildfires, while thawing ice allows microbes to more rapidly decompose soil organic matter, releasing both CO2 and methane.

Those regional changes add up to global trends. In 2023 and again in 2024, record heat reduced the land carbon sink — the difference between the amount of CO2 emitted and the amount of CO2 absorbed by all terrestrial ecosystems — and contributed to a record jump in atmospheric CO2 concentrations. In 2025, the land sink appears to have recovered to its previous strength, although tropical forests in Southeast Asia and South America flipped from sink to source due to deforestation, wildfires, and more decomposition. 

Like 2023 and 2024, 2026 will be an El Niño year, with elevated Pacific Ocean surface temperatures that are expected to boost global temperatures. These events are associated with hot, dry conditions in the Amazon, which can amplify wildfires and weaken the region’s carbon sink. The heat can also slow photosynthesis and increase the rate of decomposition in other tropical ecosystems.

Researchers are alarmed by these changes. “We’ve been worried about this for a long time,” says Rob Jackson, an environmental scientist at Stanford University and the chair of the Global Carbon Project, an international consortium of researchers tracking flows of carbon through Earth systems. Yet climate models have largely set the problem of warming-induced emissions aside: Of the 11 Earth system models used in the most recent IPCC assessment, none included warming-induced emissions from all of the main sources — wildfire, wetlands, and permafrost. Five included wildfire; just two included permafrost.

That’s largely because of the difficulty of the problem: The scientific challenge amounts to predicting the response of every ecosystem on Earth to warming temperatures. “It adds computational processing time, complexity, and there’s no good agreement about how to represent wetland emissions, permafrost, wildfire,” says Poulter. Modeling all of these emissions requires simulating diverse, nonlinear processes that behave in ways that can be difficult to predict. The basic modeling challenges are exacerbated, experts say, by a lack of monitoring in hard-to-access ecosystems, from Siberia to the Congo basin.

Just how much warming might be missing from the picture? In the new study, Poulter and a group of colleagues from leading climate modeling groups used a simplified climate model to estimate the volume of emissions that might be expected from these sources under a range of scenarios, using past estimates of the scale of the feedback for each source of warming-induced emissions. They then modeled how much those emissions would raise temperatures on top of the contribution from human emissions.

If humans quickly rein in their emissions, the researchers found, emissions from forests, wetlands, and other ecosystems might amount to anywhere between zero additional degrees and 0.4 additional degrees C this century, an amount that would accelerate warming due to human emissions by up to 60 percent. Under a more pessimistic scenario, in which human emissions peak around 2060 and then decline, they found emissions from ecosystems could raise temperatures from between 0.2 to 0.6 degrees C. The lion’s share of the uncertainty stemmed from how researchers estimated ecosystems would respond to warming, rather than from the choice of warming scenario or any inherent ambiguities in the models.

Keeping the rise in average global temperatures well below 2 degrees C is already a huge challenge. “Something like these warming-induced emissions makes that harder,” says Chris Jones, a climate scientist at the U.K. Met Office working to integrate these emissions into climate models. Warming feedbacks could also add to the danger of overshooting Paris Agreement goals, given that they may not be reversible, he says. While a coal-fired power plant can be replaced with renewables, he explains, melting permafrost will continue to melt after a certain amount of warming. 

To put the scale of the problem into perspective, the researchers estimated that by 2100, the annual contribution of warming-induced CO2 emissions could be equivalent to those from today’s power and building sectors, which together make up about half of global CO2 emissions due to direct human activities. Warming-induced methane emissions could be almost equivalent to today’s annual fossil methane emissions from Asia and North America combined, they found.

Now that the signal of these emissions has emerged, researchers say it is essential to get a better grip on how they work and what they mean for the future of the climate, and to get countries to start counting them. “We can quantify and measure these emissions, but then we also need to develop an accounting framework so that we can do something about them,” says Poulter.

Since launching last fall, Spark’s program on warming-induced emissions has connected and made grants to more than 20 independent modeling groups from around world to add those emissions to models, improve measurements, and explore ways to potentially reduce those emissions. “There’s a community of people now that are building this field,” says Poulter. Their aim is to make robust projections that can be used in the next IPCC climate assessment and thus play a role in global climate policy. Work on the assessment is already underway, with a final report due by late 2029. The modeling work also feeds into efforts to improve measurements of ecosystems — to understand how they are changing and what role warming is playing. Modelers can identify gaps in observations and make decisions about where limited resources should be placed.

One of the first projects funded through Spark’s warming-induced emissions campaign is the installation, by a team at the University of California, Los Angeles, of dozens of methane sensors at wetlands in central Africa, which are known to be major emitters of methane. Another measurement effort led by Jackson focuses on improving measurements of wetland emissions in the Amazon. In both regions, Jackson says there is a lack of data on baseline CO2 and methane emissions, which are essential to making reliable future projections. This also makes it challenging to know what is driving the change in emissions: Is it wetlands growing with more precipitation, or is it a warming-induced increase in microbial respiration, or other factors?

Unlike a leaky natural gas well, Jackson says, “you can’t turn a wrench on a wetland.” But there are ways that warming-induced emissions might be addressed directly. 

Bronson Griscom, an ecologist and founder of a natural climate solutions company called Ceiba Earth, sorts ideas for addressing warming-induced emissions into three buckets. The first promotes the continuation of projects that will work even as temperatures rise. For example, reforestation efforts in temperate woodlands might actually see more growth and therefore more carbon storage with elevated CO2. Fuel reduction efforts, such as thinning and conducting controlled burns, will also reduce emissions. 

The second bucket contains tweaks to existing projects to account for future warming. Teams working on restoration in the Amazon, for example, might consider picking tree species that are more resilient to hotter temperatures, says Griscom, something researchers are already experimenting with.

The third bucket of ideas addresses warming-induced emissions directly. In the Arctic, Cansu Culha, an adjunct professor at the University of British Columbia, is looking to slow the melting of permafrost by applying insulating “blankets” of vegetation to permafrost slumps, where the land is releasing carbon most rapidly. Jackson and other researchers have considered ways to manage wetlands to reduce the amount of methane they generate by changing cycles of wetting and rewetting, altering water chemistry to influence microbial activity, or reconnecting them with seawater. Managing a natural wetland in this way is complicated and controversial, but some of the approaches have been tested with success in rice paddies — another major source of methane.

Researchers and policymakers who are trying to tackle emissions say they would benefit from a clearer picture of feedback mechanisms as the signal of warming-induced emissions begins to emerge from the noise. “People are seeing this stuff happening,” says Buma. “It’s in front of their eyes.” Climate modelers are scrambling to catch up.