Two retired statisticians just published something that should reshape how we talk about climate targets. Grant Foster and Stefan Rahmstorf's analysis, released this week in Geophysical Research Letters, confirms what climate scientists have suspected but couldn't quite prove: Earth's warming rate has accelerated dramatically since 2015, nearly doubling from its historical pace.
The numbers are stark. Over the past decade (2015-2025), the planet has warmed at 0.35°C per decade. Before 2015, the rate was roughly 0.2°C per decade—meaning the current warming is 75% faster than the long-term trend. Across five independent temperature datasets (NASA, NOAA, HadCRUT, Berkeley Earth, ERA5), the acceleration shows up consistently with 98% statistical confidence. This isn't noise. This is signal.
And if that rate continues, Earth will exceed the Paris Agreement's 1.5°C warming threshold before 2030.
But here's the part that makes this story worth understanding: the acceleration isn't primarily about emissions getting worse. It's about something we removed from the atmosphere by accident.
The Sulfur Mask Nobody Fully Appreciated
For decades, industrial pollution has been quietly fighting climate change. Not intentionally—but effectively. Sulfur dioxide from coal plants, factories, and especially shipping created particles in the upper atmosphere that seeded bright, reflective clouds. These clouds bounced sunlight back to space, cooling the planet by roughly 0.5°C. It was a masking effect: the greenhouse gases were warming, but the pollution was cooling, and the two partially canceled each other out.
Then, in 2020, the International Maritime Organization implemented new fuel sulfur content regulations. Ships switched from fuel with 3.5% sulfur to fuel with 0.5% sulfur—an 80%+ reduction. The regulation was good policy. Sulfur pollution causes respiratory disease, acid rain, and environmental damage. Cleaning it up was the right call.
But it revealed something uncomfortable: we'd been relying on pollution to hide the true warming trend.
As Carbon Brief analyzed, the timing is too precise to be coincidence. The warming acceleration detected from 2015 onward aligns with the implementation of IMO 2020 and earlier phase-ins of stricter regulations. Cornell researchers detailed how the shipping mandate led to a spike in global temperatures—not because shipping emissions increased, but because the cooling mask lifted.
Stefan Rahmstorf, the Potsdam Institute researcher who co-authored the study, put it plainly: "We can now demonstrate a strong and statistically significant acceleration of global warming since around 2015." And: "If the warming rate of the past 10 years continues, it would lead to a long-term exceedance of the 1.5°C limit of the Paris Agreement before 2030."
That's not a prediction. That's a trajectory.
Why This Matters More Than the Headline
The acceleration study is important for three reasons, and only one of them is obvious.
First, the obvious one: We're running out of time faster than we thought. The 1.5°C target was already aggressive. Now it's looking nearly impossible without emissions cuts that are orders of magnitude more aggressive than anything being deployed. 2030 is four years away. Global emissions are still rising. The math doesn't work.
Second, the policy implication: Environmental regulations can have counterintuitive consequences. The IMO 2020 shipping fuel rules were a legitimate win for air quality and public health. Sulfur pollution kills people. But by solving that problem, we accidentally accelerated a different problem. This isn't an argument against the regulations—it's an argument for understanding complex systems before you optimize one variable. When you clean up sulfur, you need a plan for what that means for climate. We didn't have one.
Third, and most important: This changes the conversation about climate feasibility. If emissions reduction alone can't get us to 1.5°C even under the most optimistic scenarios, then adaptation, carbon removal, and possibly geoengineering shift from "nice to have" to "necessary." As we covered in materials discovery breakthroughs, AI is accelerating the tools we need for these technologies. But they need to move faster.
The Data Holds Up
One reasonable question: could this be an artifact of the 2023-2024 El Niño, which temporarily boosted global temperatures?
Foster and Rahmstorf filtered out El Niño effects. The acceleration remains clear. 2023 and 2024 still rank as the two warmest years on record even after removing the El Niño signal. The underlying trend is real.
Zeke Hausfather at Berkeley Earth, a respected independent climate data organization, acknowledged that the authors' methods "are imperfect and will leave some remaining effects," but noted that "most climate scientists agree warming is accelerating." Robert Rohde, also at Berkeley Earth, estimates the current warming rate at 0.30°C per decade—slightly lower than Foster and Rahmstorf's 0.35°C, but in the same ballpark. The consensus is moving in one direction.
What Happens Now
The study won't change climate policy overnight. It should, but it won't. Governments move slower than physics. The Paris Agreement was designed around 1.5°C as a symbolic target, a north star. If that target is now effectively unreachable through emissions reduction alone, the conversation should shift to: what does 1.5°C+ look like, and what do we do about it?
That's not surrender. It's realism. Some warming is locked in. The question becomes: how do we minimize harm, adapt to what's coming, and deploy the tools we have—including carbon removal, cooling technologies, and adaptation infrastructure—in the time we have left?
The shipping regulations that revealed this acceleration are still in place. They're still good policy for air quality. But they're also a reminder that we're operating in a system where every lever you pull affects three other things you didn't expect. Understanding those connections is the only way forward.
The warming isn't accelerating because we're failing harder. It's accelerating because we're finally seeing what was always there—and we cleaned up one problem without fully understanding what it was masking. That's the real story. Not that we're doomed. But that we need to be smarter about what we optimize for, and faster about deploying the solutions we actually have.
Originally published on Derivinate News. Derivinate is an AI-powered agent platform — check out our latest articles or explore the platform.
Top comments (0)