Global warming doubled its pace since 2015 while the world looked away
New analysis strips natural variability from temperature records to reveal a stark acceleration, even as climate has vanished from public concern
For half a century, the planet warmed at roughly 0.2°C per decade. That figure became so familiar to climate scientists that it functioned almost as a heartbeat, steady and predictable enough that any deviation from it could be attributed to the noise of El Niño cycles, volcanic eruptions, or the eleven-year rhythm of the sun. Then something changed. Strip away that noise, as a new preprint by Grant Foster and Stefan Rahmstorf of the Potsdam Institute for Climate Impact Research does, and the underlying signal snaps into focus with uncomfortable clarity. Since approximately 2015, the warming rate has roughly doubled, to around 0.4°C per decade. The planet is not merely warming. It is warming faster than at any point since systematic records began.
The finding, posted to Research Square in March 2025, has provoked fierce debate among climate scientists and almost no reckoning at all in the broader political world. That mismatch between the gravity of the signal and the volume of public response tells its own story.
The signal beneath the noise
Temperature records are noisy. A strong El Niño like that of 2015-2016 or 2023-2024 can push global temperatures well above the long-term trend for a year or two, then recede. Volcanic eruptions inject reflective aerosols into the stratosphere and temporarily cool the planet. Solar output fluctuates on roughly an eleven-year cycle. These natural oscillations create enough variability in the raw data that genuine shifts in the underlying warming rate can be fiendishly difficult to detect. A 2024 study led by Claudia Beaulieu and published in Communications Earth & Environment applied change-point analysis to the unadjusted temperature record and concluded that a recent acceleration had not yet reached the 95 per cent confidence threshold.
Foster and Rahmstorf took a different approach, one they first developed in a widely cited 2011 paper. Using five independent global temperature datasets from NASA, NOAA, HadCRU, Berkeley Earth, and ERA5, they statistically estimated and removed the contributions of the El Niño Southern Oscillation, volcanic aerosols, and solar variation. What remained was, in effect, the human warming signal with much of the natural static filtered out. The adjusted data tell a strikingly different story from the raw records. In all five datasets, a statistically significant change point appears around 2015, after which the warming rate climbed to approximately 0.4°C per decade. The most recent decade's temperature trend deviates substantially from the trend of the preceding decades.
The elegance of this method lies partly in its symmetry. During the early 2000s, when commentators proclaimed a "pause" or "hiatus" in global warming, Foster and Rahmstorf's approach showed that no statistically significant slowdown had occurred; the apparent plateau was an artefact of natural variability, particularly a sequence of La Niña events. The same analytical framework that debunked false comfort a decade ago now debunks false complacency. If the warming had not paused then, it has genuinely accelerated now.
The Faustian bargain of clean air
If acceleration is real, what is driving it? Greenhouse gas concentrations have risen steadily, not suddenly, which makes them an unlikely explanation for a sharp uptick. The most contentious candidate is an obscure regulation that few outside the shipping industry had heard of before 2023.
On 1 January 2020, the International Maritime Organisation's new fuel sulphur rules came into force, cutting the maximum sulphur content of shipping fuel from 3.5 per cent to 0.5 per cent. The regulation was a public health triumph. Sulphur dioxide from ships causes respiratory and cardiovascular disease, and the IMO estimated the change would prevent hundreds of thousands of premature deaths. But sulphur dioxide also produces sulphate aerosols, tiny particles that serve as nuclei around which cloud droplets form. More aerosols mean brighter, more reflective clouds. Shipping, responsible for nearly all sulphur emissions over the open ocean, had been inadvertently conducting a massive cloud-brightening experiment for decades. When the regulation took effect, shipping SO2 emissions fell by roughly 80 per cent almost overnight.
Here lies a dilemma that captures something essential about the human relationship with complex systems. The same particles that were killing people by the tens of thousands were also shielding the planet from the full force of the warming it had already set in motion. As The Lancet observed when the regulation was introduced, the immediate reduction of sulphur dioxide contributes to global warming if carbon dioxide is not concurrently reduced. It was not.
A 2024 study by Jordan and Henry using the UK Earth System Model estimated the regulation removed a cooling effect equivalent to approximately 0.05°C averaged over the decade, or roughly two to three years of warming. Researchers at the Science Media Centre described it as an "inadvertent reverse marine cloud brightening experiment on a global scale." A separate analysis published in Communications Earth & Environment estimated a radiative forcing of +0.2 watts per square metre over the global ocean, enough to potentially double the warming rate of the 2020s relative to recent decades.
James Hansen, the former NASA scientist whose 1988 testimony to the United States Congress first brought climate change to broad political attention, goes much further. In a February 2025 paper in the journal Environment: Science and Policy for Sustainable Development, Hansen and eighteen co-authors argued that the aerosol forcing from reduced shipping emissions is at least 0.5 watts per square metre, roughly six to ten times the estimate used by Zeke Hausfather and Piers Forster of the University of Leeds in their Carbon Brief analysis. Hansen contends that the Intergovernmental Panel on Climate Change has systematically underestimated aerosol forcing for decades, with cascading consequences for estimates of climate sensitivity. If he is right, the planet is more sensitive to greenhouse gases than mainstream models assume, and the loss of aerosol cooling has unmasked warming that was always there, waiting.
Not everyone agrees. Michael Mann of the University of Pennsylvania has argued that recent temperatures remain within the range projected by existing climate models. Kevin Trenberth of the National Centre for Atmospheric Research points out that the Southern Hemisphere has also warmed substantially, and shipping aerosol reductions cannot explain that. Trenberth suspects changes in atmospheric circulation and ocean heat redistribution play a larger role than aerosols. The Forster et al. Indicators of Global Climate Change 2024 report, involving Hausfather among others, places the human-induced warming rate at 0.27°C per decade over 2015-2024, a notable 40 per cent increase from the post-1970 average but well below Hansen's estimates. Much of that acceleration, the report notes, is driven by declining aerosol cooling unmasking greenhouse warming.
The disagreement is not academic. As Hansen himself put it, the gap between his estimate and the mainstream figure is so large that observations over the next few years should be able to resolve it. He predicted that 2025 temperatures would remain near or above 1.5°C, even without an El Niño to boost them. Early 2025 data leaned his way, with January recording 1.75°C above the pre-industrial baseline.
The great scientific contest of 2025
Rahmstorf and Foster's preprint lands in the middle of what amounts to the most consequential methodological dispute in climate science since the sensitivity debates of the 1990s. The question is not whether warming is occurring, nor even whether it has recently accelerated. A growing number of researchers now accept some degree of acceleration. Robert Rohde, chief scientist at Berkeley Earth, told the Washington Post that something has clearly changed. Hausfather, who was cautious even a year ago, wrote in his Climate Brink newsletter in mid-2025 that the consilience of evidence from surface temperatures, ocean heat content, and Earth energy imbalance measurements tips the scale towards clear acceleration. The remaining argument is over magnitude and cause.
Rahmstorf and Foster, by their own admission, do not identify a specific cause for the acceleration. Their paper is a statistical finding, not a mechanistic explanation. They note that their method of removing El Niño effects is approximate and may not completely eliminate the influence of the strong 2023-2024 El Niño on the final years of their dataset. A separate analysis by researchers using a machine-learning approach to filter internal variability, rather than ENSO indices, found no evidence of an abrupt additional acceleration beyond what declining aerosols would explain. This suggests Rahmstorf and Foster's result, while robust, might partly reflect methodological choices about how natural variability is defined and removed.
What matters for policymakers is that even the most conservative interpretation of the data points to a world warming meaningfully faster than it was a decade ago. The Forster et al. estimate of 0.27°C per decade may sound modest beside Hansen's 0.4°C, but it still represents a significant departure from the roughly 0.18°C rate that governed climate planning for decades. Budgets, adaptation timelines, and infrastructure investments were calibrated to the old rate. They need recalibrating.
Shifting baselines and the attention economy
In 1995, the fisheries scientist Daniel Pauly of the University of British Columbia published a one-page essay that became one of the most cited papers in ecology. He described what he called "shifting baseline syndrome," the tendency for each generation of researchers to accept the degraded state of the environment they inherited as normal, never recognising how much had already been lost. Pauly was writing about fish stocks, but the concept applies with uncomfortable precision to climate discourse.
When the Paris Agreement set 1.5°C as its aspirational target in 2015, the idea of breaching it felt distant, a boundary decades away. The Rahmstorf and Foster data now show four of five major temperature datasets projecting that the smoothed, adjusted warming trend will cross 1.5°C by late 2026. The ERA5 reanalysis dataset suggests it may already have crossed that threshold. Yet the public treatment of 1.5°C has quietly shifted from a bright line to a smudge, from a target to a talking point. At COP29 in November 2024, delegates reaffirmed their commitment to pursuing efforts to limit warming to 1.5°C, a formulation that was already at odds with the observational record.
The displacement of climate from public attention is itself a phenomenon worth examining. Research published in 2025 in Climate Policy by Penasco and Grossman found that wealthier nations show declining support for stringent climate policy despite maintaining high awareness of the problem. The authors attributed this partly to policy fatigue in countries that have already implemented mitigation measures. Psychologists describe a complementary dynamic. Climate change, as a threat, is poorly matched to human cognitive architecture. It is distant in time and space, diffuse in its causation, and lacks a clearly identifiable antagonist. Political psychologists call this "low-threat salience," and it is compounded by what researchers term "system justification," the tendency to defend existing arrangements as basically sound. When people feel overwhelmed, they do not mobilise; they adjust their expectations downward.
The pattern is visible across contemporary society. The prevailing public mood on climate has shifted, over the past few years, from alarm to something closer to exhausted acceptance. Wars in Ukraine and the Middle East, the turbulence of American politics, the anxieties surrounding artificial intelligence, and the lingering economic consequences of pandemic-era disruption have competed ferociously for limited reserves of collective worry. Climate change, which requires sustained attention to a slow-moving threat amidst an endless supply of fast-moving crises, has been edged off the front page not because people stopped believing in it but because they stopped believing they could do anything about it. It is worth remembering that during the early months of the pandemic, when aviation halted and commuting ceased, there was a brief flowering of environmental optimism, a sense that humanity might emerge from lockdown with greener habits and better priorities. That hope evaporated remarkably quickly.
What replaced it was not denial but a kind of practical fatalism. Individuals who accept the science nonetheless struggle to see how their choices matter when emissions are driven overwhelmingly by industrial systems, shipping fleets, and the energy infrastructure of entire nations. The result is a population that recycles conscientiously, considers buying an electric vehicle, and quietly suspects that none of it will make the slightest difference. This is not ignorance. It is a rational response to a collective action problem of staggering scale, one in which any single nation's sacrifices can be rendered meaningless by another's refusal to participate. The tragedy is that the logic applies equally to every nation, producing a global equilibrium of inaction even as the thermometer climbs.
A subtler form of fatalism holds that the billionaire class will eventually force through geoengineering solutions once their own assets come under threat, a cynical optimism that relies on self-interest to accomplish what democratic consensus could not. Meanwhile, spiking property insurance rates, increasingly uninsurable homes in fire-prone and flood-prone regions, and the creeping withdrawal of agricultural viability from once-productive land suggest the pain has already begun, only distributed too unevenly to provoke a unified response.
What acceleration means for the timeline
The practical consequences of a doubled warming rate are not merely twice as bad. They are qualitatively different. At 0.2°C per decade, the world had decades to adapt, to build seawalls, to redesign agricultural systems, to transition energy infrastructure. At 0.4°C per decade, the window for each of these adjustments roughly halves. Heat waves that were projected for the 2040s arrive in the early 2030s. Coral reef systems adapted to gradual change face thermal shocks they cannot keep pace with. Infrastructure designed for a climate that no longer exists degrades faster than it can be replaced.
The acceleration also reframes the debate about climate interventions. Geoengineering, once dismissed as a distraction from emissions reduction, enters the conversation through a back door. The IMO 2020 experience demonstrates, in the words of experts assembled by the Science Media Centre, that marine cloud brightening could be an effective cooling intervention, because the removal of shipping aerosols is essentially its inverse. If reducing sulphur pollution caused measurable warming, then deliberately reintroducing reflective particles into the atmosphere could cause measurable cooling. The logic is straightforward if uncomfortable. The danger is that such interventions become a substitute for emissions reduction rather than a complement to it, a thermostat that allows humanity to keep burning fossil fuels while periodically adjusting the shade.
There are grounds for something other than despair, even if they are easily drowned out by the drumbeat of record temperatures. Renewable energy costs have plummeted to the point where new solar and wind capacity is cheaper than new fossil fuel plants in most of the world. China, for all the criticism it attracts, has installed more renewable energy capacity in recent years than any other nation. The central estimate of twenty-first century warming under current policies has fallen below 3°C, down from the 4°C projections that prevailed as recently as the early 2010s. Emissions growth has slowed meaningfully. These are not trivial achievements. They represent the accumulated effect of decades of policy, investment, and technological improvement, and they matter.
But the acceleration documented by Rahmstorf, Foster, Hansen, Forster, and their various colleagues changes the arithmetic of how much those achievements buy. The margins are thinner than anyone assumed. Hansen's team, while warning that current policies amount to wishful thinking, argues for temporary solar radiation management alongside aggressive emissions cuts and improved aerosol monitoring. Hausfather, while more sanguine about the emissions trajectory, agrees that less time remains than previously assumed.
There is something clarifying, if deeply uncomfortable, about the Rahmstorf and Foster finding. For years, the debate about warming acceleration was genuinely uncertain, and reasonable scientists disagreed about whether the signal had emerged from the noise. That uncertainty provided a kind of shelter. If acceleration was not yet statistically proven, it could be treated as a possibility rather than a present reality. The adjusted data have removed that shelter. The signal is there, confirmed across five independent datasets. Whether the cause is primarily aerosol unmasking, increased climate sensitivity, cloud feedback changes, or some combination of all three, the planet is demonstrably warming faster than it was.
The appropriate response to this finding is a subject on which reasonable people can and do differ. But the finding itself is no longer in serious dispute. As Rahmstorf wrote on his blog in late May 2025, the most important insight from these adjusted data is that there is no longer any doubt regarding a recent increase in the warming rate. The world may not continue warming at this pace. It could equally continue accelerating. What it cannot do is return to the comfortable baseline that shaped a generation of climate planning. That baseline has shifted, and this time we watched it happen.