After filtering out El Niรฑo, volcanic eruptions, and solar variability, researchers found that Earth has warmed at roughly 0.35ยฐC per decade over the past ten years โ nearly twice the average rate seen from 1970 to 2015.
Global warming appears to have shifted into a faster gear.
A new statistical analysis from the Potsdam Institute for Climate Impact Research (PIK) finds strong evidence that the rate at which Earth is warming has accelerated since around 2015.
Over the past decade, global temperatures have risen at an estimated rate of roughly 0.35ยฐC per decade, depending on the temperature dataset examined. Between 1970 and 2015, the average warming rate was just under 0.2ยฐC per decade. The recent pace is higher than that of any previous decade since instrumental global temperature records began in 1880.
Published in Geophysical Research Letters, the study is significant because the researchers did not simply compare recent hot years with earlier ones. Instead, they mathematically removed several major sources of natural temperature variability that can temporarily push global temperatures higher or lower.
Once those effects were filtered out, the underlying acceleration became much clearer.
Separating long-term warming from natural fluctuations
Earth's global temperature does not increase smoothly from one year to the next.
Natural processes can cause substantial short-term fluctuations, sometimes temporarily amplifying warming and sometimes masking it.
One of the most important is El Niรฑo, a recurring climate pattern in the tropical Pacific that releases large amounts of ocean heat into the atmosphere and can temporarily raise global temperatures.
Large volcanic eruptions can have the opposite effect. Sulfur-containing particles injected high into the atmosphere can reflect sunlight back into space, producing temporary global cooling.
Changes in solar activity also contribute smaller variations.
These effects create what statisticians often describe as noise around the longer-term climate trend.
To reveal the underlying signal more clearly, researchers Grant Foster and Stefan Rahmstorf estimated and removed the influence of El Niรฑo, volcanic activity, and solar variability from the observational temperature records.
The result was a clearer picture of how the planet's underlying warming rate itself has changed.
Five major temperature records tell the same story
The researchers did not rely on a single global temperature record.
They examined five major datasets produced by independent scientific organizations and research groups:
NASA, NOAA, HadCRUT, Berkeley Earth, and ERA5.
Although these datasets differ in how measurements are collected, processed, and combined, they all showed the same broad pattern once natural variability was accounted for.
The adjusted records indicated an acceleration in global warming with more than 98% statistical confidence.
Importantly, the finding remained consistent across all five datasets and did not depend on which of the study's statistical techniques was used.
That consistency strengthens the argument that the observed shift is not simply an artifact of one particular temperature record or analytical method.
The acceleration becomes visible around 2013 to 2015
The timing of the change is also notable.
Across the temperature datasets, indications of a faster warming rate begin appearing around 2013 or 2014, with the broader transition becoming clear around 2015.
That timing matters because the last several years have included extraordinary global temperatures, particularly 2023 and 2024.
Those record-hot years raised questions about whether Earth had entered a period of more rapid warming or whether the extreme temperatures were largely the result of temporary natural fluctuations.
The researchers tested that possibility directly.
After adjusting the records for influences including El Niรฑo and the solar maximum, the temperatures of 2023 and 2024 were reduced somewhat.
Yet both years still remained the warmest in the instrumental record.
More importantly, the statistical evidence for acceleration did not disappear.
That indicates the result is not simply being produced by a few exceptionally hot years at the end of the dataset.
Two different statistical tests reach the same conclusion
To test whether Earth's warming trajectory has genuinely changed, Foster and Rahmstorf applied two independent statistical approaches.
The first involved a quadratic trend analysis.
A simple linear trend assumes temperatures rise at approximately the same rate over time. A quadratic analysis, by contrast, can detect whether the temperature curve is bending upward โ meaning the rate of warming itself is increasing.
The researchers also used a piecewise linear model.
Rather than assuming a smooth acceleration, this method divides the temperature record into different periods and determines whether there is an identifiable point at which the warming rate changes.
Both approaches supported the conclusion that warming has accelerated.
Finding the same pattern using different statistical methods reduces the likelihood that the result depends on a particular mathematical assumption.
The recent warming rate is roughly 0.35ยฐC per decade
The magnitude of the change is substantial.
From 1970 to 2015, global temperature increased at an average rate slightly below 0.2ยฐC per decade.
Over approximately the past ten years, the estimated rate has climbed to around 0.35ยฐC per decade, although the precise value varies somewhat depending on which observational dataset is used.
That does not mean Earth will necessarily continue warming at exactly 0.35ยฐC per decade indefinitely.
Climate trends measured over relatively short periods can still change as natural variability, greenhouse-gas emissions, aerosols, and other influences evolve.
But the analysis indicates that the recent acceleration is statistically distinguishable from the slower trend that dominated the previous decades.
The study detects acceleration โ not its cause
One important limitation is that the research was designed primarily to answer a statistical question:
Has the underlying rate of global warming increased?
It was not designed to determine precisely why the acceleration has occurred.
The researchers therefore do not attribute the recent change to a single physical cause.
Climate models, however, are capable of producing periods in which the rate of warming increases, meaning an acceleration is not inherently inconsistent with current understanding of the climate system.
A range of factors can influence how quickly surface temperatures rise over particular periods, including greenhouse-gas concentrations, aerosol pollution, ocean heat uptake, clouds, changes in atmospheric circulation, and natural climate variability.
Determining the relative contributions of these factors requires separate attribution studies.
The new work instead establishes evidence that the warming rate itself has changed significantly.
Why filtering natural variability matters
Detecting an acceleration in global warming is statistically difficult because Earth's climate system contains so much year-to-year variation.
A strong El Niรฑo can temporarily push temperatures upward, while volcanic aerosols or La Niรฑa conditions can pull them downward.
If researchers analyze the raw temperature series without accounting for these effects, short-term fluctuations can obscure gradual changes in the long-term trend.
Removing known sources of variability reduces that interference.
The approach does not erase climate events from history. Instead, it estimates how much those temporary influences contributed to global temperatures so scientists can examine the underlying warming trajectory more clearly.
After this adjustment, Foster and Rahmstorf found that the evidence for acceleration exceeded the 98% confidence level.
What the faster rate could mean for 1.5ยฐC
The findings also have implications for the 1.5ยฐC warming threshold associated with the Paris Agreement.
The agreement seeks to hold the increase in global average temperature well below 2ยฐC above preindustrial levels while pursuing efforts to limit warming to 1.5ยฐC.
A single year exceeding 1.5ยฐC does not mean that the long-term Paris temperature threshold has permanently been crossed. Scientists generally distinguish temporary annual exceedances from a sustained long-term warming level.
But if the warming rate observed during the past decade were to continue, Rahmstorf says a long-term exceedance of 1.5ยฐC could occur before 2030.
That would make the rate of future emissions reductions increasingly important.
Future warming still depends on emissions
Although the recent acceleration is already visible in observational data, its future trajectory is not predetermined.
The amount of additional warming ultimately depends strongly on how much more carbon dioxide and other greenhouse gases humanity releases.
Carbon dioxide emitted by burning coal, oil, and natural gas accumulates in the atmosphere and contributes to Earth's long-term energy imbalance.
Reducing net carbon dioxide emissions toward zero would eventually halt the additional warming caused by continued COโ accumulation.
As Rahmstorf emphasized, how rapidly Earth continues warming therefore depends substantially on how quickly global fossil-fuel carbon dioxide emissions decline.
The new findings show that understanding that trajectory is becoming increasingly urgent.
For decades, global temperatures rose at a relatively steady long-term pace beneath large natural fluctuations.
Now, after those fluctuations are statistically stripped away, the underlying trend appears to be changing.
Earth is not simply continuing to warm.
According to the new analysis, the warming itself is accelerating.
Journal reference
Grant Foster and Stefan Rahmstorf. โGlobal Warming Has Accelerated Significantly.โ Geophysical Research Letters, 2026, 53(5), e2025GL118804. The DOI and article details were verified against the official PIK publication record and PIK's direct journal listing.