SerpApi.Org
For a few years in the early 2020s, something unusual appeared to be happening to the Arctic's winter sea ice: the steady, decades-long decline seemed to be slowing down. Satellite measurements showed that the trend was losing its statistical power. Some researchers raised the cautious but striking possibility that wintertime ice — not summer, which gets all the attention, but winter — had found a temporary floor.
That apparent stabilization is now definitively dismantled.
A new study published Monday in the Proceedings of the National Academy of Sciences (PNAS), led by researchers at the University of Southampton and the National Oceanography Centre (NOC), finds that two consecutive record-breaking winters have put the question beyond statistical doubt. Once the 2025 and 2026 data are folded into the 47-year satellite record, the 20-year trend in Arctic winter sea ice extent is again showing statistically significant decline. The pause was never a genuine recovery. It was natural variability masking a trend that never stopped.
"What looked like a pause in the early 2020s now appears to have been a temporary slowdown within a longer-term decline associated with global warming," said Dr. Duo Chan, lead author of the paper from the University of Southampton's School of Ocean and Earth Science.
Record Drop of Nearly Six Percent in a Single Winter
The number that anchors the study is extraordinary. Between the winter of 2024 and the winter of 2025, Arctic sea ice extent during the peak phase — roughly February through March — fell by 5.8%, the largest year-to-year wintertime drop in the entire 47-year satellite record.
That translates to a concrete measurement: on March 22, 2025, Arctic sea ice reached its winter maximum extent for the year at 14.33 million square kilometers, the lowest winter maximum since satellite monitoring began in 1978. The previous low had been set on March 7, 2017, at 14.41 million square kilometers. The 2025 maximum beat that by 80,000 square kilometers — an area roughly the size of South Carolina.
Then, in 2026, winter sea ice recovered slightly but remained the second-lowest extent ever recorded. Two consecutive winters at or near all-time lows — back to back — were sufficient to decisively shift the statistical weight of two decades of satellite observation. The 20-year trend that had weakened to statistical insignificance came roaring back.
"This sharp drop changes the way we interpret recent Arctic winter sea ice change," Dr. Chan said.
"This new record low is yet another indicator of how Arctic sea ice has fundamentally changed from earlier decades," said Walt Meier, a senior research scientist at the National Snow and Ice Data Center, in a separate statement on the 2025 maximum.
What Satellite Monitoring Actually Measures
The study's central tool is a 47-year record of Arctic sea ice extent — defined as the area where satellite passive-microwave sensors detect at least 15% ice cover. Monitoring began with the Nimbus-7 satellite, jointly operated by NASA and NOAA, and has continued through successive Defense Meteorological Satellite Program satellites.
The researchers divided each seasonal cycle into four phases: preconditioning (August through October), growth (November through January), peak (February through March), and decline (April through July). Their analysis showed that while the 2025 and 2026 cycles did not begin from the lowest late-summer starting points, both stood out sharply because their growth-phase and peak-phase extents lay largely outside the historical envelope of the full satellite record.
This is the key statistical insight: the 20-year trend diagnostic requires enough data points to distinguish a genuine long-term signal from random year-to-year variation. When several years of anomalously high natural variability are added to a thinning dataset, even a real trend can temporarily lose statistical significance. The early 2020s pause was not a sign of ice recovery — it was an artifact of exactly this masking effect.
To verify that the 2025 drop was not an irreversible threshold crossing — a sign of something fundamentally new rather than an unusual fluctuation within a warming trend — the research team analyzed CMIP6 climate model simulations. They searched for analogue events: simulated years in which a sharp single-winter drop comparable to 2025 occurred under warming levels similar to what the Arctic is experiencing today. Such events exist in the model record, though they are rare.
"By identifying these analogous events, we found that such a sharp drop in 2025 was unusual, but physically plausible under current Arctic warming," said Dr. Alessandro Silvano, co-author at the University of Southampton.
Why Winter Ice Matters Beyond the Headlines
Arctic winter sea ice receives far less public attention than the summer minimum — the annual "shrinkage" measurement reported each September. But winter ice performs a structural function that summer open water cannot replicate.
The ice cover acts as an insulating lid between the relatively warm Arctic Ocean below and the frigid winter atmosphere above. When that lid thins or shrinks, more ocean heat escapes into the air — altering pressure patterns, influencing storm development, and affecting the large-scale atmospheric circulation that connects the polar region to weather systems across the Northern Hemisphere.
The ice-albedo feedback compounds the problem. Sea ice is highly reflective, bouncing incoming solar radiation back into space. Dark open ocean, by contrast, absorbs that energy, warming the water and accelerating subsequent ice loss. The Arctic is already warming nearly four times faster than the global average — the phenomenon known as Arctic amplification. Reduced winter ice cover feeds directly into this feedback.
An important nuance: the September 2025 Arctic summer minimum — which came after the record-low winter — reached 4.60 million square kilometers, making it only the 10th lowest on record. The record-breaking winter was not followed by a record-breaking summer loss. This is consistent with the CMIP6 model findings, which show that dramatic single-winter drops do not automatically trigger corresponding summer extremes.
But there is a separate and potentially more significant metric: while winter extent recovered slightly from 2025 to 2026, total Arctic sea ice volume in March 2026 hit its lowest point on record, approximately 15% lower than mid-March 2024. Thinner ice is more vulnerable ice — and volume matters for how much ice survives the summer melt season.
Near-Term Outlook: Lower, Longer
The CMIP6 analogue analysis does more than confirm that the 2025 drop was plausible — it also provides a statistical window into what typically follows.
"The model analogues suggest that, under current levels of Arctic warming, Arctic winter sea ice is more likely to fluctuate around a lower level over the next few years than quickly rebound to early 2020s conditions," said Professor Simon Josey, co-author of the study at the National Oceanography Centre.
In other words: the early 2020s apparent plateau is unlikely to return. The new baseline is lower, the year-to-year volatility is higher, and the long-term direction remains firmly downward.
The IPCC Sixth Assessment Report assessed with high confidence that Arctic sea ice area will likely drop below 1 million square kilometers in at least some September minimum extents before 2050, under the five illustrative emissions scenarios it examined. The 2025 and 2026 winters are consistent with that trajectory — and with the IPCC's separate finding that summer sea ice loss does not exhibit irreversible hysteresis, meaning it can recover from near-ice-free summers if climate conditions allow, but that the threshold for such extreme events will be crossed more frequently with each fraction of a degree of additional warming.
Short-Term Pauses Will Happen Again — That Is Now the Point
Dr. Chan's closing observation carries an implication that extends beyond this specific study: "Short-term recoveries can temporarily obscure the longer-term direction of Arctic sea ice change. The low winter sea ice extents in 2025 and 2026 clearly show that Arctic sea ice loss has not stopped — it is continuing in a warming climate," he told the journal.
The mechanism that produced the early 2020s apparent pause — natural variability in ocean heat transport, atmospheric pressure patterns, and the interplay of the Atlantic and Pacific oceans with the Arctic basin — has not disappeared. Under a more volatile, lower-baseline Arctic, those same mechanisms will temporarily mask the trend again. The statistical signature of the 20-year trend test will again weaken in some future run of years when natural variability runs against the warming signal.
What has changed is the context in which those future pauses will be read. The early 2020s pause occurred when the idea of genuine Arctic stabilization was scientifically defensible enough to prompt careful journal articles. The 2025 and 2026 winters have closed that window. A future apparent pause now arrives in a scientific record that includes two back-to-back winters that shattered records and restored a trend — meaning the prior history is available to place any future pause in its proper context.
The study was funded by a UKRI Future Leaders Fellowship and the Natural Environment Research Council (NERC). It was submitted to PNAS on April 22, 2026, accepted June 1, 2026, and published July 20, 2026.
Frequently Asked Questions
What does it mean that the Arctic sea ice "pause" was not real?
Between roughly 2019 and 2024, the standard statistical test for detecting a meaningful trend in Arctic winter sea ice extent — measuring 20 years of data against natural variability — weakened to the point of losing significance. Scientists were right to report this as a genuine finding based on the data available at the time; it was not fabricated or exaggerated. What the 2025 and 2026 winters revealed is that the weakening was an artifact of natural variability temporarily running against the warming signal, not a sign that warming had paused or reversed. The statistical test requires enough data to distinguish signal from noise — when noise dominates a stretch of the record, even a real trend can temporarily vanish from the test results.
Why is winter sea ice important if the summer Arctic gets more attention?
Summer sea ice extent is the most dramatic indicator of Arctic warming because it shows the annual minimum — the point of greatest exposure. But winter sea ice serves a physical function summer open water cannot: it acts as an insulating barrier that suppresses the transfer of heat and moisture from the ocean to the frigid winter atmosphere. When that barrier shrinks, the heat escapes — altering pressure patterns, disrupting atmospheric circulation, and potentially affecting weather systems across the Northern Hemisphere at lower latitudes. A record-low winter maximum also sets a lower starting point from which the summer melt season begins.
Will the Arctic appear to stabilize again, even if the long-term decline continues?
Almost certainly yes. The natural variability mechanisms that produced the early 2020s apparent pause — ocean heat transport fluctuations, atmospheric pressure patterns, the interaction of Atlantic and Pacific ocean cycles with the Arctic basin — are still operating. In a more volatile, lower-baseline Arctic, those mechanisms will again temporarily run against the warming signal and weaken the 20-year trend statistic. The difference is that future apparent pauses will arrive in a scientific record that now includes the 2025 and 2026 winters, making it much harder to interpret a short-term apparent stabilization as evidence of genuine recovery. The CMIP6 model analogues used in the new PNAS study confirm that the ice is more likely to fluctuate around a lower new baseline than to rebound to early 2020s levels.
Does satellite sea ice extent fully capture what is happening to the Arctic ice pack?
Extent — the area with at least 15% ice cover — is the most widely used metric because it has been measured consistently since 1978. But it does not capture thickness or total volume. International Cryosphere Climate Initiative data shows that total Arctic sea ice volume in March 2026 was the lowest on record, approximately 15% below mid-March 2024, even as extent recovered slightly from the 2025 record low. Thinner ice melts more rapidly in summer and provides less insulating capacity in winter. Volume may be the more significant long-term indicator — and its trajectory in 2026 is, if anything, more alarming than the extent data alone.
