On the bridge of the RRS Discovery, Captain Antonio Gatti scans the horizon for a surfacing RAPID mooring.
NATIONAL OCEANOGRAPHY CENTRE
From Sciences by Paul Voosen Sarah Crespi, Kevin McLean, Paul Voosen
After decades of warnings, new data suggest the Atlantic’s vital circulation may withstand climate warming better than feared 11 Jun 2026
Off the coast of the Canary Islands—In calm waters here off the northwestern coast of Africa, the crew of the RRS Discovery, a U.K. research ship, was scanning the horizon, waiting for a sentinel to return from the deep.
An acoustic ping had triggered the release of a mooring holding 2 years of precious ocean measurements from its anchor 2000 meters below.
More than 20 minutes had elapsed, and there was still no sign of the bright orange float that would lift the mooring to the surface.
But Ben Moat, the cruise’s chief scientist and an oceanographer at the United Kingdom’s National Oceanography Centre (NOC), wasn’t worried.
He had been here before.
On the bridge, Moat glanced at a black Casio watch attached to his clipboard: 22 minutes.
There was more competition than usual to be the first to spot the float.
Moat, the captain, and the third officer were joined by NOC’s CEO, as well as several members of a U.K. TV news crew.
“Is it still off to port?” Moat asked, peering through binoculars for a mote of orange against a sea of azure.
The crowd on the bridge reflected the importance of the mooring, one of 10 in a vital climate observatory called the RAPID array.
For more than 2 decades, RAPID’s instrument-packed moorings, spaced across the Atlantic Ocean at 26°N between the Bahamas and the Canary Islands, have monitored the changing strength of ocean currents called the Atlantic Meridional Overturning Circulation, or AMOC.
The currents usher tropical waters and heat to the northeastern Atlantic, allowing cabbage palms to flourish in Ireland and keeping Norwegian ports ice-free in winter.
As the waters move north, they cool and become saltier as sea ice forms and rejects brine.
The resulting cold, salty water becomes dense enough to sink to the abyss, carrying heat and carbon dioxide down with it.
The water returns south along the floor of the Atlantic, heading to the Southern Ocean and beyond.
Climate models have long warned that global warming could weaken “deep-water formation”—the density-driven sinking that is the engine of the AMOC.
The logic is straightforward: As Greenland’s ice sheets melt and sea ice formation declines, North Atlantic waters will freshen.
Combined with warmer sea temperatures, the freshening makes surface waters more buoyant.
The AMOC was thought to have shut down abruptly during past climate warmings, and a handful of researchers now argue such a tipping point could occur this century.
A sputtering AMOC could trigger a sharp cooldown in northwestern Europe, rising seas along the U.S.
east coast, and shifts in tropical rainfall.
“It is a risk that would really have severe impacts,” says Stefan Rahmstorf, a climate scientist at Potsdam University and a prominent voice warning of the threat.
Yet for all the alarming headlines, most climate researchers think the AMOC is more resilient than these worst case scenarios make it seem.
Emerging evidence suggests the AMOC may not have actually collapsed in the warm climates following ice ages.
More detailed climate models suggest it could weaken but not collapse in the current surge of warming.
And studies of the AMOC’s present behavior do not yet show any clear signs of trouble.
They’re also exposing new facets of the circulation that could buffer any eventual weakening.
“The paradigm has been, if we warm and freshen these areas, we’ll get less dense water and AMOC will slow down,” says Susan Lozier, an oceanographer at the Georgia Institute of Technology.
“That paradigm isn’t holding up.”
By retrieving the RAPID moorings and harvesting their data, the cruise to the Canaries in February should sharpen the picture.
Back on the Discovery bridge, 25 minutes after the mooring’s release, experience won the informal competition to spot it.
“There it is,” Moat announced.
The orange float bobbed several hundred meters off the port bow, with a ribbon of smaller yellow floats and instruments trailing behind it.
From a young age, Moat was drawn to the ocean.
He grew up the son of a fisherman in Bridlington, on England’s northeastern coast, and by age 8 was piloting his father and friends to North Sea fishing grounds—standing on a box to see over the bow.
A math whiz, he later studied computational fluid dynamics and joined NOC, where he spent years at sea studying how wind stirs the ocean’s surface.
Conditions were often punishing.
“We were seeking that big bad weather,” he says.
In a brief period of calm, Ben Moat, RAPID’s chief scientist, prepares for a TV news interview in the RRS Discovery’s storage hangar.National Oceanography Centre
The circulation he now studies operates on a very different scale.
For more than a century, oceanographers had pieced together an understanding of the Atlantic’s slow, density-driven churning from shipboard measurements.
In the 1980s, famed climate scientist Wallace Broecker dubbed the system the “great ocean conveyor belt”—a vivid shorthand that stuck.
He also helped popularize the more troubling idea that this conveyor belt could collapse if large pulses of freshwater from melting ice sheets flooded the North Atlantic, disrupting the sinking of dense waters.
Evidence from ice cores and seafloor debris suggested it happened at the end of the last ice age—and perhaps more than once.
Such shutdowns, scientists proposed, could explain abrupt temperature swings seen in paleoclimate records.
Around the same time, climate models began to suggest global warming could also weaken the AMOC, not so much by melting ice, but by warming waters and keeping them buoyant.
But ship-based observations were too sparse to show any change in the AMOC.
So instead of relying on occasional ship surveys, researchers set out to continuously monitor the circulation with moorings spanning the Atlantic.
They settled on a line at 26.5°N—a transect where the overturning signal would stand out against other currents.
Backed by U.K. and U.S. funding agencies, the array was deployed in 2004.
After steaming out from Santa Cruz de Tenerife, Moat and his team got straight to work.
RAPID cruises run each year, alternating between the eastern and western boundaries of the Atlantic.
This one would be brief: Over the course of 1 week, the team would recover four moorings along the continental slope near the Canaries—like the one now bobbing alongside them—and replace them with fresh instruments.
Each mooring is studded with MicroCATs, compact sensors that measure water characteristics including temperature, pressure, and salinity.
They can be fickle, so the team had to calibrate new ones before deploying them.
The team sank the MicroCATs thousands of meters alongside a reference sensor, then retrieved them and compared the readings.
After the Discovery’s crew winched a set back up onto the deck, engineers hustled to remove them like a NASCAR pit crew, readying the winch for another submersion.
Some of the instruments were wonky “and had to be sent back for bad behavior,” said Yvonne Firing, a physical oceanographer at NOC.
But in the end, the team had 37 MicroCATs ready for their 2-year tour of duty, with one good spare ready to step in.
Fresh from the deep, each MicroCAT instrument is rushed to the lab, where 2 years of data are downloaded. P. Voosen/Science The sensors do not directly measure the currents, which creep along in many places with the speed of a tortoise, but rather the structure of the ocean itself.
From subtle changes in temperature, pressure, and salinity across the basin, scientists can estimate the “dynamic height” of the ocean—the equivalent of atmospheric pressure.
Just as winds flow from regions of high to low pressure, the tilt of these ocean heights, across the basin, leads to a calculation of flow.
Combined with other measurements, those data yield RAPID’s estimate of the AMOC’s strength.
What the currents lack in velocity they make up in flow rate, which is measured in sverdrups, millions of cubic meters per second.
A sverdrup (Sv) is roughly the combined flow of all the world’s rivers; the Amazon carries about 0.2 Sv.
RAPID measurements show the northward, surface leg of the AMOC, which flows as part of the Gulf Stream in the western Atlantic, transports about 17 Sv.
(The Gulf Stream, which is mostly driven by winds, carries 90 Sv in total.) A similar volume oozes southward above the ocean floor.
That stately flow actually swings wildly year to year, masking any long-term trend, the first RAPID measurements showed.
Swings between apparent decline and recovery have since become a hallmark of AMOC monitoring, and a recurring source of alarm and reassessment.
The Gulf Stream is one of Earth’s strongest ocean currents.
It is resilient to climate change because it is primarily driven by surface winds and the planet's spin.
But the Gulf Stream is one small part of the Atlantic Meridional Overturning Circulation (AMOC), a complex network of currents that may be vulnerable to global warming.
The AMOC redistributes heat across the North Atlantic Ocean, warming northwest Europe.
Surface flows (red) usher tropical heat north before cooling, sinking, and returning south along the sea floor (blue)
The AMOC redistributes heat across the North Atlantic Ocean, warming northwest Europe.
Surface flows (red) usher tropical heat north before cooling, sinking, and returning south along the sea floor (blue)
The AMOC is powered by a process called deep-water formation, which occurs in the North Atlantic.
Climate change could threaten it, triggering a sharp cooldown in Europe and rising seas along the U.S.
east coast.
Deep-water formation, also known as overturning, is driven by density and salinity changes of North Atlantic waters.
1
Deep-water formation begins when warm surface waters move north, cool, and get saltier from brine expelled as sea ice forms.
2
The resulting cold, salty water is denser and sinks.
3
Cool water returns south along the ocean floor.
4
But global warming, combined with the disappearance of sea ice and ice sheets, could cause waters to warm and freshen.
The resulting surface waters would be more buoyant, potentially disrupting deep-water formation.
Scientists rely on sensors suspended along moored lines at a range of depths to monitor the AMOC's strength.
One mooring array in the North Atlantic is called OSNAP.
One mooring array in the North Atlantic is called OSNAP.
|
Some data suggest that deep water can form farther north in the Arctic, perhaps fortifying the AMOC.
Farther south, at 26.5ºN, the RAPID mooring array has collected more than 20 years of data on the AMOC’s strength.
The AMOC’s flows are measured in sverdrups, millions of cubic meters per second—a unit roughly equivalent to the combined flow of all the world’s rivers.
RAPID’s estimates of the AMOC’s strength, adjusted to remove seasonal variations and the effects of wind, show that its flows swing wildly.
Scientists extract a smoothed signal from the noisy data.
Swings between periods of decline and recovery have become a recurring source of alarm and reassessment.
Whether climate change is responsible for the slight, gradual weakening of the AMOC is uncertain.
(GRAPHIC) M. Hersher and N. Burgess/Science; (DATA) Map sensors: Darren Rayner/National Oceanography Centre; Map arrows: AMOC flows are a simplified approximation developed in consultation with Susan Lozier; Chart: G. D. McCarthy et al., Geophysical Research Letters, 52, e2025GL115055 (2025)
Gerard McCarthy remembers well the first time he saw an AMOC decline.
It was 2011, and McCarthy, now a climate scientist at Maynooth University, had just joined the RAPID team.
His first task was calculating AMOC’s strength.
Beginning in 2009, it plunged.
“Everyone was like, ‘The new guy made a mistake,’” he recalls.
Others checked the numbers.
The drop held.
“We all realized that something dramatic had happened.”
What happened was not caused by climate change, but rather the weather.
That winter, unusual swings in air pressure weakened the jet stream and shifted wind patterns, disrupting the AMOC’s flow.
The decline likely contributed to a frigid European winter in 2009 and, by leaving more heat in tropical basins, also led to an active Atlantic hurricane season the following summer.
A permanently crippled AMOC would lead to bigger problems.
In the U.K., it could blunt rising temperatures, partially offsetting global warming.
Farther north, however, Scandinavia would sit “in the bull’s eye,” says Marius Årthun, a physical oceanographer at the University of Bergen: Temperatures could plunge, even as the planet warms, and sea ice could creep south from the Arctic.
Across northwestern Europe, storm tracks could shift and rainfall could falter.
The impacts would ripple far beyond Europe.
If the AMOC’s flows weaken, sea levels on the U.S. east coast could rise by up to 30 centimeters, says Denis Volkov, an oceanographer at the U.S.
National Oceanic and Atmospheric Administration’s Atlantic Oceanographic and Meteorological Laboratory.
Globally, tropical rain belts might shift south, weakening the Asian monsoon and rainfall over the Amazon.
And if the AMOC’s currents stop sequestering heat and carbon dioxide into the ocean’s deep waters, overall planetary warming could rise by an extra 0.2°C.
Freshwater melt from the Greenland Ice Sheet may pose less of a threat than scientists once feared to deep-water formation, which drives the Atlantic Meridional Overturning Circulation. Paul Souders/Getty Images
To some scientists, those dire scenarios can be a distraction.
“It is taking our attention from impacts of climate change that we are sure are happening,” says Fiamma Straneo, an oceanographer at Harvard University.
But even a weakening AMOC could have consequences, like the cold European winter McCarthy remembers.
The question is just how weak it might get.
The RAPID array isn’t alone in searching for clues.
High in the North Atlantic is a second array of moorings, called the Overturning in the Subpolar North Atlantic Program (OSNAP), spanning the ocean from Newfoundland to Greenland to Scotland.
Funded since 2014 by the U.S.
National Science Foundation (NSF) along with international partners, OSNAP was sold on the idea that its location would give a better view of the deep-water formation that drives the AMOC.
Its first 10 years of data, expected to be published soon, will show a slight increase in the AMOC’s strength—but can’t yet say much about the effect of climate change, says Lozier, its longtime leader.
Comparing RAPID and OSNAP has already led to some puzzles.
Scientists expected AMOC changes would be apparent throughout the entire circulation, so any speedup or slowdown would register at both arrays.
“We had somewhat naïve expectations,” Lozier says.
In fact there’s little coherence in the system—when one part of the AMOC weakens, another might kick up, or not respond at all.
It seems the AMOC is not a single conveyor belt, but a belt of belts, each part operating semiautonomously.
Straneo wants to ditch the conveyor belt analogy entirely.
“It has hindered our ability to imagine how these systems might evolve,” she says.
Encrusted with barnacles after 2 years in the abyss, a float reappears (first image) as it lifts a mooring to the surface.Marine organisms cover other parts of the moorings (second image).
P.
Voosen/Science
OSNAP has changed the picture in other ways, including by showing that overturning occurs not so much in the Labrador Sea, as models suggested, as it does farther north, in the Irminger and Iceland basins.
Additional data suggest deep-water formation is migrating even farther north, into the Arctic Ocean, following the retreat of sea ice, Årthun says.
“You’re expanding the reach of this cooling machine.” The northward migration could make the AMOC more resilient to warming, although the Arctic areas in which these cold, salty waters can form is not infinite, he says.
“How long can the compensation last? That’s the natural question to ask.”
OSNAP also acts as a sentinel for the AMOC’s response to injections of freshwater.
There are signs that, for the past 2 decades, the Beaufort Gyre, a loop of current in the Arctic Ocean, has been trapping freshwater from Arctic rivers, fat with thaw from permafrost.
This freshwater now seems to be migrating south into the Atlantic, a potential threat to deep-water formation, Straneo says.
But she notes that such a freshwater invasion has happened before, most famously during the “great saline anomaly” of the late 1960s, only to be flushed out of the system several years later.
Other omens come from sea surface temperature measurements, particularly those showing a “warming hole” in the subpolar Atlantic, where the ocean has cooled despite global warming.
Rahmstorf and others have argued the hole is the result of a weakened AMOC ushering less heat northward.
In 2023, Peter and Susanne Ditlevsen, sibling researchers at the University of Copenhagen, went further, reporting variability in this record that they said suggested the AMOC was nearing a tipping point within the next few decades.
Many researchers argue these warnings rest on shaky ground.
Sea surface temperatures, they note, are strongly shaped by the atmosphere and may not reliably track the AMOC.
Others found that the statistical indicators used by the Ditlevsens tend to point to AMOC collapse regardless of the underlying data.
Even so, hints of change are mounting.
In February, a study led by René van Westen, an oceanographer at Utrecht University (UU), suggested the Gulf Stream has shifted north in recent decades—a potential sign of AMOC weakness.
That’s because a weak AMOC would shrink the subpolar gyre—a broad circulation of currents south of Greenland—allowing the Gulf Stream to drift north.
With each piece of evidence, “you can quibble that it’s not AMOC,” Rahmstorf says.
“But when you have all this evidence pointing in the same direction, then it starts to become robust.”
Some climate modeling backs him up.
Research led by van Westen and UU oceanographer Henk Dijkstra has recently suggested a collapse will begin when the buoyancy of surface waters in the North Atlantic turns positive.
Climate models, even if they don’t show AMOC collapse, tend to predict changes in temperature, salinity, and other factors that would flip the buoyancy to positive by 2063.
Van Westen thinks a slow, centurylong collapse could begin then.
“We say it tips at 2.5°C,” he says.
“There’s quite a substantial risk.”
Yet such outcomes often require freshwater inputs far beyond what Greenland is expected to deliver.
New climate model runs that capture more realistic melt from the Greenland Ice Sheet are less dire.
In two preprints posted online in the past year—one led by Chuncheng Guo, a climate scientist at the Danish Meteorological Institute (DMI), the other led by Oliver Mehling, an ocean modeler at UU—researchers created multiple simulations where carbon emissions continued until 2250 and temperatures rose by up to 7°C.
In both studies the AMOC weakened, losing about 40% of its strength.
But it never collapsed.
Both studies also suggest the weakening is reversible, says Marion Devilliers, a DMI climate modeler.
Once emissions stop, she says, the “AMOC comes back.”
In a third, even more sophisticated simulation, that resilience persisted even in the face of catastrophic warming.
This modeling effort, presented in February at the Ocean Sciences Meeting, found that even if atmospheric carbon dioxide levels quadrupled, driving extreme warming, the AMOC would decline by 40% after 20 or so years—but once again, it would rebound.
The result seemed to reflect the high resolution of the model, which could resolve currents flowing through narrow straits in the North Atlantic better than earlier models.
When a low-resolution version of the model simulated the same conditions, the AMOC simply collapsed.
The past, too, is proving more complicated than once thought.
Evidence from past ice ages seemed to suggest the AMOC switched off entirely when massive pulses of freshwater from the melting of the North American ice sheet poured into the Atlantic.
But new work, also presented at Ocean Sciences, suggests the AMOC may not have collapsed at all during these periods.
Using marine cores collected off the U.S. east coast, David Thornalley, a paleoceanographer at University College London, and colleagues measured differences in the grain size of sediments to calculate the AMOC’s ancient flows.
They found that even at the end of the last ice age—seen as a likely moment for an AMOC shutdown—the Atlantic circulation didn’t change much.
If anything, Thornalley says, “it may have been stronger.”
Yet even if the classic scenario of AMOC collapse doesn’t hold up, there is still reason to worry.
In work also presented at Ocean Sciences, Thornalley and his colleagues analyzed marine cores from a time some 400,000 years ago, when the climate was slightly warmer than today’s and Greenland experienced strong melt.
Those cores show the AMOC weakened by 30%, cooling the North Atlantic region by up to 7°C.
“This is the poster child for where we might be heading,” Thornalley says.
RAPID aims to provide an early warning.
After the first mooring floated to the surface near the Discovery, Moat and the crew set about recovering it.
Captain Antonio Gatti carefully approached the chain of floats, bringing it to the ship’s starboard with a kind of parallel parking maneuver.
Precision was essential—a wrong approach could entangle the propellers in hundreds of meters of line.
With grappling hooks, the crew snared the tangled-up mooring and dragged the mess to the ship’s aft, where it was winched aboard.
Neon red crabs and alien-looking snails festooned the mooring, and the MicroCATs were slick with algae and seaweed.
After an inspection, they were sent off to the instrument bay, where 2 years of data would be immediately downloaded.
The work was no easier on the ship’s final day at the array.
In rising swells, the crew set out to recover one last mooring and a seafloor lander from 3000 meters down.
The lander came up cleanly.
The mooring did not, and the crew faced another tangle.
On deck, technicians methodically cut through the snarls with knives.
“It’s a right mess,” said John Hopley, an engineer overseeing the work.
Piece by piece, they freed the line and retrieved the instruments.
By afternoon, a replacement mooring had vanished serenely into the depths.
The ship would soon be turning back to shore.
Was Moat, visibly tired, also relieved? Not yet, he said.
“When I step off the ship.”
Tenerife and the other Canary Islands shielded the RRS Discovery from swells.National Oceanography Centre
The new AMOC calculations from the recovered data won’t be released until September.
But the circulation is beginning to reveal a pattern.
Late last decade, it surged, erasing the decline RAPID initially measured.
In the past few years, however, the readings have slipped again, suggesting an overall decline of nearly 2 Sv since 2004.
Some of that drop appears in the cold deep waters flowing south in the western Atlantic, though it has been partially offset by strengthening southward currents near the Canaries.
Meanwhile, strong warming in the western Atlantic has caused some northward flows to slow, though not the Gulf Stream.
This weakening is consistent with model predictions, but it is not yet statistically significant, nor clearly tied to global warming, Moat says.
Another decade of measurements is needed to make such a call.
By then, the array may be able to rule out the most extreme scenarios of imminent collapse this century, Moat and others reported in a study published last year in Geophysical Research Letters.
For now, Moat is wary of terms such as collapse and shutdown.
“Shutdown,” he says, “is not meaningless, but it’s not a good term.”
RAPID’s U.K. funding is secure until 2029, but OSNAP’s funding ends in January 2027, and hopes are dwindling that NSF will renew it.
Just last week, NSF said it would remove several OSNAP moorings in the Irminger Sea as part of cuts to the agency’s wider Ocean Observatories Initiative.
Efforts are underway to get the measurements more cheaply.
This month, the RAPID team is testing an experimental mooring off the Bahamas that can transmit its data to ships or autonomous submersibles.
Others are exploring whether satellite altimetry, which measures subtle sea-surface topography, combined with autonomous Argo floats can reproduce RAPID measurements.
And NOC is investigating whether subsea telecom cables could double as sensors.
“Eventually, we would move away from the heavy engineering of moorings,” Moat says.
Meanwhile, climate scientists are still struggling to describe the AMOC’s future.
At a U.K. workshop this month titled “Collapse of the Atlantic Ocean circulation: Can it? Has it? Will it?” researchers debated the evidence.
A broader assessment, commissioned by the European Union and modeled on United Nations climate reports, is now underway.
Most agree the media-stoked drama of collapse has outpaced the evidence.
“People may be inclined to be more inflammatory,” says Eleanor Frajka-Williams, a physical oceanographer at the University of Hamburg who previously led RAPID.
But that risks oversimplifying a system that resists easy narratives.
“There’s a reason we don’t have a unified simple theory for the AMOC,” she adds.
“It doesn’t exist.”
Aboard the Discovery, the crew continued to plumb the currents.
Recovered MicroCATs were lowered back into the water to verify their measurements, another set of checks in a project built on patience and repetition.
Between shifts came dart games in the bar, sunrises at breakfast, and the rare clear night when the clouds parted and the sky opened wide.
Beneath it, the ship motored on in the darkness, a small point on a vast, moving ocean that still hid its secrets.
