AMOC Collapse: How Will We Know When the Atlantic Is Approaching a Tipping Point?

Oct 9, 2026 · @Paul

By Paul Harrison, Orbis Weather

Scientists are getting better at identifying the warning signs, but there may not be a simple countdown before one of Earth’s most important ocean circulation systems undergoes a profound change.

The Atlantic Meridional Overturning Circulation, or AMOC, carries warm water north through the Atlantic and keeps northwest Europe several degrees milder than its latitude would suggest. Climate scientists agree it is very likely to weaken this century. They disagree about whether, and when, it could cross a threshold beyond which it shuts down for centuries.

The harder question is whether anyone would see that threshold coming. Researchers now watch for it in several ways: statistical tremors in ocean temperature records, the flow of fresh water at the southern edge of the Atlantic, a cold patch south of Greenland, and direct measurements from instruments strung across the ocean. Each has given cause for concern. None gives a date.

“We are never going to get a [fully] reliable ‘early warning’,” Stefan Rahmstorf of the Potsdam Institute for Climate Impact Research told Carbon Brief this year. A UK government research agency is nevertheless funding an attempt to build one, with a prototype due by 2030.

Key facts

What the AMOC is and why it matters

The AMOC is a conveyor of heat. Warm, salty water flows north near the surface of the Atlantic. In the seas around Greenland and Iceland it gives up its heat to the air, becomes cold and dense, sinks, and flows back south along the ocean floor.

Salt is what makes it work. Salty water is denser than fresh water, so it sinks more readily when it cools. Anything that freshens the far North Atlantic, such as more rain, melting sea ice or meltwater from Greenland, makes the surface water lighter and the sinking weaker.

That creates a feedback. A weaker circulation brings less salty water north, which makes the northern water fresher still, which weakens the circulation further. It is this self-reinforcing loop that gives the AMOC the potential to tip.

It is not the Gulf Stream

The two are often confused. The Gulf Stream is a wind-driven current along the U.S. East Coast. It is part of the route the AMOC’s warm water takes, but it would keep flowing in some form even if the overturning stopped.

It has shut down before

The geological record shows the circulation can change state. Studies of shells in sea-floor sediment found that far less surface water reached the deep ocean during the Younger Dryas, a cold period that began about 12,900 years ago and lasted about 1,300 years. The oceanographer Wallace Broecker warned in the journal Nature in 1987 that human-caused warming could disrupt it again. “We play Russian roulette with the climate,” he said.

What a tipping point means here

A tipping point is a threshold past which a system keeps changing on its own, even if the push that started it stops.

For the AMOC the idea goes back to 1961, when the oceanographer Henry Stommel showed with a simple model that the circulation could have two stable states, one strong and one weak. Push it far enough and it flips from the first to the second. Getting it to flip back is much harder than tipping it in the first place.

Three points are easy to miss.

• Weakening is not collapse. A gradual decline of a third or a half by 2100 is the mainstream expectation and would have serious effects of its own. Collapse means settling into the weak state, at roughly 3 to 6 sverdrups against about 17 today. Niklas Boers of the Potsdam Institute calls the exact definition a “matter of convention”.
• Tipping is not instant. The change scientists have in mind “happens over decades or maybe a century”, said Laura Jackson of the UK Met Office. The threshold could be crossed long before the circulation actually winds down.
• Crossing may be invisible. Because of that delay, the point of no return and the visible collapse are different moments. “We might be committed to centuries of change without knowing it,” Paul Holland of the British Antarctic Survey told New Scientist.

The warning signs scientists watch

There is no single gauge. Researchers look at five kinds of evidence, each with its own weakness.

The statistical signals

The mathematics of tipping points predicts a tell-tale sign. As a system loses stability it takes longer to settle after each disturbance, and its fluctuations grow. Scientists call this critical slowing down. A 2021 study in Nature Climate Changelooked for it in long records of Atlantic temperature and salinity and found what it described as strong evidence that the circulation is approaching such a transition. A 2023 study in Nature Communications went further and extrapolated a date range.

The physical signals

A group at Utrecht University in the Netherlands has argued for indicators grounded in ocean physics. Their 2024 paper in Science Advances tracked the transport of fresh water at the Atlantic’s southern boundary and concluded the circulation is on course to tip. A 2025 paper in the Journal of Geophysical Research: Oceans estimated that collapse could begin as early as 2063 under a medium-emissions pathway, and put the threshold near 2.5°C of global warming. A 2026 study pointed to abrupt shifts in the path of the Gulf Stream as another signal.

“The risk of reaching the tipping point… is too high to ignore,” René van Westen, one of the Utrecht authors, told New Scientist.

The meltwater

Much of the uncertainty lies in Greenland. Kelly Hogan of the British Antarctic Survey told New Scientist that about 200 fjords there are now releasing icebergs and meltwater about five times faster than three decades ago. Her colleague Pierre Dutrieux described the retreat of fjord glaciers as “an unstable, self-sustaining process”, which makes it hard to model. Icebergs also drift thousands of kilometers before they melt, so the fresh water does not enter the ocean where it left the land.

How the AMOC is measured today

Direct measurement began only in 2004, which is the root of much of the disagreement.

RAPID, at 26.5°N

A line of moored instruments runs across the Atlantic from Florida to the Canary Islands. It has measured the overturning continuously since 2004. “RAPID was designed specifically to be close to the maximum of that heat transport,” said Ben Moat of the UK National Oceanography Centre.

• The average strength was 18 to 19 sverdrups in 2004 to 2008 and 15 to 17 in 2011 to 2020.
• A 2025 study put the decline over 2004 to 2023 at about 1 sverdrup a decade, similar to the pace climate models project for the rest of the century.
• Much of that fall came as a step around 2008 to 2009, not as a steady slide, according to a 2026 analysis.

The subpolar arrays

Farther north, the OSNAP array has measured the overturning between Canada, Greenland and Scotland since 2014. A study published in May 2026 in the journal Ocean Science combined OSNAP with ship and float data to extend that northern record back to 2004. It found no statistically significant trend over 20 years, while confirming the decline at RAPID over the same period.

So the two best direct records do not yet tell the same story.

Why 20 years is not enough

The circulation swings naturally from year to year and decade to decade. A 2023 review of the RAPID data said the record was “still too short” to separate a climate-driven trend from those swings. An earlier study estimated that about 29 years would be needed, a mark the array reaches around 2033.

Reaching back before 2004

For earlier decades scientists rely on indirect evidence. A reconstruction posted as a preprint in March 2026, built from ocean temperature and salinity observations, estimated a weakening of 16% to 23% since the mid-20th century across the whole Atlantic. It has not yet been peer reviewed. The IPCC, in its last assessment, had low confidence in estimates of 20th-century weakening.

What recent studies say, and why they disagree

Published estimates range from a collapse beginning within a few decades to no collapse this century at all.

Summaries from Carbon Brief, Yale Environment 360 and ArcticToday.

Why the answers differ

• They measure different things. Some look for statistical signals in temperature records. Others run climate models forward. The two approaches have different blind spots.
• They define collapse differently. A study can find a 50% weakening and still conclude there was no collapse.
• They stop at different dates. “The most interesting AMOC dynamics happen after 2100, but most [climate model] simulations are terminated at 2100,” van Westen said.
• The models may be too stable. Many climate models are too fresh in the South Atlantic, a bias that can make the simulated circulation harder to tip than the real one. Correcting for it was the main reason the 2026 Science Advances study projected a larger slowdown.

The official assessment

The IPCC’s 2021 report said the AMOC will “very likely decline” this century under every scenario, and that it had “medium confidence” there would be no abrupt collapse before 2100. It had low confidence in the numbers. Most of the studies above were published after that report.

Why there may be no simple countdown

The warning signs can show that risk is rising. They cannot, so far, be turned into a date. Scientists give five reasons.

1. The records are too short and too indirect. Twenty-two years of direct measurement cannot yet separate a trend from natural swings. Longer records use surface temperature as a stand-in for a current that runs kilometers deep. Rahmstorf says the data are simply too limited for a fully reliable warning.

2. Statistical signals indicate direction, not timing. Niklas Boers, whose 2021 study found evidence of an approaching transition, is blunt about what that evidence can do: “So, early warning in terms of a prediction? No way. It just doesn’t work.”

3. The threshold itself may be blurred. An AMOC collapse has never been observed with instruments. “It may be that there is a very definite threshold in the system,” Jon Robson of the University of Reading told New Scientist, but it is also possible that there is not one sharp line. A sudden shock, such as a pulse of fresh water from Arctic sea ice, could also bring a transition forward.

4. The point of no return and the collapse are decades apart. The circulation responds slowly. A June 2026 preprint led by Philip Holden of the Open University estimated a 10% to 23% chance that collapse is already committed, depending on how much meltwater Greenland releases. “There is a significant probability that we’re already committed to collapse, and we can’t change that even now,” he told New Scientist. That study has not been peer reviewed.

5. The biggest driver is the hardest to forecast. How fast Greenland’s glaciers will shed ice, and where that fresh water will go, is poorly constrained.

What a warning system could still do

The UK’s Advanced Research and Invention Agency, ARIA, is funding an attempt to build an early-warning system for climate tipping points, with a prototype planned for 2030. As described to New Scientist, it would not give a date. It would publish a monthly risk level on a five-step scale, from no elevated threat to a tipping point crossed, alongside forecasts of what adaptation would then be needed.

The program includes new measurements at the source. The research ship Sir David Attenborough is carrying instruments, an autonomous submarine and an uncrewed surface vessel to Greenland’s fjords to measure meltwater and iceberg calving directly.

The model is closer to a storm watch than a countdown clock: a statement that conditions are becoming dangerous, issued without knowing exactly when or whether the event will arrive.

What a collapse or sharp weakening would do

The effects would reach far beyond the North Atlantic. “You’re getting a whole-system shock,” van Westen said.

European summers would stay nearly as hot as today, so the continent would face colder winters and hot, dry summers together. “The winter is like being in the ice age,” Tim Lenton of the University of Exeter said of Scotland in such a scenario.

The IPCC’s wording is that a collapse “would very likely cause abrupt shifts in the regional weather patterns and water cycle”.

A weakening short of collapse would produce smaller versions of the same effects. The 2026 Science Advances study notes that every sverdrup lost by 2100 removes about 0.05 petawatts of northward heat transport.

What is known, what is uncertain, and what to watch

Known

• The AMOC can exist in a strong state and a weak one, and has switched in the past.
• It is very likely to weaken this century.
• Direct measurements at 26.5°N show a decline since 2004.
• A collapse would be severe and, on human timescales, irreversible.

Uncertain

• Whether the measured decline is a climate trend or natural variability.
• How close the circulation is to a threshold, and whether that threshold is sharp.
• Whether climate models are too stable.
• How much fresh water Greenland will add, and how fast.
• Whether the threshold has already been crossed.

What to watch

• Around 2033: the RAPID record becomes long enough, by one estimate, to distinguish a forced trend from natural swings.
• By 2030: ARIA’s prototype early-warning system.
• The northern arrays: whether the subpolar record begins to show the decline seen farther south.
• Peer review of the June 2026 “already committed” study and the March 2026 reconstruction.
• Governments: Iceland designated a possible AMOC shutdown a national security threat in November 2025. In April 2026 its Climate Council urged faster emission cuts, citing projections of roughly 50% weakening by the end of the century.
• The next IPCC assessment, which will weigh the studies published since 2021.

What can be done

The risk rises with warming. The 2025 Utrecht study placed the threshold near 2.5°C, and model runs show shutdowns becoming far more common at high emissions. “Let’s do everything in our power to get to net-zero as quickly as possible,” Lenton said.

About this article

This article is an explainer on how scientists look for warning of an AMOC tipping point. It draws on Carbon Brief’s review of the research published in April 2026, on New Scientist’s reporting from September 2026, on a 2026 paper in Ocean Science, and on summaries of other recent studies. It was written on October 9, 2026.

Terms used

Notes on the evidence

• Two studies cited here are preprints and may change: the June 2026 estimate that collapse may already be committed, and the March 2026 reconstruction of weakening since 1960.
• Dates are onsets, not endpoints. A study that says collapse “could begin” in a given decade is describing the start of a process that takes decades more.
• Percentages of weakening use different baselines. Some compare with 1850 to 1900 and others with recent decades.
• Impact figures come from model experiments in which the circulation is made to collapse. They show what is possible, not what is forecast.

Limits

• Most studies are described here from summaries by Carbon Brief and other outlets. The original papers were not all reviewed directly.
• The science is moving quickly. Several findings are less than a year old and have not yet been tested by other groups.
• This article does not assign a probability to collapse. The published estimates are too far apart to combine.

Frequently asked questions

What is the AMOC?

It is the system of Atlantic currents that carries warm surface water north and returns cold water south at depth. It moves heat toward Europe and affects rainfall patterns around the world.

Is the AMOC collapsing now?

No collapse has been observed. Measurements show it has weakened at one latitude since 2004, and most scientists expect it to weaken further this century. Whether it is close to a tipping point is disputed.

When could it collapse?

Published estimates range from the 2030s to after 2100, and some studies find no collapse this century. The IPCC has medium confidence that an abrupt collapse will not happen before 2100.

How would we know it was about to happen?

Scientists watch for statistical changes in ocean records, shifts in freshwater flow in the South Atlantic, the cold patch south of Greenland and direct measurements of the current. These can show rising risk. Leading researchers say they cannot give a reliable date.

Could the tipping point already have been passed?

Possibly. One 2026 study, not yet peer reviewed, put the chance at 10% to 23%. Because the circulation responds slowly, a threshold could be crossed decades before the effects are obvious.

Is the AMOC the same as the Gulf Stream?

No. The Gulf Stream is a wind-driven current off the U.S. East Coast and is one part of the system. It would not stop if the overturning collapsed.

Would Europe freeze?

Northern Europe would become much colder in winter, by 5 to 15°C in one study. Summers would stay hot. The Hollywood version of an overnight ice age is not what the science describes.

How would it affect the United States?

The main effects would be faster sea level rise on the East Coast and changes in storm tracks and rainfall.

Can a collapse be prevented?

The risk rises with global temperature, so limiting warming lowers it. One study put the threshold near 2.5°C of warming.

Has it happened before?

Yes. The circulation weakened sharply about 12,900 years ago, during a cold period called the Younger Dryas that lasted about 1,300 years.

Sources

• Carbon Brief: AMOC: Is global warming tipping key Atlantic ocean currents towards “collapse”?, April 24, 2026
• New Scientist, as reposted by GeoGarage: The collapse of the AMOC would be catastrophic. How will we know when it’s coming?, September 28, 2026
• Ocean Science: The Scotland–Canada overturning array (SCOTIA): twenty years of meridional overturning in the subpolar North Atlantic, May 8, 2026
• Yale Environment 360: Collapse of Atlantic currents may already be “locked in”, July 7, 2026
• ArcticToday: AMOC collapse may already be locked in, new study suggests, July 8, 2026
• Ocean to Climate: observational constraints project a 50% AMOC weakening, April 16, 2026, on Portmann and colleagues in Science Advances
• Ocean to Climate: a reconstruction of the AMOC since 1960, March 13, 2026, on a preprint by Li and colleagues
• Annual Review of Marine Science: The probability of an AMOC collapse onset in the twenty-first century, Dijkstra and van Westen, 2026

Orbis Weather compiled this article from the public sources above.