Atlantic meridional overturning circulation

• 1. Introduction

  ›

  The Atlantic meridional overturning circulation moves warm water north through the Atlantic Ocean and returns cold water south along the deep sea floor. In October 2024, forty four climate scientists signed an open letter warning that the risk of its collapse has been badly underestimated. They said a collapse could happen within the next few decades, with severe effects especially for Nordic countries. The scientists urged those nations to hold to the Paris Agreement to help prevent it. What exactly keeps this circulation running, and why would losing it reshape the climate of an entire continent? How do scientists even know whether it is slowing down, and could it really stop?

• 2. How The Conveyor Works

  ›

  In the Nordic Seas, surface water that has grown salty and cold sinks into the depths. This sinking forms what oceanographers call the North Atlantic Deep Water. The process draws on named regional water masses: the Denmark Strait Overflow Water, the Iceland-Scotland Overflow Water, and the Nordic Seas Overflow Water. Labrador Sea Water was once thought to feed this system too, but growing evidence suggests it mostly circles within the North Atlantic Gyre instead. Beneath it all sits the Antarctic Bottom Water, the coldest, densest layer in any Atlantic basin deeper than 4,000 meters, its upper reaches melding into the sinking layer above.

  Off the coasts of northwest and southwest Africa, wind-driven Ekman transport pulls water back toward the surface. This happens in the Canary Current and the Benguela Current, balancing the sinking happening in the Nordic Seas. Upwelling runs stronger at the Canary Current today, though the pattern ran the other way before the Central American Seaway closed in the late Pliocene. Farther east, upwelling depends more on sea surface temperature than on wind. It shows up only during certain months, loosely syncing with the El Nino and La Nina cycle.

  About 80 percent of the North Atlantic Deep Water eventually rises back to the surface in the Southern Ocean. This links the system to the separate Southern Ocean overturning circulation. Water that resurfaces near Antarctica gets chilled by sea ice and sinks again, some of it rejoining the Antarctic Bottom Water and some traveling on toward the Pacific and Indian Oceans. Water that resurfaces at lower, ice-free latitudes drifts north instead. It eventually returns to the Atlantic past the coast of Africa and through the Indonesian archipelago, where it cools, grows dense, and sinks once more.

  The Pacific Ocean receives so much rainfall that its surface water cannot sink past several hundred meters. That is why deep water formation simply does not happen there. Trade winds carry Atlantic moisture across Central America before it can fall back as rain, and mountain ranges including the Andes and the Rockies block any return flow to the ocean. The saltier water this leaves in the Atlantic does more than drive a current. Wherever it rises again, it also carries dissolved oxygen, carbon, and nutrients that ocean ecosystems depend on.

• 3. Climate's Thermostat

  ›

  Heat radiating from the equator moves toward the poles through the oceans, sometimes north and sometimes south. The Atlantic is the only ocean where that heat flow always runs north. Much of that transport rides on the Gulf Stream, a surface current that carries warm water north from the Caribbean, driven by wind alone. Its northern arm, the North Atlantic Current, instead draws heat from the AMOC's thermohaline exchange. Together this delivers up to 25 percent of the total heat carried into the northern hemisphere.

  There is a consensus that the AMOC keeps northern and western Europe between 4 and 10 degrees Celsius warmer than it would be otherwise, depending on location. It is generally considered incorrect that Europe would be as cold as northern North America without this transport, since atmospheric patterns matter too. One modeling study suggested a collapse could bring Ice Age-like cooling and even glacier growth within a century, though its accuracy is questioned. Studies of the Florida Current suggest the Gulf Stream ran around 10 percent weaker between 1200 and 1850, likely contributing to the Little Ice Age.

  The AMOC also turns the Atlantic into an effective carbon sink in two ways. Upwelling delivers nutrients that fuel phytoplankton growth and raise overall marine photosynthesis. The water that rises is typically about 1,000 years old, meaning it was never exposed to the atmosphere's recent buildup of human-caused carbon. That lets it absorb far more carbon before being pulled back down. The Southern Ocean remains the strongest carbon sink of all, but the North Atlantic is the largest single sink in the entire northern hemisphere. That balance is a fairly recent, stable arrangement. It was anything but steady during the last ice age.

• 4. Ancient Whiplash

  ›

  Twenty five abrupt temperature swings rippled between the hemispheres during the Late Pleistocene, the epoch running from about 126,000 to 11,700 years ago that included what is colloquially called the last ice age. Willi Dansgaard and Hans Oeschger identified these swings, now called Dansgaard-Oeschger events, by analyzing Greenland ice cores in the 1980s. Greenland could warm by between 8 and 15 degrees Celsius within a few decades during one of these events. The Southern Ocean cooled at the same time, consistent with the AMOC redistributing heat between hemispheres.

  That northern warming melted ice sheets, and many Dansgaard-Oeschger events ended when huge streams of icebergs broke away from the Laurentide ice sheet. These episodes are called Heinrich events. As those icebergs melted, the ocean grew fresher, the circulation weakened, and the warming stopped. Researchers still do not agree on why these swings happened only during this glacial period. Proposed explanations include cyclical shifts in North Atlantic salinity and wind patterns driven by the growth and retreat of ice sheets large enough to alter atmospheric circulation.

  The penultimate of these oscillations, Dansgaard-Oeschger event 1, occurred roughly 14,690 years ago. It marks the shift from the Oldest Dryas period into the Bolling-Allerod Interstadial, a warm spell lasting until about 12,890 years before the present. It takes its name from two sites in Denmark, chosen because fossil vegetation there could only have survived during a comparably warm stretch in the northern hemisphere. That transition also triggered a burst of ice-sheet collapse called Meltwater pulse 1A, sending sea levels up, while concurrent southern-hemisphere cooling kept the net change in global temperature small.

  Two centuries of the opposite pattern then interrupted the warm spell. This cold snap was named the Older Dryas, after the Arctic flower Dryas octopetala, which spread wherever northern forests had briefly grown. The interstadial finally ended with the Younger Dryas, a period from about 12,800 to 11,700 years ago, when northern hemisphere temperatures crashed back to near-glacial levels, possibly within a single decade. Meltwater poured through what is now Canada's Mackenzie River, rather than a mass iceberg release like a true Heinrich event, freshening the ocean and slowing the AMOC. The Intertropical Convergence Zone shifted south, drying Europe and South America while soaking North America, before long-term post-glacial warming finally resumed.

• 5. A System With Two Settings

  ›

  For much of Earth's history, the planet's overturning circulation ran through the North Pacific rather than the Atlantic Ocean. Paleoclimate evidence points to a switch at the Eocene-Oligocene transition, roughly 34 million years ago, when the Arctic-Atlantic gateway closed off. Some researchers think continued climate change could eventually reverse that switch. It might send the overturning circulation back to the Pacific once the AMOC itself shuts down. Warmer surface water from Earth's energy imbalance is one driver of change. Added fresh water, mostly meltwater from Greenland, is another. Both widen the gap between the ocean's surface and its depths, making it harder for the circulation to sink and rise.

  In the 1960s, the oceanographer Henry Stommel built what became known as the Stommel Box model. With it, he introduced an idea called the Stommel bifurcation. Under this idea, the AMOC could sit in a strong state like the one seen throughout recorded history. Or it could collapse into a much weaker state it would not leave again unless the original warming or freshening reversed. Ordinary fluctuations could push a weakened circulation past that tipping point on their own. Whether the AMOC truly behaves as a bistable system, switching abruptly between on and off, has been debated by scientists ever since.

  Some models built on Stommel's foundation allow for one or more in-between stable states. This pattern shows up more often in simplified Earth Models of Intermediate Complexity than in full general circulation models, the field's gold standard. General circulation models typically find the AMOC has only one stable state and rarely collapse it at all. Researchers suspect this is partly because those models funnel meltwater toward the North Pole in ways that do not happen in nature. In 2024, three researchers ran a simulation using a Community Earth System Model that did produce a classic collapse. They gradually raised meltwater over more than 1,700 years until it reached an input equivalent to 6 centimeters of sea level rise per year, about 20 times the actual 2.9 millimeter-per-year rate measured between 1993 and 2017. The researchers said the exaggerated input was needed only to counter their model's unrealistic stability. Other scientists called the result useful mainly for calibrating future, more realistic studies.

  Statistical analysis of lower-complexity models produced the most-cited collapse date, around 2065, revised from an earlier estimate of 2057 in August 2025, with 95 percent confidence of collapse occurring sometime between 2037 and 2109. Many scientists remain skeptical, partly because the analysis leans on proxy temperature data from the Northern Subpolar Gyre, which other researchers think may follow a separate tipping point of its own. The IPCC held only medium confidence that a full collapse would be avoided before 2100 in its 2021 assessment, down from high confidence of avoiding it in its 2014 report. High-quality Earth system models suggest collapse only becomes likely once warming of 4 degrees Celsius or more persists well past the year 2100.

• 6. Watching It Slow

  ›

  A Nature Geoscience study published in November 2024 reconciled two conflicting pictures of the AMOC: direct observations showing a slowdown, and climate models showing stability. Using what the researchers called Earth system and eddy-permitting coupled ocean-sea-ice models, they found the observations and models could agree after all. The result pointed to a slowdown of 0.46 sverdrups per decade since 1950.

  Direct measurements have existed since 2004, when the RAPID mooring array began recording the circulation's strength at 26 degrees north in the Atlantic. In May 2005, submarine-based research led by Peter Wadhams found downwelling in the Greenland Sea running at under a quarter of its normal strength, measured through giant water columns nicknamed chimneys. A separate 2004 measurement of the North Atlantic Gyre showed a 30 percent decline compared with 1992, which some read as a sign of AMOC collapse. RAPID data later showed this was a statistical anomaly, and the gyre had recovered by 2007 and 2008. Scientists now consider the gyre largely separate from the rest of the AMOC, capable of collapsing on its own.

  By 2014, processed RAPID data through the end of 2012 showed a decline ten times larger than the best models of the time had predicted. This reignited debate over how much of it reflected climate change rather than natural variability. Separate heat-transport estimates from NASA's CERES satellites and international Argo floats, made in 2017 and 2019, suggested 15 to 20 percent less heat movement than RAPID implied, pointing to a fairly stable flow instead. The Florida Current, meanwhile, has measured as stable for four decades once corrections for Earth's shifting magnetic field are applied.

  Between 1900 and 2020, a patch of the North Atlantic Gyre near Greenland cooled by 0.39 degrees Celsius even as the rest of the ocean warmed substantially. The cooling is strongest in February, reaching 0.9 degrees Celsius at its center. Between 2014 and 2016 the area stayed cool for 19 months before warming again, a pattern the media nicknamed the cold blob. Later research traced part of this local cooling to atmospheric shifts, including more low cloud cover and a strengthened North Atlantic Oscillation. That means the cold blob alone cannot settle how much the AMOC itself has slowed.

• 7. If It Stops

  ›

  A full AMOC shutdown would be largely irreversible, with recovery likely taking thousands of years. Its clearest effect would be sharp cooling across Britain, Ireland, France, and the Nordic countries. A 2002 study compared such a shutdown to the ancient Dansgaard-Oeschger events and estimated local cooling of up to 8 degrees Celsius in parts of Europe. A 2022 review of tipping points estimated a full collapse would lower the global average temperature by around 0.5 degrees Celsius.

  A 2020 study modeled the effect of a collapse on farming in Great Britain. It found an average temperature drop of 3.4 degrees Celsius once background warming was subtracted out. Growing-season rainfall would fall by around 123 millimeters, shrinking the land suitable for arable farming from 32 percent to just 7 percent. The researchers put the resulting loss to British farming at around 346 million pounds a year, more than 10 percent of the sector's value in 2020.

  A 2024 study modeling a collapse against a pre-industrial climate found even sharper cooling. Northwest European sea surface temperatures fell by 10 degrees Celsius, and February land temperatures dropped between 10 and 30 degrees Celsius within a century. Sea ice would reach into the territorial waters of the British Isles and Denmark each winter under this scenario, even as Antarctic sea ice shrank. A 2015 study led by James Hansen concluded a shutdown would intensify winter storms by speeding up mid-latitude winds by 10 to 20 percent, potentially fueling hurricane-force winter cyclones nicknamed superstorms, though this paper too has drawn controversy.

  Beyond Europe, researchers disagree on how a collapse would reshape the El Nino-Southern Oscillation. Estimates range from no change to a shift toward dominant La Nina conditions, a roughly 95 percent drop in El Nino extremes, heavier rainfall in eastern Australia, and worsened drought and wildfire seasons in the southwestern United States. A 2021 study found a collapse would actually increase rainfall in the southern Amazon by shifting the Intertropical Convergence Zone, potentially helping stabilize that part of the rainforest against dieback, while a 2024 study found the region's entire wet and dry seasons could reverse. A 2005 paper warned a severe AMOC disruption could cut North Atlantic plankton populations to less than half their normal levels, as increased ocean stratification chokes off the nutrient exchange those populations depend on.