When it comes to tipping points, one of the greatest worries is the status of what is known as the Atlantic Meridional Overturning Circulation, or AMOC—a key player in making the northern part of Europe habitable. The AMOC has been described as a sort of conveyor belt, that uses currents such as the Gulf Stream to circulate saltwater and heat from south to north and back again in a long cycle within the Atlantic Ocean, warming up the land masses it passes on its northward journey in the process—and with effects on the rest of the globe as well.

To very loosely paraphrase the words of one observer, if this heat-transporting mechanism was to go out of business, then raising crops in Great Britain would be like trying to grow potatoes in northern Norway (near the Arctic Circle).

Fortunately, this oceanic circulatory system was generally regarded as more-or-less stable.

But new methods of data-collecting and computer-modelling have revealed that the AMOC has been unstable in the past, and its earlier collapses led to some of the most dramatic and abrupt climate shifts ever known. These days, melting ice in the North Atlantic—caused by increased global warming—has steadily caused more freshwater to be introduced into the system, changing its salinity and further pushing it toward collapse. At some point, this steady diet of change might no longer produce effects in a neat linear fashion but reach that moment of critical mass—an abrupt threshold, or tipping point—beyond which there could be large, accelerating, and possibly irreversible changes in the system and change its salinity and further pushing it toward collapse (or, at least irreversible changes within the span of a human lifetime).

But the evidence for the implosion of the AMOC is still relatively new—only a few decades old—and the computer models themselves have sometimes been called into question.

To help explore the complexities of the AMOC situation, I spoke with physicist/oceanographer Henk Dijkstra of Utrecht University, who specializes in dynamic oceanography. In Dijkstra’s words, this field is basically physical oceanography but with a focus on the mechanisms that cause change: “We try to answer questions like why El Nino has a four-year period, why the Gulf Stream is in the west of the basin—all the ‘Why?’ questions.”

(Editor’s note: This interview has been condensed and edited for brevity and clarity.)

Dan Drollette Jr.: Glad we have a chance to talk. Hope all’s well in Utrecht.

Henk Dijkstra: Actually, I’m in Italy at the moment. I have an adjunct appointment at the University of Trento and am there usually a few months per year—and this is a good time to be in Italy. Utrecht is pretty far up in the Netherlands, and the winters there can be horrible: a lot of gray skies, rain, snow, and hail.

When I was younger—in the 1970s—there were winters there when it was so cold that you could actually drive a car on the frozen rivers. We have never seen that happen again in the last few decades. But in the ‘60s and ‘70s, it was often really cold in the winter.

Now, people in the Netherlands are happy to have one day when they can skate outside—and even that’s on a very thin layer of ice.

Drollette: That goes into something I wanted to ask: How is the climate expected to change in Europe—particularly with respect to the Atlantic Meridional Overturning Circulation? I hope I pronounced that correctly.

Dijkstra: Yeah. We just call it the AMOC (“a-mock”) for short.

Drollette: How would you explain to a general-interest audience what the AMOC is?

Dijkstra: Probably everybody knows of the Gulf Stream. In the North Atlantic, the Gulf Stream runs near the East Coast of the USA and transports heat and salt water northwards, along the ocean’s surface. There’s also a return flow, where the flow goes deeper down and southward.

The AMOC is the number of cubic meters per second of water which go through a given geographic section of the North Atlantic, between the ocean surface and a given depth.

Drollette: I’ve heard the AMOC described as a “conveyor belt” that brings warm salty water from the tropics to the north, where it is separated out into what becomes freshwater icebergs on the surface and very cold, very salty, super-chilled water, down below. Then, that cold deep water heads south and the cycle begins again. Is the conveyor belt still considered a good metaphor?

Dijkstra: You have to be a bit careful using it in the geophysics community, because “conveyor belt” sketches the image of a very coherent, whole-Atlantic Ocean circulation.

Meanwhile, the reality is much different, and much less coherent. That’s why we avoid the conveyor belt analogy in our community.

But you can use it in broad terms for the general public: “The AMOC is a conveyor belt for volume transport.” And that matches up with what I just said, right? It’s just a means of transport of sea water through a certain section.

And what we call “AMOC strength” is sort of the maximum number of cubic meters of seawater per second that goes northward through a specific section, such as at the 26 degrees North latitude mark in the Atlantic[1].

Drollette: My understanding is that researchers have learned that the AMOC has a major effect on climate—not just locally, but all over the globe?

Dijkstra: Yes, it does have a global effect. But of course, most of it impacts the North Atlantic and its boundaries—the continents immediately adjacent to the North Atlantic. The reason is that the AMOC transports heat northwards: It’s warmer in the tropics and cooler at the poles, so the AMOC is the flow northward of surface water that takes heat from the equatorial region to the north.

And the warm water that the AMOC delivers to Europe is then eventually released in the form of heat to the atmosphere, which makes northern Europe a lot warmer than at the same latitude in the Pacific.

Consequently, if you compare the climate of Alaska with Norway—which are both on similar latitudes—then you see a big difference in winter temperatures, due to the AMOC.

Drollette: I hadn’t realized—Alaska and Norway are on the same line of latitude?

Dijkstra: Well, not exactly the same, but roughly. And don’t forget that Norway is very long, running from north to south.

Drollette: That ties in with something that I’d run across when I had lived and worked in France and Switzerland. As someone from the northeastern USA, I was always amazed at how long the summer days were in that part of the world, and how short the winter days got—and I was also startled to realize how far north cities like Paris were on the map, compared to New York City.[2] But Paris—and London and Amsterdam—had comfortable-enough weather.

Dijkstra: Part of that is because of ocean circulation, which has quite an influence on that part of Europe.

And the northern hemisphere is, on average, about 1.5 degrees [Celsius] warmer than the southern hemisphere, because the AMOC transports heat over the equator.

And of course, there’s other processes involved. There’s also, for example, the land/sea contrast, which partly determines the temperature on the continents.

Drollette: How stable is all this? Doesn’t what they call “paleoclimatic data”[3] show that the AMOC has been unstable in the past?

Dijkstra:  In the past, there have been relatively dramatic and abrupt shifts in it.

A good example is what are called Dansgaard–Oeschger Events, or “D-O cycles.”

Willi Dansgaard was a Danish scientist in Copenhagen, and Hans Oeschger was a Swiss researcher, and together they discovered that there were rapid changes in isotopes of oxygen in ice cores they took in Greenland. By looking at these isotopes of oxygen, they could then reconstruct the temperature at the time of the ice’s deposition—and they found that there were these rapid transitions, usually with a gradual cooling and then followed by a very rapid warming. This phenomenon appears in the Greenland ice cores in particular.

The mechanism behind it is still under discussion; there are many different theories. It still has not been settled, partly because the data is still relatively limited. But it is commonly accepted in the [geophysics] community that changes in the AMOC played a major role.

Drollette: When these researchers say “abrupt,” what kind of time-scale are they talking?

Dijkstra: There’s a dominant period of about 1,500 years for a D-O event. So that one’s pretty long.

And then within that period you can have a longer cooling event, which is where most of the decay of the AMOC occurs.

To complicate things, the rapid, centennial-long recovery of the temperature is associated with the strengthening of the AMOC.

But those are the typical timescales.

In any case, most of these fluctuations occurred between 60,000 and 20,000 years ago.

Drollette: I get the impression that it’s only really in the past couple of decades that researchers have come to understand the influence of these type of events?

Dijkstra: Well, the D-O events were really discovered in the early ‘90s. That’s when people could actually start to do this ice-coring at levels deep enough that they cover enough of a scale to enable you to do reconstructions.

But the idea that the AMOC can change rapidly goes all the way back to the 1960s, when people looked at early computer models and saw that temperature and salinity can work together—or compete against each other—to change the density of the sea water and speed up or slow down its circulation. These first, very, very conceptual models indicated that there could be very rapid transitions in the AMOC.

At first, these models were considered to be purely theoretical—not really something of value —until in the mid-1980s somebody demonstrated it in more convincing detail in a sophisticated ocean model. Or, at least, sophisticated for the 1980s. And that got the search for physical evidence going in the late ‘80s.

Drollette: From the admittedly limited background reading that I’ve done, it sounds like the very first models in the ’60s were… I hate to use the word “primitive”… but they were simplistic?

Dijkstra: Well, yes. But we call them “conceptual.”

So, for instance, back in those pioneering models of the 1960s, the Atlantic was described as one northern box, with a single temperature and salinity. And then you had an equatorial box, which also had a single temperature and salinity. The circulation between them was just solely driven by the density difference between the northern box and the southern box. So if the density in the northern box was higher, water would sink there and moved on southward at depth, and then at the surface it would move northwards.

Those levels of conceptuality were used in the ‘60s, but they didn’t contain any ocean dynamics—they had no Gulf Stream dynamics, no rotation of the Earth, no winds, no underwater topology, no continental geography.

So that’s why the AMOC and its possible weakening was considered basically a curiosity—one that wasn’t realistic.

But that changed in 1986, as I mentioned, when one PhD student in Colorado, Frank Bryan, used  a much more sophisticated model that included more realistic features. And this model showed basically the same thing—that you can have these transitions, and that they can happen relatively quickly. So then this subject really got going in the ‘90s and 2000s; it’s now getting lots of attention.

Drollette: In your view, how close is the AMOC to collapse?

Dijkstra: I would say the current situation is very uncertain.

We don’t have adequate observations to locate where the AMOC is in respect to its collapse points.

We don’t even know for sure that collapse points exist in the real system. We have to be honest about this: We don’t know for sure.

Although there are some things we do know for certain: We have direct observations of the AMOC at 26 [degrees of latitude] North and in the northern part near Greenland and Iceland—for example, in the Labrador Sea. And we have some direct measurements at 35 South.

And of course, there is what we know from computer model simulations. Though the models have their biases and uncertainties.

Having said all that, there are strong indications that there is a collapse point in the AMOC system, and that we’re approaching it.

But we still don’t know what the distance is to the collapse point.

I think that’s the best way to express it.

And you know, there will be people in the geophysics community—researchers like Stefan Rahmstorf[4]—who think that we may be close to the point of collapse of the AMOC.

At the same time, there are also researchers who will say “This is only a feature of the models, and it’s not in the real system. Because the system is much more complicated than we think. And so the models are not fit to analyze it.”

Drollette: That ties into what I’d read in Real Climate. The headline said it all: “The AMOC is slowing, it’s stable, it’s slowing, no, yes, …”[5]

It’s hard for us journalists to keep track; some call it “media whiplash.”

Dijkstra: Well, you know, we scientists try to be as careful as possible, so we try not to claim too much.

And everyone’s had papers submitted where they forgot that maxim—and then you’re immediately shot down by the referees.

Consequently, there is a negative feedback mechanism—to not overstate the claims from models and from observation analysis.

Drollette: Would it be fair to say that the AMOC has slowed in the past two decades?

Dijkstra: I am quite careful about even saying that.

All we know is from observations from 26 North, that we started in 2004. We have just a little bit over 20 years of data, and even from that we can already see that the AMOC has quite some variability. It decreased from 2004 to 2014, but after that, it recovered. The AMOC strength is currently not up to the level of 2004, but it came up. Those are just the natural fluctuations that occur on that particular time scale.

To make up for this, what some people have tried to do is connect the AMOC’s strength to all kinds of other variables which we do have observations of, like sea surface temperature—and then by making reconstructions based on those sea surface temperature data, they concluded that the AMOC decreased by 15 percent since 1950.

That was done a few years ago, but it’s been contested, because there’s a problem as to whether those temperature fluctuations correlate very well with the AMOC strength.

There was another paper two weeks ago in which they looked at heat fluxes at the higher latitudes—and it showed that the AMOC has not decreased from 1963 to 2017.

Drollette: This was the piece in Nature Communications?[6]

Dijkstra: Right.

So it all shows that the relationship between the AMOC and the observations we have of sea surface temperature and sea surface salinity are not straightforward. The bottom line is that we should be careful with interpreting AMOC reconstructions before 2004; they are very uncertain.

Drollette: What makes 2004 special?

Dijkstra: That’s when new instrumentation was put in. The year 2004 is when the RAPID-MOCHA array went into the water at 26 North, which gave us really detailed observations of the strength of the AMOC from 2004 onwards.

This array is a system of highly sophisticated buoys, with one set in the east part of the ocean and another in the west part. And they effectively measure the density difference between the water in the two locations. The system also incorporates data from an electric cable strung between Bermuda and Florida, to measure Gulf Stream strength. And then there’s some pressure sensors on the Mid-Atlantic Ridge.

From a design like that, you can really measure the AMOC’s strength.

It’s a very ingenious instrument, first developed in computer models and then tested in the water, and it worked. So this was a fantastic result.

It was all under one umbrella program: The RAPID part is the UK contribution, and it focused more on water volume at the beginning[7]. And the US contribution is called MOCHA; it concentrated more on heat transport.

Now it’s sort of a unified program—and it really changed things a lot.

When we didn’t have those measurements, we could only guess at what was happening.

Now the AMOC is monitored, and it will hopefully be monitored for a long time. If there are any worrying changes in the near future, we will know about it.

Drollette: So, it really sounds like a lot of the data about the AMOC is just based on what’s happened in the past 21 years—which to an outsider like me seems like kind of a small span of time.

Dijkstra: Yes, it is. Of course, we also have data from other sources, such as a lot of mostly temperature data from about 1850 to now. And a satellite record from maybe the late 1980s. And a global array of free-floating instruments called ARGO[8], which was deployed in the early 2000s.

So there are more observations, but these are either limited to the surface or they don’t really home in on the AMOC the way that RAPID-MOCHA does.

Drollette: If there were to be an AMOC collapse, what are the most likely consequences—and how much time do we have?

Dijkstra: The cleanest model simulations come from when you look at pre-industrial model configurations—the average 1850-1900 situation—and then collapse the AMOC in those models. That is what we did in our Science Advances paper that appeared last February.[9] And we found that when the AMOC collapses, you get an enormous cooling over Greenland, Britain, Scandinavia, and countries over most of Western Europe, with up to 15-degrees Celsius of cooling on average, over the span of a century—although that is just an average, because there is a tremendous variation, geographically speaking.

For example, the February temperatures in Bergen, Norway, would decrease by greater amounts—about three degrees cooler per decade, or twice the average.

A lot will depend on what part of Europe you are in, because some regions cool faster than others—and that will also affect the overall temperature change in Europe as a whole.

And it’s important to realize that while these changes are abrupt by the standards of geological time scales, they can appear longer when measured in terms of the human life-span: 100 years would likely pass from the onset of the collapse of the AMOC to its eventual total collapse.

To complicate matters, climate models like ours have to take into account that at the same time, there is this warming going on that is due to the increase of greenhouse gases. The scale and the timing of any AMOC collapse would depend on what will happen in the near future with these gases: Will they continue to be emitted at the same volume? Decline? Increase?

Drollette: What is the takeaway from all this? What should people be thinking about the AMOC?

Dijkstra: The main message is that climate change can lead to unexpected consequences, which will have an enormous impact—so much so that you may not be able to adapt to it anymore.

I think the main message is that we know that there are critical phenomena in the climate system: If you increase the amount of greenhouse gases, that it can pass a critical forcing on to the system that leads to these unexpected, relatively rapid changes which are difficult to adapt to. That is an important message. Because it makes it hard to plan how to build resiliency into your infrastructure if you don’t really know what’s going to happen or when.

Drollette: Any last comments?

Dijkstra: I think it’s important to realize that there is still a lot of uncertainty in this kind of a problem. There is a lack of observational data, so a lot is based on computer models—which inevitably have biases.

However, the models are getting more sophisticated. Computers get bigger; you can do more of these simulations.

It will take a while, but eventually we’ll have a lot better representation of ocean circulation, and then the estimates on the probability of a possible future AMOC collapse will get better.

We researchers don’t want to scare people, but the fact is that there are these tipping phenomena. Rapid changes can occur—and the chance of them occurring increases with increasing greenhouse gas emissions.

Endnotes

[1] Scientists often use the latitude of 26 degrees North to measure the strength of the AMOC by monitoring the transport of water masses across this line. This latitude crosses nears the Mid-Atlantic Ridge, a significant geological feature, and is close to the Azores.

[2] The northerliness of the region’s geography and how it compares to the United States is a recurring theme in David McCullough’s 2011 book, The Greater Journey: Americans in Paris.

[3] Paleoclimate date is the use of sea-floor sediments, ice sheets, corals, cave formations, ancient trees, and alpine glaciers to gain clues to the past climate, For more about the role of paleoclimate data in the AMOC, see Stefan Rahmstorf’s article, “Is the Atlantic Overturning Circulation Approaching a Tipping Point” in the April 10, 2024 issue of Oceanography at https://tos.org/oceanography/article/is-the-atlantic-overturning-circulation-approaching-a-tipping-point

[4] Oceanographer and climatologist Stefan Rahmstorf of the Potsdam Institute for Climate Impact Research in Germany, whose research focuses on paleoclimate, ocean circulation, sea level, extreme weather events, and Earth System modeling. He co-founded the award-winning blog Realclimate to help explain the impacts of climate change.

[5] See January 26, 2025 issue of Real Climate titled “The AMOC is slowing, it’s stable, it’s slowing, no, yes, …” at https://www.realclimate.org/index.php/archives/2025/01/the-amoc-is-slowing-its-stable-its-slowing-no-yes/

[6] For more, see the January 15, 2025 article by Jens Terhaar et al in Nature Communications titled “Atlantic overturning inferred from air-sea heat fluxes indicates no decline since the 1960s” at https://www.nature.com/articles/s41467-024-55297-5

[7] More on RAPID can be found at https://rapid.ac.uk/ while more on MOCHA is at https://www.aoml.noaa.gov/rapid-mocha-atlantic-circulation-story/

[8] See https://sealevel.nasa.gov/missions/argo

[9] For more, see the February 9, 2024 issue of Science Advances, titled “Physics-based early warning signal shows that AMOC is on tipping course.” https://www.science.org/doi/10.1126/sciadv.adk1189