The collapse of the Atlantic Meridional Overturning Circulation (AMOC)—the engine that drives our oceans—could happen as soon as 2100, according to a study from Utrecht University published last month in the journal Science Advances. And while the world may not suffer a fatal flash freeze like in the 2004 action flick The Day After Tomorrow, an AMOC breakdown would probably reduce average air temperatures in the Northern Hemisphere by around 40 degrees Fahrenheit or more, spelling out disaster for life on our planet.

But last year, a brother-sister science team in Denmark made their own prediction about the tipping point. Their model found that the AMOC could shut down as early as 2025 (though the duo says it’s more likely to happen toward the middle of the century). The controversial study, published in Nature Communications in June 2023, sparked a media frenzy and a sizable amount of backlash.

If correct, either of those studies’ findings would mean catastrophic consequences for our climate. But why do the two studies arrive at such drastically different timelines for the AMOC’s demise, and which conclusion is more credible?

The Beating Heart of the World’s Oceans

The Atlantic Meridional Overturning Circulation is a branch of the Ocean Conveyor, a global circulation system in the world’s oceans that moves warm water from the Southern Hemisphere to the Northern Hemisphere and returns the cold water back again; it impacts temperatures at the surface, resulting in the climates we now recognize around the globe. It moves ten times more water every second than all of the world’s rivers combined, and yet it may take one drop of water 1,000 years to travel the whole route. In the Atlantic, the branch of the Ocean Conveyer known as the AMOC is important to this system because it hosts the beating heart of the system known as the Greenland Pump.

Topographic map of the Nordic Seas and subpolar basins with schematic circulation of surface currents (solid curves) and deep currents (dashed curves) that form a portion of the Atlantic Meridional Overturning Circulation. Colors of curves indicate approximate temperatures.© Wikimedia Commons

Water moves up from the Southern Hemisphere of the United States, across the Atlantic, and up to Greenland via the Gulf Stream, which is partly driven by ocean currents like the AMOC and partly by the wind and the movement of Earth. As the Gulf Stream travels, some of it evaporates, releasing warmth into the air and cooling the water. The water is now not only cooler, but saltier, because of the evaporation. In the past, some of it would freeze when it hit the cold temperatures of Greenland, and the remaining salty water would sink down and begin traveling south, again, toward the equator. That’s the pump; it moves in the AMOC like a heartbeat moves blood through the circulatory system.

The Tipping Point Is Coming

But that circulation has begun to slow. As Greenland warms due to climate change, less water is freezing and the existing ice is melting faster than expected, dumping fresh, cold Arctic water into the salty water. Without so much heavy, salty water moving to the bottom and heading south, the pump is now weaker. If the amount of fresh water hits a tipping point, it could stop working altogether.

The AMOC hasn’t shut down for 12,000 years. But when it shut down in the past, it preceded dramatic climate changes. If it does so again, it could drop temperatures in the Northern Hemisphere, raise sea levels, interrupt rain flows, and cause droughts, famine, and other disasters.

Susanne Ditlevsen, Ph.D., a professor in the department of mathematical sciences at the University of Copenhagen; and her brother, Peter Ditlevsen, Ph.D., a professor of physics in ice, climate, and earth science at the Neils Bohr Institute at the University of Copenhagen, are focused on that tipping point.

Two years ago, the Ditlevsens were talking about Peter’s work studying tipping points. They knew that the AMOC collapsed and restarted repeatedly during the ice ages from 115,000 to 12,000 years ago. They knew that because, since the 1950s, scientists have been drilling down into Greenland’s ice and pulling up ice plugs of roughly 3–24 feet in length and five inches in diameter that tell the story of the climate when each of those layers of ice formed. They have drilled all the way down into the land below the ice, some 1,500 feet. Scientists can read Greenland’s climate history in those ice cores in the same way you can read the rings of a tree. They can read it from bubbles of gas in ice frozen thousands of years ago that reveal information about the greenhouse gases in the atmosphere at the time. They can read it from the chemical composition of organisms trapped in the ice. Through reading these ice cores, paleoclimatologists can see clearly when the AMOC broke down in the past, and how dramatic climate shifts always followed.

An ice core sample extracted at McMurdo Sound, Ross Island, Antarctica.© Getty Images

As Susanne Ditlevsen tells Popular Mechanics, they knew that the models used in climate studies—like the ones used in the Utrecht University team’s February 2024 report—are very complex, with enormous numbers of variables and very high resolution. They take a long time to run, and researchers must average the predictions from some 100 different models. And, like weather models, which are subject to chaotic change as a result of the merest change, climate models can go in all kinds of directions depending on the data given. The Ditlevsens wanted to create a much simpler model based solely on understanding the mathematical structure that signifies you’ve hit a tipping point.

“Tipping points exist in a lot of systems, not only in the climate, and they have a specific mathematical structure,” Susanne Ditlevsen says. “You don’t only observe this mathematical structure, of course, you observe a high-dimensional system with a lot of noise. So we wondered, how do we cut through this noise with statistical analysis to figure out how to use this specific structure that this tipping point has?”

“We said, ‘Let’s focus on the data we can measure and make as few assumptions as possible. Then we throw away a lot of the details. (And of course, in that sense, we are wrong because we don’t have the details). And then we take the most essential parts and fit them to the data.”

With Peter Ditlevsen’s expertise in physics and Susanne’s work on dynamical systems, random processes, and statistics around climate change, including working with biologists to study the impact of human activities on marine mammals, they started to build a model around the tipping point itself, also known as a bifurcation.

“It’s like the skeleton,” Susanne Ditlevsen says. “We took this skeleton and said, ‘This has a very specific structure which is universal … the road toward the bifurcation is different, but the bifurcation itself is always according to this structure.”

Their mathematical model showed them a slope indicating that the tipping point could happen much sooner than 2100—as early as 2025, but more likely in the mid-century. It’s similar, she says, to guessing at what precise moment ice will melt into water using math.

“You can compare it a little bit to phase transitions,” she says. “Imagine that you have ice and it melts and then it is water; that’s a phase transition from solid to fluid.”

Guess Wrong, But Act More Quickly

Compared to the robustness of the Utrecht University team’s models, the Ditlevsens’ model seems rather simple. On the other hand, without having to focus on so many factors, they may be better able to focus more on what matters. Guessing wrong, and postponing action, at any rate, seems riskier than guessing wrong and acting more quickly.

“No one, neither those who make the models nor the [Intergovernmental Panel on Climate Change], believe that any of these models is the correct model,” Susanne Ditlevsen says. “... If they knew one of the models were correct, they would only need one, they wouldn’t need 100 models.”

The problem with climate models, she says, is that they only project out so far. When predicting weather, you can look back a week to learn whether your models were accurate, and then improve them. But there’s no time for us to wait even 10 years to see whether our climate models are correct. We have to take action on the information that we have, before it’s too late.

“The collapse of the AMOC has huge implications, and we can’t just sit back and say, ‘I don’t know, maybe we’re wrong,’” Susanne Ditlevsen says, shrugging. “I hope we’re wrong. But I don’t think we’re wrong. The more data we collect and the more we work on the model, the more conf

ident we are in our results. And we have to stop fossil fuel usage and greenhouse gas emissions if we want to stop this.”