It’s a winter’s day late this century and London is engulfed in a December blizzard, its third of the season. The air is sharp with ice. Oxford Street is deserted. For nearly two days, Heathrow’s runways — all four of them — have stood empty. Visibility on the roads is near zero.
In the following months, many English football pitches will lie idle, the groundsmen having given up trying to protect the surfaces from the elements. Bookmakers no longer take bets on a white Christmas: the probabilities have shifted too far.
This future is so much colder — and indeed so much drier — that British farming has dramatically shrunk. The complaints of the 2020s, such as Brexit and farm taxation, turned out to be minor shocks compared to the rapid cooling of the following decades. A proposal for irrigation — piping water from Scotland southwards — was deemed too expensive. The land of the second agricultural revolution no longer grows crops at scale.
Restaurants do not boast of local produce. Nearly all the UK’s vegetables and fruits are imported, as are its wheat, oats and sugar. To make matters worse, fish stocks have collapsed in the north Atlantic.
This future Britain has adapted to some aspects of the cold and dry. After a hiatus in the early 21st century, the country did build new reservoirs. Even in summer, when rain doesn’t arrive, London does not run short of drinking water.
But infrastructure is not Britain’s strong point. Despite efforts to learn from Scandinavia, its rail network cannot withstand the cold. Disruption is constant. The wealthy insulated their houses but millions of homes and offices, not to mention schools and hospitals, remain suited only to the old, moderate winters that now rarely occur.
To predict the future, at least in any meaningful way, requires both science and imagination. For decades, the scientific advice has been that the world should brace for a warmer climate. By 2050, according to one study, London’s summers will resemble those of Barcelona. Already temperatures of 40C have led to the installation of air conditioning and water fountains. But for a growing group of scientists, another scenario is becoming increasingly plausible.
One day, possibly in our lifetimes, the temperature could start to drop in northern Europe. When this is forecast in isolation, without the offsetting warming effects of climate change, London would end up as much as 10C colder. Once climate change is taken into account, the scenario is for London’s average temperature to rise and then gradually fall, ending up perhaps 2-3C cooler than before industrial times. Such a shift may not sound dramatic. But the storms, wildfires and heatwaves of recent years have taken place on an average temperature shift of just over 1C. As averages move, the extremes become more extreme.
What’s more, recently published modelling estimates that the cooling could start in the next couple of decades, although it would take perhaps a century to play out in full.
Some content could not load. Check your internet connection or browser settings.
One of the most alarming — and uncertain — fields of climate research, the explanation lies in the oceans. Today, currents known as the Atlantic Meridional Overturning Circulation (Amoc) bring vast amounts of water from the tropics to the northern regions. The Amoc is a conveyor belt or perhaps a central heating system. The warmth that it transfers from south to north is about 50 times all the energy that humans use.
In places, the Amoc forms part of the Gulf Stream. It represents about one-fifth of the water that the Gulf Stream transports, but most of the heat. It keeps London several degrees warmer than places at similar latitudes, for example on the Pacific coast of Canada. In recent decades, however, those currents have been weakening. A point may come when they no longer circulate. They may pass a tipping point, beyond which there is no return.
A full collapse of the Amoc would be a “massive, planetary-scale disaster”, says Stefan Rahmstorf, a climate scientist at Germany’s Potsdam Institute and one of the world’s leading experts on the system. “It is quite difficult to know exactly what the impacts will be, but my feeling is that they will be quite devastating.”
They would certainly be global. An Amoc collapse would push the band of rainfall around the equator southwards, damaging some of the Earth’s most vital ecosystems and the food security of billions of people in Africa and Asia. Heatwaves in mainland Europe would worsen: as the north Atlantic cooled, hot air from the south would be stuck there.
As for the UK, “in simple terms, [it would be] fundamentally less habitable than it is at the moment”, says Tim Lenton, a professor of climate change at the University of Exeter.
The IPCC, the intergovernmental body whose reports on climate change are taken as gold standard, has long downplayed the risk of Amoc collapse. Rahmstorf once agreed. The possibility of passing the tipping point in the next century was below 10 per cent. But new research changed his mind. Last August, he told CNN that the risk was “probably even greater than 50 per cent”. He and Lenton are part of a network of scientists who complain that an existential threat is being overlooked. If they are right, then climate change could be even more disruptive than we are currently planning for.
A tidy, bespectacled 64-year-old, Rahmstorf started studying the Amoc in the 1990s. His focus was on the distant past. Sediment samples and Greenland ice cores revealed that the climate underwent abrupt changes more than 25 times in the last ice age, which ended some 11,000 years ago. Northern Europe had seen dramatic changes in vegetation, in periods that North America had not. Antarctica had warmed by between 8C and 16C in just 50 years, while the northern hemisphere had cooled. Both pointed towards the Amoc as the likely culprit.
In 1987, Wally Broecker, a pioneering geochemist and a key influence on Rahmstorf, warned that global warming could lead to an Amoc collapse. “We play Russian roulette with climate, hoping that the future will hold no unpleasant surprises,” he wrote. “No one knows what lies in the active chamber of the gun, but I am less optimistic about its contents than many.”
Experts largely agreed with Broecker’s suggestion. They also agreed that global warming increased the chances of it happening. This is because the saltier water is, the heavier it is. Today the Amoc brings warm, salty water from the subtropics northwards at the surface. Some of the warm water evaporates as it meets cold air over the north Atlantic; the remaining water is so salty that it sinks to a depth of 2,000-3,000 metres. (It has also cooled, making it heavier again.) The downwards flow is equivalent to 6,000 Olympic swimming pools, every second.
The water returns southwards at this depth for thousands of miles, past Brazil. But if the north Atlantic changes — with more freshwater from rainfall or from melting ice — then the flow no longer becomes so salty and heavy. The conveyor belt slows down. At some point, the system passes a tipping point. The Amoc shuts off completely. The question was how likely was a collapse, and when might it happen?
In 2004, the disaster movie The Day After Tomorrow imagined an Amoc collapse. In it, much of the northern hemisphere is gripped by a new ice age. The filmmakers played loose with the science. In the movie, the Amoc collapses within days (in reality, scientists estimate it will take up to a century for the effects to play out) and a tsunami hits Manhattan (an Amoc collapse would have no such effect). Rahmstorf wrote a review of the film on his blog, urging scientists not to dismiss it despite the artistic licence taken. He also spoke to the screenwriter, Jeffrey Nachmanoff, at a preview of the film. “He told me, ‘If we were making films for a few million people, we would stick to the law of physics, but we are making films for a few hundred million people, and there we stick to the law of Hollywood.’”
Rahmstorf was pleasantly surprised by Nachmanoff’s grasp of climate politics. The film conveyed basic truths. At one point a politician tells a scientist, “Maybe you should stick to science and leave policy to us.” The scientist responds, “Well, we tried that approach.” (The scientist works at the National Oceanic and Atmospheric Administration, the same institution where Rahmstorf held a post.) Like other climate scientists, Rahmstorf was used to being ignored.
By 2013, IPCC reports showed a map of the world with rising surface temperatures indicated by shades of orange and red. Global warming was happening fast. But, amid all the red, there was a surprising blob of blue, representing not rising but cooling temperatures. The blob was located around the Irminger Sea, in the north Atlantic, exactly where the Amoc should be transporting heat. The models used by the IPCC mostly did not predict the cold blob. Since the 19th century, the northern Atlantic was the only region in the world to have become colder. “I thought this must be Amoc, why doesn’t the IPCC discuss this? It was really staring into my eyes, but nobody seemed to make a big deal out of this,” recalls Rahmstorf.
Rahmstorf’s research had moved on to other aspects of climate change, but now he turned back to the Amoc. In 2015, he published a paper with the US climate scientist Michael Mann and others, which catalysed the idea that an Amoc slowdown was not a future possibility but a present-day reality. It concluded the Amoc had weakened and called it “an unprecedented event in the past millennium”. If Greenland’s ice sheet continued to melt, the Amoc would diminish further. Mann describes Rahmstorf as “a fiercely independent scientist and thinker, who has often proved ahead of his time”. A common theme in his research is that “key climate impacts are being observed earlier than predicted by most climate models”.
The problem with studying the Amoc is a lack of direct evidence. Precise measurements of the current, via a series of moorings with instruments stretching down thousands of metres, date back only to 2004. This is not long enough for scientists to disentangle any slowdown from short-term variation. (After 2012, for example, the Amoc strengthened.)
Instead scientists must look for indirect evidence, or “fingerprints”. One is sea surface temperatures, such as the cold blob. Temperatures were once measured by sailors dipping buckets over the side of their boats while crossing the Atlantic; this method continued well into the 20th century. Now satellites help. Using climate models, Rahmstorf and co-authors estimated that the Amoc had weakened about 15 per cent since the mid-20th century. Another fingerprint is the declining salinity of the Amoc. Salt concentrations are at their lowest for 120 years.
Think of a boxer losing a fight. Every time he gets knocked down, it takes a little longer for him to get back to his feet. Similarly, as natural systems near collapse, they often lose their resilience and take longer to return from short-term fluctuations to their previous equilibrium. One of Rahmstorf’s colleagues at the Potsdam Institute showed that the Amoc was exhibiting such a lack of resilience. The results helped to convince Rahmstorf that a collapse was more likely than had been assumed.
In 2023, two Danish scientists, the brother and sister team of Peter and Susanne Ditlevsen, identified increasing fluctuations in the sea surface temperature of the North Atlantic as evidence of a pending Amoc collapse. Their model concluded that the Amoc was likely to pass the tipping point, after which collapse is inevitable, between 2025 and 2095. The most likely year was 2057.
Another group, based at Utrecht University in the Netherlands, showed that a good indicator of potential Amoc collapse was the freshwater transported at the circulation’s southern boundary in the Atlantic. The modelling was gruelling. It “took us six months or so on a substantial chunk of the supercomputer in Amsterdam”, says Henk Dijkstra, the team’s leader. Their conclusion, published in February 2024, was that the Amoc “is on route to tipping”. Such stark findings caused headlines globally, including in the FT. For Rahmstorf, different modelling was now pointing to the same alarming conclusion.
In another paper, still undergoing peer review, the Utrecht group used salinity changes to put the start of collapse between 2037 and 2064. This study is currently undergoing a major revision, following critiques.
Indeed, within the scientific community, the most alarming warnings about the Amoc are fiercely contested. Niklas Boers, Rahmstorf’s colleague who authored pioneering studies into the Amoc’s loss of resilience, says that tipping time estimates don’t factor in uncertainties in the data records, or uncertainties about future emissions and how these affect the climate. It matters, for example, not just how fast Greenland’s ice sheet melts, but how much of the meltwater flows to the areas south of Greenland where the Amoc sinking happens. “I really don’t think we can make any safe estimates on this tipping time,” says Boers. When data uncertainties are factored in, he adds, the Ditlevsens’ model results in a tipping point that could be imminent — or as far as 6,000 years away.
Other scientists question how Rahmstorf and others use sea surface temperature in the north Atlantic as a proxy for the strength of the Amoc. How warm the water is may not accurately represent the flow, especially over short time periods. Denis Volkov, an oceanographer at the University of Miami, has studied the Florida current, part of the Amoc, through its effect on the voltage in underwater telecoms cables. He concluded that, at least over a short time period, the Amoc was slightly more stable than previously thought. The likelihood of a complete Amoc collapse is “still uncertain”, says Volkov, calling for more research.
The IPCC, which effectively sets the scientific climate consensus, remains conservative. It has warned that a weaker Amoc could “cause a decrease in marine productivity in the North Atlantic (medium confidence), and more storms in Northern Europe (medium confidence)…” among other effects. But its last big report professed only “low confidence” that the Amoc had declined last century. Although it was “very likely” that the circulation would now decline, the panel had “medium confidence” that there would be no abrupt collapse by 2100.
Rahmstorf blames the IPCC process: the lack of Amoc experts involved and the inbuilt biases in major climate models. Anecdotally, he heard that, when several climate models were first designed, “the Amoc collapsed even in present-day climate . . . What do you do as a climate scientist, when you compare your model result for present-day climate with observations [which contradict it]? You conclude it’s wrong, we must change this model. However, if by design your model has an Amoc that is too stable, everything looks fine. So there is asymmetry there — which model errors get fixed, and which model errors don’t get fixed.”
Some content could not load. Check your internet connection or browser settings.
The models relied on by the IPCC do not include the effect of meltwater from Greenland changing the salinity of the North Atlantic. In November, a study by scientists in Sydney estimated that meltwater would be responsible for 40-50 per cent of the Amoc’s future slowdown, if the world did not dramatically reduce greenhouse gas emissions. And this was under an “optimistic” scenario, in which the Greenland ice sheet did not itself become rapidly more unstable. Greenland is the “big gorilla”, says Rahmstorf. “It’s not contributing a huge amount [today], but the share of meltwater from Greenland will increase as the world warms further.”
The IPCC concedes that climate models have such biases, which produce “unrealistic stability”, but says that this is taken into account in its conclusions. As a result, the panel “does not underplay the risk of Amoc collapse”, says Robert Vautard, a climate scientist and lead IPCC author. Nonetheless, its last major report only covered studies up to 2020; new research means the next report is likely to devote “significant space” to Amoc, he adds.
Rahmstorf has responded to some critics, but he also seems tired by resistance to the idea of looming Amoc collapse. “I have encountered a reluctance among colleagues to speak about uncertain risks, for fear of sounding alarmist.” He compares the risk of Amoc collapsing to the risk of a plane crashing. “You’re not going to ask, are you 99 per cent sure that it crashes . . . Even if the possibility of a shutdown happening is only 10 per cent, that is far too high to accept. And I now think it’s larger than 10 per cent.”
Lenton, of the University of Exeter, argues that scientists need to warn more about risks, given how the climate is already defying their expectations. “We sit after the end of more than 12 months of unprecedented sea surface temperatures and global average temperatures, and there isn’t a consensus in the scientific community as to why this is happening. Everyone’s got to have a bit of humility, and say, ‘Oh dear.’” Very unlikely scenarios are occurring. In 2021, the UK government’s climate risk assessment said the chance of Britain experiencing 40C heat by 2040 was “less than 0.02%”. That temperature was hit the following year.
In October, Rahmstorf, Lenton and 41 other experts signed an open letter, highlighting the possibility of an Amoc collapse this century. (Boers, who argues we can’t accurately predict when any tipping point will occur, also signed.) The letter also noted that the bigger possibility is that the Amoc passes the point of no return this century — and the effects fully unfold in the 2100s. The impact would be “catastrophic”, impacting “the entire world for centuries to come”. Rahmstorf presented the letter to Iceland’s climate minister. But is society capable of processing long-term, uncertain risks?
“In the long run, we are all dead,” said John Maynard Keynes. He wasn’t being flippant. Indeed, he was pushing back at economists’ tendency to downplay short-term suffering. The idea that society thinks too long-term is now almost funny. Companies fixate on their share prices. Politics works to the beat of the next election, in which future generations have no say. Economists’ accounting gives much less importance to future costs, thereby making long-term harm seem palatable. Scientific climate forecasts end at 2100, even though a baby born today in the west is expected to live beyond that date. Last September, in response to a parliamentary question, the UK government said it had not “assessed the effect of any slowing or collapse of the Atlantic Meridional Overturning Circulation (Amoc) on economic planning.”
The long-term thinking that does exist often doesn’t age well. Germany’s debt brake, justified as a way of stopping financial burdens being placed on future generations, has meant years of under-investment that the current generation is having to mop up. The Oxford philosopher William MacAskill, who championed long-term thinking, soon found his message tarnished by his close association with crypto entrepreneur Sam Bankman-Fried, now a convicted fraudster.
The warnings of catastrophic climate tipping points have not gone entirely unheeded. In September, the Advanced Research and Invention Agency, a new UK research agency, announced £81mn in funding for research into tipping points — including the Greenland ice sheet and the north Atlantic subpolar gyre, a system of currents that connects with the Amoc. The subpolar gyre is thought to be more vulnerable to collapse than the Amoc. Its effects would also hit Europe quicker.
The winning teams, to be announced in March, will be expected to develop new sensing systems, along with physical and statistical models. The hope is that the methods could then be extended to the Amoc and other tipping systems, like the West Antarctic Ice Sheet, whose melting could increase sea levels by more than three metres. Rahmstorf is involved in one group of bidders. Their theory is that, as the Amoc weakens, deep convection will slow — water will stop plunging thousands of metres deep in the north Atlantic, perhaps only plunging hundreds of metres. That would show the Amoc is doomed. There would be no way back on human timescales. But as the Amoc slowly wound down, society could begin a race to adapt. Whether the climate in northern Europe gets warmer or colder, it will get drier. That means investments in water storage should be worthwhile. So too insulation, which helps buildings both in hot and cold weather.
For now, unable to say with precision if or when the Amoc will collapse, scientists can merely flag the possibility and are unable to rule out a different scenario. Some environmentalists are wary of even talking about the Amoc collapse and other catastrophe risks. The public can already hear the alarm about climate change, argues Alex Evans, a former climate official in the UK government and the UN. “It makes them feel overwhelmed, so they tune it out.” Instead of pushing people into a “fight-or-flight state”, environmentalists need “a story of rebirth — of how we make it through”.
We implicitly assume that our society is smarter than the ones that have gone before — that it is building things that will last for longer. In London, St Paul’s Cathedral is now more than 300 years old. If the worst projections of future extreme events are correct, many things that we build now will not last half as long.
Some content could not load. Check your internet connection or browser settings.
In his own field, Rahmstorf had hoped to be able to say, with precision, where the Amoc’s tipping point lay. Despite the recent advances, he concedes that question is unresolved. He is due to retire in 2026, before any tipping point project completes. “Data uncertainty is substantial. But uncertainty is not our friend. Uncertainty could mean the tipping point is passed early,” he says.
There is something even more important than more climate research: stopping emissions to reduce the risk of Amoc collapse. In a study awaiting peer review, Rahmstorf and other scientists have extended the IPCC’s models beyond 2100, and found that, in all scenarios where society fails to cut emissions sharply, the Amoc collapses. Even with dramatic emissions cuts, some models show an Amoc collapse.
“Would any great leaders in the late Middle Ages have taken decisions that, for example through sea level rise, wiped London and other great cities off the map? How would we look back at them?” he says, quietly, urgently. “My children are expected to live past the year 2100, and what about my grandchildren? They will really very strongly see the consequences of what we decide now.”
After the re-election of Donald Trump as US president, I asked him whether he can escape the implications of his work. “I try my best to not let the bad news about the escalating climate crisis take over my life, although this is increasingly hard,” he replied.
Back to the future. In a dramatically cooler Britain, when the winter storms come, higher sea levels mean that water surges in from the coast. The flooding is intense. The winds rip across roads, tearing down power lines. Those without battery storage are frequently cut off. Instead of the UK being a refuge for others escaping climate disruption, we might struggle to support even the population we have now.
In isolation, and in theory, many of the effects of climate change might seem manageable. The overarching problem is society. Humans can adapt to some disruption, but will they? Imagine a Britain where extreme weather bites deeper, food prices rise higher and young people choose to leave. If recent politics are any guide, that is a recipe for anger and nihilism, not co-operation and science. Radical politicians would mock the idea of global warming, arguing that Britain should burn more fossil fuels to stay warm and spray as much fertiliser as possible to eke out cropland.
A smart, resilient society could react in lots of ways to an Amoc collapse. If we were such a society, wouldn’t we already be doing a lot more to lessen the risk of it happening in the first place?
Henry Mance is the FT’s chief features writer
Data visualisation by Jana Tauschinski
Find out about our latest stories first — follow FT Weekend Magazine on X and FT Weekend on Instagram
