July 25, 2023

expert reaction to paper warning of a collapse of the Atlantic meridional overturning circulation

A paper published in Nature Communications warns of a forthcoming collapse of the Atlantic Meridional Overturning Circulation (AMOC).

Prof Andrew Watson FRS, Royal Society Research Professor at the Global Systems Institute, University of Exeter, said:

“This study highlights that the North Atlantic circulation is showing signs of instability, which might indicate that a collapse of the overturning could occur, with major climate implications. However, the instability could also be less dramatic, not a full-scale shutdown  but a change in the sites of deep water formation for example. This might be more likely, since most models currently suggest a slowing, but not a shutdown of the overturning during this century.

“While their tipping point analysis is robust and suggests the system is approaching a transition, it doesn’t give a clue as to what lies beyond the transition. I’m unconvinced that this would be a catastrophic collapse. The models they show in figure 2 are “intermediate complexity” and less reliable in general, than the CMIP models used by the IPCC. The N. Atlantic is a complex system with at least three possible regions of deep water formation (The Labrador, Irminger and the Greenland / Iceland Seas) and it could shift between these as major formation regions of deep water, which would probably have a similar tipping point signal.”

Prof Richard Pancost, Professor of Earth Systems at the University of Bristol Cabot Institute for the Environment, said:

“A change in ocean circulation has long been considered to be a critical climate change induced tipping point. This paper provides evidence that our carbon emissions have pushed us towards this tipping point.  What is particularly important about this paper is how it highlights the non-linear responses to climate change – we are not sure what degree of warming will cause key aspects of the Earth system to change. This is further illustrated if we consider how this change in ocean circulation will have impacts on other parts of the Earth system. How will it affect the climate of different Atlantic regions?  How will it affect Arctic or deep sea ecology? As we continue to burn fossil fuels, we risk crossing a variety of thresholds for both our climate and nature – potentially irreversibly.”

Dr Ben Booth, Senior Climate Scientist, Met Office Hadley Centre, said:

“The new work is interesting as it looks to move past the limitations of our short observational record of the overturning.  Whilst there is definitely a role for papers like this, the conclusions are far from settled science.  Other factors, such as volcanic and industrial aerosols, also project onto the same historical records that this study uses as purely a fingerprint for past historical change in overturning.  This new study neglects these factors, so a lot of caution needs to be taken in interpreting the findings as a definitive inference of the future overturning change.”

Prof Penny Holliday, Head of Marine Physics and Ocean Circulation at the National Oceanography Centre, and Principal Investigator for OSNAP, an international programme researching AMOC processes, variability and impacts, said:

“Confidence in the validity of the conclusions are undermined by our knowledge that sea surface temperature of the North Atlantic subpolar gyre is not a clear indicator of the state of the AMOC, and that there is no evidence that the AMOC has dramatically weakened in the past 50-75 years. A collapse of the AMOC would profoundly impact every person on Earth but this study overstates the certainly in the likelihood of it taking place within the next few years.”

Does the press release accurately reflect the science?

“On the whole it does – the title of the paper is more sensational than the actual statements within it, and the press release does make that clear. However there are two statements that are not accurate as follows:

‘The strength of the AMOC has only been monitored continuously since 2004 and these observations have shown AMOC to be weakening’

“This is stated in the paper but it is not correct information. The observations since 2004 show that the AMOC goes through fluctuations of being in a stronger or weaker state that last for about 10 years. The observations since 2004 show the subtropical AMOC getting slower from 2004 to 2012, but gradually becoming stronger since then.  The only data from AMOC observations shown in the paper are from 5 sparse ship surveys and are used out of context – the authors use them to argue for a severe decline in the AMOC, but that interpretation has long been discredited in the scientific literature (including in the reference they cite for it).

‘The authors found early warning signals of a critical transition of the AMOC system and suggest that it could shut down or collapse as early as 2025 and no later than 2095.’

“This is not quite as the paper states. In the paper the time period of potential collapse depends on choices they have made in how they construct the time series of sea surface temperature which they use as evidence for change. They present three versions of the temperature records, and the three resulting model predictions suggests a collapse is ‘likely’ at any time from 2024 to 2180. The 2025-2095 is the period of time their statistical model predicts that a full or partial collapse is most likely.

Is this good quality research?  Are the conclusions backed up by solid data?

“The use of statistical models to test whether a tipping point can be detected is something that previous authors have done using datasets from climate models. So the approach is valid, and the topic is of great interest because of the potentially very serious climate impacts that a collapse of an AMOC would bring to the world. However there are some assumptions in the methodology which means that the results are not be quite as solid as the title and abstract suggest (see below). The authors acknowledge in the Discussion that there is large uncertainty in their conclusions.

How does this work fit with the existing evidence? 

“The conclusions are different to the consensus derived from climate projections as described by the IPCC AR6 assessment. The averaged AMOC projections from climate models under all the IPCC emissions scenarios all show an AMOC decline, but not a collapse (a “high confidence” conclusion). Some individual climate model runs do show a future collapse in the AMOC, so the possibility cannot be entirely ruled out.

Are there any important limitations to be aware of? 

“There are some questionable assertions and decisions in the methods as follows. The authors state confidently that the sea surface temperature (SST) of the North Atlantic subpolar gyre can be used as a proxy for the strength of the AMOC. The validity of an SST proxy for AMOC strength is a matter of ongoing scientific debate however, because it is based on model behaviour and is not proven using real-world data. There is solid evidence that there is no such clear relationship, especially on timescales of less than 30 years.

“I believe the authors have overstated the pattern of subpolar North Atlantic SST change by subtracting two (and three) times the global mean surface temperature trend. This is not the usual approach for highlighting North Atlantic regional temperature trend (instead it is more usual to subtract just 1 x the global trend). The choice means that some of the SST data they use in the statistical model has exaggerated decline since the 1970s when the global SST began to sharply rise. In the version of the statistical model for which the global mean SST trend is removed, the predicted likely time of a partial or complete collapse becomes later and over a wider window of time.

“As mentioned above, the actual observations of AMOC since 2004 have long-since discredited the evidence that the authors are using to validate their modified SST temperature record. The 5 data points they show in the paper were collected several years apart by ship surveys, and it is well known and well established that they give a highly misleading impression of AMOC decline. All the observational evidence we have shows no evidence of dramatic decline in the AMOC over the past 50-75 years.

How uncertain are the uncertainties?

“The authors say that the model’s 95% confidence interval is 2025-2095. This is a measure of statistical uncertainty and they state in the discussion that they cannot rule out slowing rather than a collapse, as well as listing other reservations and caveats. Because of the limitations of their use of modified SST as a proxy for AMOC, the uncertainty in the stated message in the title and abstract is high.

What are the implications in the real world? 

“The potential for the AMOC system of currents to collapse under global warming is a high impact, low likelihood scenario, and policymakers and planners do need to be aware of it. NOC and international partners are investing in ongoing observations of the AMOC in order to determine how closely changes in AMOC contribute to changes in SST and consequential climate and social and economic impacts on people. The strength of the out-of-sight ocean currents of the AMOC has surprisingly direct impacts on food, water and energy security, infrastructure risk, biodiversity, and human health. The paper demonstrates that decades of observations are needed to be able to detect a major tipping point in the AMOC, and the authors call for continued measurements of these great Atlantic ocean currents for long enough to do so.

What are the tangible impacts of a collapse? 

“The AMOC carries huge amounts of heat northwards through the whole Atlantic Ocean, setting climate conditions for all the Earth’s continents. If it switched off, the result after a few decades would be much lower surface temperatures and stronger winds across the whole northern hemisphere (land and ocean). Heat would pool in the Southern Ocean and South Atlantic, but over the southern continents, temperatures would also decrease. Major rainfall zones would shift, leading to far less rainfall over Europe, North and Central America, north and central Africa and Asia, and more over the Amazon, Australia and southern Africa. Sea ice would extend southwards from the Arctic into the subpolar North Atlantic, and the Antarctic sea ice would extend northwards. For people and governments this would lead to dramatic change in every nation’s ability to provide enough food and water for its population. Energy supply and demand would change rapidly with new climate conditions and infrastructures would need heavy investment to adapt and cope. The patterns of vector-borne disease and health (including mental health) would be profoundly affected. World-wide many land and marine ecosystems would be unable to cope and adapt to such fast changing climate conditions and biodiversity would be severely impacted.

Is there any overspeculation?

“The authors are appropriately careful not to overspeculate in the text, although the title of the paper and the text of the abstract is rather more dramatic and conveys a more certain outcome than the authors actually write in the results and discussion. They describe the potential for AMOC collapse within a few years as “worrisome” and the evidence as something that we should not ignore. It’s hard to disagree with that.”

Dr Jon Robson, National Centre for Atmospheric Science and the University of Reading, said:

“There is significant evidence that the AMOC could undergo a significant and abrupt weakening, which would have severe consequences globally, and for the UK, including changes in rainfall, winds and sea level.

“However, there are large uncertainties involved in predicting if and when an abrupt weakening of AMOC could occur. For example, this particular analysis is based on only one time series of sea surface temperature, which might not provide the best indication of the AMOC. It also ignores a range of processes that may be important for shaping the observed changes over the North Atlantic.

“Nevertheless, it is clear that the warning lights are flashing on for the North Atlantic climate system and further greenhouse gas emissions are only going to increase the likelihood of abrupt changes in the North Atlantic. Given the slow progress in reducing greenhouse gas emissions, it’s crucial that we continue to bring a range of observations and models together in order to be able to make more accurate predictions of the potential for rapid North Atlantic climate change.”

Prof Jeffrey Kargel, Senior Scientist, Planetary Science Institute, said:

“The authors, Peter Ditlevsen and Susanne Ditlevsen, developed a model that describes a rapid shut-down of the Atlantic Meridional Overturning Circulation (AMOC), which connects to the publicly better known Gulf Stream ocean current. The Gulf Stream bathes the U.S. Eastern Seaboard, the U.K., Iceland, and Scandinavia in warm water, thereby moderating their winters, so what the authors have modelled is no small matter. The AMOC system of ocean currents is regulated in part by melting ice in Greenland and the seasonal melting of Arctic sea ice. Outstanding questions about the modelled shutdown of this climate regulating system are: to what degree, how fast, and when will these changes occur? 

“Many scientists have examined this climate-critical part of the Earth and have found evidence for occasional weakening and shut down of AMOC, then re-establishment after several centuries. This also happened during the Ice Age. A European climatologist, Niklaus Boer, and many others, including the two authors of this new study, point to evidence that the AMOC system of ocean currents is weakening and becoming more erratic. This behaviour is similar to the Ice Age fluctuations that were driven by ice sheet melting, but this time industrial emissions of greenhouse gases are driving it instead of natural changes in greenhouse gases. The authors do not press the causes, but melting ice is involved then and now. One important difference is in the time scale: today’s melting has had a sudden onset due to fossil fuel combustion just in the past 150 years and primarily the last 50 years; however, in nature, when similar North Atlantic thermal events happened, the triggering climate disturbances took place over many centuries or millennia.

“The authors point out that recent erratic, anomalous fluctuations in North Atlantic sea surface temperatures may presage a complete or partial shutdown of the AMOC system. I see an interesting parallel in stock market behaviour: when a bull market nears its end, volatility increases, and then there is a rapid plunge to a different stable state of collapsed stock prices. 

“The authors show that all the related disturbances to ocean currents, like the Gulf Stream, and consequent climatic changes, could happen–and they claim will happen, with 95% confidence–some time this century, even this decade, and do so suddenly in just a few years. This modelled imminence is the newest and most worrisome part of this research paper.”

Prof Tim Lenton, Director of the Global Systems Institute, University of Exeter, said:

“The authors make important improvements to the method of providing early warning of a climate tipping point directly from data.

“They provide an alarming result, forecasting the passing of a tipping point for collapse of the Atlantic Meridional Overturning Circulation (AMOC) in the middle of this century – with it becoming possible in just a couple of years’ time and almost certain to have occurred before the end of the century, if human-caused global warming continues at the current rate.

“Once past the tipping point, the collapse of the AMOC would be irreversible. The collapse and its impacts will take time to unfold – but how long is uncertain. Past abrupt changes in the AMOC during the last ice age happened on a decadal timescale, but some models predict a slower winddown of the overturning on a century timescale.

“The impacts of a collapse of the AMOC would be profound, including radically changing the climate in the North Atlantic region, and seriously disrupting the West African and Indian monsoons. For the UK the resulting drying of the climate is predicted to eliminate arable farming and trigger a water crisis (https://www.nature.com/articles/s43016-019-0011-3). Worldwide the viable areas for growing wheat and maize are predicted to shrink considerably (https://www.oecd.org/environment/climate-tipping-points-abc5a69e-en.htm).

“It is possible that the tipping point would lead to a partial rather than total collapse of the AMOC. It could involve a collapse of deep convection in the Labrador Sea region (but not the Greenland-Iceland-Norwegian Seas region). But that can have major impacts on a decadal timescale in many of the same regions.

“This study is yet another warning that we should be doing everything in our power to accelerate action to decarbonize the economy and get to net zero greenhouse gas emissions as soon as possible.”

Prof Meric Srokosz, National Oceanography Centre, University of Southampton, said:

“While the possible collapse of the AMOC with significant climatic impacts is a concern, providing a warning of its collapse is problematic as a long set of observations is required. In this paper the warning depends on using proxy AMOC data (here based on sea surface temperature, SST) as direct continuous AMOC measurements are only available since 2004. The warning comes from applying statistical techniques to a long time series (over a century) of proxy AMOC data, but the warning is only as good as the proxy data are in representing the true AMOC. So, this warning needs to be treated with caution as there is no consensus as to which proxies can accurately capture the behaviour of the AMOC over the long term.”

Prof Jonathan Bamber, Director of the Bristol Glaciology Centre, University of Bristol, said:

“This is an interesting study looking at an extremely important, and complex, part of the climate system. The strength of the Atlantic Meridional Overturning Circulation (AMOC) is extremely important for the climate of the Northern Hemisphere so it has received a lot of attention from researchers. Some studies suggest that it could be bi-stable and has an “on” state and an “off” state and that it can switch rapidly between these two. Other, mainly more recent studies, suggest that the AMOC responds more smoothly as other parts of the climate system change. In this study, the authors assume it is bi-stable and make several other assumptions to determine how long it will take for the AMOC to switch “off”. If their assumptions are correct, and the model they develop is correct, then their conclusions have profound and serious implications for the climate of NW Europe and North America.  A collapse of the AMOC would be disastrous. Other plausible models and assumptions could have been used, however, which would result in a different conclusion. This study highlights how important it is to continue to monitor AMOC variability and to improve our understanding of its stability under present-day and future climate conditions.”

Comment sent out by our colleagues at SMC Spain:

Prof Niklas Boers, Professor of Earth System Modelling at the Technical University of Munich, said:

“I do not agree with the outcome of this study. While the qualitative statement that the AMOC has been losing stability in the course of the last century is true and supported by the data, uncertainties are too high to reliably estimate a time of tipping. In particular, the uncertainties in the heavily oversimplified model assumptions by the authors are too high. Moreover, the uncertainties in the underlying datasets are huge and would make the extrapolation carried out by the authors far too uncertain to actually report a year or even a decade for the AMOC tipping.”

Comments sent out by our colleagues at SMC Germany:

Prof. Dr. Jochem Marotzke, Director of the Department Ocean in the Earth System, Max-Planck-Institute for Meteorology, Hamburg, Germany, said:

“The work provides no reason to change the assessment of the 6th IPCC Assessment Report: ‘There is medium confidence that there will be no abrupt collapse before 2100′. The statement so confidently made in this paper that collapse will occur in the 21st century has feet of clay. The maths are solid, but the starting point is highly dubious: the essential equation – marked with (1) in the paper – relies on the simplified models representing bifurcation – i.e. AMOC collapse – also being correct. But the more comprehensive models do not show this very bifurcation. In this respect, the paper does not live up to its self-imposed claim: ‘The strategy is to infer the evolution of the AMOC solely on observed changes in mean, variance and autocorrelation.’ The interpretation relies to an enormous extent on the authors’ theoretical understanding being correct, and there are huge doubts about that.

“It must be added that there is considerable doubt as to whether surface temperature measurements are a valid proxy for the AMOC. Again, the paper addresses these uncertainties inadequately.

“When reporting about this study, it is important to include the key aspects in which this paper fails to include the scientific uncertainties.

Prof. Dr. Johanna Baehr, Institute of Oceanography, University of Hamburg, Germany

“The paper can be a valuable contribution to the discussion on mathematical models of climate variations. The purely statistical investigation of the AMOC focuses exclusively on surface temperatures in the North Atlantic. Thus, it does not do justice to the complexity of the climate system in many respects. Furthermore, with regard to the reference to current model simulations and extensive measurements, the study does not deliver what it initially suggests.

“Personally, I am surprised that the authors derive such far-reaching conclusions regarding the future development of the AMOC from this purely mathematical analysis. In my view, the results of the study are not transferable to the actual future development of the AMOC. An abrupt collapse of the AMOC – as described in the 6th IPCC Assessment Report – is still not to be expected in the foreseeable future.”

“Such a purely mathematical analysis, which ignores the physical perspective, does not help in drawing more attention to climate change. We are already seeing right now a whole series of events that are also effects of anthropogenic, i.e. human-made, climate change. These effects are reason enough to act. It is clear that greenhouse emissions must be drastically reduced – the sooner the better. Acting now will make the difference between 1.5 or 2 or 3 degrees of global warming, with drastic consequences for us all.

Dr. Levke Caesar, Research Associate in the Department of Oceanography, Institute of Environmental Physics (IUP), University of Bremen, and Research Associate in the Department of Oceanography, MARUM – Center for Marine Environmental Sciences

“Tipping elements and tipping points can be described mathematically. Such mathematical models show that systems approaching their tipping point change their behaviour. In particular, both the variance of the system (in simple terms, how much does the strength of the system fluctuate around its mean) and its autocorrelation (in simple terms, how much does the strength of the system in one year influence the strength in the following year) increase. The measurements of the changes in variance and autocorrelation can therefore be used to derive early warning signals that may indicate a ‘tipping‘ of the AMOC – i.e. a more or less unstoppable transition of the AMOC from its current flow strength to a significantly lower flow strength.

“The difficulty lies in the choice of observational data that is used to ‘feed’ the mathematical model. The AMOC is a complex system of surface and deep (around two to three kilometers depth) ocean currents that cannot simply be measured directly. Measuring instruments have been installed at various locations in the Atlantic since the beginning of the 2000s (e.g. the RAPID and the OSNAP projects), measuring the flow strength of the AMOC at different latitudes as well as the deep water formation that is an important driver of the AMOC. But the resulting time series are too short to be able to make statements about the stability of the AMOC. Therefore, the authors of the study use a so-called ‘AMOC index‘ or ‘AMOC proxy‘, i.e. they use other observational data from which the state of the AMOC can be derived indirectly or approximately. For the AMOC, the measurement of sea surface temperatures in the subpolar North Atlantic in comparison to the rest of the planet has proven useful. While these data suggest that the AMOC has been weakening over the last 70 years [1], they are not yet sufficiently perfected and tested to make reliable statements about the future of the AMOC.

“The question of whether and, if yes, when the AMOC might reach its tipping point has been debated for decades. Paleo data from ice and ocean sediment cores indicate that the AMOC was significantly weaker at earlier times in Earth’s history than it is today, and that the North Atlantic and the adjacent land masses were correspondingly much colder. Also, both simple mathematical and complex climate models show that the AMOC can – in theory – collapse.

“Future simulations of climate models all show a weakening of the AMOC over the course of the 21st century, with the extent of the weakening also depending on the amount of future carbon dioxide emissions and being around 30 to 45 percent by 2100. Yet there are very few models that simulate a collapse of the AMOC before 2100, so we are relatively certain that the probability of this is low. However, we also know that models are not able to model the AMOC perfectly and there is evidence that most models overestimate the stability of the AMOC. Investigating the stability of AMOC using observational data is therefore very important.

“It is possible that the AMOC has not one but several tipping points – where a tipping point refers to the transition from one stable state to another stable state that is significantly different from the first. That is, it could be that crossing the first tipping point does not lead to a complete collapse of the AMOC, but only to a weakening of the system compared to today.

“So far, we can only estimate the consequences of a significant weakening of the AMOC. For example, we know that a collapse of the AMOC could increase sea level rise on the east coast of the USA by up to several decimeters and temperatures in and around the North Atlantic could drop by several degrees. A weaker AMOC has also been linked to stronger as well as more frequent winter storms in the UK. Additionally, a 2016 study found that heatwaves in Europe could also be favored by a weaker AMOC. This connection, which seems somewhat paradoxical compared to the cooling caused by a complete collapse, is due to the fact that a weaker AMOC cools the subpolar North Atlantic. This cooling in the North Atlantic sea surface temperatures can then favor an atmospheric circulation pattern that directs hot air from the south into Europe.

“Yet the AMOC affects not only Europe and the USA, but also the equatorial region, and seems to have a particular influence on the probability and quantity of precipitation in parts of Africa and Asia. Thus, there is a large number of people who would be affected by a weaker AMOC. On the whole, however, there is still need for additional research to better quantify the impact of AMOC weakening on both the global and the local scale.

“When reporting about this study, it is very important to distinguish between the Gulf Stream and the AMOC. The AMOC is equal to the Gulf Stream System, but AMOC is not equal to the Gulf Stream. The Gulf Stream is a wind-driven current that flows along the east coast of the USA, separates from it at about 40 degrees north latitude and flows towards the center of the Atlantic, from where it is called the North Atlantic Current. The Gulf Stream system – the AMOC – refers to the northward transport of warm, salty surface currents, and southward return flow at depth and covers the whole length of the Atlantic. Gulf Stream and Gulf Stream system certainly overlap, but they are not identical.

“In summary, it is important to understand that the study does not show that the AMOC will collapse in 2050. However, it does suggest – as does another study by Niklas Boers in 2021 [2], who examined several AMOC indices – that the AMOC may have already lost some of its stability. We should take this as a serious message to further study the system and reduce our greenhouse gas emissions, which are the main cause of anthropogenic climate change.

“The authors have developed a method to determine the tipping point of the AMOC as soon as we have sufficiently good observational data for the past variations of the AMOC. However, at the moment, these data are still based on indirect measurement: in this study the sea surface temperatures in the North Atlantic. While this data indicates that the AMOC has weakened over the last 70 years [1], it has not yet been sufficiently perfected and tested to be able to make reliable statements about the future of the AMOC. It is therefore important that we continue to measure the AMOC directly, ideally at several locations in the Atlantic. We also need to increase our efforts in developing better proxy data for past variations of the AMOC, as we cannot travel back in time and take additional direct observations. As a matter of fact, scientist, including myself, are working at answering these questions at the moment, as part of the EPOC (Explaining and Predicting the Ocean Conveyor) project that is funded by the European Union.”

Prof. Dr. Stefan Rahmstorf, Head of the Research Area Earth System Analysis, Potsdam Institute for Climate Impact Research (PIK), Potsdam, and Professor of Physics of the Oceans, Institute of Physics and Astronomy, University of Potsdam.

“The work by Ditlevsen & Ditlevsen coming out tomorrow fits in with earlier studies looking with somewhat different methods and data sets at the issue of early warning signals.

“The new paper comes to similar conclusions: there is still large uncertainty where the tipping point of the AMOC is, but the new study adds to the evidence that it is much closer than we thought just a few years ago.

“As always in science, a single study provides limited evidence, but when multiple approaches lead to similar conclusions this must be taken very seriously. Especially when we’re talking about a risk that we really want to rule out with 99.9 percent certainty. The scientific evidence now is that we can’t even rule out crossing a tipping point already in the next decade or two.

[1] Caesar L et al. (2018): Observed fingerprint of a weakening Atlantic Ocean overturning circulation. Nature. DOI: s41586-018-0006-5.

[2] Boers N (2021): Observation-based early-warning signals for a collapse of the Atlantic Meridional Overturning Circulation. Nature Climate Change. DOI: 10.1038/s41558-021-01097-4.

‘Warning of a forthcoming collapse of the Atlantic meridional overturning circulation’ by Peter Ditlevsen and Susanne Ditlevsen was published in Nature Communications at 4pm UK time on Tuesday 25 July.

DOI: https://doi.org/10.1038/s41467-023-39810-w

Declared interests

Dr Ben Booth: “No financial conflicts. I have published about other drivers of Atlantic SST changes.”

Prof Jonathan Bamber: “I am not aware of any competing interests.”

Prof Meric Srokosz: “I have been responsible for the RAPID programme as science coordinator since it started.”

Prof Jeff Kargel: “I have no conflicts of interest.”

For all other experts, no reply to our request for DOIs was received.