RealClimate logo

Atlantic circulation change summary

Filed under: — gavin @ 19 January 2006 - (Français)

Nature this week has an excellent summary of the state of the science with regards to possible changes in the ocean thermohaline (or meridional) circulation in the Atlantic and its impact on climate. Even though it quotes a couple of us, it’s still worth reading if you want to understand how results like the Bryden et al paper – that suggested that the Atlantic overturning had reduced by 30% in recent decades – are assimilated into the scientific picture.

The [Bryden et al] result came as a surprise to those in the field. Few scientists had thought that such dramatic slowing of the thermohaline circulation could happen so soon. Models suggest that the increase in fresh water needed for a conveyor shutdown would not be expected without a global warming of 4–5 C; warming in the twentieth century is currently put at 0.6 C (ref. 3). The most complex computer models of the climate and oceans, the sort used to make climate predictions for the Intergovernmental Panel on Climate Change (IPCC), suggest that the flow might be expected to slow by an average of 25% by the end of the twenty-first century, but not to shut down completely.

The reason for this surprise is that other data of relevance, such as the changes in salinity or deep convection or lack of observed cooling in the North Atlantic, aren’t obviously consistent with such a large change:

[Ruth Curry’s] recent analysis of 1950 to 2005 salinity data suggests that 4,000 cubic kilometres — eight times the annual outflow of the Mississippi river — of fresh water have accumulated in the upper ocean layers since the 1960s. “The extra freshwater input is beginning to affect density,” she says. But the amount of fresh water needed to shut down the thermohaline circulation in Rahmstorf ’s comparisons is an order of magnitude greater than the flux reported by Curry, and she agrees that the circulation will not be unduly affected this century. Peter Wadhams, an oceanographer at the University of Cambridge, UK, last year reported a substantial weakening of convection ‘chimneys’ down which surface water flows in the Greenland sea, but it is unknown how much of the observed effect is due to natural variability.

It is possible that the Bryden result was an unlucky statistical artefact: “The results are based, after all, on just five snapshots of an extremely noisy and under-sampled system,” says Carl Wunsch. His own analysis doesn’t suggest any significant change over the last decade or so (though it’s probably worth pointing out that Wunsch’s opinions of the relative (un)importance of the Atlantic circulation for climate change are probably not (yet?) mainstream in the community).

Everyone quoted is however agreed on one thing: “the notion that [a future change in the themohaline circulation] may trigger a mini ice age is a myth”. The evidence of previous changes for instance at the Younger Dryas or during the 8.2 kyr event is quite strong, and significant coolings were observed particular around the North Atlantic, but even such localised coolings are not predicted to occur if the circulation slows as an effect of global warming.

It is however a complicated business, and the stability of this circulation depends on many aspects of climate that are poorly observed and uncertainly modelled. So it may yet be some time before new observations (such as the permanant monitoring array recently installed along a section of the ocean), better modelling, and a better appreciation of the paleo-climatic data add up to a coherent understanding of this interestingly counter-intuitive aspect of climate change.

33 Responses to “Atlantic circulation change summary”

  1. 1
    John L. McCormick says:

    The impact of Hollywood and media iterpretation of consequences of slowdown or halting of the THC has frozen the lay discussion of this extremely complex heat transport system.

    Though not a scientist, I can only follow the conversation while reading journals and reports that indicate there is a lot more at stake with a change in the flow of the THC. That includes worsening droughts in West Africa and other precip and weather anomalies RealClimate scientists are tracking. Perhaps a more comprehensive look at aspects of the THC aside from its providing some winter warmth to Northern Europe would get the discussion of impacts on to a more enlightening track.

    Dr. Broecker, for one, has a wider view of THC’s contributions to the planet’s weather than most of the opinions I have read thus far. Seems we are not giving the public the full treatment on a changing THC if we are not also discussing impacts being felt (future impacts as well) in the Southern Hemisphere

  2. 2
    Mauri Pelto says:

    The single statement included here is something many scientists I have talked with are still not up to speed on “the notion that [a future change in the themohaline circulation] may trigger a mini ice age is a myth”. This is an idea that Broecker let out of the bag to which counter evidence has not been offered in equal measure to disuade. Putting that aside we can focus on the true implications of changes in the gulf stream and thermohaline circulation and their impacts. You mention Ruth Curry’s results, where is this freshwater preferentially found in the ocean or is it well distributed.

    [Response: The freshwater is mainly in the North Atlantic (graphic here). -gavin]

  3. 3
    Maynard Handley says:

    “the notion that [a future change in the themohaline circulation] may trigger a mini ice age is a myth”

    What are we to make of this? One interpretation is that a future change in the themohaline circulation will not trigger *global* lower temperatures, but I was unaware that anyone was arguing for this.
    What I thought the concern was was the temperature of Europe, and that a future change in the themohaline circulation *would* substantially reduce temperatures in Europe. Am I to interpret the above statement as saying that this is no longer believed to be true?

    [Response: All other things being equal, a decrease in the THC will lead to a cooling around the North Atlantic and Europe. That isn’t to say that the glaciers will return – that would require a much larger cooling. However, a THC change as currently projected as a consequence of increasing greenhouse gases is much more likely to only moderate the rate of warming (i.e. Europe will still warm, but possibly not as fast as the rest of the planet). – gavin]

  4. 4


    Perhaps you and Mike will now see fit to cut Bill Gray a little slack for not jumping at your previous request that he immediately retract his thoughts linking hurricane activity to the the THC based upon the Bryden et al. results as you all called for here.

    “By Gray’s very clearly articulated reasoning, there should have been a downturn, not the observed upturn in major Atlantic hurricane activity over the past several decades (in the absence of other—including anthropogenic–influences on tropical Atlantic climate) if Bryden et al.’s results are correct. It will be interesting to see if Gray, and others, will change their line of argument in the face of this new study. Today, the last day of the official 2005 Atlantic Hurricane season, might be a fitting opportunity for them to do so.”


    [Response: What we were criticising was Gray’s dogmatic insistence that hurricanes were definitely linked to the THC when he has no independent evidence whatsoever that the THC was strongly increasing. As the above-linked article shows, there is a lot of uncertainty in what is going on to sea surface temperatures (in which both natural cycles and global warming play a role) and absolute insistence that either player must be negligible is unsupportable. I would still like to see him make that statement. – gavin and mike]

  5. 5
    Mauri Pelto says:

    Gavin: That was my understanding that most of the freshwater addition was in the North Atlantic. Has salinity increased with higher SST in the origination are of the Gulf Stream as well, which would increase the north, south surface density gradient?

    [Response: Density gradients appear to be remarkably stable since the salinity changes have been compensated (to a large degree) by temperature changes. That is, saltier water is also slightly warmer, and the freshening water is also getting cooler. – gavin]

  6. 6
    Pete Best says:

    Let me attempt to get this straight. The Sun heats the earth and ocean at the equator (and everywhere else) and this in turn heats the atmosphere. Evaporation from the oceans is not happenning faster as the amount of energy from the sun is constant but we are saying that due to increased CO2 levels the water vapour is increasing relative to the amount falling back to earth due to the atmospheres ability to hold more water vapour ? So in effect the seas are becomming saltier around the equator. Now further north where the water vapour goes we get more precipitation than evaporation and because more fresh water is falling then indeed over time the northern seas are becomming fresher.

    Now get me if I am wrong but this sounds like the fundamental reshaping of a complex interplay between heat and water transfer between the equator and the sub tropical and polar regions to me or am I overeacting.

  7. 7
    Lynn Vincentnathan says:

    I read last year about findings that plankton & sea life in the Atlantic had decreased in some areas due to less overturning or churning of nutrients, I think due to thermohaline decline (?). So even without an ice age, there are other problems.

    I also read some 15 years ago about a plankton decline in some area of the Pacific, due to warmer waters, with concomitant fish decline.

    Eventually we need to get all the various reports, even those not related to GW (such as those dealing with regular pollution), and get a better picture of it all.

  8. 8
    JohnLopresti says:

    The Woods Hole Institute graphic [Curry et al] reminded me on macro scale of several factors pertaining to buffered solutions and their exceptional propagation of electromagnetism compared to fresh water; additionally, looking at the ocean as a fairly confined liquid mass, I would look for the effect induced by the eastward force of the planet’s westward rotation. Also the ice cap melt entering from polar portals shows nicely in the WHI progression of diagrams.

  9. 9
    Dietmar Temme says:

    The most complex computer models of the climate and oceans, the sort used to make climate predictions for the Intergovernmental Panel on Climate Change (IPCC), suggest that the flow might be expected to slow by an average of 25% by the end of the twenty-first century, but not to shut down completely.
    The crucial word is average. From paleontology there are well documented “Heinrich-Events” at the end of some ice-age, indicating an oszillating behaviour between an Europe under ice and free, within small periods of time. So the question is: can by very hot summers and unusually warm winters (cf. news from Spitzbergen) the THC shut down for a season or two?

    [Response:You probably mean Dansgaard-Oeschger events, not Heinrich events. But I think you’ve been misled by the term “rapid”: D-O event onset may be fast but they last for quite a time: there is no way to fit them into a season or two – William]

  10. 10
    David Bassendine says:

    Re:#3. Gavin, you say that a decline in the Atlantic THC will most probably moderate the rate of warming in Europe. I suppose you are talking about annual means here, but would such as slowdown also cause changes in seasonality towards a more ‘continental’ climate (i.e. more extreme winters) as the oceanic contribution to heat transport diminishes?

    [Response: You generally see a bigger effect in winter, so an increase in seasonality could be expected – but for the future projections, a more rapid warming of summers is the more likely expression of that. I haven’t seen any results that suggest that winters will get colder even as the annual mean temperature increases. This is really a function of the rate of change though – for the moderate slowdowns that the models project for the future this doesn’t happen, in ‘8.2 kyr event’ simulations (for instance) it does. -gavin]

  11. 11
    Hank Roberts says:

    Gavin’s “for instance” link in his reply above is to the abstract; I recommend going to the “Science Brief” page you can reach via that abstract.

    This is
    (Gavin, edit me if I’m misreading this please! I’m just an amateur reader)

    a test of a climate model, to see how well it could simulate the well known brief climate change event 8000+ years ago (when the big North American glacier receded and its meltwater lake spilled into the North Atlantic).

    As Gavin says, that reconstruction modeled a change much like the real climate did change — modeled-Europe gets modeled-cold, as real Europe got cold — due to sudden large inputs of cold fresh water to the North Atlantic.

    Lonnie Hancock (quoted in “Thin Ice”) says the same thing Gavin does above — that we don’t expect cooling in the immediate future — today, with no huge lake behind a melting glacier about to bust loose — fresh water has to melt from ice or fall as rain to dilute the North Atlantic salt water. But a slowdown in the Atlantic circulation could reduce the speed at which Europe heats up, for a while (and have other consequences).

    The “Science Brief” shows how nine different models are in fair agreement about the past 1850-present Atlantic circulation — and how they differ in modeling possible futures through 2100:

    This is good info — congratulations to whoever writes the Science Brief pages. It makes one think, and gives enough footnotes and references to pursue this.
    (I’m basing comments solely on the “Science Brief” and reading as a nonscientist here, it just seens a very interesting perspective to be able to start testing models against this fairly well known event 8000+ years ago and getting useful results).

    [Response: My student (Allegra LeGrande) and I wrote the Science Brief, but I recommend reading the full paper – it’s not that obscure! – gavin]

  12. 12
    Lynn Vincentnathan says:

    Okay, maybe the North Atlantic & Europe will only cool a bit, if at all, from a THC slowdown/halt, due to the GW offset, but where does that leave me (lat. 26, S. Texas, near the Gulf of Mexico)? If the world is warming on average, and some places cool at bit, then that means some other places might really sizzle. Would it get a lot hotter here, sort of like the Arctic enhancement effect. Could we be going from 112 degree F days to some 120 degree days? (Of course, I wouldn’t expect it in my lifetime.)

  13. 13
    Almuth Ernsting says:

    The present cold spell in Europe probably just shows how a rare event severe winter weather has, on the whole, become in Europe – and might support the Met Office’s guess that rising SSTs are, in the long term, making the North Atlantic Oscillation ever more strongly positive in winter (I think last May SSTs in the Azores were a bit more like they had been in 1995 which, whereas on the whole they are rising). In any case, Scotland would cool quite a bit if the THC weakened significantly and quickly – and our primroses and forsythia have been blooming just nicely for weeks.

    Anyway, I just wondered – if the THC really was quite suddenly starting to show (ie if what Wadhams and Bryden found was both accurate and the beginning of a new trend, which the Nature article says is unlikely) – what would that look like on a weather chart? Obviously, more temperatures below zero in the winter. Would it actually show itself in terms of the winter NAO becoming first less positive and soon more strongly negative (less hot air from the tropics pushing over Europe), and in a more northerly and easterly winds developing? Would it show itself in more and more intense droughts developing over countries like Kenya, Tanzania, Rwanda, Burundi?

    Almuth Ernsting

  14. 14

    As a journalist living on the Southeast U.S. coast, I’m following climate change issues with considerable interest.

    I’ve read a great deal about the possibility of changes in Atlantic circulation popularly described as “the conveyor belt.” All of the writing in the general press has been about how cold it could get in Northern Europe, but no one seems to have commented on whether there would be an opposite effect at the southern end of this circulation. Would there be a corresponding increase in average temperatures?

    [Response: Models suggest a slight increase in the Southern Atlantic, but it’s not as significant as in northern cooling – see the figure in Stocker (2000) for some idea. -gavin]

  15. 15
    Steve Bloom says:

    Re #4: Folks following the hurricane “controversy” involving Bill Gray may be interested in the exchange documents Judy Curry (of Webster et al) has posted on her web site: . I can’t tell for sure, but it appears that the Gray comment may have been submitted to Science in hopes of publication. Since the Webster team’s response left Gray with no leg to stand on, publication may be unlikely. In any case it’s an interesting read.

  16. 16
    Tom Brogle says:

    I’ve been looking at
    These clearly show the tne Gulf Stream pentrating the Arctic as far as Novaya Zemblya.
    If the Gulf Stream was weakening surely it would be shortening as well.

  17. 17
    Brian Forbes says:

    The Gulf Stream is the major carrier of heat from the sub-tropics to the Arctic.
    If is has slowed down substantially why is Arctic melting accelerated?
    The two statements contradict each other
    If the Gulf Stream has collapsed to the extent they say it has it should have caused the Arctic to cool.
    Something is WRONG.

    [Response:See our comments here – gavin]

  18. 18
    Dan Welch says:

    Forgive me if this is a naieve question but is it possible that slowing down or shutting down of the North Atlantic conveyor could offset temperature rises from global warming to the extent that this would stop the melting of the Greenland ice cap. To the lay person media reports seem to alternately present the danger of plumetting temperatures in northern Europe as a knock on effect of increased fresh water melting from the Greeland ice, and the danger of sea levels rises due to increased temperatures melting the Greeland ice, but I am struggling to find speculation on how the two might interact.

    [Response:Fair enough point. In the model scenarios where there is a slow down in the circulation, the relative cooling of the area (compared to the rest of the world) is not sufficient to actually cool Greenland, and so Greenland continues to warm, albeit at a slower pace than if the circulation hadn’t changed. -gavin]

  19. 19
    tom brogle says:

    RE response to 18
    So Greenland will carry on melting when the rest of the Arctic gets colder as the THC slows down.
    Are you sure?

  20. 20
    Brian Forbes says:

    re response to 17
    It does not invalidate my point I quote from the article
    “it is a long way from the Greenland Sea to the Gulf Stream ”
    It is even longer way from Novaya Zemblya which is at the end of the tongue of water that juts into the Arctic to the north of Russia.
    I surmise that if the GS(NAD) had slown down then this area would have shown signs of it. It does not appear that it has.

  21. 21
    Eric Swanson says:

    RE #16 and others

    Here’s another look at SST’s:

    Note the rather warm pool south of Greenland and the cold pool off Norway. That’s been there for several weeks now, so one is left to wonder whether the lack of convective chimneys reported by Peter Wadhams last Spring may have been the cause.

    The rather warm area seen in the NAVY graphic for the Barents Sea may be the result of processing of the data. Areas which happened to exhibit sea-ice during the base period used to calculate the anomaly but which now are free of sea-ice will appear to be much warmer to satellite instruments. Of course, if the sea-ice is melting, this may be an important realization. A comparison of NH sea-ice extent with last year may be found here, along with data for previous years:

  22. 22
    Brian Forbes says:

    Re 21
    is interesting it shows that the oceans of the Southern Hemisphere are mostly cooling(quite substantially in places).
    Does this show up on your models ?

    [Response:An instantaneaous snapshot? no. The longer term trends which are warming? yes. – gavin]

  23. 23
    Brian Forbes says:

    Re Gavin’s comment on 21
    Well then it does not show the long term trend in your hot and cold spots in the N Atlantic. I’d rather accept the official navy wersion.

  24. 24
    Hank Roberts says:

    Brian, the Unisys page goes back six days. Is there some Navy page you found you meant to refer to that shows long term trends (and that doesn’t show warming)?

    Here’s a deepsea question —

    That’s a year old, mentions a tiny fraction of a degree centigrade warming in very deep water that wasn’t expected to show global warming effects. (Volcanos are ruled out, they are too scattered to warm this whole expanse).

  25. 25
    Brian Forbes says:

    I read the article.
    The upwelling in the N Pacific is the other end of the downwelling in the N Atlantic
    How long does it take for the journey; 8 years, highly unlikely;80 yrs ,I doubt.But 800 yrs brings us back to the MWP which certainly affected the N Antlantic.
    The navy page has shown a slow warming of the N. Hemisphere and a slow coolng in the S. Hemisphere for the last 2 years.It doesnt show the hot and cold spots as on Unisys.
    I know the basis for the navy differential temperatures.
    Enquiries to Unisys as to the basis for theirs have been
    So I trust the official navy site to be truthful rather than an apparently commercial organisation

  26. 26
    smyyga says:

    This is indirectly related to the issue at hand:

    Here are the presentation slides of Eric Rignot about the mass-balance of the whole Greenland ice-sheet (I brought up some of his results on realclimate in December):

  27. 27
    Nigel Williams says:


    You make a stab at the total journey time for an element of water to complete the circuit of the THC, and relate that process to the swings in N H and S Hemisphere warming.

    I suggest that the more significant figure is the travel time for a deep-water wave to do the circuit of the THC. This is important as it defines the rate of propogation in any change in the impetus of the THC.

    A drop (say) in the driving head at any point in the THC will propogate as a wave travelling at something over 700kph (assuming an average depth of 4000m). This will take about 90 HOURS to run the full 65,000km or so circuit of the THC. Thus any event altering the drive of the THC (for example the failure of Wadham’s chimneys in the north Atlantic – and probably off the Antarctic also) will be reflected in a corresponding drop in the flow rate of the entire THC within the time of travel of a wave around the circuit. Days; not years.

    It doesn’t matter how long a molecule of water takes to do the journey. A drop in the driving head (hence flow) of the THC at any one location will be expressed in a correspoinding drop in flow at every location within a few days.



  28. 28
    Brian Forbes says:

    Rubbish !
    A wave is water going up and down. Waves are propogated without a flow of water (dropping a pebble in a pond will demonstrate this).

  29. 29
    Nigel Williams says:

    A wave front propogates laterally. In deep water the vertical component is just a boundary manifestation of the underwater wave – as happens with a tsunami.

    Indeed a wave is propogated with out overall flow, BUT a change in flow is propogated as a wave through the medium, hence the influence of any change in flow will propogate rapidly, not over centuries.

    Your stone in the water shows very nicely how the ‘information’ travels.

    [Response: Different kinds of information travel at different rates. The information you have highlighted is the barotropic sea level response (which travels fast). The infomation that controls the speed of deep currents in the overturning circulation is related the density contrasts at depth (baroclinic structure) and which travels much more slowly. There is also a transport affect (since density is a function of the active tracers temperature and salinity) and that needs the full advective time scale to equilibriate. Bob Dickson published a nice paper a few years back (Dickson et al, 1999) that looked at how current speeds propagated from Fram Strait to Denmark Strait and beyond. He found it took years – close to the advective timescale. -gavin]

  30. 30
    Nigel Williams says:

    Thanks very much for that Gavin – most helpful.

    I guess the ‘integrity’ of the deep sea current streams is increased when the difference in density / temperature is greater across the boundary layer between the current and the adjacent ‘static’ ocean, while where the difference is small the ‘bleeding’ between these elements is greater; so cold currents will persist through warming deep water, while shallower warm currents may dissipate more rapidly in warming surface water?

  31. 31
    Alisa Brooks says:

    I’m new to the website and have been very afraid to ask questions because of my limited knowledge and I don’t want bog your site down with questions that may not be relivant to this site (too basic).

    However, while “doing my homework” I stumbled across this article:

    I know this is a very lengthy article but I was wondering about this particular area:

    III.1. North Atlantic circulation as a trigger or an amplifier in rapid climate changes:

    The article sites:

    The circulation of the North Atlantic Ocean probably plays a major role in either triggering or amplifying rapid climate changes in the historical and recent geological record (Broecker 1995, Keigwin et al., 1994, Jones et al., 1996; Rahmstorf et al., 1996).

    Am I reading old information? Has this been disproved? Is the statement : “the notion that a future change in the themohaline circulation] may trigger a mini ice age is a mythâ??. True?

  32. 32
  33. 33