This story is the dream of every science writer. It features some of the most dramatic and rapid climate shifts in Earth’s history, as well as tenacious scientists braving the hostile ice and snows of Greenland and Antarctica for years on end to bring home that most precious material: kilometre-long cores of ancient ice, dating back over a hundred thousand years. Back in their labs, these women and men spend many months of seclusion on high-precision measurements, finding ingenious ways to unravel the secrets of abrupt climate change. Quite a bit has already been written on the ice core feat (including Richard Alley’s commendable inside story “The Two Mile Time Machine”), and no doubt much more will be.
It was the early, pioneering ice coring efforts in Greenland in the 1980s and 90s that first revealed the abrupt climate shifts called “Dansgaard-Oeschger events” (or simply DO events), which have fascinated and vexed climatologists ever since. Temperatures in Greenland jumped up by more than 10 ºC within a few decades at the beginning of DO events, typically remaining warm for several centuries after. This happened over twenty times during the last great Ice Age, between about 100,000 and 10,000 years before present.
The latest results of the EPICA team (the European Project for Ice Coring in Antarctica) are published in Nature today (see also the News & Views by RealClimate member Eric Steig). Their data from the other pole, from the Antarctic ice sheet, bring us an important step closer to nailing down the mechanism of the mysterious abrupt climate jumps in Greenland and their reverberations around the world, which can be identified in places as diverse as Chinese caves, Caribbean seafloor sediments and many others. So what are the new data telling us?
These data connect the Antarctic ups and downs of climate to the much greater ones of Greenland. This is hard, as dating an ice core is a difficult art (no pun intended). If one makes an error of only 5% in determining the age of an ice layer, for 40,000-year-old ice that’s an error of 2,000 years. But to understand the mechanisms of climatic changes, one needs to know the sequence of events – for example, one needs to know whether a particular warming in Antarctica happens before, after, or at the same time as a warming in Greenland.
To get around this problem, Thomas Blunier and colleagues nearly ten years ago pioneered an ingenious method to synchronise the ice cores of Greenland and Antarctica by analysing changes in the amount of methane in air bubbles in the ice. Changes in methane are recorded at both poles, and they should occur almost exactly in step as gases are quickly mixed through the whole atmosphere. After the ice cores are synchronised by aligning the methane variations, the relative timing of Greenland and Antarctic temperature changes can be seen.
While Blunier and colleagues were originally able to connect only a handful of large climate events, the results published today take this method to a new level by applying it to the new, high-resolution Dronning Maud Land ice core. The new data confirm with unprecedented precision what Blunier found: Antarctica gradually warms while Greenland is cold. But as soon as Greenland temperatures jump up in a DO event, Antarctic temperatures start to fall (see graph). This happens for every DO event, and it is a peculiar and tell-tale pattern that is also found in model simulations of these events (see graph).
Figure: The top two panels show idealised model DO events on an arbitrary time axis (in years), highlighting the phase relationship between Greenland and Antarctic temperatures: when a DO event hits Greenland, Antarctica switches from warming trend to cooling trend. The bottom panels show the “real thing”, the noisy data from ice cores. Note the expanded scale for Antarctica in both cases. Time here runs from left to right – normal for regular folks, but somewhat unusual for the ice core experts (my apologies to these).
It is (at least in the model) a result of a big change in northward heat transport in the Atlantic. If the heat transport by the Atlantic thermohaline circulation suddenly increases for some reason (we’ll come to that), Greenland suddenly gets warm (an effect amplified by receding sea ice cover of the seas near Greenland) and Antarctica starts to cool. Changes in Antarctica are much smaller and more gradual, as it is far from the centre of action and the vast reservoir of ocean around it acts as a heat store. The basic physics is illustrated very nicely in a simple “toy model” developed by Thomas Stocker and Sigfus Johnsen.
There is still debate over what kind of ocean circulation change causes the change in heat transport. Some argue that the Atlantic thermohaline circulation switches on and off over the cycle of DO events, or that it oscillates in strength. Personally, I am rather fond of another idea: a latitude shift of oceanic convection. This is what happens in our model events pictured above: during cold phases in Greenland, oceanic convection only occurs in latitudes well south of Greenland, but during a DO event convection shifts into the Greenland-Norwegian seas and warm and saline Atlantic waters push northward. But I am biased, of course: my very first Nature paper (1994) as a young postdoc demonstrated in an idealised model the latitude-shift mechanism. Other oceanic mechanisms may also agree with the phasing found in the data. In any case, these data provide a good and hard constraint to test models of abrupt climate events.
But irrespective of the details: the new data from Antarctica clearly point to ocean heat transport changes as the explanation for the abrupt climate changes found in Greenland. We are thus not talking about changes primarily in global mean temperature (these are small in the model results shown above). We are talking about what I call a climate change of the second kind: a change in how heat is moved around the climate system.
As an analogy, think of your bath tub and the types of change to the water level you can get there. A change of the first kind would be a change in mean level, e.g. if you add water. A change of the second kind would be the changes you get by sloshing around the water in the tub.
There are very few possibilities to change the global mean temperature, a climate change of the first kind: you have to change the global heat budget, i.e. either the incoming solar radiation, the portion that is reflected (the Earth’s albedo), or the outgoing long-wave radiation (through the greenhouse effect). Temporarily, you can also store heat in the ocean or release it, but the scope for changes in global mean temperature through this mechanism is quite limited.
Changes of the second kind are due to changes in heat transport in the atmosphere or ocean, and these can occur very fast and cause large regional change. Think of your tub: if you want 10 cm higher water level at one end, you can achieve this by turning on the tap – but you can get there much faster by pushing some water over there with your hand, albeit temporarily and at the expense of the water level at the other end. That kind of “see-saw” (but with heat, not water) apparently happens during DO events, as the new data confirm.
The two kinds of climate change are sometimes confounded by non-experts – e.g., when it is claimed that DO events represent a much larger and more rapid climate change than anthropogenic global warming. This forgets that our best understanding of DO events suggests they are changes of the second kind. The same error is made by those who claim that the 1470-year cycle associated with the DO events could lead to an “unstoppable global warming”. A global warming of 3 or 5 ºC within a century, as we are likely causing in this century unless we change our ways, has so far not been documented in climate history.
One crucial point has been left unanswered thus far. If DO events are due to ocean circulation changes, what triggers these ocean circulation changes? Some have argued the ocean circulation may oscillate internally, needing no trigger to change. I am not convinced – the regularity of the underlying 1470-year cycle speaks against this, and especially the fact that sometimes no events occur for several cycles, but then the sequence is resumed with the same phase as if nothing happened. I’d put my money on some regularly varying external factor (perhaps the weak solar cycles, which by themselves cause only minor climate variations), which causes a critical oceanic threshold to be crossed and triggers events. Sometimes it doesn’t quite make the threshold (the system is noisy, after all), and that’s why some events are “missed” and it takes not 1,500, but 3,000 or 4,500 years for the next one to strike. But the field is wide open for other ideas – the cause of the 1470-year regularity is one mystery waiting to be solved.
Alley, R.B., 2002: The Two-Mile Time Machine: Ice Cores, Abrupt Climate Change, and Our Future. Princeton University Press.
Blunier, T. and E. J. Brook, 2001: Timing of millennial-scale climate change in Antarctica and Greenland during the last glacial period. Science, 291, 109-112.
Blunier, T., J. Chappellaz, J. Schwander, A. Dällenbach, B. Stauffer, T. F. Stocker, D. Raynaud, J. Jouzel, H. B. Clausen, C. U. Hammer, and J. S. Johnsen, 1998: Asynchrony of Antarctic and Greenland climate climate change during the last glacial period. Nature, 394, 739-743.
Braun, H., M. Christl, S. Rahmstorf, A. Ganopolski, A. Mangini, C. Kubatzki, K. Roth, and B. Kromer, 2005: Solar forcing of abrupt glacial climate change in a coupled climate system model. Nature, 438, 208-211.
Stocker, T. F. and S. J. Johnsen, 2003: A minimum thermodynamic model for the bipolar seesaw. Paleoceanography, 18, art. no. 1087.