Abrupt Climate Changes and the Ocean
The Greenland ice cores showed that the climate of the last ice age jumped by ten degrees within decades, and the ocean cores showed that the Atlantic overturning switched with it. This unit explains the events, the bipolar seesaw, the two-box model with two stable circulations, the Younger Dryas, and the evidence on whether the overturning is weakening today.
Until the late 1980s most climate scientists assumed that the climate of the Earth changes slowly, at the pace of the orbital cycles of Unit 14, and that a human lifetime is far too short to see it move. The ice cores of Greenland ended that assumption. They showed that during the last glacial period the temperature over Greenland jumped by five to sixteen degrees within a few decades, about 25 times, and the sediment cores of the North Atlantic showed that the overturning circulation of Unit 8 switched with it. This unit follows that discovery from the ice to the seafloor, explains with a model of two boxes why the overturning can have two stable states, watches the switch at work in the last deglaciation, and ends with the question that has moved from the literature into the news: has the Atlantic overturning already begun to weaken, and could it collapse?
1 The surprise in the ice
A Greenland ice core is an archive of snow. Each year’s snow is buried by the next, compressed first into firn, the grainy packed snow of the upper tens of metres, and then into ice, and the ratio of the oxygen isotopes in it records the temperature at which the snow formed, by the fractionation that Unit 13 explained for the ocean: the colder the cloud, the less \(^{18}\mathrm{O}\) reaches the snow, and the more negative its \(\delta^{18}\mathrm{O}\). Today the snow at the summit of Greenland has a \(\delta^{18}\mathrm{O}\) of about \(-35\) ‰ against the VSMOW standard of Unit 13. The ice from the depth of the last glacial period has about \(-42\) to \(-44\) ‰, and at face value that difference means a Greenland colder by more than ten degrees. The ice holds a second thermometer, better than the oxygen isotopes alone, in the air bubbles trapped in it. When the surface warms quickly, the heavier isotopes of nitrogen in the porous firn sink a little toward the cold bottom of the firn column, a process called thermal fractionation, and the air that is sealed into bubbles at that depth carries the anomaly with it. The size of the anomaly gives the size and speed of the warming independently of the snow, which is why this method is called the nitrogen thermometer. Figure 1 shows why such a record can be dated to the year.

Figure 2 shows the NGRIP record of the last 100,000 years. The slow trends of Unit 14 are in it, from the warm last interglacial at the left to the Last Glacial Maximum and the warm Holocene at the right, but what strikes the eye are the spikes. Twenty-one times in these 100,000 years, and about 25 times in the whole glacial period, the record jumps from a cold state to a warm state. Each of these jumps is a Dansgaard-Oeschger event, named after Willi Dansgaard (1922-2011), the Danish geophysicist who first measured such a record, and Hans Oeschger (1927-1998), the Swiss physicist who measured the gases trapped in the ice. The warm phases are called Greenland interstadials and the cold phases Greenland stadials, and both are numbered from the youngest, so that Greenland Interstadial 1 (GI-1) is the warm interval that began 14,700 years ago and opened the end of the glacial period, and GI-8, the large event at 38,200 years, is one of the most studied.