Ice Core Science and Fluctuating Temperatures

Cartoon of water molecules, with and without deuterium. Remember that the oxygen atom in each of these can also vary its isotope value as well, either as O-16 or O-18. Since the coldest snow is the lightest (containing the least D and O-18), analysis of these isotopes in polar ice allows for the reconstruction of past temperatures.
Ice & Isotopes
The best way to get an impression of what Canada and Scandinavia looked like during the last ice age is to fly over Greenland. From high up, the frozen white endless desert in the interior seems utterly hostile to the observer; the eye wanders in search of markers to fasten on but fails to penetrate the cold haze. The margins, on the other hand, are spectacular, with dramatic crags and steep-walled ice-filled valleys where the glacier tongues leave the main body and seek their way to the ocean, feeding it with icebergs. A geologist can readily sense the relentless motion of these ice streams running down into the fjords, at an imperceptibly slow speed, but powerful in purpose, removing what is in their way and carving ever deeper ravines into the hard host rock.

The icy wastelands of Greenland and Antarctica have something extremely precious to offer, which makes some scientists want to spend months at a time there, even while enduring hard labor: the memories of the ice. As was described in Lesson 4, year after year the layers of snow turn to ice, as new layers are added on top. The frozen water contains a record of the conditions under which the snow fell, in the form of the isotopic composition of the ice. The isotopes of interest are hydrogen (H) and its heavy sibling deuterium (D), as well as oxygen-16 and oxygen-18, which have been described previously in connection with the deep-sea record in foraminifers. Water vapor turns to precipitation over the polar ice sheet more readily when it has the composition HOD and H18OH than if it is normal water, H16OH. As air cools upon climbing up an ice shield, water changes phase from vapor to liquid, thus losing D and 18O preferentially. This means that the coldest snow has the least D and 18O in it.

With this basic information (and some statistics and isotope chemistry) we can extract a temperature record from the ice on Greenland for the last 100,000 years. For Antarctica, a record going back 400,000 years has been reconstructed. To this end, scientists employ heated drills, which penetrate the ice layers one after one from the surface on down, and recover cylindrical cores of ice one after another until the bottom of the ice is reached. The ice cores are then sampled, and the composition of the water is measured in the laboratory using mass spectrometers. Then, when plotting the ratio of D/H or 18O/16O, a record of temperature change emerges. (Actually, as described in Section 7.0, what is plotted is the percent deviation of the isotope ratio from the value of a standard, which simplifies the relationship to temperature.)

Ancient Glacial Ice Composition
The founding event of ice-core science may be taken as the publication of several articles in 1969 by the Danish chemist Willi Dansgaard, professor in Kopenhagen, Denmark, and his collaborators. Their first important result was the determination of the composition of Greenland ice for the last 100,000 years. Assuming that glacial-age ice had a similar ratio of O-18 to O-16, this is a much better guess than Emiliani's Chicago snowfall model, and the composition of the glacial ocean could now be guessed with much greater confidence. It turned out that the deviation from present-day conditions was really twice greater during the glacial period than had been postulated by Emiliani, so that much of the variation of the oxygen isotope signal detected by Emiliani had to be ascribed to the effects of ice-buildup and decay rather than to temperature.Just as important was the second finding that the oxygen isotope composition of glacial ice varies with time, and that it faithfully reflects glacial and interglacial conditions. To bring out this pattern, Dansgaard and his co-workers had to date the raw ice record with simple flow model based on the movement of large glaciers. As the ice sheet maintains the balance between thickening on top, and moving horizontally toward the sea, the patterns of flow are such as to maintain thick layers in the upper half of the glacier (which gets carried along by the flow below) and greatly thinning the layers in the more mobile lower half, and especially on approaching bedrock.

The third important result is that the temperature variations are much larger for glacial time than for interglacial time. This suggests that the climate was much more fickle when large ice sheets covered Canada and Scandinavia, compared with periods when these areas were free of ice. The presence of ice sheets did not stabilize climate (as one might expect) but introduced instability. Many geologists believe that this instability came from the ice itself. From time to time small changes in climate led to sudden surging of large glaciers which covered much of the North Atlantic with icebergs. Such events, presumably, came suddenly and unannounced. They produced severe winters and poor summers, which was bad for growing things, and resulted in starving mammoths and other large mammals. If this is so, the herds of the large mammals in North America and in Eurasia were in worse shape than ever at the end of the last ice age, around 15,000 years ago.

A record of temperature change in Greenland, based on isotopes in ice.
Other Information from Ice Cores
Besides temperature, many other types of information can be extracted from ice cores by analyzing wind-blown dust, volcanic matter, and trapped air. The dust content of glacial ice in Greenland is more than ten times greater than that in ice made during warm intervals. This suggests that glacial periods tended to have dry weather and strong winds, in the regions south of the ice shields. (Cold air holds much less water vapor than warm, so this is not surprising.) During a cold dust storm in the plains, the thick matted hair of mammoth must have been a real life-saver.

A record of volcanic activity is preserved in the amount of acid found in the ice. Some layers are very rich in acid, presumably from the sulfuric matter entrained by the snow at the time, in the atmosphere, after an eruption. These intervals also tend to show low temperature! This relationship strongly suggests that volcanism influences climate. More volcanism means cooler summers and more severe winters. The resulting catastrophic weather, following such eruptions, must have greatly stressed the populations of mammoth and other large mammals that need much food and cannot hide in burrows and caves. (In light of this data it cannot be entirely excluded that Icelandic sagas refer to such events in a legend about the end of the world, which includes a terrible winter lasting three years. If they do, their tribal tradition would have to reach back 11,000 years! However, the legend could easily refer to more recent events such as the eruption of Santorini, 3500 years ago.)

In addition, the air trapped in the ice can be analyzed for trace gases such as carbon dioxide and methane. This was done in laboratories in France (in Grenoble, by the physicist Claude Lorius and his co-workers) and in Switzerland (in Bern, by the physicist Hans Oeschger and his team). Results show that the carbon dioxide content of the atmosphere closely follows the ups and downs of temperature. Whenever it was cold, carbon dioxide and methane were low in concentration, whenever it was warm, they were both high.