by Jean Jouzel
Much More Information
Much more information has been revealed by the ice of Vostok, all of which are “firsts,” since, up until 2002, that glacial ice core was the only one that allowed us to obtain reliable records beyond 100,000 years. We will mention just a few of the results that have undeniably advanced our knowledge of the conditions that prevailed since 420,000 years ago. We have already mentioned the spike in production of beryllium 10 that occurred around 40,000 years ago, discovered by Grant Raisbeck and Françoise Yiou. Jean-Robert Petit has shown that the fallout of desert dust—which depends on the extension of arid zones and the intensity of atmospheric circulation—was systematically different between cold and warm periods with very high levels at the moments when glacial conditions were the most intense, up to forty times those observed in current conditions.8 The geochemical composition of desert dusts enables us to identify their source: South America and more specifically Patagonia, where variations in climate have led to a particularly intense aridity and erosion in cold periods.9 Françoise Vimeux has focused on the analysis of two isotopes present in the ice, deuterium and oxygen 18, whose joint study gives access to the way in which the temperature of ocean regions where the snow of Vostok comes from has varied throughout time. In 1983–84 Michael Bender, then an associate professor of oceanography at the University of Rhode Island, worked at the Centre des Faibles Radioactivités in Gif-sur-Yvette, near Paris. He had thought for some time that oxygen 18 in air bubbles, not ice, should contain a message about the changes in volume of the continental ice and about the sea level during a glacial-interglacial transition. He discussed his idea with Laurent Labeyrie, a paleoceanographer at Gif, and then developed an approach at the Gif laboratory to measure the isotopes of oxygen in air bubbles. The idea attracted Dominique Raynaud in Grenoble, who proposed to prepare samples corresponding to the last climatic transition from the ice core from Dôme C. The results, published in 1985 in Nature, demonstrated a link between the oxygen 18 of the bubbles, the deglaciation, and the productivity of the continental and oceanic biosphere. Bender quickly attracted a following for that idea among his graduate students and in his own laboratory with Todd Sowers and at the LSCE in Saclay with Bruno Malaizé, Amaëlle Landais, Gabrielle Dreyfus, and Emilie Capron. It is interesting to note, moreover, that the air trapped in the Antarctic ice, through variations in oxygen 18 of the air and methane, contains information on the evolution of the monsoons that developed in the lower latitudes of the Northern Hemisphere, which in turn has been used to improve the dating of the ice cores of Vostok; this example illustrates perfectly the wealth of ice core archives.
A Huge Lake under the Ice
The 420,000-year-old ice is located at a depth of 3,310 meters, beyond which the core drilling at Vostok continued for more than 300 meters. Signals were then clearly perturbed. The reason for this is the mixing of layers of ice whose correct piling had just been reconstituted to an age of 440,000.10 Beyond 3,350 meters, a depth reached in 1996, it appeared that all hope was lost to reconstruct a climatic record. The deepest ice then lost much of its interest for our community of paleoclimatologists. Of course, it was still analyzed with as much care. One never knows what one might find. Although in 1999 core drilling was voluntarily halted 120 meters above its surface, the bottom of the drilling held a very nice surprise for us, which turned out to be associated with the existence of Lake Vostok.
We will return later to the discovery and the characteristics of this lake, but imagine our surprise and that of our Russian colleagues when we analyzed the deepest part of the ice core itself. There was nothing very interesting up to 3,538 meters. Then, abruptly, over a few dozen centimeters, the properties changed dramatically.11 Below that transition, the ice did not contain any gas, very large crystals appeared, from 10 centimeters to even one meter in length, and the electrical conductivity dropped below the threshold of detection of the apparatus. All these properties are characteristic of frozen water from a reservoir of liquid water. This ice thus did not come from precipitations accumulated in the region of Vostok which, after a few hundred thousand years, would have settled into the depths of the Antarctic ice sheet—or at least not directly. The content of isotopes of that very deep ice provided proof of the existence of “lake ice,” that is, resulting from the freezing of the lake water. Over less than a meter, we noted a true jump in the concentrations of deuterium and oxygen 18 with very different values above and below 3,538 meters. Not only did the values change but the relationship between the two isotopes did as well; the parameter that we call “deuterium excess” was modified just as abruptly. A mystery at the very beginning, which was easily explained if one admitted that this ice had gradually accumulated, this time from below, following refreezing on the surface of a lake. Located directly below the drilling site, the surface of the lake was expected approximately at a depth of 3,750 meters, and there should have been thus more than 200 meters of lake ice in that location above Lake Vostok.
This was a unique situation, one that could not have been conceived of before it was discovered. The ice sheet, along with the lake, forms an inclined interface creating a difference in level of 600 meters between the Vostok site and the northern part of the lake, where the ice is thicker and the conditions propitious to melting because the temperature at which it takes place drops when the pressure increases—by a few tenths of a degree, but it is enough to make a difference. And since the lake has a constant volume, the quantities of ice that melts to the north are equivalent to those that come out of the refreezing in the southern part. In retrospect, the choice of the Vostok Station site, made at a time when we knew nothing about subglacial lakes, proved to be a very good one. No one at present has yet accessed the water of the lake—which may be the case later if Russian drillers succeed in sampling the ice formed in the Vostok hole from the lake water that surged up when the lake surface was hit and then froze, as reported earlier—but the 85 meters of ice already cored between 3,538 and 3,623 meters were indeed formed from the water that has passed through the lake and has thus retained some of its properties. That ice is available for the scientists of the three countries involved in that project: the United States, France, and Russia. It is a true gold mine for extreme microbiologists, as we will discuss in the last part of this book.
CHAPTER 8
Dôme C
800,000 YEARS AND THE REVOLUTION OF THE RHYTHM OF GLACIATIONS
When, in 1994, the scientific document that would serve as a foundation for the EPICA project was written, the stated primary objective was to drill in a place that would provide ice older than that found at the Vostok site. The choice of a dome was important because, with the same accumulation, the age of the deepest layers of ice would be older there than at a site located on a flowline, as is the case for Vostok. There is greater thinning of the deep layers, and even if the accumulation at the site chosen at Dôme C was 30% higher than that at Vostok, the age calculated, far enough from the bedrock to limit the risks of perturbation of the ice layers, was estimated at 500,000 years. We thus started with the hope of covering five climatic cycles, twice as many as the drilling at Vostok, which at this time had just barely reached 250,000 years, and we still did not know that the fifth attempt, which was in progress, would prove successful. The EPICA project was then put into motion. A dozen years later we realized, with some satisfaction, that our prediction of 500,000 years was conservative. We have been able to access 800,000 years of archives at Dôme C—twice what Vostok would have provided—but above all, Dôme C has been the entrance into a different world from a climatic point of view.
Antarctica was formed thirty-five million years ago at a time when the opening of passages between Antarctica and Australia, on the one side, and South America, on the other, enabled the formation of a circum-Antarctic ocean current that isolated that continent from the rest of the world. That isolation, probably combined with a decrease in the concentration of carbon dioxide in the atmosphere, led to the formatio
n of an ice sheet that has remained relatively stable since then. The ice sheets of the Northern Hemisphere were formed only much later, around 2.5 million years ago, perhaps there, too, in response to geographical modifications and weaker concentrations of carbon dioxide. Unlike Antarctica, these ice sheets were subject to very marked cyclical variations seen in the records of the volume of ice deduced from the oceanic isotopic data, with a rhythm clearly linked to variations in insolation but which has gradually changed. Up until 900,000 years ago, the glacial-interglacial cycles are less pronounced and were dominated by a periodicity of 41,000 years, probably connected to the variation in obliquity. A second transition, 430,000 years ago, initiated a period characterized by a rhythm of 100,000 years and variations of much greater amplitude, giving birth to those four great glaciations that have been very well documented for the continents, in the oceans, and, thanks to the Vostok core drilling, in the ice of Antarctica. Between these two periods, the many marine records available are a bit murkier because the periodicities linked to the variation in insolation are less clearly identifiable. The EPICA drilling came just in time because it offered the possibility of penetrating into that intermediary zone and, we hoped, of clarifying the picture.
Ice Older than That at Vostok
At Dôme C, the more than 430,000-year-old ice was very deep, below 2,790 meters. We thus had to be patient in accessing it—that depth would be reached only in 2002—but the study of the first kilometers of ice was in itself interesting. For example, it was important to determine whether the variations in temperature such as we reconstructed at Vostok were representative of a larger area of Antarctica. These reconstructions were henceforth available at three different sites because our Japanese colleagues, with whom we were collaborating, had completed their analysis of the cores from Dôme Fuji, located on the other side of East Antarctica, which covered three climatic cycles, from up to 330,000 years ago.1 Taken as a whole, the three isotopic series indeed gave a very homogeneous image of the climate on Antarctica. For example, all three told us that in Antarctica, during the warm periods that culminated 130,000 and 330,000 years ago, temperatures were as much as 5°C higher than during the Holocene, the period in which we have been living for 11,000 years (Figure 8.1). Another confirmation: the Holocene is the most stable period that Antarctica—and probably the entire planet—has experienced in the last 400,000 years. The two warm periods mentioned above were characterized by a peak of temperature that lasted only around 4,000 years, followed by a cooling period, rapid early on, then slower, before the next glaciation began. And the peak in the intermediary warm period, 240,000 years ago, was also very short. This climatic stability of the Holocene has probably played a role in the evolution and development of our civilizations.
Figure 8.1. Obliquity in degrees. The top curve indicates the level of the Earth’s incline; the middle curve, the variation in temperature at Dôme C; and the bottom curve, the concentration of oxygen 18 in the foraminifers, which indicates sea level.
Ice core drilling at Dôme C was continued to the end. The analysis of all the parameters we have discussed—in the ice, in the air trapped within it, in the impurities it contained—was in full swing. A certain number of results were quickly available, including for the deepest ice whose analysis was exciting for whoever wanted to go back in time: imagine, the first 400,000 years were covered by close to three kilometers, and the 400,000 years that preceded them by approximately ten times less. This thinning of the layers near the bedrock nevertheless raised some difficulties. First of all, the existence of perturbations in the last 300 meters of the drilling done at the center of Greenland, GRIP and GISP2, and at Vostok necessarily led us to question the integrity of the equivalent part of the drilling at Dôme C. The risk that these last 10% of the core samples were unusable for climatic ends, as was the case for these three sites, could not be ruled out.
Fortunately that did not occur. One way of demonstrating this was to compare, during a deglaciation, the variations in the concentration of carbon dioxide and methane to that of the temperature. As we had seen at Vostok, these three parameters increased almost at the same time. But synchronous events are not recorded at the same depth in the ice and in the air that was trapped in it. To be convinced of this we have only to look at what was happening in the present. The temperature is determined by the composition of isotopes of the most recent snow right on the surface of the ice sheet, whereas the air, with its current composition, is trapped at the base of the névé at about hundred meters below. As the layers become thinner that “distance” of a hundred meters between the temperature signals and those of the greenhouse gases diminish in depth but remain sufficient to be recorded, unless the layers have been mixed, in which case their properties would be as well. Important variations in all of the parameters are then likely to be measured at the same depth; this was the case in the perturbed zone of the Vostok drilling. Such a situation up to then had not been seen in the drilling at Dôme C, at least as far as 60 meters from the bedrock, a depth at which the deglaciation that occurred a bit less than 800,000 years ago, the ninth as we go back in time, was recorded. We are confident of the integrity of the data provided by the ice of Dôme C over this entire period, which covers eight complete climatic cycles and the preceding deglaciation (Figure 8.1). The situation deteriorates below that level: the variability in all of the indications, whether they are recorded in the ice or in the air, becomes much weaker than what was expected beyond 800,000 years, and everything indicates that a mixture of layers of ice of different origins is the cause of this. We are resigned to abandoning these last 60 meters of ice, at least insofar as information on the evolution of the climate is concerned.
The second difficulty concerns the dating of the ice. As we went deeper into the ice sheet it became increasingly difficult to calculate with precision the way in which the layers thinned and thus to determine the age of the ice using a glaciological model. Fortunately some of the parameters we analyzed were strongly influenced by variations in insolation. The data already available on the Vostok core drilling indicated that the same was true of the content of oxygen 18 in the air influenced by the rhythm of monsoons and thus by precession; Gabrielle Dreyfus in collaboration with our colleagues from Bern extended that record beyond 400,000 years on the oldest part of the EPICA drilling.2 The drillings at Vostok and Dôme Fuji revealed periodicities in the variations in the oxygen-nitrogen ratio, which most likely resulted from the influence of the local insolation on the metamorphism of the snow in the upper layers of the névé. A similar local insolation signature has also been revealed in the variations in the quantity of air trapped in the ice. We thus have a battery of indicators that enable us, together with the information available on the accumulation of snow and the flow of ice, to propose a viable dating even if it is uncertain by a few thousand years.3
Inversion of the Magnetic Field
The Dôme C measurements also offered us, for the first time in an ice core, the possibility of identifying an inversion in the Earth’s magnetic field. We all know that the Earth acts as a magnet, that it possesses a magnetic field, and that this is the reason why the needle of a compass lines up following a north-south axis. But it has not always been that way; at certain periods in the Earth’s history our compass would have indicated geographic south. This discovery goes back to 1906; it is owed to the French physicist Bernhard Brunhes, then director of the Puy-de-Dôme observatory, who showed that volcanic lava kept the memory of the magnetic field that existed at the time of its formation. By analyzing lava, more or less ancient, he deduced that there were inversions in the Earth’s magnetic field, but it would be necessary to wait another fifty years for the chronology of those inversions to be established, showing that the last inversion took place 780,000 years ago. Since then we have been in the Brunhes period, called normal, with a magnetic field oriented toward the north. Before, the opposite was true, and we must go back more than two million years to rediscover the magnetic
field oriented to the north.
Has the ice core sample from Dôme C retained a trace of this inversion that occurred 780,000 years ago? Not directly. In spite of the extreme sensitivity of existing recording methods, there are not enough magnetic particles in the ice for magnetism to be measured. So we must be creative. The Earth’s magnetic field forms a sort of shield that protects the Earth from cosmic radiation and affects the rate of production of certain elements—carbon 14, beryllium 10, chloride 36—formed as a result of the interaction between those rays and the molecules present in the upper layers of the atmosphere. When the field disappears, which happens when it is reversed, the production of these elements can be multiplied by a factor of two. Using very detailed analyses, Grant Raisbeck researched the peak of beryllium 10 corresponding to the last inversion that could be located in the ice whose age predicted by the dating model was indeed very close to 780,000 years.4 We were fortunate because that peak that tells of the magnetic reversal during the so-called Brunhes-Matuyama transition was located only a dozen meters or so above the depth at which the layers of ice are perturbed. But it was a wonderful confirmation: with the ice core drilling at Dôme C we have viable climatic records covering the last 800,000 years, nearly 400,000 more years compared to Vostok.
