The White Planet: The Evolution and Future of Our Frozen World

by Jean Jouzel

  Consequences on a Planetary Scale

  These rapid variations were also identifiable on the other side of the ocean in the United States and as far as Brazil. But it took only a few months after the extraction of the core sample at GRIP for us to realize that the rapid climatic variations had repercussions beyond these regions close to the Atlantic: Western Europe, America, and Africa. We owe this to the work carried out by Jérôme Chappellaz; as soon as the first part of the GRIP core sample, covering the last 45,000 years, was available, it showed that a significant increase in the content of methane in the air bubbles extracted from the ice generally corresponded to these variations. The production of methane is connected to the extent of flooded zones. Consequently, these increases likely prove variations in the continental hydrological cycle at low latitudes, which suggests that the rapid events influenced the climate of the Northern Hemisphere overall. The variations in methane have henceforth been documented over all of the glacial period. What is more, the analysis of stalagmites taken from Chinese caves enabled scientists to establish a precise connection between the Dansgaard-Oeschger events and the rhythm of monsoons, which were more active at the time when the temperature in Greenland was relatively mild and the concentrations of methane were high. The circle was closed. It became very clear that the entire Northern Hemisphere was perturbed during the last glacial period: temperatures, precipitation, winds, the intensity of the monsoons—all those parameters seem to have varied more or less simultaneously, seemingly over the entire hemisphere. Paleoceanographers also discovered traces of these rapid events in the Baltic Sea, in the Mediterranean, in the Indian Ocean, and even in the distant Pacific.

  But what about the Southern Hemisphere? In Brazil, as we’ve just mentioned, changes associated with the Dansgaard-Oeschger events were detectable but not very marked. They were also relatively difficult to identify in the marine sediments of the Southern Hemisphere. In fact, it was the ice core samples from Antarctica that provided the most detailed information. Granted, the absolute age has not been precisely established there, but the ice of Antarctica and Greenland can be placed on a common scale of time thanks to successive peaks in methane recorded in the north and in the south. Thanks to the Byrd core samples and, more recently, to those of Dronning Maud Land and Dôme C, we now have a good idea of what occurred in Antarctica during the glacial period. Each Dansgaard-Oeschger event had a counterpart in Antarctica but of a different form—the phase of warming and that of cooling were relatively symmetrical there, whereas in the north the warming was rapid and the cooling slower—and of lesser intensity.19 The events there did not exceed 2–3°C, while it reached as much as 16°C in Greenland. The south began to warm up when the north was very cold, and the rapid warming occurred in Greenland at about the time when the temperature reached its maximum in Antarctica. Thus, for the most intense events, warming could have begun there, in Antarctica, more than 1,000 years before Greenland began to warm.

  This structure, which is sometimes erroneously called a “seesaw” and can also be used for the last deglaciation, clearly proves the key role of the Atlantic Ocean and its circulation in the transfer of climatic variations between each of the hemispheres, but many questions remain. Is the south a simple “slave” to what occurred in the north, the absence of thermohaline circulation in a cold period favoring the gradual warming of the south and its resumption during the rapid warming of Greenland provoking a cooling in the south?20 On the contrary, was the south, as might be suggested by the fact that the warming started earlier there, a true agent through changes in winds from the west or from the ocean circulation in the Southern Ocean or even from the injection of freshwater coming from Antarctica? Finally what about the role of tropical regions? Some work suggests that a change in rhythm of the El Niño phenomenon would be likely to modify the amount of freshwater in the Atlantic and as a consequence the thermohaline circulation. The abrupt and intense nature of the variations recorded in Greenland may suggest that mechanisms that took place in the North Atlantic were at the origin of the very great variability of the climate observed during the glacial period, at least for the most part. Despite the progress made in recent years, a progress to which the data acquired on the core sample of Greenland and the Antarctica have greatly contributed, no one has a definitive answer for now. For the debate to be settled, it would be necessary for inherently complex models to account for the interactions among the atmosphere, the ocean, and the polar ice caps on these scales of time. A lot of ground remains to be covered.

  CHAPTER 10

  The Last 10,000 Years

  AN ALMOST STABLE CLIMATE

  The calm after the storm: this is the image we have from the climatic records of polar regions. In Antarctica the contrast between the hills and the valleys that followed each other during the last glacial period, then the deglaciation, and the stability of those same records since the beginning of the Holocene, a bit more than 10,000 years ago, is clear. And above all, there is nothing in common between the highs and the lows, rapid warming and slower cooling, which punctuated the temperature variations in Greenland between 100,000 years ago and the end of the Younger Dryas, 11,500 years ago, and the “flatness” of the same records since then. Climatically speaking, they are two different worlds.

  Not entirely, however. Around 10,000 years ago we were already in an interglacial period—temperatures were even a bit warmer than they are today, at least in Antarctica, but our planet was not yet completely free of the jolts characteristic of colder periods. The reason for this is the slow pace at which the enormous glacial ice sheet that covered the northern part of North American disappeared. A good third of it still existed 10,000 years ago. Mild temperatures favored the melting of the ice, which sometimes led to the formation of “glacial” lakes imprisoned by ice barriers. This was the case 8,200 years ago in northeastern Canada.1 But the barrier that closed Lakes Agassiz and Ojibway gave way, freeing enormous quantities of freshwater, on the order of 200,000 km3, which poured into the North Atlantic in less than a hundred years. The consequences of a stop in the thermohaline circulation that followed were very widely felt in the Northern Hemisphere, in the Atlantic Ocean, in Western Europe, in Africa, and as far as the regions affected by monsoons. This is also seen in the decrease in the concentration of methane analyzed in the ice of Greenland in which the chronicle of that event is most faithfully recorded. The climate cooled there by around 5°C in a few dozen years; this colder climate, characterized by a decrease in precipitation, continued for about seventy years, and the return to earlier conditions again took a few decades.

  Other events, considered abrupt, have marked the Holocene period, with rapid variations in the covering of sea ice and in precipitation in the high latitudes of the hemisphere 4,000 to 5,000 years ago. Then there was rapid cooling in western Europe, periodic droughts over a large part of North America, and notable climate changes in South America. However, the most remarkable event remains that which occurred 8,200 years ago because in the second part of the Holocene, there were no longer ice sheets on North America likely to quickly free large quantities of freshwater.

  Throughout the Holocene, conditions of insolation changed in response to variations in the Earth’s orbit. Nine thousand years ago summer insolation, as well as the contrast between summer and winter, reached a maximum in the Northern Hemisphere and remained high until the middle of the Holocene. This strong seasonal contrast increased the difference in temperature between the continent, which warmed or cooled rapidly, and the ocean, which had difficulty following because of its great thermal inertia; as a consequence, monsoons in the tropical regions intensified. This was observed from the beginning to the middle of the Holocene in North Africa, India, and Southeast Asia. The seasonal variations of insolation would also be at the origin of the gradual intensification of the rhythm and the intensity of the El Niño phenomenon, which was less marked at the beginning of the Holocene than in the second part. Finally, and we now return to the i
ce, strong summer insolations could be at the origin of the weak expansion—indeed the absence—of many glaciers from the Northern Hemisphere, which did not begin to expand until around 5,000 years ago.

  Volcanism and Solar Activity: Natural Climatic Forcings

  As we get closer to the current period, our knowledge of the evolution of the climate becomes increasingly detailed, both in time and in space. But just as much as the most precise description possible of the evolution of the climate, it is important to look carefully at the phenomena that could be at the origin of it, to specify what we call climatic forcing. In this period of relative stability, certain forcing, which retreats from the forefront when we are looking at the great changes of the last glacial period, must be taken into consideration. This is the case for volcanic eruptions and variations in solar activity, forcings that are recorded in polar ice.

  The chemical composition of the ice was modified at each sufficiently large volcanic eruption with, for example, an increase in the content of sulfates. The eruptions in the Northern Hemisphere are clearly recorded in the ice of Greenland, those of the Southern Hemisphere in the Antarctic ice. Some eruptions occurring close to the equator or in the tropics can leave traces in both the north and the south when they are very violent. Beyond the historical archives, polar ice sheets thus enable us to establish a calendar of eruptions that have marked our recent and more distant past, a calendar that in some cases proves more precise than that kept in historical archives. For a long time it was believed that the eruption of the volcano of Santorini, in Greece, whose climatic consequences were felt as far as China and California, occurred in the sixteenth century; the ice in Greenland, dated year by year, place that eruption a century earlier; it is that chronology, confirmed by other methods, that is now considered fact.

  The influence of these volcanic eruptions on the climate results from the presence of ash, which darkens the atmosphere for a few weeks, but above all from the formation of sulfur aerosols that result from the oxidation of gaseous compounds emitted during the eruption. These aerosols have an effect over longer periods, as much as several years. Volcanoes are thus related to cooling and, for a limited time, they can, as a result of an increase in the optical thickness of the atmosphere, have negative forcings equivalent to that associated with an increase in the greenhouse effect of anthropogenic origin. This forcing was on the order of 3 Wm−2 for the eruptions of Krakatoa in 1883 and of Pinatubo in 1991 and weaker (2 Wm−2) for those of Agung in 1963–64 and El Chichon in 1982, all of which are relatively well documented. It is more difficult to estimate the variations in optical thickness for eruptions only recorded in the ice as well as the geographic distribution and the duration of associated perturbations, but this is the only approach available for going back in time. This effort of reconstruction has concentrated on the last millennium; with more than 10 Wm−2, the prize goes to the eruption of 1259, to which no named volcano has up to now been associated. Next in line are the eruptions of Laki in 1783–84 and Tambora in 1815 (both around 5 Wm−2), which, a year later, led to the year without summer. We can imagine that the year 1260 did not have a summer either.

  The climatic role of variations in solar activity is the subject of much debate. Some scientists suggest that the Sun is at the origin of the rather regular rhythm of the rapid variations of the last glacial period, which are separated by a duration close to 1,500 years or by a multiple of that duration; others see the imprint of the Sun in some records covering the Holocene in which that same periodicity, generally fleetingly, is present. As we will see, we cannot attribute the recent warming to a variation in solar activity, but it is undeniable that those variations have an influence on our climate; thus they could be at the origin of the Little Ice Age that affected Europe between the fifteenth and nineteenth centuries. Whatever the case, we understand the importance of knowing the past variations in solar activity with sufficient precision. Here, too, polar ice is of precious help because solar activity has been precisely measured for only a few decades, and the observations, for example, those of sunspots, which enable us to estimate such activity, were not available before 1610. Cosmogenic isotopes took over, formed, like carbon 14 or beryllium 10, through the action of the cosmic rays on the upper layers of the atmosphere. Their rate of production depends, among other parameters, on solar activity. Beryllium 10, which can be analyzed in polar ice, is particularly useful for reconstructing this activity over millennia because concentrations of carbon 14 are also influenced by different processes intervening throughout the carbon cycle.2 There is a lot of agreement among the estimates based on the observation of sunspots and those deduced from beryllium 10. There is, however, a serious problem of calibration that can cause those estimates to vary by a factor of 4 at least. The most recent work indicates that the variations in solar activity are weaker than anticipated a few years ago; its increase between the Maunder Minimum, in the seventeenth century, and the current period would not exceed 0.1%, or in terms of climatic forcing, less than 0.2 Wm−2.

  How Long Has Human Activity Been Changing the Composition of the Atmosphere?

  Another debate is the one that surrounds the variations in concentrations of carbon dioxide and methane throughout the Holocene. Everyone agrees regarding the data that are viable and very well documented: the concentration of carbon dioxide slightly diminished until 8,000 years ago, by 7 parts per million by volume (ppmv), then increased by 20 ppmv between that period and the beginning of the industrial revolution, while that of methane diminished until 6,000 years ago, from 730 to 580 parts per billion by volume (ppbv), then increased again to a level of 730 ppbv, also at the beginning of the industrial revolution. The variations in nitrous oxide are parallel to those observed for carbon dioxide. In terms of radiative forcing, the variability of greenhouse gas effect over this entire period was 0.5 Wm−2. This represents around 20% of the increase observed in the last 200 years in response to the emissions resulting from human activity. The American researcher Bill Ruddiman has posed the hypothesis that the increases observed—in the last 8,000 years for carbon dioxide and 6,000 years for methane—are due to those activities. This hypothesis is rightly controversial, as human activity, in particular at the level of deforestation, does not appear sufficiently great to lead to the increases observed.

  Once again, it was arguments based on the analysis of air bubbles trapped in the polar ice that were decisive: the variations in carbon 13 from carbon dioxide, as revealed in the Antarctic ice for the Holocene, are not compatible with the idea advanced by Ruddiman.3 Ruddiman built his argument around the data from the Vostok ice cores, the only ones available when he defended his hypothesis in December 2003 during the fall meeting of the American Geophysical Union, which was held in San Francisco. He compared what happened in the last 10,000 years with the data available for the three warmest periods of the preceding interglacial periods, which were as warm or warmer than the Holocene and that mark the beginning of each new climatic cycle. The last interglacial period and the two that preceded it began around 130,000, 245,000, and 335,000 years ago, with warm periods for which, given the similarities between the climatic records deduced from ice cores extracted in Antarctica on the one hand and the records obtained from marine sediments on the other, the community of glaciologists adopted the terminology defined by paleoceanographers, Marine Isotopic Stage (MIS) 5.5, 7.3, and 9.3, respectively. Ruddiman noted that the highest concentrations of carbon dioxide and methane were observed at the beginning of those three warm periods—MIS 5.5, 7.3, 9.3, and 11.3—then decreased. Only the Holocene has escaped that trend; this is the argument Ruddiman used to see the hand of man in the increase in concentrations of those two greenhouse gases in the last 8,000 years for the former and 6,000 years for the latter.

  The full length of the preceding interglacial period, MIS 11.3, which the marine records indicate began around 420,000 years ago, was not recorded, at least continuously, in the Vostok ice. The EPICA ice core drilled at Dôme C cover
ed all of the 11.3 stage, and the data obtained from those samples would disprove Ruddiman’s hypothesis two years later. Moreover, in the meantime, it was possible to extend the records from Vostok by 20,000 additional years by reconstructing a correct chronology of the corresponding layers of ice affected by a process of reversal due to the ice flow disturbance occurring at very great depth. All of stage 11.3 and of the transition that preceded it were thus indeed present at Vostok as was confirmed by the comparison with all of the data available from the EPICA drilling. What was observed there? Let’s first take the case of methane. The highest levels were observed at the very beginning of the interglacial period, then, rather similarly for the Holocene, decreased for 5,000 years, and then began to increase. This increase could obviously not be attributed to human activity, which was practically nonexistent 400,000 years ago, and it weakened Ruddiman’s argument, which was also not obviously supported by the records of concentrations of carbon dioxide recorded at Vostok and at Dôme C throughout that interglacial period.

  Why was this counterexample provided to us by stage 11.3 so crucial? After all, over the four preceding interglacial periods, after the initial decrease, only stage 11.3 showed an increase in methane, but it was then necessarily of natural origin. In fact, among these four preceding interglacial periods the most important one for comparison with the Holocene is indeed stage 11.3 because it is the one for which variations in astronomical parameters are most directly comparable. The reason for it is that we have been, for tens of thousands of years, in a period of weak eccentricity during which the Earth’s orbit is round, or almost. The direct consequence of this is that the variations in insolation received in the summer in the Northern Hemisphere, that which according to the astronomical theory governs the rhythm of glaciations, are little marked: we must indeed go back 400,000 years to find similar astronomical conditions.