by Jean Jouzel
A few decades appears quite short for a major climate change. And yet the ice in Greenland revealed that certain characteristics of the climate can be modified even more rapidly. Our American colleagues focused on studying in great detail other indications that proved this. At the end of the Younger Dryas it took only five years for the content of dust, a direct proof of the atmospheric circulation, to go from high values of a glacial type to much smaller values characteristic of the current climate. During this transition, which lasted fifty years, the accumulation doubled almost instantaneously in three years and perhaps even from one year to the next.
The similarities observed between GRIP and Dye3 throughout the deglaciation existed during all of the last 40,000 years and erased any doubt that still existed. And at GRIP there was no longer any question of evoking the proximity of the bedrock: 30,000-year-old ice was close to a kilometer away from it. Furthermore, for this ice core we have a sufficiently precise chronology of the sequence of events that occurred in a glacial period. In a few dozen years the climate went from a glacial state to milder intermediary conditions between those of a glacial climate and the current climate. A return to cold conditions was much slower, from 500 to 2,000 years (Figure 9.1).7 These “sawtooth” sequences were repeated a dozen times. We then estimated that the associated warming was 5–7°C, half of which corresponded to the passage from a glacial climate to the current climate.
Figure 9.1. GRIP temperature in °C. The isotopic record (oxygen 18 content of the ice) obtained from the GRIP ice core drilling for the last 20,000 years reveals two extremely rapid warmings, one around 14,500 years ago, the other around 11,500 years ago (upper curve). These same variations are recorded in the lake sediments of Lake Ammersee in Bavaria, indicating that these rapid variations also affected western Europe (middle curve), while over that same period variations were much more smoothed in Antarctica (bottom curve). ACR: Antarctic Cold Reversal.
Rapid Events during a Warm Period?
July 12, 1992: GRIP core drilling reached the bedrock at a depth of a bit more than three kilometers. A preliminary dating allowed us to hope that those 3,000 meters of ice would cover 250,000 years. Up to around 100,000 years ago, the isotopic profile revealed an entire series of rapid and large variations that were completely in line with those already identified in the glacial period after 40,000 years ago. This was not surprising because around 100,000 years ago we were still in that glacial period. We counted twenty-four of those events to which, a few years later, Wally Broecker gave the name Dansgaard-Oeschger, in homage to the key role played by those two researchers.
In the field, using different analyses, it was easy to estimate the approximate age of the ice as the drilling progressed. At a depth of 2,800 meters, all indications suggested that we were entering into the Eemian, the warm period that preceded our own. This was around 110,000 years ago. This interglacial ice was close to 100 meters thick. As the core samples were extracted, we felt we were in for some surprises: the dielectric conductivity did not present the stability characteristic of the ice from the last 10,000 years, and the size of the crystals and various proxies showing the contribution of continental aerosols provided evidence of rapid variations. The isotopic curve was thus impatiently awaited. It was available in the month of October; there were indeed many surprises.
The isotopic curve revealed the existence of rapid—indeed catastrophic—events during the last interglacial period, the Eemian, which began around 135,000 years ago and ended around 115,000 years ago (Figure 9.2).8 No one expected that this period, which was slightly warmer than the current climate, would be completely different. But it was interrupted several times by shifts toward conditions rather close to those of a glacial period. Transitions were very rapid (a few dozen years). Cold conditions persisted, depending on the case, for periods of 70 to 5,000 years. This was the first time that such transitions were observed during that period, which had sometimes been considered analogous to the climate of our planet in terms of climatic warming due to an increase in greenhouse gases. The stability of the current climate, which has prevailed for around 10,000 years, thus appeared to be unusual.9 This was an enormous surprise for specialists in climates of the past, especially since the detailed analysis of one of these rapid variations yielded completely unexpected results. A cooling estimated at 14°C, which took place over the course of twenty years, lasted only seventy years and was followed by a return to a warm climate, also over the course of twenty years. And, simultaneously, there were just as abrupt and spectacular modifications of different indicators of atmospheric circulation that showed drastic changes in the atmospheric circulation and supported the image of a shift toward a glacial climate right in the middle of a warm period.10
Figure 9.2. Variation in the oxygen 18 content along the GRIP and North GRIP ice cores; beyond 105,000 years the recording of the GRIP cores was perturbed as a consequence of a mixing of layers linked to the proximity of the bedrock.
These results raised many questions, the first of which concerned their viability. Wouldn’t there have been a mixing of different layers of ice? Was the isotopic content of the ice indeed representative of the temperature of the site and its chemical composition of that of the atmosphere? To address all of these questions the GRIP team put forward solid arguments and quickly published these results. And even though we were cautious, we did propose an erroneous interpretation in Nature. There were real instabilities in the GRIP records, but we had believed then (wrongly) that we could interpret them as climatic variations occurring in a warm period. We were disappointed to learn not too much later that they were in fact the result of a “common” blending of layers of ice linked to the proximity of the bedrock.
The warning came from our American colleagues who had less luck than we did in their drilling operation. Their new drill, of impressive size, allowed them to extract core samples 6 meters long and of greater diameter (12 centimeters versus 10 with Istuk). This drill was entirely satisfactory, but the cable was a source of problems. As already reported in chapter 6, the 1992 season had to be interrupted at a depth of 2,200 meters, and suddenly our colleagues were lagging. A new cable was brought to the GISP2 site in May 1993. The Americans reached the base at the beginning of July, one year after the Europeans, whose record—with an ice core at 3,054 meters—they beat. In the meantime we published the GRIP results, but it was with some anxiety that we awaited confirmation of this surprising structure of the last interglacial period.
The results of the GISP2 drilling11 indeed raised some serious questions about the validity of our interpretation. Beyond 100,000 years ago the records obtained from each of the drillings began to diverge, indicating that the stratigraphy of the cores, or at least of one of them, was modified due to distortions linked to the flow of the ice close to the bedrock. The presence of inclined layers, generally associated with the phenomenon of folding, was observed at GRIP only at 130,000 years ago, whereas it was noticed much earlier at GISP2, leaving a slim hope that only that second drilling was perturbed. But analysis of the composition of air bubbles will definitively reject the idea of rapid climatic variations during the Eemian. Let’s take, for example, methane, which mixes rapidly in the atmosphere; in the absence of perturbations, we should, for a given period, obtain a record that is practically identical in Greenland and in Antarctica. But beyond 100,000 years, this was not the case at either GRIP or GISP2: neither record could be compared to that then available at Vostok, which we knew had not been perturbed because the corresponding ice sequence was more than 1,500 meters above the base. We had thus gone too far in the interpretation of the rapid variations recorded at GRIP during the Eemian. They were unquestionably the result of a blending of ice layers formed at different periods in the past; in no way did they testify to rapid climate changes.
A striking proof of this was provided by the results obtained at North GRIP (Figure 9.2), which indicate that the deepest layers remained in chronological order, probably because the fl
ow of the ice was facilitated there by the presence of liquid water interfacing with the bedrock. This ice core sample, whose age was estimated at 123,000 years, covers part of the Eemian but shows over that period no change of the type recorded at GRIP and GISP2. At most it reveals a new Dansgaard-Oeschger event, the twenty-fifth, around 110,000 years ago, at the moment when the last glaciation was beginning. Everything was in order: the preceding interglacial period was indeed a climatically stable period, at least compared to the last glacial period, which throughout (or almost throughout) its 80,000 years was very agitated. Indeed, between 100,000 years ago and the Last Glacial Maximum 20,000 years ago, isotopic variations recorded at GISP2 are exactly the same as those revealed a year earlier by the ice of GRIP. This comparison held just as true for the North GRIP drilling, and these three isotopic records are also very similar over the entire last deglaciation, which leads us to the beginning of the current Holocene period, which began around 12,000 years ago.
Initially Underestimated Changes in Temperature
However, it quickly became apparent that the way in which we were evaluating the variations in temperature from the isotopic composition of the ice in Antarctica led us to underestimate it in Greenland. The warning signal was sounded by our Danish and American colleagues, who at GRIP and GISP2 had measured the temperature along the bore holes extremely precisely and meticulously. The diffusion of heat was sufficiently slow for this temperature profile to partially keep the memory of conditions that ruled on the surface in the past. Mathematical methods then enabled us to estimate the mean temperature for periods such as the Last Glacial Maximum. And then a new surprise: in the last 20,000 years the climate at the center of Greenland has warmed by close to 25°C, or more than twice the 12°C that we had announced using the isotopic composition of the ice.12
What happened to the rapid warming associated with the Dansgaard-Oeschger events? Was their amplitude also underestimated by a factor of two? The temperature profiles measured in the bore holes could not answer that question because they involved events that were too short to be kept in the archives. Here, too, we had to shift gears. But thanks to an American team—that of Jeff Severinghaus in San Diego—that rapidity became an ally.13 We have already noted this: air bubbles are definitively trapped in the ice at the base of the firn at about a hundred meters in depth. The snow is packed under its own weight, becoming less and less porous the deeper it gets until it becomes completely airtight. Imagine that a rapid warming occurred on the surface. The effect would gradually travel into the firn but take dozens of years for it to be completely felt at a hundred meters in depth. During that time, it would be warmer on the surface than at the base of the firn, whose depth would change slightly. And in such a porous medium there would be a “fractionation” of the gaseous compounds due to gravity on the one hand and to that difference in temperature on the other. Severinghaus was interested in the nitrogen isotopes and in those of argon whose proportions had no other cause to be modified except by the processes that took place in the firn. As theoretically expected, those isotopic compositions changed measurably each time the temperature varied rapidly on the surface. A new thermometer was born. It enabled the evaluation of the amplitude of the successive warmings, which are systematically greater than those estimated from the isotopic analysis of the ice and could reach as much as 16°C. Such a high value was surprising, but it appeared very viable thanks to analysis done independently at Bern14 and by Amaelle Landais at Saclay.15
The Connection with the Ocean Henceforth Demonstrated
Fifteen years went by, during which an interest in these rapid variations—their geographic extent, the mechanisms that were at their origin, the possibility that they would occur in the coming centuries—continued to grow. In fact, in the 1980s specialists in marine sediments weighed in. If the seductive scenario of a connection between the violent changes recorded at Dye3 and changes in the ocean currents in the North Atlantic proposed by Oeschger and Broecker a dozen years earlier were true, we should have found a trace of those events in marine sediments. The paleoceanographers got to work. The first indications that this was indeed the case were revealed by sedimentological analysis. Some core samples were interrupted by a series of layers that came from the base of the Canadian continent. They were known as “Heinrich layers,” after the German scientist who discovered them. And Gerry Bond, a colleague of Broecker at Lamont, demonstrated the cyclical variations in the color of the sediment. These teams and other researchers joined together and in 1993 proposed a coherent image of these different observations.16
Six Heinrich layers were identified between 70,000 and 14,000 years ago and were mapped. Their continental origin was confirmed. Furthermore, a decrease in the amount of oxygen 18 in the seawater was associated with them. These two elements suggested that they were connected to massive discharges of icebergs (poor in oxygen 18) breaking off from the North American ice sheet when it grew to the point of instability. These massive detachments occurred at the end of the cooling phase, a period that lasted 5,000 to 10,000 years. We thus find again the “sawtooth” structure of the ice of Greenland, and it was tempting to associate the events identified in the marine sediments of the North Atlantic with the most marked rapid changes recorded in the ice. In fact, this correspondence has been demonstrated by examining in detail the different marine and ice core records. There was, through the ocean, a connection between the massive discharges of icebergs and at least some of the rapid variations recorded in Greenland. It has been proposed that these discharges, whose volume could be on the order of a million cubic kilometers—or the equivalent of nearly two kilometers of water over all of France—were produced at a time when it was very cold in the North Atlantic and when the Gulf Stream was practically stopped. The discharges were conveyed by a retreat of the edges of the glacial ice sheet toward the interior of the continent, an ice sheet that for a few hundred years provided much less freshwater to the ocean. The surface waters became a bit denser, and that would have been sufficient for the sinking of the waters of the Gulf Stream to start up again in the North Atlantic with, as a consequence, a rapid warming of those regions, Greenland in particular.
The relationships between the climate of Greenland and the conditions that existed in the North Atlantic are even closer than these initial comparisons might allow us to see. Paleoceanographers have since then undertaken increasingly detailed analyses over different parameters (temperature, salinity, ocean circulation), which have revealed that each of the Dansgaard-Oeschger events was systematically associated with changes in these properties. But the atmosphere was also involved. This is seen in the extremely rapid variations of the quantity of dust present in the Greenland ice; it was very elevated during the cold phases that preceded warming but 10 to 100 times less once that warming began. These differences correlate with drastic modifications in the winds. There were, moreover, true upsets that accompanied each rapid event both in the atmosphere and on the surface of the North Atlantic, inasmuch as the oceanic source of the water vapor, from which the snow that fell at the center of Greenland was formed, also moved rapidly. A study conducted by Valérie Masson-Delmotte from the combined analysis of deuterium and oxygen 18 at North GRIP has clearly demonstrated the conditions prevailing at the surface of the ocean.17 At first sight, the concentrations of these two isotopes in the snow vary in a parallel fashion. But if we look closer, slight differences come to light, which can be used to identify the conditions that prevailed on the ocean surface in regions from which masses of air, which arrived on Greenland, originated. Jorgen Peder Steffensen and his colleagues have shown that these conditions can change very abruptly during periods of rapid warmings, with shifts of the atmospheric circulation in the Northern Hemisphere resulting in changes of 2–4°C in the temperature of the Greenland moisture source from one year to the next.18
In France we are very aware of the influence of the weather that exists in the North Atlantic on the climate of a good part of our coun
try and more generally over Western Europe. This is simple good sense even without the available data, which would suggest that the upsets characteristic of the glacial climate of the Atlantic were noticed by our ancestors. Records exist and confirm that intuition. Furthermore, the Dansgaard-Oeschger events were accompanied by clear modifications in precipitation, which was less abundant in periods of intense cold than during milder phases. The repercussions on vegetation are obvious and were reconstructed through an analysis of pollen in continental series like those of the Grande Pile in the Vosges and the Échets near Lyon, as well as in marine core samples taken near the coasts. The isotopic composition of the carbon of the stalagmites in our caves in the Massif Central, formed of calcite (Ca CO3), was influenced by that of the vegetation that covered the ground and retains a record of it, while that of its oxygen reflects the isotopic composition of precipitation. Stalagmites (we have an example of this in the Villars cave in Dordogne) are thus excellent records of Dansgaard-Oeschger events that, thanks to isotopes of uranium and thorium, are generally able to be dated more precisely than is possible from polar ice or marine sediment.