An understanding of long-term climate changes, with some surprises, is emerging from analysis of ice core samples.

DEBORAH SCHOEN

     Physical and chemical--particularly isotopic--analyses of ice cores have provided information on a wide array of parameters, including age, temperature at the time of ice formation, aerosol deposition, and composition of the atmosphere. In the case of the Vostok ice core samples, such measurements enabled glaciologist Claude Lorius and his colleagues at the French laboratories Laboratoire de Glaciologie et Géophysique de l'Environnement, Grenoble, France (LGGE) and Laboratoire des Sciences de l'Environnement, Saclay, France (LSCE) to estimate the atmospheric temperature at the time that the ice formed and, through analysis of air bubbles trapped in the ice, to describe the composition of the ancient atmospheres. They found a strong positive correlation between temperature changes and changes in carbon dioxide and methane; this has contributed empirical evidence of the magnitude of climatic feedback between increasing levels of greenhouse gases and temperature (1).

     Greenland ice cores are not as deep as those in Antarctica, but they provide greater resolution with respect to time. The Greenland cores best reflect the rapid temperature fluctuations of the last ice age, characterized by increases of up to 6 ^oC in a few years or decades. The discovery of these fluctuations, known as Dansgaard-Oeschger oscillations after the Danish and Swiss investigators who first documented them, has given rise to intense interest in their causes and speculation that current increasing levels of greenhouse gases in the atmosphere could trigger such rapid change in the coming decades.

     The importance of the ice core record was recognized by the North American environmental science community in 1996, when Claude Lorius, Willi Dansgaard, and Hans Oeschger received the Tyler Prize for Environmental Achievement. Their work has stimulated efforts for continued polar ice core research and more investigation into other paleoclimate records, such as marine sediments, and provided essential input for the modeling of past climates. Researchers argue that paleoclimate models, by addressing the full range of variability documented by ice core records, can advance our understanding of climate mechanisms--mechanisms that may produce only gradual changes over the coming millennia or, on the other hand, bring about an abrupt and fundamental disruption of the relatively benign climate that we know today.

Antarctic ice cores
Ancient ice has been extracted and analyzed at research stations throughout Antarctica for more than three decades. In the coastal areas, however, lateral ice flow makes it difficult to obtain undisturbed records. To obtain continuous, long-term data from the last glaciation (ice age) and beyond, researchers have gone inland to the high plateau, to Byrd Station in western Antarctica, and to Dome C and Vostok stations in eastern Antarctica.

     In the 1970s, the Soviet Mining Institute set up the drilling station at Vostok. Because of a personal friendship with V. M. Kotlyakov, the lead scientist at Vostok and the director of the Institute of Geography of the Russian Academy of Sciences, Claude Lorius was able to arrange French participation in the drilling project. "The mining institute had experience drilling in ice and measuring ice deformation and temperature in situ. The French and the Americans had the laboratory technology for analyzing the ice," he explained. "When we first arrived at Vostok in the early 1980s, there were two kilometers of ice already extracted, and we took samples back to France for analysis."

At the Vostok station in Antarctica, scientists are obtaining ice core samples down to ice depths exceeding 3600 m. (Courtesy Claude Lorius, LLGE)

     Lorius and his colleagues analyzed the chemical and isotopic composition of the ice and its trapped air bubbles. Proxy data for temperature were obtained from the deuterium (D)-to-hydrogen ratio. This was based on a previously observed linear relationship between deuterium concentration in Antarctic ice and local temperature, an observation attributed to kinetic and vapor pressure differences between DHO and H₂O. They determined the composition of the ancient atmospheres through direct analysis of trapped air bubbles; an age correction factor was estimated to account for the time lapse between the formation of the precipitation and the isolation of air bubbles from the atmosphere (1). The scientists dated the ice core based on ice flow modeling using electrical conductivity measurements, ice accumulation changes, and correlation with other paleoclimatic records (2).

     The pattern of temperature climate change recorded in the Vostok ice core supports the orbital theory of ice ages, in which the timing of glaciation cycles is attributed to the periodicity of changes in the shape of the Earth's orbit (eccentricity), the tilt of the Earth's axis (obliquity) and the timing of its closest approach to the Sun (precession). The astronomical effect is evident in the Vostok record, with a strong eccentric signal, noted Lorius. "The amount of energy coming to the Earth doesn't change much but the [latitudinal] distribution of the energy does, and this can affect the building or decay of northern hemisphere polar ice sheets. A decrease in solar input at high latitudes, for example, can lead to building ice sheets. This, in turn, reinforces the orbital effect, as more radiation is reflected away."

     It is the ice core data on greenhouse gases, however, that led Lorius and his colleagues to conclude that these gases, as well as the ice sheets, "played a significant part in the glacial-interglacial climate changes, by amplifying the relatively weak orbital forcing and by constituting a link between the Northern and Southern Hemisphere climates." The magnitude of the change in greenhouse gas concentrations was in itself unexpected (see figure on next page). At the last glacial maximum (21,000 years before present), for example, carbon dioxide is estimated to be 190-200 ppm, compared with the 270- to 280-ppm preindustrial average for the Holocene (last 10,000 years). The warming that occurred between the glaciation and the Holocene was approximately 10 ^oC over Antarctica or 4-5 ^oC when averaged globally (1).

The changeable climate of Earth

     Where did the carbon dioxide come from? "This is one of the grand unsolved puzzles in climate research," said Thomas Stocker, a climate modeler at the Physics Institute of the University of Bern. "About 50% of the 80-ppm glacial-to-interglacial increase can be explained by a change in the solubility of carbon dioxide. Warmer ocean water carries less carbon dioxide than colder water. However, there are complicated biochemical processes in the ocean, such as pH, the depth of the dissolution level for calcium carbonate, and the net primary productivity of the marine carbon cycle that are also playing a role."

     Another puzzle concerns the sequence of forcing events--what climate researchers refer to as the "chicken and egg" problem. Lorius argues, however, that our lack of knowledge of the mechanisms and timing of climate change does not preclude an assessment of the role of greenhouse forcing in past climatic shifts. Through a multivariate analysis in which the temperature variance was decomposed into forcing factors--greenhouse gases, dust and non-sea-salt sulfate, ice volume, and local insolation--Lorius and his colleagues estimated that greenhouse gases were responsible for at least 40% of the glacial-interglacial temperature change, or roughly 2 ^oC of the global average. This estimate implies a considerable climate feedback resulting from increased greenhouse gas concentrations, in that direct greenhouse radiative forcing with no climate feedback was estimated to have an amplitude of only 0.7 ^oC (1). "This was a first attempt at evaluating the magnitude of the feedback from the ice core data," said Lorius. "But estimates derived independently from general circulation models give rather close results."

The changeable climate of Earth

Greenland ice cores
Ice coring in Greenland dates back to the 1950s, when scientists with the U.S. Cold Regions Research and Engineering Laboratory in Hanover, N.H., set up drilling stations. Many of the investigations at this time were in conjunction with the International Geophysical Year, an international research effort that focused on the Arctic and Antarctica. In the late 1960s and early 1970s, glaciologists Willi Dansgaard, of the University of Copenhagen, and Hans Oeschger, of the University of Bern, pioneered developing techniques for dating the ice, measuring temperature, and analyzing the chemical composition of the ice and trapped air bubbles.

     Greenland cores do not date back as far as those from the Antarctic high plateau, although recent drillings date to the last interglacial, the Eemian (135,000-115,000 years ago). However, because of greater snow accumulation and less compression of the ice, the climatic history of the Northern hemisphere ice sheets is preserved with a years-to-decades resolution in contrast to the century-to-millennial scale resolution of the Antarctic cores. Scientists can directly date the Greenland ice for the past 14,500 years by counting annual layers. Beyond that time, the ice is dated through ice flow modeling; proxy temperature data are derived from the ratio of ¹⁸O/¹⁶O (3). Greenland also receives considerably more dust than Antarctica. Dust particles, through chemical reactions with the trapped air, can interfere with the atmospheric carbon dioxide record. They provide unique information, however, on the atmospheric circulation at the time of deposition (4).

     The discovery of abrupt climatic shifts, or Dansgaard-Oeschger (D-O) oscillations, was the most surprising feature of the Greenland ice core data. The 24 D-O oscillations of the last glaciation are marked by temperature increases over Greenland of up to 6 ^oC, in a time span of less than a decade, followed by a more gradual return to the cooler glacial climate (3). These events are not clearly manifested in the Vostok record. However, diverse paleoclimatic evidence, including marine sediment data, pollen profiles, and glacial snow line data, indicates that the effects of at least some events were felt on a global scale (5).

The thermohaline circulation
The reports of the Dansgaard-Oeschger fluctuations have since generated tremendous excitement in the climate sciences community. "They indicate a climate capable of violent change," said Lawrence Mysak, a climate dynamicist at McGill University's Center for Climate and Global Change Research in Montreal. One area of Mysak's research focuses on understanding the mechanisms for such abrupt climate shifts through coupled modeling of atmospheric and ocean processes (6). "Today, we think the only way such events could have happened is through major changes in the ocean's thermohaline circulation."

     The thermohaline circulation is a pole-to-pole overturning circulation in the Atlantic, which drives warm water north from the tropics, raising Europe's temperature an estimated 5 ^o to 10 ^oC. At high latitudes the water is cooled by the winds off of Canada and Greenland and, because cool water is denser than warm water, it sinks to the bottom. While storms add freshwater, making the cooled water lighter, the cooling effect nonetheless overcomes the buoyancy effect of the freshwater input. The deep-bottom flow then turns south, eventually reaching the Antarctic circumpolar current. The North Atlantic downwelling draws more warm replacement water up from the tropics. Thus, a cooler Europe would be the most direct manifestation of a weakening or shutdown of the thermohaline circulation.

     What could have caused a major shift in the operation of the thermohaline circulation? Wallace Broecker of Columbia University's Lamont-Doherty Earth Observatory has hypothesized that six Heinrich events--the "melting of huge armadas of icebergs"--that curred during the last glaciation could have provided the freshwater to shut down the thermohaline circulation (7). Syukuro Manabe and Ronald Stouffer, climate modelers with the Geophysical Fluid Dynamics Laboratory/NOAA at Princeton University, investigated the effects of such a massive flux of freshwater to the North Atlantic using a coupled ocean-atmosphere model (8). They concluded that in response to the flux of freshwater, "the thermohaline circulation weakens abruptly, intensifies, and weakens again, followed by a gradual recovery, generating episodes that resemble the abrupt changes [D-O oscillations] of the ocean-atmosphere system recorded in ice and deep-sea cores."

     "This is why it's so important to monitor the Arctic sea ice cover," says Mysak. "As the ice shrinks due to global warming, the exposed ocean warms the atmosphere, leading to more breakup of the ice and a greater flux of freshwater to the North Atlantic." Scientists monitoring the Arctic have indeed observed a net decrease of 0.6% per year in the summer sea ice cover for the period between 1978 and 1995. The change, however, has not been uniform across the Arctic Ocean, as most of the decrease occurred in the Siberian sector. The trend is compatible with observed increases in temperature, but because of the high interannual climatic variability of the region, researchers have not yet determined if the decrease in sea ice cover is part of an Arctic-wide response to global change (9).

Predictive modeling
Understanding how the ocean's thermohaline circulation and its interaction with other climate processes could produce the extreme variability that has been recorded in ice cores is part of an increasingly influential area in climate change research known as paleoclimatic modeling. Models used to simulate past climates are generally simpler than three-dimensional general circulation models, allowing finer tuning through a greater number of computer runs. The simpler models can also be run over thousands of years to track the climatic response to an experimental condition, such as freshwater input.

     Modeling the past, in effect, provides the link between the ice core data and models currently used to forecast climate change. As Anthony Broccoli, a climate modeler with the Geophysical Fluid Dynamics Laboratory explained, "The ice core records contain information about past concentrations of greenhouse gases, volcanic aerosols, and atmospheric dust, all of which influence the radiation balance of the earth. This information is essential as input to model simulations of ancient climates. Ice cores also tell us about past temperature, precipitation, atmospheric circulation, and other climatic variables. We can use such information to determine if our simulations can reproduce important climatic features of the past. This is an important way in which we evaluate the performance of climate models."

     How well do current models recreate past climates? Stocker believes the major processes are well described but that there is considerable uncertainty regarding the effects of cloud cover and height and condensation processes, as well as regarding the response of small-scale processes, such as heat and freshwater fluxes across the air-sea interface, the behavior of sea ice, and deep water formation. In addition, he argues for greater attention to the role of the tropics and low latitudes, particularly the hydrological cycle (10).

An unprecedented rate of increase
The Vostok core showed that atmospheric carbon dioxide increased by approximately 80 ppm over a span of 10,000 years, between the Last Glacial Maximum and the pre-industrial times of the Holocene. In the past 200 years the atmospheric carbon dioxide has again increased by 80 ppm to the present 360 ppm, most of the increase in the past 50 years. "What is new is the rate of increase," Mysak emphasized. Adding that the scientific uncertainties regarding climate change are not limited to the physical processes, Mysak commented, "How robust are ecosystems in handling this carbon dioxide? We don't know."

     Stocker, as well, underlines the significance of the rate of increase in greenhouse gas concentrations. He and colleague Andreas Schmittner looked at the problem through experiments with a simple, coupled atmosphere-ocean climate model in which a final carbon dioxide concentration of 750 ppm was attained over different time spans. They found that the thermohaline circulation weakens when the increase in carbon dioxide to 750 ppm is relatively slow, spanning several centuries or more. However, when the rate of increase in atmospheric greenhouse gases (expressed as CO₂) is similar to today's rate of growth (1% per year)--or the concentration of 750 ppm is reached in 100 years--the thermohaline circulation permanently shuts down (11). The quantitative results are dependent on model parameters and their associated uncertainties. Nonetheless, the lesson from the modeling experiments is that the climate is sensitive not only to the level of carbon dioxide in the atmosphere, but also to the rate at which this level is attained. This is relevant to policy decisions concerning the timing of reductions in greenhouse gas emissions, noted Stocker. "It demonstrates that early reductions of emissions make sense."

References

(1) Lorius, C.; Jouzel, J.; Raynaud, D.; Hansen, J.; Le Treut, H. Nature 1990, 347, 139-145.

(2)  Jouzel, J., et al. Clim. Dyn. 1996, 12, 513-521.

(3)  Dansgaard, W.; Johnsen, S. J.; Clausen, H. B.; Dahl-Jensen, D.; Gundestrup, N. S.; Hammer, D. U.; Hvidberg, C. S.; Steffensen, J. P.; Sveinbjörnsdottir, J. J.; Bond, G. Nature 1993, 364, 218-220.

(4)  Mayewski, P. A., et al. Science 1994, 263, 1747-1751.

(5)  Broecker, W. GSA Today 1997, 7(5), 1-7.

(6)  Bjornsson, H.; Mysak, L. A.; Schmidt, G. A. J. Clim. 1997, 10, 2412-2430.

(7)  Broecker, W. Nature 1994, 372, 421-424.

(8)  Manabe, S.; Stouffer, R. J. Nature 1995, 378, 165-167.

(9)  Maslanik, J. A.; Serreze, M. C.; Barry, R. G. Geophys. Res. Lett. 1996 23, 1677-1680.

(10)  Stocker, T. F. Science 1998, 282, 61-62.

(11)  Stocker, T. F.; Schmittner, A. Nature 1997, 388, 862-865.

Deborah Schoen is a science writer based in Montreal, Quebec, Canada.