Greenland ice cores

Greenland ice cores At 12.30 on 12 July 1992 the forty or so field scientists working on the Greenland Ice Core Project (GRIP) funded by the European Science Foundation, celebrated as the drill bit penetrated ice over 200 000 years old and finally reached bedrock at a depth of 3028.8 m. The GRIP core was located at Summit, some 3208 m above sea level situated on the ice divide at the highest part of the ice sheet in central Greenland. The project entailed a great deal of logistical and technical expertise. The base camp had been constructed in 1989; drilling started the following year and took three field seasons to complete. Over 1300 individual core segments 2.3 m long and 100 mm in diameter were recovered by an electromechanical drill 11.5 m long. (Less than 1 m of core was lost in total during the drilling process.) The cores were partly analysed in an underground laboratory 50 m long and were kept on site before being transferred to a purpose-built storage facility in Copenhagen, where they will be available for further study.

Almost exactly a year later, on July 1993, the Greenland Ice Sheet Project 2 (GISP2), funded by the United States, completed a five-year drilling programme at a site only 28 km to the west of the GRIP core. The GISP2 core con-sisted of segments up to 6 m in length and 13.2 cm in diameter. At 3053.44 m (not including the 1.5 m drilled into bedrock), it was the deepest ice core ever recovered.

The completion of the deep GRIP and GISP2 cores heralded a new era in palaeoenvironmental research. They provided the most detailed, high-resolution data on climatic change yet produced, and have yielded information that is invaluable for unravelling past climate history, for testing models and theories of climatic change, and for predicting the future rate and direction of climate change. But GRIP and GISP2 are not the first deep cores available from Greenland. Why are they considered to be so important?

In 1966 the USA Cold Regions Research Engineering Laboratory completed the first deep ice core at Camp Century in north-west Greenland, where the ice was 1390 m thick. Further deep cores were subsequently completed at Renland (1981) in eastern Greenland and at the Dye 3 radar station (1987) in southern Greenland.

Ice cores potentially contain a wealth of information on atmospheric composition, circulation, and pollution, volcanic activity, climatic change, and changes in solar output. The main techniques used in studying them include stable isotope analysis (particularly deuterium and ¹⁸O), analysis of physical features (such as melt horizons and particulate matter), and chemical and electrical variations. However, the Camp Century, Renland, and Dye 3 cores had important limitations. The Camp Century core was located at an altitude of 1885 m above sea level (some 1345 m below GRIP) at a considerable distance from the ice divide. Because of ice flow, the deeper the ice, the further it will have travelled and the greater the potential for a disturbed stratigraphy. Renland also had this problem, lying east of the ice divide. The Dye 3 core was located about 100 km from the ice divide south of the Arctic Circle in a zone of summer melting. Melting introduces an additional problem because the meltwater absorbs soluble gases from the atmosphere before refreezing, and this alters the chemical composition of the ice. Allowances have therefore to be made for these factors in interpreting the data from these cores.

At only one point on the Greenland ice sheet, at Summit, is there potential for recovering a complete undisturbed record with no horizontal movement or melting problems. Even here, however, there is the possibility that at lower levels the great pressure of ice might have squeezed out part of the record. It is also possible that even the location of the ice divide itself may have changed through time. This is why it is so important to have two high-resolution deep ice cores, GRIP and GISP2, so near to one another. By comparing the two cores it is possible to assess the amount of disruption to each record caused by ice flow.

Much work still remains to be completed on GRIP and GISP2, but some remarkable results have already been published. Perhaps the most amazing is shown by the oxygen isotope record, which gives a reliable estimate of the prevailing air temperature over the ice sheet when it was deposited. These data, together with the chemical and electrical signatures, show that during the past 10 000 years (i.e. during the Holocene) the climate has been relatively stable. Most striking, however, is the evidence of rapid climatic change; for instance the temperature rise of 7 °C that marked the most recent climatic shift at the end of the Younger Dryas some 11 700 years ago seems to have taken only 50 years.

There is also remarkable evidence for twenty or so large-scale, rapid, and short-lived cyclical fluctuations from warm interstadial to cold glacial conditions within what is generally termed the last glacial period, between 20 000 and 100 000 years ago.

These warm–cold oscillations, termed ‘Dansgaard– Oeschger’ events (after two prominent ice-core scientists), had previously been observed near to bedrock in the Camp Century, Renland, and Dye 3 cores but it was considered a possibility that these changes might have been caused by disturbed stratigraphy. The almost exact correlation of the ¹⁸O profiles between the Summit, Dye 3, Camp Century, and Renland ice cores shows, however, that these inferred rapid changes in temperature were real.

These events seem to have occurred at irregular intervals. Each comprised a rapid warming phase and a longer, step-like cooling phase. Indeed, the abrupt warming to interstadial conditions probably reflects temperature rises of 5–8 degrees C within a few decades—which is more than half the total glacial-to-interglacial temperature difference.

It is believed that the driving force behind these cyclical events is probably connected with the changing strength of and/or direction of the North Atlantic warm surface ocean current associated with changing amounts of sea ice and deep water formation. The Greenland climate is greatly influenced by the amount of heat released by the sea. The direction and amount of this heat release is closely linked to the global oceanic thermohaline circulation, which acts rather like a conveyor belt. In the North Atlantic the thermohaline circulation is driven by the formation of cold, dense, saline water which sinks and travels southwards, inducing a return flow of surface warm water. When the ‘conveyor’ is working, warm surface water is introduced into the North Atlantic. It seems likely that the rapid changes seen in the Greenland ice cores relate to the switching on and off of this conveyor.

The mechanism that switched the conveyor off may in part be due to the release of huge ‘armadas’ of icebergs from the huge Laurentide ice sheet which spread across the northern Atlantic ocean. The result would be a rapid cooling of the surface water and the creation of a freshwater ‘lid’. This would have the effect of reducing surface salinity and hence density, and would result in the thermohaline circulation being switched off. Evidence for these ‘armadas’, known as Heinrich events, comes from the analysis of deep ocean cores in the North Atlantic. Layers in these cores have been found to contain ice-rafted debris from Canada and low numbers of foraminifera, which reflect low oceanic productivity.

The Heinrich events seem to be spaced at intervals of about 10 000 years and occur during the coldest of a series of Dansgaard–Oeschger phases. The apparent regularity of these cycles may suggest a link with the Earth's precessional cycle. However, others argue it might be due to the internal dynamics of the Laurentide ice sheet, alternately building up and discharging large amounts of ice. The fact that similar and synchronous discharge events have now been recorded from the Scandinavian ice sheet make it more likely that a factor external to the ice sheets is responsible, since it would be unlikely that the internal dynamics of two dissimilar ice sheets would be in harmony.

The GRIP and GISP2 cores also penetrated through the previous interglacial (Eemian) and revealed further evidence for climatic instability. In the GRIP core the Eemian period comprises about 80 m thickness of ice some 163 to 238 m above bedrock. Only in the first part of the interglacial, equivalent to Marine Isotope Stage 5e, do temperature appear to reach interglacial warmth. Even within this stage there seem to be many short-lived returns to interstadial conditions (5e2, 5e4; Fig. 5) interspersed with warmer episodes (5e1, 5e3, 5e5; Fig. 5). And even within these warmer episodes there are very short-lived changes; for instance, within warm period 5e1 there is a cooling event (event 1) which lasted about 70 years, during which temperatures dropped from interglacial to mid-glacial and back again.

The Eemian at its warmest was on average 2 degrees C warmer than the peak of our present interglacial and has been used as a model for possible climates in a future warmer world. These abrupt shifts within the Eemian have not been found in other cores. If valid, they may indicate the possibility of future rapid climate changes on a scale not seen before during the present interglacial.

There is some debate over these Eemian fluctuations because they occur lower in the core and may be the result of disturbed stratigraphy. Indeed comparisons of the GRIP and GISP2 cores using ¹⁸O stratigraphy and electrical conductivity confirms excellent agreement apart from the bottom 10 per cent of the cores. Below about 2700 m, from interstadial 22 and through the Eemian, there is a breakdown in the correlation between the two cores. This could mean that either or both the GRIP and GISP2 cores have disturbed stratigraphy.

There are two ways in which the stratigraphy might be disturbed: through large-scale overturn folds and through small-scale squeezing out of some layers (a process comparable to boudinage in rocks). One method of detecting disturbance is to look at the dip of cloudy bands within the ice. Cloudy bands consist of microbubbles that form round microparticles of dust, particularly in the colder ice, once the core has been retrieved from great depths and the pressure is relaxed. The microparticles tend to be deposited in early summer and therefore show the position of former ice surfaces. While there is some evidence for disturbance in the GRIP core between 2900 and 2954 m, this is below the Eemian layers.

On the smaller scale, boudinage might result in the squeezing out of some layers and thickening of others, but this would only stretch some parts of the timescale and contract others; the broad outline of the timescale would still be valid and it could not change an originally stable ¹⁸O record into an unstable one.

The evidence for rapid Eemian temperature changes is not found elsewhere. There could be three possible reasons for this: (1) disturbed stratigraphy (thought unlikely by GRIP scientists); (2) the evidence observed may represent changes of limited geographical extent that are consequently not found in other areas; and (3) the changes might have occurred outside the North Atlantic region but the effects were damped because of low sensitivity or time resolution of the techniques applied in other areas. The search is now on for comparable abrupt events in the oceanic and continental records, including North Atlantic ocean cores and north-west European lake and peat deposits, and a more detailed reanalysis of the Vostok ice core. Plans are also being made to drill new ice cores in Antarctica. It is hoped that the European Project for Ice Coring in Antarctica (EPICA) programme will focus on the periods of rapid climatic change through several glacial cycles.

B. A. Haggart

Bibliography

Bond, G.,, Broecker, W.,, Johnsen, S.,, McManus, J.,, Labeyrie, L.,, Jouzel, J.,, and and Bonami, G. (1993) Correlations between climate records from North Atlantic sediments and Greenland ice. Nature, 365, 143–7.