Global sea levels would rise by several metres if the Greenland Ice Sheet melted completely. Two studies have examined its past behaviour in an effort to evaluate its vulnerability in a warming world — and have come to seemingly conflicting conclusions. Two geochemists and a glaciologist discuss the issues. See Letters p.252 & p.256
The papers in brief
Knowledge of the ancient history of ice sheets is needed to inform predictions of their future response to climate change.
From their data, they propose that Greenland has undergone one or more episodes of deglaciation during the past 1.1 million years.
They conclude that the East Greenland Ice Sheet remained present during the Pleistocene (the epoch that lasted from 2.6 million to 11,700 years ago), but grew and shrank dynamically in response to climate.
Pierre-Henriblard & Guillaume Leduc
It is difficult to find evidence for past waxing and waning of the Greenland Ice Sheet (GrIS), because the glacial debris that can provide a record of ice-sheet dynamics (Fig. 1) is erased each time glaciers advance after a period of deglaciation. Schaefer et al. and Bierman et al. now report the presence of two rare nuclides (10Be and 26Al) in Greenland's geological archives. These 'cosmogenic' nuclides are produced by the interaction between atoms in minerals and high-energy cosmic neutrons that probably originated in supernovae, and they provide much-needed evidence about the GrIS's ancient past.
A key characteristic of these isotopes is that they are produced in detectable quantities in only the first few metres below Earth's surface. If they become buried below hundreds of metres of ice, their concentrations decrease through radioactive decay. The half-lives of 10Be and 26Al are 1.4 million and 0.7 million years, respectively, so the presence of these nuclides in bedrock under ice sheets is direct evidence that one or several ice-free episodes occurred in the recent past (less than 5 million years ago).
Bierman et al. report variations in 10Be and 26Al from marine sediments deposited near the eastern margin of today's GrIS over the past 7.5 million years. The low 10Be concentrations suggest that the sedimentary material originates from locations that were efficiently shielded from cosmic rays. The authors therefore conclude that large ice caps probably existed continuously across East Greenland — the source of the sedimentary material — during the past 7.5 million years. They also observe an additional long-term decrease in 10Be and 26Al concentrations during this period, which indicates a gradual increase in the volume of the GrIS.
In parallel, Schaefer et al. analysed bedrock samples recovered from beneath several kilometres of ice in the centre of the present-day GrIS. These rocks cannot have been transported to their current position from elsewhere. The presence of 10Be and 26Al in these samples is therefore direct evidence for past deglaciation episodes at this location.
Schaefer and colleagues used the observed concentrations and ratios of 10Be and b26Al in a numerical model to explore plausible glaciation–deglaciation scenarios. They firmly conclude that the sampled bedrock must have been ice-free for at least one episode of about 280,000 years during the past 1.1 million years. Other glaciation–deglaciation scenarios are more plausible, however, such as several shorter disappearances (several thousand years) of the GrIS during the warmest interglacial periods of the Pleistocene3.
The deglaciation proposed by Schaefer et al. seems at odds with Bierman and colleagues' conclusions. But small ice caps might have persisted on the East Greenland highlands during such deglaciated episodes, and could be the origin of the 10Be-poor material observed in Bierman and co-workers' marine sediments. Indeed, modelling experiments4 estimate that ice caps 1,000 metres thick might persist on the southeastern Greenland highlands in the future even if 95% of the GrIS disappears (Fig. 2).
Could the GrIS undergo similar deglaciations in the future? Average global warming of just 2 °C compared with preindustrial temperatures could be enough to melt more than 75% of the GrIS3. If this high sensitivity to warming is confirmed, then the current worst-case scenarios5 for future sea-level rise associated with anthropogenic global warming will need to be revised.
One of the most powerful methods for predicting future behaviour is to look for clues from the past. But interpreting the past is rarely straightforward, as illustrated by the seemingly contradictory conclusions of Schaefer et al. and Bierman and colleagues. The findings can perhaps be reconciled, but that would require glaciologists to have a major rethink about current theories of ice-sheet behaviour.
Schaefer and colleagues' analysis of the terrestrial bedrock beneath the GrIS leads them to conclude that the ice sheet was sometimes reduced to less than 10% of its current volume, so that the only ice to be found in Greenland was in the peripheral mountain ranges. By contrast, Bierman and co-workers' study of the marine record indicates that warming was short-lived, and that it seldom caused substantial and lengthy reductions in the extent of the East Greenland Ice Sheet. So how do these conclusions fit in with what is known about ice-sheet dynamics?
One source of information is numerical modelling of ice sheets. Simulations6 based on such models tend to predict that the GrIS existed continuously throughout the Pleistocene. The models rarely predict extended periods during which Greenland was nearly ice-free.
Glaciologists also know from observations that the GrIS is dynamic, and that its peripheral outlet glaciers are responding rapidly to climate change (Fig. 1). But most of these observations come from areas around the periphery of the ice sheet; the central areas are considered to be relatively stable. The conclusion that the GrIS shrank to less than 10% of its current volume, leaving ice-free areas at its centre, is therefore particularly difficult to explain glaciologically because it requires the entire ice sheet to have been removed and completely 'rebuilt', possibly on several occasions.
Perhaps part of the explanation lies in Greenland's subglacial topography7. Earth's crust is depressed at the centre of Greenland because of the great weight of the ice sheet, whereas the periphery is surrounded by mountain ranges (Fig. 2). Deglaciation right to the centre of the GrIS — rather than just changes in the peripheral outlet glaciers — would therefore require massive, rapid changes of ice dynamics. This calls for glaciologists to rethink how the ice sheet interacts with subglacial topography and sea level. Could the same deep valleys that give the central GrIS its great thickness also open up the underside of the ice sheet and make it vulnerable to warming oceans as sea levels rise?
These new papers throw down three immediate challenges. First, we must seek ways to reconcile the two seemingly contradictory records of the ice sheet's past behaviour. Second, we must try to understand the dynamical processes of the ice sheet that make possible the required huge and rapid variations in the size and volume of the GrIS. And third, we need to assess whether such variations could happen again in the near future, with all the attendant social and economic consequences that would accompany a rapid rise in global sea level.