During the last ice age, huge numbers of icebergs were episodically discharged from an ice sheet that covered North America. Numerical modelling suggests that these events resulted from a conceptually simple feedback cycle. See Letter p.332
Understanding abrupt changes in ice sheets and climate is a long-standing issue in palaeoclimate science1. In the last glacial period, North America's Laurentide Ice Sheet episodically discharged large armadas of icebergs through the Hudson Strait into the North Atlantic2. This expulsion of fresh water in the form of icebergs not only raised the global sea level by a few metres, but also weakened ocean circulation in the North Atlantic and therefore affected climate. The cause of these 'Heinrich events' remains hotly debated. On page 332, Bassis et al.3 use an ice-sheet model to demonstrate that such events could have been triggered by relatively small fluctuations in subsurface ocean temperature — a result that has implications for the survival of present-day ice sheets.
Evidence for Heinrich events is found as distinct layers of coarse glacial debris in ocean-sediment records2. Such particles were incorporated at the base of the Laurentide Ice Sheet, and were then transported with the discharged icebergs into the North Atlantic. After the icebergs melted, these particles accumulated on the ocean floor.
Heinrich events occurred every few thousand years during the glacial periods, but their interval and intensity varied. Interestingly, the timing of these events coincides with cold phases of Dansgaard–Oeschger (DO) cycles4 — millennium-scale climate oscillations recorded in ice cores drilled in Greenland5. Rapid ice-sheet decay during cold periods seems surprising, given present-day observations of warming and rapid glacier retreat in Greenland (Fig. 1).
In the 1990s, a 'binge–purge' mechanism6 was proposed to explain Heinrich events. In this explanation, the Laurentide Ice Sheet slowly increased in mass by gradually accumulating ice (the 'binge' phase). When the ice was sufficiently thick, the base of the ice sheet reached its melting point. The subglacial sediment acted as a lubricant, causing the ice sheet to slide, become unstable and discharge icebergs at an enhanced rate (the 'purge' phase). Although this mechanism does not require the involvement of external factors such as climate change, it cannot explain the synchronization between Heinrich events and DO cycles.
Other explanations7,8 assume that a large, protruding ice shelf, situated between Canada and the southern tip of Greenland, buttressed the Laurentide Ice Sheet. Subsurface ocean warming during the cold phases of DO cycles, as revealed by proxy data9, could then have melted and destabilized the ice shelf, triggering Heinrich events. However, the existence of a large ice shelf is incompatible with evidence10 of a mostly open ocean during the last glacial period. Moreover, the models that describe the ice-shelf collapse are poorly constrained.
Bassis and colleagues consider an alternative mechanism in which pulses of subsurface ocean warming act as an external trigger, but the existence of a large ice shelf is not required (see Figure 1 in the paper3). The Laurentide Ice Sheet starts at its full extent, depressing the sill — a raised frontal section of the glacier bed — at the mouth of the Hudson Strait. A pulse of subsurface ocean warming of a few degrees then enhances iceberg calving. Because the glacier bed deepens upstream, this causes an unstable, rapid ice-sheet retreat in the form of a Heinrich event. The glacier bed gradually adjusts to the ice sheet's decreasing mass by rising,which reduces iceberg calving and elevates the sill, cutting off contact with the warm subsurface water. The ice sheet can then slowly advance to its full extent and the entire process is repeated, giving rise to a pulsating ice sheet.
The authors simulate Heinrich events generated by this mechanism using a numerical ice-sheet model. They find that their results are fully consistent with the observational record, in both timing and magnitude. Crucially, their events occur only for the ocean-warming pulses for which Heinrich events are actually observed. This implies that full ice-sheet recovery and subsequent depression of the sill are prerequisites for ocean warming to trigger a Heinrich event, and also control the interval between events.
A strength of the authors' model is that it is quantitative, physically based and includes the dominant dynamic processes and feedbacks expected for such ice sheets. Furthermore, although the model is simple, it is robust against a wide choice of model parameters. Nevertheless, future studies should certainly explore more-sophisticated models.
Bassis and colleagues' proposed mechanism relies on subsurface ocean warming, which is the factor that seems to drive today's dynamic mass loss from ice sheets in Greenland and Antarctica11. However, it is not known to what extent glacier-bed adjustment preconditioned today's ice sheets for abrupt retreat. Bassis et al. focus on explaining Heinrich events and do not resolve the cause of ocean warming. Nor do the authors consider how freshwater flux from Heinrich events affects the global climate system. Answering these enigmatic questions in palaeoclimate science and understanding Heinrich events in the context of DO cycles would require the proposed mechanism to be integrated into coupled atmosphere–ocean-circulation models.
The authors' work is a reminder of the complexity and nonlinearity of ice-sheet–climate systems, showing how small perturbations can trigger extreme responses through feedback mechanisms. This is relevant for understanding the palaeoclimate, as well as for discussions on global warming, rapid ice-sheet retreat and future sea-level rise1.Footnote 1
Stocker, T. F et al. (eds) Climate Change 2013: The Physical Science Basis. Working Group I Contribution to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (Cambridge Univ. Press, 2013).
Heinrich, H. Quat. Res. 29, 142–152 (1988).
Bassis, J. N., Petersen, S. V. & Cathles, L. M. Nature 542, 332–334 (2017).
Bond, G. et al. Nature 365, 143–147 (1993).
Dansgaard, W. et al. Nature 364, 218–220 (1993).
MacAyeal, D. R. Palaeoceanography 8, 775–784 (1993).
Hulbe, C. L., MacAyeal, D. R., Denton, G. H., Kleman, J. & Lowell, T. V. Palaeoceanography 19, PA1004 (2004).
Alvarez-Solas, J. et al. Nature Geosci. 3, 122–126 (2010).
Marcott, S. A. et al. Proc. Natl Acad. Sci. USA 108, 13415–13419 (2011).
de Vernal, A., Hillaire-Marcel, C., Turon, J.-L. & Matthiessen, J. Can. J. Earth Sci. 37, 725–750 (2000).
Joughin, I., Alley, R. B. & Holland, D. M. Science 338, 1172–1176 (2012).