Ice Sheet Hysteresis

Coupled climate-ice sheet models have been reasonably successful in simulating sudden Cenozoic glaciation events such as Oil and Mil (DeConto and Pollard, 2003a, b). For example, beginning with an ice-free continent and assuming gradually declining greenhouse gas concentrations and accounting for orbital forcing, DeConto and Pollard (2003a) simulated the sudden stepwise glaciation of East Antarctica within a 200-ky interval. The simulated ice sheet was comparable in volume to the modern EAIS, but significantly smaller than the volume of Mil ice reconstructed from the proxy isotope and sea-level records discussed above. Subsequent modelling work, including a representation of ice shelves not included in their earlier simulations (Pollard and DeConto, 2007), have shown that an Antarctic Ice Sheet b 20-25% bigger than today would have required a glaciated West Antarctica, and ice grounding lines extending close to the continental shelf break around much of the margin. An Mil ice sheet of this size would have been similar in geometry to the ice sheet that existed at the Last Glacial Maximum (Huybrechts, 2002). However, the presumably warmer ocean at Mil might have been unconducive to the seaward migration of grounding lines, so this scenario maybe difficult to reconcile from a modelling perspective. Furthermore, the cold south polar conditions implied by such an Ice Sheet also implies global temperatures low enough to have allowed significant glaciation in the Northern Hemisphere, especially during orbital periods which produced cold boreal summers. While Greenland may have contained some glacial ice as early as the Eocene (Eldrett et al., 2007), the Oligocene-Miocene boundary is ~20myr before the onset of the fist significant Northern Hemisphere glacial cycles. Clearly, some important problems remain in terms of reconciling the magnitude of the Mil event.

While the rapid growth of Antarctic ice at Mil can be explained through a combination of decreasing greenhouse gas concentrations and orbital forcing (with other possible influences from mountain uplift and/or ocean circulation), the ephemeral nature of the event and subsequent variability of ice volume are also problematic from a modelling perspective (Pollard and DeConto, 2005). As shown in both simple and sophisticated numerical icesheet models, (Weertman, 1961; Huybrechts, 1994; Pollard and DeConto, 2005), the high albedo and elevation of large polar-centred ice sheets produce considerable hysteresis. In a scenario of cooling climate, a polar ice sheet can grow suddenly, once the snow line intersects sufficient land area in mountains and high plateau. The non-linear jump in ice volume is facilitated by height-mass balance and albedo feedbacks, as the ice sheet spreads horizontally (albedo feedback) and more of the parabolic ice surface rises above the snow line and out of the ablation zones around its margins (height-mass balance feedback) (Abe-Ouchi and Blatter, 1993; DeConto and Pollard, 2003a). The high elevation and albedo of the ice sheet inhibit the ice sheet from disappearing during subsequent warming interval, unless temperatures (snow lines) rise far above their initial values (elevation) at the time of glacial onset (Huybrechts, 1994). Pollard and DeConto (2005) studied this hysteresis effect in a coupled GCM-ice sheet model and in a simple flowline model with parameterised mass-balance forcing. They concluded that the hysteresis effect is strong enough to preclude orbital forcing from driving the range of Cenozoic ice-volume variability seen in the oxygen isotope and sea-level records described above, unless the orbital forcing is accompanied by significant changes in greenhouse concentrations. During favourable (cold austral summer) orbital periods, the atmospheric CO2-glaciation threshold for Antarctica is ~2 x pre-industrial levels, while CO2 must approach ~ 4 x pre-industrial levels during a warm austral summer orbital period to trigger the collapse of the interior EAIS. If the sensitivity of the models to orbital and greenhouse gas forcing is reasonable, the short duration of the peak Mi1 event would require a significant perturbation to the carbon cycle, producing significant global warming soon after the peak glacial interval. Greenhouse gas variability of this magnitude is not evident in existing proxy reconstructions of early Miocene CO2 (e.g. Pagani et al., 2005), however, higher-resolution records will be required to resolve this type of CO2 variability across key climatic events like the Oligocene-Miocene boundary.

0 0

Post a comment