IPCC Fourth Assessment Report: Climate Change 2007
Climate Change 2007: Working Group I: The Physical Science Basis Can Climate Models Simulate these Abrupt Changes?

Modelling the ice sheet instabilities that are the likely cause of Heinrich events is a difficult problem because the physics are not sufficiently understood, although recent results show some promise (Calov et al., 2002). Many model studies have been performed in which an influx of freshwater from an ice sheet instability (Heinrich event) or a melt water release (8.2 ka event; see Section 6.5.2) has been assumed and prescribed, and its effects on ocean circulation and climate have been simulated. These experiments suggest that freshwater input of the order of magnitude deduced from palaeoclimatic data could indeed have caused the Atlantic MOC to shut down, and that this is a physically viable explanation for many of the climatic repercussions found in the data (e.g., the high-latitude northern cooling, the shift in the ITCZ and the hemispheric seesaw; Vellinga and Wood, 2002; Dahl et al., 2005; Zhang and Delworth, 2005). The phase relation between temperature in Greenland and Antarctic has been explained by a reduction in the NADW formation rate and oceanic heat transport into the North Atlantic region, producing cooling in the North Atlantic and a lagged warming in the SH (Ganopolski and Rahmstorf, 2001; Stocker and Johnsen, 2003). In freshwater simulations where the North Atlantic MOC is forced to collapse, the consequences also include an increase in nutrient-rich water in the deep Atlantic Ocean, higher 231Pa/230Th ratios in North Atlantic sediments (Marchal et al., 2000), a retreat of the northern treeline (Scholze et al., 2003; Higgins, 2004; Köhler et al., 2005), a small (10 ppm) temporary increase in atmospheric CO2 in response to a reorganisation of the marine carbon cycle (Marchal et al., 1999) and CO2 changes of a few parts per million due to carbon stock changes in the land biosphere (Köhler et al., 2005). A 10 ppb reduction in atmospheric N2O is found in one ocean-atmosphere model (Goldstein et al., 2003), suggesting that part of the measured N2O variation (up to 50 ppb) is of terrestrial origin. In summary, model simulations broadly reproduce the observed variations during abrupt events of this type.

Dansgaard-Oeschger events appear to be associated with latitudinal shifts in oceanic convection between the Nordic Seas and the open mid-latitude Atlantic (Alley and Clark, 1999). Models suggest that the temperature evolution in Greenland, the seesaw response in the South Atlantic, the observed Irminger Sea salinity changes and other observed features of the events may be explained by such a mechanism (Ganopolski and Rahmstorf, 2001), although the trigger for the ocean circulation changes remains undetermined. Alley et al. (2001) showed evidence for a stochastic resonance process at work in the timing of these events, which means that a regular cycle together with random ‘noise’ could have triggered them. This can be reproduced in models (e.g., the above), as long as a threshold mechanism is involved in causing the events.

Some authors have argued that climate models tend to underestimate the size and extent of past abrupt climate changes (Alley et al., 2003), and hence may underestimate the risk of future ones. However, such a general conclusion is probably too simple, and a case-by-case evaluation is required to understand which effects may be misinterpreted in the palaeoclimatic record and which mechanisms may be underestimated in current models. This issue is important for an assessment of risks for the future: the expected rapid warming in the coming centuries could approach the amount of warming at the end of the last glacial, and would occur at a much faster rate. Hence, melt water input from ice sheets could again become an important factor influencing the ocean circulation, as for the Younger Dryas and 8.2 ka events. A melting of the Greenland Ice Sheet (equivalent to 7 m of global sea level) over 1 kyr would contribute an average freshwater flux of 0.1 Sv; this is a comparable magnitude to the estimated freshwater fluxes associated with past abrupt climate events. Most climate models used for future scenarios have thus far not included melt water runoff from melting ice sheets. Intercomparison experiments subjecting different models to freshwater influx have revealed that while responses are qualitatively similar, the amount of freshwater needed for a shutdown of the Atlantic circulation can differ greatly between models; the reasons for this model dependency are not yet fully understood (Rahmstorf et al., 2005; Stouffer et al., 2006). Given present knowledge, future abrupt climate changes due to ocean circulation changes cannot be ruled out.