Examples of abrupt change have been identified in the Holocene as well. Between about 12 and 5 thousand years ago the last phases of deglaciation led to repeated reroutings of glacial lake meltwater into the convecting regions of the North Atlantic (e.g., Teller and Leverington, 2004). It has been argued that each of these events weakened the North Atlantic THC enough to cause abrupt cooling of the northern North Atlantic, Greenland, Iceland, and Europe, perhaps by as much as a few degrees. Yet proxy data from the northern North Atlantic so far have revealed no compelling evidence for changes in THC (Keigwin and Boyle, 2000).

Even if, as many assume, threshold behavior in THC is the most likely cause of abrupt climate change, the processes that lead to such behavior cannot be quantified. The ways in which the North Atlantic THC interacts with other oceans is not well known (Whitworth et al., 1999); it is not certain whether freshwater forcing actually weakens THC under all conditions (Marotzke, 2000). Further, the way in which convective activity is linked to THC is not fully understood (Marotzke and Klinger, 2000).

An abrupt climate change would have more severe impacts on natural and human systems than more gradual change. In particular, it could result in more extremes in climate variables (e.g., heat index, storm intensities and frequencies). Knowledge of what triggers abrupt climate changes is still quite limited; more research is needed to determine whether human-caused increases in greenhouse gases or land-use changes might lead to an abrupt change. Indeed, past abrupt climate changes have been especially common when the climate system itself was being altered (NRC, 2002).