The most extreme manifestation of climate change in the geological record is the transition from full glacial to full inter-glacial conditions. During the most recent glacial cycle, peak glacial conditions prevailed from about 25 to 18 ky BP. Temperatures close to those of today were restored by approximately 10 ky BP. However, warming was not continuous. The deglaciation was accomplished in two main stages, with a return to colder conditions (Younger Dryas/Antarctic Cold Reversal) or, at the least, a pause in the deglaciation.
The central Greenland ice core record (GRIP and GISP2) has a near annual resolution across the entire glacial to Holocene transition, and reveals episodes of very rapid change. The return to the cold conditions of the Younger Dryas from the incipient inter-glacial warming 13,000 years ago took place within a few decades or less (Alley et al., 1993). The warming phase, that took place about 11,500 years ago, at the end of the Younger Dryas was also very abrupt and central Greenland temperatures increased by 7°C or more in a few decades (Johnsen et al., 1992; Grootes et al., 1993; Severinghaus et al., 1998). Most of the changes in wind-blown materials and some other climate indicators were accomplished in a few years (Alley et al., 1993; Taylor et al., 1993; Hammer et al., 1997). Broad regions of the Earth experienced almost synchronous changes over periods of 0 to 30 years (Severinghaus et al., 1998), and changes were very abrupt in at least some regions (Bard et al., 1987), e.g. requiring as little as 10 years off Venezuela (Hughen et al., 1996). Fluctuations in ice conductivity indicate that atmospheric circulation was reorganised extremely rapidly (Taylor et al., 1993). A similar, correlated sequence of abrupt deglacial events also occurred in the tropical and temperate North Atlantic (Bard et al., 1987; Hughen et al., 1996) and in Western Europe (von Grafenstein et al., 1999).
A Younger-Dryas type event is also recorded in a Bolivian ice core (Thompson et al., 1998; Sajama, South America in Figure 2.24) and in a major advance of a mountain glacier in the Southern Alps of New Zealand (Denton and Hendy, 1994). However there is recent evidence against a significant Younger Dryas cooling here (Singer et al., 1998) and at other sites of the Southern Hemisphere (reviewed by Alley and Clarke, 1999). Instead, the Antarctic (and Southern Ocean) climate was characterised by a less pronounced cooling (the Antarctic Cold Reversal: Jouzel et al., 1987) which preceded the Younger Dryas by more than 1 ky (Jouzel et al., 1995; Sowers and Bender, 1995; Blunier et al., 1997). Curiously, one coastal site in Antarctica, Taylor Dome (Steig et al., 1998) exhibited cooling in phase with the North Atlantic. Recent series obtained at Law Dome, another coastal site of East Antarctica, show instead a cold reversal preceding the Younger Dryas as in other Antarctic records. This suggests that the Taylor Dome record is of limited geographical significance but it also suggests that there is more to be discovered about this cooling event in the Southern
The inception of deglacial warming about 14.5 ky BP was also very rapid, leading to the Bölling-Alleröd warm period in less than twenty years (Severinghaus and Brook, 1999). Almost synchronously, major vegetation changes occurred in Europe and North America with a rise in African lake levels (Gasse and van Campo, 1994). There was also a pronounced warming of the North Atlantic and North Pacific (Koç and Janssen, 1994; Sarnthein et al., 1994; Kotilainen and Shackleton, 1995; Thunnell and Mortyn, 1995; Wansaard, 1996; Watts et al., 1996; Webb et al., 1998).
The rate of temperature change during the recovery phase from the last glacial maximum provides a benchmark against which to assess warming rates in the late 20th century. Available data indicate an average warming rate of about 2°C/millennium between about 20 and 10 ky BP in Greenland, with lower rates for other regions. Speleothem data from New Zealand, and positions of mountain glacier moraine termini suggest warming rates of 2°C/millennium from 15 to 13 ky BP (Salinger and McGlone, 1989). Speleothem data for South Africa suggest a warming rate of 1.5°C/millennium (Partridge, 1997) over the same time period. On the other hand, very rapid warming at the start of the Bölling-Alleröd period, or at the end of the Younger Dryas may have occurred at rates as large as 10°C/50 years for a significant part of the Northern Hemisphere.
Oxygen isotope measurements in Greenland ice cores demonstrate that a series of rapid warm and cold oscillations called Dansgaard-Oeschger events punctuated the last glaciation (Figure 2.23, see North Atlantic SST panel, and Dansgaard et al., 1993). Associated temperature changes may be as high as 16°C (Lang et al., 1999). These oscillations are correlated with SST variations in several North Atlantic deep-sea cores (Bond et al., 1993). There was clearly a close relation between these ice core temperature cycles and another prominent feature of North Atlantic deep-sea core records, the Heinrich events. Heinrich events occurred every 7,000 to 10,000 years during times of sea surface cooling in the form of brief, exceptionally large, discharges of icebergs from the Laurentide and European ice sheets which left conspicuous layers of detrital rocks in deep-sea sediments. Accompanying the Heinrich events were large decreases in the oxygen isotope ratio of planktonic foraminifera, providing evidence of lowered surface salinity probably caused by melting of drifting ice (Bond et al., 1993). Heinrich events appear at the end of a series of saw-toothed shaped, near millennial temperature cycles. Each set of millennial cycles is known as a Bond cycle. Each cycle was characterised by a succession of progressively cooler relatively warm periods (interstadials) during the Ice Age period. Each cooling trend ended with a very rapid, high amplitude, warming and a massive discharge of icebergs. The impact of these Heinrich events on the climate system extended far beyond the northern North Atlantic. At the time of major iceberg discharges, strong vegetation changes have been detected in Florida (Grimm et al., 1993; Watts et al., 1996), oceanic changes occurred in the Santa Barbara Basin off California (Behl and Kennet, 1996) and changes in loess grain-size, associated with atmospheric circulation changes, have been detected in China (Porter and An, 1995; Ding et al., 1998).
Deep-sea cores also show the presence of ice rafting cycles in the intervals between Heinrich events (Bond and Lotti, 1995). Their duration varies between 2,000 and 3,000 years and they closely coincide with the Dansgaard-Oeschger events of the last glaciation. A study of the ice-rafted material suggests that, coincident with the Dansgaard-Oeschger cooling, ice within the Icelandic ice cap and within or near the Gulf of Saint Lawrence underwent nearly synchronous increases in rates of calving. The Heinrich events reflect a slower rhythm of iceberg discharges, probably from the Hudson Strait.
Air temperature, SST and salinity variations in the North Atlantic are associated with major changes in the thermohaline circulation. A core from the margin of the Faeroe-Shetland channel covering the last glacial period reveals numerous oscillations in benthic and planktonic foraminifera, oxygen isotopes and ice-rafted detritus (Rasmussen et al., 1996a). These oscillations correlate with the Dansgaard-Oeschger cycles, showing a close relationship between the deep ocean circulation and the abrupt climatic changes of the last glaciation. Warm episodes were associated with higher SST and the presence of oceanic convection in the Norwegian Greenland Sea. Cold episodes were associated with low SST and salinity and no convection in the Norwegian Greenland Sea (Rasmussen et al., 1996b). Cores from the mid-latitudes of the North Atlantic show that the iceberg discharges in Heinrich events resulted in both low salinity and a reduced thermohaline circulation (Cortijo et al., 1997; Vidal et al., 1997).
These rapid climatic events of the last glacial period, best documented in Greenland and the North Atlantic, have smoothed counterparts in Antarctica (Bender et al., 1994; Jouzel et al., 1994). A peak in the concentration of the isotope beryllium-10 in ice cores (Yiou et al., 1997a), changes in the concentration of atmospheric methane (Blunier et al., 1998) and in the isotopic content of oxygen in ice cores (Bender et al., 1999) indicate links between the Northern and Southern Hemisphere climates over this period. Large Greenland warming events around 36 and 45 ky BP lag their Antarctic counterparts by more than 1,000 years. This argues against coupling between northern and southern polar regions via the atmosphere but favours a connection via the ocean (Blunier et al., 1998).
New evidence suggests that the North Atlantic has three modes of operation. These are: deep-water sinking in the GIN (Greenland-Iceland-Norwegian) Seas and the Labrador Sea, deep-water sinking in the North Atlantic or in the Labrador Sea but not the GIN Seas (Duplessy et al., 1991; Labeyrie et al., 1992) in the cold phase of Dansgaard-Oeschger events and at glacial maximum, and little deep-water sinking in the GIN or Labrador Seas (Heinrich events) (Sarnthein et al., 1994; Vidal et al., 1997, 1998; Alley and Clark, 1999; Stocker, 2000). The first type corresponds to modern, warm conditions. Shut-down of convection in the GIN Seas has a strong effect on the high latitude Atlantic atmosphere and on areas that respond to it such as the monsoon regions of north Africa (Street-Perrott and Perrott, 1990). However, cross-equatorial Atlantic ocean surface transport that supplies the water for the formation of the Labrador Sea deep-water continues to remove heat from the South Atlantic under these conditions. The additional "Heinrich shut-down" of the North Atlantic and Labrador Sea deep-water formation allows this heat to remain in the South Atlantic (Crowley, 1992), and may increase deep-water formation either south of the area affected by melt-water injection (Vidal et al., 1997, 1998) or in the Southern Ocean (Broecker, 1998). This reorganisation could cause warming of regions of the South Atlantic and downwind of it (Charles et al., 1996; Blunier et al., 1998) through a seesaw relationship with the North Atlantic. However, the behaviour of Taylor Dome in the Antarctic and several other southern sites (see above) which exhibit cooling in phase with the North Atlantic argue for an additional atmospheric link to some southern regions.
Other reports in this collection