Ice core, marine and terrestrial records show that the Holocene was marked by a millennial-scale mode of variability (Meese et al., 1994; O’Brien et al., 1995; Bond et al., 1997; Yiou et al., 1997a,b). These variations affect both atmospheric (Mayewski et al., 1997) and oceanic (Bianchi and McCave, 1999) indicators. The occurrence of very large floods in the south-western United States also reflects substantial low-frequency variability (Ely et al., 1993). SSTs reconstructed from analyses of a sub-tropical, high sedimentation rate site off West Africa might indicate a remarkably high amplitude Holocene variability of 5 to 8°C on a time-scale about 1,500 years (deMenocal, 1998). During the later Holocene, New Zealand speleothems indicate a lowering of temperature after about 7 ky BP, with small advances of the mountain glaciers in the Southern Alps near about 4 and 2.5 ky BP (Salinger and McGlone, 1989). Speleothem records also indicate a temperature decrease of about 1.5°C some 2 to 3 ky ago (Williams et al., 1999). These indications are consistent with cooler periods at these times shown by South African speleothems (Partridge, 1997). By contrast, temperature peaks appeared in China at about 7 ky BP and at 5.5 to 6 ky BP (Wang and Gong, 2000).

Figure 2.24: Records of climate variability during the Holocene and the last climatic transition, including the 8.2 ky BP event (adapted from Johnsen et al., 1992; Hughen et al., 1996; Thompson et al., 1998; von Grafenstein et al., 1999; Jouzel et al., 2001). The shaded areas show the 8.2 ky BP event, the Younger Dryas event and the Antarctic Cold Reversal. The grey scale used in the Tropical North Atlantic record is a measure of sea surface temperature, deduced from the colour of plankton rich layers within an ocean sediment core.

Central Greenland ice cores and European lake isotopic records show correlated temperature variations within the Holocene, with a roughly 50% higher amplitude at Summit Greenland, compared to Europe (Figure 2.24). The most prominent event in both records occurred about 8,200 years BP (Alley et al., 1997; von Grafenstein et al., 1998; Barber et al., 1999) when annual mean temperatures dropped by as much as 2°C in mid-Europe and the European alpine timberline fell by about 200 m (Wick and Tinner, 1999). The event may be related to a significant decrease of SST in the Norwegian Sea (Klitgaard-Kristensen et al., 1998). Lake records from the southern border of the Sahara indicate extremely dry conditions during this time, and probably also during other cool but less dramatic events of this kind (Street-Perrot and Perrot, 1990 ; Gasse and Van Campo, 1994). The about 8,200 year cooling may also have been worldwide (Stager and Mayewski, 1997), although abrupt early Holocene climate changes recorded in a North American lake are thought to reflect a different event (Hu et al., 1999). Thus cooling is indicated in the New Zealand Southern Alps, with small advances of the mountain glaciers at about 8,000 years BP (Salinger and McGlone, 1989).

Further abrupt climatic changes and reversals on millennial time-scales during the Holocene are documented from pollen and lake level records e.g., in Europe (Magny, 1995; Pazdur et al., 1995; Combourieu-Nebout et al., 1998), North Africa (Gasse et al., 1990; Lamb et al., 1995), North America (Jacobson et al., 1987; Overpeck et al., 1991) and Australia (Kershaw et al., 1991). Holocene lake level changes in Europe have been shown to correlate (Magny, 1995; Yu and Harrisson, 1996) with millennial-scale changes in North Atlantic SST and salinity records (Duplessy et al., 1992; Gasse and van Campo, 1994), suggesting a possible link between millennial thermohaline circulation variability and atmospheric circulation over Europe.

The early Holocene was generally warmer than the 20th century but the period of maximum warmth depends on the region considered. It is seen at the beginning of the Holocene (about 11 to 10 ky BP) in most ice cores from high latitude regions e.g., north-west Canada (Ritchie et al., 1989), central Antarctica (Ciais et al., 1992; Masson et al., 2000) and in some tropical ice cores such as Huascaran in Peru (Thompson et al., 1995). It is also seen during the early Holocene in the Guliya ice core in China (Thompson et al., 1998) but not in two other Chinese cores (Dunde, Thompson et al., 1989; and Dasuopu, to be published). North Africa experienced a greatly expanded monsoon in the early and mid-Holocene, starting at 11 ky BP (Petit-Maire and Guo, 1996), and declining thereafter. In New Zealand the warmest conditions occurred between about 10 to 8 ky BP, when there was a more complete forest cover than at any other time. Glacial activity was at a minimal level in the Southern Alps and speleothem analyses indicate temperatures were about 2°C warmer than present (Salinger and McGlone, 1989; Williams et al., 1999).

By contrast, central Greenland (Dahl-Jensen et al., 1998), and regions downstream of the Laurentide ice sheet, did not warm up until after 8 ky BP (including Europe: COHMAP Members, 1988; eastern North America: Webb et al., 1993). The East Asian monsoon did not commence its expanded phase until after 8 ky BP (Sun and Chen, 1991; Harrison et al., 1996; Yu and Qin, 1997; Ren and Zhang, 1998). A more detailed description of the climate at 6 ky BP as well as of the mechanisms involved is given in Chapter 8. Long-term climate changes during the Holocene are consistent with the effects of orbital forcing, modified by the persistence of the Laurentide ice sheet (which finally disappeared around 6 ky BP).

Seasonal to interannual climate variability may also have varied its character during the Holocene. This is a period for which a variety of palaeo-proxies and archaeological investigations (e.g., Sandweiss et al., 1996; Rodbell et al., 1999) provide evidence for past variations in the strength and frequency of ENSO extremes. A 16-year long time-series of temperature and hydrological balance from a coral dated at 5,370 years BP from the Great Barrier Reef (Gagan et al., 1998) implies that ENSO, or its teleconnections to Australia, were substantially different in the mid-Holocene than today. Mid-Holocene changes in the spectrum of ENSO variability have also been implicated by sedimentary palaeoclimatic records in Australasia (McGlone et al., 1992; Shulmeister and Lees, 1995) and South America (Sandweiss et al., 1996; Rodbell et al., 1999).

To sum up, the Holocene shows both long-term trends (including changes in the nature of ENSO) and millennial time-scale variability although the amplitude of the variability is small compared with that characteristic of Ice Ages. As more detailed information becomes available, the timing of the Holocene maximum warmth is seen to differ across the globe. There appears to be a south to north pattern, with southern latitudes displaying maximum warming a few millennia before the Northern Hemisphere regions. Interestingly, the Holocene appears by far the longest warm “stable” period (as far as seen from the Antarctic climate record) over the last 400 ky, with profound implications for the development of civilisation (Petit et al., 1999).

IPCC Homepage