IPCC Fourth Assessment Report: Climate Change 2007
Climate Change 2007: Working Group I: The Physical Science Basis

9.3.2 What Can be Learned from the Last Glacial Maximum and the Mid-Holocene?

Relatively high-quality global terrestrial climate reconstructions exist for the LGM and the mid-Holocene and as part of the Global Palaeovegetation Mapping (BIOME 6000) project (Prentice and Webb, 1998; Prentice and Jolly, 2000). The Climate: Long-range Investigation, Mapping and Prediction (CLIMAP, 1981) reconstruction of LGM sea surface temperatures has also been improved (Chapter 6). The LGM climate was colder and drier than at present as is indicated by the extensive tundra and steppe vegetation that existed during this period. Most LGM proxy data suggest that the tropical oceans were colder by about 2°C than at present, and that the frontal zones in the SH and NH were shifted equatorward (Kucera et al., 2005), even though large differences are found between temperature estimates from the different proxies in the North Atlantic.

Several new AOGCM simulations of the LGM have been produced since the TAR. These simulations show a global cooling of approximately 3.5°C to 5.2°C when LGM greenhouse gas and ice sheet boundary conditions are specified (Chapter 6), which is within the range (–1.8°C to –6.5°C) of PMIP results from simpler models that were discussed in the TAR (McAvaney et al., 2001). Only one simulation exhibits a very strong response with a cooling of approximately 10°C (Kim et al., 2002). All of these simulations exhibit a strongly damped hydrological cycle relative to that of the modern climate, with less evaporation over the oceans and continental-scale drying over land. Changes in greenhouse gas concentrations may account for about half of the simulated tropical cooling (Shin et al., 2003), and for the production of colder and saltier water found at depth in the Southern Ocean (Liu et al., 2005). Most LGM simulations with coupled models shift the deep-water formation in the North Atlantic southward, but large differences exist between models in the intensity of the Atlantic meridional overturning circulation. Including vegetation changes appears to improve the realism of LGM simulations (Wyputta and McAvaney, 2001). Furthermore, including the physiological effect of the atmospheric CO2 concentration on vegetation has a non-negligible impact (Levis et al., 1999) and is necessary to properly represent changes in global forest (Harrison and Prentice, 2003) and terrestrial carbon storage (e.g., Kaplan et al., 2002; Joos et al., 2004; see also Chapter 6). To summarise, despite large uncertainties, LGM simulations capture the broad features found in palaeoclimate data, and better agreement is obtained with new coupled simulations using more recent models and more complete feedbacks from ocean, sea ice and land surface characteristics such as vegetation and soil moisture (Chapter 6).

Closer to the present, during the mid-Holocene, one of the most noticeable indications of climate change is the northward extension of northern temperate forest (Bigelow et al., 2003), which reflects warmer summers than at present. In the tropics the more vegetated conditions inferred from pollen records in the now dry sub-Saharan regions indicate wetter conditions due to enhanced summer monsoons (see Braconnot et al., 2004 for a review). Simulations of the mid-Holocene with AOGCMs (see Section for forcing) produce an amplification of the mean seasonal cycle of temperature of approximately 0.5°C to 0.7°C. This range is slightly smaller than that obtained using atmosphere-only models in PMIP1 (~0.5°C to ~1.2°C) due to the thermal response of the ocean (Braconnot et al., 2000). Simulated changes in the ocean circulation have strong seasonal features with an amplification of the sea surface temperature (SST) seasonal cycle of 1°C to 2°C in most places within the tropics (Zhao et al., 2005), influencing the Indian and African monsoons. Over West Africa, AOGCM-simulated changes in annual mean precipitation are about 5 to 10% larger than for atmosphere-only simulations, and in better agreement with data reconstructions (Braconnot et al., 2004). Results for the Indian and Southwest Asian monsoon are less consistent between models.

As noted in the TAR (McAvaney et al., 2001), vegetation change during the mid-Holocene likely triggered changes in the hydrological cycle, explaining the wet conditions that prevailed in the Sahel region that were further enhanced by ocean feedbacks (Ganopolski et al., 1998; Braconnot et al., 1999), although soil moisture may have counteracted some of these feedbacks (Levis et al., 2004). Wohlfahrt et al. (2004) show that at middle and high latitudes the vegetation and ocean feedbacks enhanced the warming in spring and autumn by about 0.8°C. However, models have a tendency to overestimate the mid-continental drying in Eurasia, which is further amplified when vegetation feedbacks are included (Wohlfahrt et al., 2004).

A wide range of proxies containing information about ENSO variability during the mid-Holocene is now also available (Section 6.5.3). These data suggest that ENSO variability was weaker than today prior to approximately 5 kyr before present (Moy et al., 2002 and references therein; Tudhope and Collins, 2003). Several studies have attempted to analyse these changes in interannual variability from model simulations. Even though some results are controversial, a consistent picture has emerged for the mid-Holocene, for which simulations produce reduced variability in precipitation over most ocean regions in the tropics (Liu et al., 2000; Braconnot et al., 2004; Zhao et al., 2005). Results obtained with the Cane-Zebiak model suggest that the Bjerknes (1969) feedback mechanism may be a key element of the ENSO response in that model. The increased mid-Holocene solar heating in boreal summer leads to more warming in the western than in the eastern Pacific, which strengthens the trade winds and inhibits the development of ENSO (Clement et al., 2000, 2004). Atmosphere-Ocean General Circulation Models also tend to simulate less intense ENSO events, in qualitative agreement with data, although there are large differences in magnitude and proposed mechanisms, and inconsistent responses of the associated teleconnections (Otto-Bliesner, 1999; Liu et al., 2000; Kitoh and Murakami, 2002; Otto-Bliesner et al., 2003).