From: Wanner, H., Beer, J. Butikofer, J., Crowley, T. J., Cubaschd, U., Fluckiger, J., Goosse, H., Grosjean, M., Joos, F., Kaplan, J.O., Kuttel, M., Muller, S., Prentice C., Solomina, O., Stocker, T. F., Tarasov P., Wagner M., Widmann, M. (2008). "Mid- to late Holocene climate change - a comprehensive review, Quaternary Science Review 27(19-20): 1791-1828.
Historical descriptions of glacier variations, dates of moraines and lake sediment properties in the glacial areas are major types of records providing information of former glacier fluctuations. The historical descriptions are limited in space and time, while geological reconstructions are generally suffering from the low accuracy of the dating. One of the most serious limitations is the uncertainty of the relationship between the timing in moraine deposition and the organic material providing the actual date of advance, which in most cases allows the estimation of the maximum or minimum date of advance only. The accuracy of the methods used for moraine dating ranges from several decades to centuries and is, therefore, rather low for high resolution reconstructions. The cross-dating of trees damaged or killed by glacier advances allows an improvement of the accuracy up to annual resolution, but these occasions are limited to regions where the upper tree limit reaches the Holocene moraines, such as the Alps, Patagonia etc. Glacio-fluvial sediment properties calibrated against other measures of glacier activity and size provide continuous records of past variability and individual advances or retreats of glaciers (Dahl et al. 2003). This method is intensively used in Scandinavia and provided very detailed reconstructions of equilibrium line altitude variations over the whole Holocene for both, glacier advances and retreats. Until very recently, little has been known about the receding glaciers during the earlier warmer climatic phases. The uncertainty of the glacier sizes during the contracted stages still remains. However, due to the modern shrinkage of the glaciers, new reconstructions of past variability based on the analysis of tree rings and organic material buried by the former glacier advances and released in the glacier forefield recently became possible (Holzhauser et al. 2005, Hormes et al. 2001, Koch et al. 2005, Jorin et al. 2006, Grosjean et al. 2007). Several aspects of glacier dynamics should be kept in mind when this data is used for climatic reconstructions. The response time of an advancing or retreating glacier front to a climatic signal differs for different glaciers depending on type, size and morphology of the glacier. For typical alpine valley glaciers the response time is estimated as 10 to 50 years (Oerlemans 1998). The best climatic indicators are non-surging mountain glaciers of moderate size and simple shape, which are located on-land (without floating tongue), and have a regular accumulation (rather than provided by avalanches or snow re-distributed by wind).
Glaciers depend from both temperature and precipitation, so the problem often arises to discriminate these two parameters. In most cases summer temperature plays the major role in controlling the mass balance and, hence, glacier size variations in the high and mid latitudes (Oerlemans 2005, Steiner et al. 2006). In the tropics however the glaciers are strongly dependent on precipitation (Grosjean et al. 1998, Kaser and Osmaston 2002). Recent improvement in glacial chronologies, coupled with high resolution multiproxy comparisons and climatic reconstructions, in several cases allowed to identify the climatic signal and attribute several advances to precipitation even in temperate regions (Luckman 2000, Wanner et al. 2000, Luckman and Villalba 2001, Nesje and Dahl 2003). Despite of many challenges in the dating as well as the interpretation of glacier variations, glaciers have been successfully used as indicators of past millennial to decadal-scale climate changes (Denton and Karlen 1973, Mayewski et al. 2004, Oerlemans, 2005). Due to the the wide range of advance and retreat dates however, this data can also be misleading and has to be treated with extreme caution. Figure 1 shows several well constrained glacier histories based on both, continuous (lake sediments) and discontinuous (moraine based) chronologies.
The glaciers from the mid to high latitudes in the NH, namely the Alps (Jorin et al. 2006), Scandinavia (Bakke et al. 2007) and the Canadian Cordillera (Luckman 2000, Koch and Clague 2007) were reduced in the EH to MH period, but increased in sizes and experienced numerous Neoglacial advances after ca 6 ka BP, reaching their maximum in the LIA (Fig. 1). This major trend in the Holocene glacier variations in the NH seems to be related to the gradually decreasing summer insolation driven by orbital forcing. Koch and Clague (2006) suggest that a gradual reduction of the glacier sizes through the Holocene in the SH is in agreement with the increase of the austral summer insolation, opposite to the NH. However, there are evidences that in several regions of the SH many glaciers were contracted during the EH to MH, including those in Patagonia (Glasser and Harrison 2004, Kilian et al. 2007), Antarctica (see Grove 2004, and references therein) and the tropical Andes (Abbott et al. 2003). Thus the millennia-long trend in some regions of the SH could probably be masked by the centennial to multidecadal regional climatic variations. To explain the long-term trend of glacier variations in the tropical Andes (small or absent glaciers in the EH to MH and large in the LH) Abbot et al. (2000 and 2003) suggested the following climatic mechanism. The lower summer insolation (January), driven by orbital forcing in the EH to MH, resulted in decreased summer precipitation and arid climate. The high insolation in winter contributed to more intense melting of glaciers. As a result the glaciers were nonexistent in the catchments lower than 5500 m until the LH (around 2400 cal yr BP; Abbot et al. 1997). The gradual re-appearance of the glaciers from north to south is explained by the onset of wetter conditions related to orbitally driven summer insolation increase during the LH, which resulted in a progressively southward shift of the location and strengthening of wet-season convection.
Shulmeister (1999) explained the lack of moraines between 8-5 kyr BP in New Zealand and their appearance afterwards by the changes in westerly circulation due to changes in seasonality as predicted by the precessional cycle. Markgraf (1993) suggested the same mechanism in the sub-tropics of South America between 30 and 34° S where the westerly flow also increased in the MH. Recent modeling however does not support the assumption of a strong link between mean SH westerlies and precipitation over south-eastern Patagonia (Wagner et al. 2007). Porter (2000), analyzing the dates of Holocene moraines in the Andes, west Antarctica and New Zealand, concluded that the first Neoglacial advances culminated there between ca. 5400 and 4900 cal yr BP, i.e. at the time close to the beginning of the Neoglacial in the NH (Grove 2004). However, the austral summer insolation having a different trend cannot be the reason for a cooling and subsequent Neoglacial glacier advances in the whole SH. Hodell et al. (2001) suggested that the rapid global cooling between 5400 and 4900 cal yr BP recorded in the Taylor Dome Ice Core (Antarctica), increased IRD in the North Atlantic, and coinciding with the end of the African humid period, can be explained by a nonlinear response of the Earth’s climatic system to gradual changes in NH insolation. It was suggested that the process may have been initiated in the tropics and subtropics (de Menocal et al. 2000) and then exported to high latitudes in both hemispheres. This very attractive explanation however conflicts with the evidence that the first Neoglacial moraines in different parts of the World span a broad interval up to several millennia and the oldest Neoglacial advances are actually almost 1000 years older than the cooling around 5000 cal yr BP. Namely the advances in the Alps and possibly in the Pyrenees, in Scandinavia, the Cascades and the Canadian west coast ranges occurred shortly after 6000 cal yr BP (Grove 2004). This discrepancy can also be explained by the poor accuracy of the dates, or a different regional forcing for the earliest Neoglacial advances.
Fig. 1. Timing and relative scale of eight timeseries of glacier advances and retreats during the last 6000 years, covering important glacial regions of the globe. Most series, except for the two Scandinavian curves, are discontinuous because they are at based on geomorphic evidence. For easier interregional comparison all series are presented as curves. The curve for the Alps is the best constrained by documentary and tree-ring data, especially for the last 3500 years (Holzhauser et al., 2005). Except for the Alps and Scandinavia, the precise periods of the glacier retreats are not known and are shown on the picture arbitrarily. The coloured dots mark possible simultaneous Neoglacial advances. The brown shaded areas indicate indirect evidence (wood above the modern tree limit, buried soils etc.). Franz Josef Land: Lubinsky et al., 1999, calibrated. Spitsbergen: curve from Svendsen and Mangerud (1997), corrected with Humlum et al. (2005). Northern Scandinavia: Nesje et al. (2005), Bakke et al. (2005), IPCC (2007). Southern Scandinavia: Matthews et al. (2000, 2005), Lie et al. (2004), IPCC (2007). Alps: Holzhauser et al. (2005, JoЁ rin et al. (2006). Brooks Range: Ellis and Calkin (1984), calibrated.Western Cordillera–North America: Koch and Clague (2006). Western Cordillera–South America: Koch and Clague (2006).