Internal variability is an inherent part of the climate system. On sub-orbital time-scales internal variability of the ocean and ocean-atmosphere systems has been used to attribute the Atlantic Multi-decadal Oscillation (e.g. Sutton and Hodson 2005) to multi-century to millennial variability (e.g. Schulz et al. 2007), primarily suggesting an origin from low-frequency variations in the Atlantic Meridional Overturning Circulation.
Some studies have suggested that there may have existed different prevailing atmospheric patterns in the early to mid-Holocene than in recent times, indicating that atmospheric forcing might have influenced the ocean response and vice versa. Rimbu et al. (2004), comparing the alkenone data-set for the Northern Hemisphere of Kim et al. (2004) with climate model experiments, note that the SST anomaly pattern appears to be similar to the tripole pattern of the positive phase of the NAO, suggesting a prevailing situation with a stronger tendency for meridional ocean and atmosphere heat transport in the eastern sector of the high-latitude North Atlantic realm. This pattern of model response to 6 ka orbital forcing is, however, not robust when comparing the coupled AOGCMs used in the PMIP2 experiments. A PMIP2 model-model intercomparison shows that three of the nine models support a positive NAO-like atmospheric circulation in the mean state for the mid-Holocene as compared with the pre-industrial period without significant changes in simulated NAO variability (Gladstone et al. 2005). Some observational evidence comes from lake studies in northern Scandinavia, indicating a generally more maritime climate in the early to mid-Holocene than in the late Holocene (Hammarlund et al. 2002). A general feature of NAO-related responses in northern Europe is a strong increase in winter precipitation in years of positive NAO index, i.e. the pattern indicated by Rimbu et al. (2004). When combining several glacier mass balance records and calculating the winter precipitation contribution to Norwegian glaciers, Nesje (personal communication) (Figure 5.7) finds no strong early to mid-Holocene winter precipitation maximum, thus documenting lack of evidence for a general NAO positive state. Rather, the glacier-based precipitation records follow the pattern of cold intervals punctuating the overall trends and occurring at century to millennial time-scales. This is comparable to the pattern of warm and cold phases seen in many other records.
Figure 5.7 (a) Mean winter (October-April) precipitation in percent of 1961-90 mean (= 100%) calculated from Jostedalsbreen, Folgefonna, Sp0rteggbreen, and Bj0rnbreen (compilation by Atle Nesje, personal communication). (b) Relative percent Neogloboquadrina pachyderma (sin) in MD95-2011 (Andersson et al. 2003; Risebrobakken et al. 2003).
(c) Ice-rafting proxy data (quartz/plagioclase ratio) from cores MD95-2011 and
100 80 60 40 20 0
Average from 4 Norwegian glaciers
Average from 4 Norwegian glaciers t—I—I—I—I—I—I—I—I—I—I—I—I—I—I—I—I—I—I—I—I—I—r
100 80 60 40 20 0
A feature of many Holocene climate records is that variability on century to multi-century time-scales often displays somewhat lower amplitudes between about 8 and 6 ka, and a tendency for an increased amplitude of the variability towards present (see Figures 5.5 and 5.7). In a number of records there is strong variability in the early Holocene with the 8.2 ka event being probably the most pronounced. Since much of this variability and the 8.2 ka event may plausibly be linked with the final demise of the large Northern Hemisphere ice sheets and thus have a dynamical origin from Ice-Age processes, we focus here on the evolution after the 8.2 ka event. In Figure 5.5 it is apparent that the amplitude of multicentury to millennial scale variability is not stationary through the Holocene. Risebrobakken et al. (2003) showed that there is an increase in the amplitude of century to millennial scale variability after the mid-Holocene, and that there is very little persistence in the main frequencies of variability through the Holocene. The noticeable increased amplitude after about 4 ka in the Nordic Seas thermocline temperature (Figure 5.5) and sea-ice proxies (Figure 5.7) indicates increased climate variability affecting winter-time conditions. This pattern is also generally consistent with the onset of neoglaciation in Europe (Nesje et al. 2000) and an increase in Arctic sea-ice cover (Ko^ and Jansen 1994).
We do not have evidence of stronger external forcing throughout the pre-industrial Holocene. It is therefore likely that the nonstationary aspect of the variability indicates that the stronger century to millennial scale variability is excited by changes in boundary conditions, which amplify processes occurring at time-scales of more than 100 years. The amplitude of millennial to sub-millennial scale variability appears potentially linked with sea-ice extent, Northern Hemisphere snow cover, and ocean surface temperature, and indicates that increased sea-ice cover following the reduced summer insolation may have put in place amplification mechanisms leading to stronger ocean temperature variability. One plausible mechanism, drawing on the general trends of orbital forcing through the Holocene, is that the reduction in boreal summer insolation and a less pronounced summertime surface ocean stratification induced sea-ice/snow albedo feedbacks which drove the overturning circulation past a threshold into a more variable mode of operation. In the absence of a clear attribution of this variability to external forcings (e.g. solar, volcanic; e.g. Risebrobakken et al. 2003), it appears most likely that the century to millennial scale variability is primarily caused by the long time-scale internal dynamics of the climate system.
Moros et al. (2006) (Figure 5.7) published records of ice-rafted debris (IRD) in the form of quartz content of marine sediments, presumed to originate from melting sea-ice. The data show a strong increase in the drift-ice occurrence off East Greenland in the latest half of the Holocene. This is coherent with orbital forcing, which led to an increased sea-ice cover in the marginal ice zone. Ko^ and Jansen (1994) found a similar pattern based on sea-ice diatom records, although this study had a much lower temporal resolution. The IRD record from the central Nordic Seas basically follows the foraminiferal temperature data (Figure 5.7), indicating that sea-ice here is related to protrusions of colder waters to the south-east and probably related to changes in atmospheric dynamics. The IRD record from the central Nordic Seas is, in a number of aspects, similar to the IRD data of Bond et al. (2001), although the different methods and temporal resolution do not make a detailed correlation possible. Overall, the data of Bond et al. as well as the IRD data from the Nordic Seas indicate that there is not one single sea-ice response in the region and that the general sea-ice response appears as one following the orbital forcing, but punctuated by colder intervals at millennial to century scales with increased presence of sea-ice and colder thermocline temperatures. As first noted by Risebrobakken et al. (2003), there is not a single persistent variability in the high-latitude North Atlantic through the Holocene. The emergence of higher amplitude millennial scale variability in many records appears to be linked to a threshold reached after the thermal optimum, and caused by decreased summer insolation and seasonality. The winter-time precipitation records are in line with this evidence, indicating that the cold intervals in the ocean records are linked with a colder, less moist winter-time climate over Scandinavia.
Was this article helpful?