Ocean Circulation Changes And Climate Over The Last Glacialinterglacial Cycle

Analysis of 813C and Cd/Ca in benthic foraminifera and 8lsO in planktic forams (reflecting temperature and/or salinity changes in surface waters) have enabled circulation changes to be reconstructed over the last glacial-interglacial cycle in some detail; longer-term changes in 813C are examined by Raymo et al. (1990). These studies indicate that significant changes in the thermohaline circulation of the oceans have occurred. Although production and circulation of NADW was similar to today in the last interglacial (5e), glacial periods were characterized by a reduction (or even cessation) of NADW production, or perhaps a change towards less dense Intermediate Water forming in the central North Atlantic and/or Labrador Sea (Duplessy et al., 1988, 1991; Keigwin et al, 1994; Oppo and Lehman, 1995). The extent to which deepwater formed in the Norwegian Sea in isotope stages 2 or 6 is unclear,

22 Organisms living in upwelling areas where they develop organic tissue in equilibrium with "old water" will have an apparent radiocarbon age several hundred (and in some areas more than two thousand) years older than "modern."

with some studies suggesting none, others indicating intermittent production, or perhaps a shifting source area across the Greenland-Iceland-Norwegian Sea region, in close connection with the sea-ice margin (Veum et al., 1992; Oppo and Lehman, 1995). Antarctic Bottom Water occupied most of the deep basins of the world oceans at those times (see Fig. 6.45) (Duplessy and Shackleton, 1985).

Deepwater changes are clearly related to changes in surface ocean conditions in the North Atlantic; when the surface waters were cold and less saline (as recorded by a high percentage of the cold-water foraminifera Neogloboquadrina pachyderma [sinistral]) deepwaters did not form, at least not in the areas where they form today. However, in the last interglacial when the marine polar front was far to the north, and North Atlantic surface waters were saltier and warmer, NADW filled the deep basins of the Atlantic Ocean (Broecker et ai, 1988b). What then could have brought the interglacial to a close? Of course, orbital variations were slowly bringing about a reduction in summer insolation and an increase in winter insolation (in the northern hemisphere) but other internal factors may have also played an important role. A higher interglacial sea level (+ 6 m) would have led to an increased flux of water through the Arctic Ocean from the North Pacific, bringing more low salinity water into the North Atlantic (Shaffer and Bendtsen, 1994). Higher temperatures may have increased evaporation and precipitation rates at high latitudes, adding additional freshwater to the North Atlantic and its surrounding drainage basins. If all of these factors were enough to lower surface water density, deepwater production (and its attendant changes in compensatory surface inflow) may have been reduced or even eliminated, setting the stage for renewed continental glaciation (Cortijo et al., 1994). Once temperatures began to fall, evaporation rates would also have decreased, allowing salinity levels to remain low. The problem is that all of these factors are intimately related in a positive feedback loop that makes identifying "cause" and "effect" extremely difficult, especially when the resolution of sedimentary records is low and dating uncertainties are relatively high (W. Berger, 1990). This is particularly problematic during times of very rapid change, as occurred at the end of the last glacial period (see Section 6.10.2). Nevertheless, where high resolution records exist, there is evidence that changes in surface water temperatures and deepwater circulation often occurred simultaneously, and often very rapidly even within the coldest intervals (Boyle and Rosener, 1990; Lehman and Keigwin, 1992a; Oppo and Lehman, 1993; Keigwin et al., 1994).

6.10.1 Heinrich Events

One of the most intriguing aspects of the Greenland ice-core records from the last glacial period is the evidence for very rapid changes in 8lsO during the last 75,000 yr (marine isotope stages 2 to 4). There is now convincing evidence that these rapid changes were also recorded in marine sediments from the North Atlantic (Rasmussen et al., 1996). Heinrich (1988) first noted large changes in the percent of lithic materials in the >180 pm fraction of marine sediment cores from the northeastern North Atlantic during the last glacial period (Fig. 6.46). These percentage increases were to some extent related to an increase in the occurrence of

Ice-rafted debris (%) Foraminifers (%)

FIGURE 6.46 Ice-rafted debris (IRD) as a percentage of the total number of forams + ice-rafted particles in a sediment core from the North Atlantic (-47° N, 20° W). Isotopic stage boundaries and volcanic ash layers are shown (Ash I was deposited approximately 10,800 yr B.P. and Ash II -54,000 B.P.).Times of maximum ice-rafted debris are now known as Heinrich events (see Table 6.5) (Heinrich, 1988).

FIGURE 6.46 Ice-rafted debris (IRD) as a percentage of the total number of forams + ice-rafted particles in a sediment core from the North Atlantic (-47° N, 20° W). Isotopic stage boundaries and volcanic ash layers are shown (Ash I was deposited approximately 10,800 yr B.P. and Ash II -54,000 B.P.).Times of maximum ice-rafted debris are now known as Heinrich events (see Table 6.5) (Heinrich, 1988).

N. pachyderma (sin.) which makes up the rest of this size fraction and, as noted earlier, is characteristic of cold polar waters. Further studies showed that these peaks of lithic material (now known as Heinrich events) could be traced over wide areas of the North Atlantic Ocean (Grousset et al., 1993). Six events (Hj to H6) have been identified in numerous sediment cores; younger events were dated by bracketing AMS 14C dates (Table 6.6). Each event appears to have been an episode of very rapidly accumulating ice-rafted detritus (IRD) at times associated with a drop in foram concentration, due to lower productivity and/or increased dissolution (Broecker et al., 1992; Broecker, 1994). A further four events within marine isotope stage 5 have been recognized in two cores from the North Atlantic, as well as a peak of IRD in stage 6. Bond et al. (1993) also identify IRD in sediments of Younger Dryas age from the Labrador Sea, which they characterize as an additional Heinrich event (H0 in Table 6.6) (Keigwin and Jones, 1995).

Heinrich originally described the IRD as mainly angular quartz grains, but cores from farther south and west have a relatively high percentage of distinctive limestone and dolomite in each detrital layer, suggesting a single source for these

TABLE 6.6 Age of Heinrich Events'

TABLE 6.6 Age of Heinrich Events'

11,000 14,300 21,000 27,000 35,500 -52,000 -69,000 -71,000 -76,000 -85,000 -105,000 -133,000

" H0 to H, bracketed by AMS 14C dates on foraminifera. H4 to H6 based on extrapolation of sedimentation rates in upper sections of sediment cores and therefore subject to revision (Bond et al., 1992, 1993). Older events (designated here as "H?" to "Hn") are based on two cores studied by McManus et al. (1994) with ages subject to uncertainties of probably ±5%.

materials. Also, Heinrich layers 1 and 2 increase in thickness westward, in a belt from 43-55° N, towards the Labrador Sea, suggesting that material originated from the Laurentide ice sheet and was dispersed across the Atlantic by icebergs (Fig. 6.47). Furthermore, Nd/Sr isotope ratios and K/Ar dates of -900 Ma on detrital clays point to a source in the Pre-Cambrian shield rocks of northwest Greenland or northeastern Canada (Bond et al., 1992; Grousset et al., 1993; Andrews et al., 1994). Heinrich events 3 and 6 differ from the others in that the IRD distribution is largely confined to the western Atlantic, perhaps because these events occurred when the Laurentide ice sheet was smaller (at the start of Stages 2 and 4, respectively) so the delivery of icebergs and entrained debris would have been more limited (Gwiazda et al., 1996b).

It is significant that several of the Heinrich events occurred at the end of prolonged cooling episodes, as recorded by increased percentages of N. pachyderma leading up to the event (Fig. 6.48). Furthermore, these longer-term cooling cycles can be correlated with similar variations in 5180 in the GRIP Summit ice core from Greenland, indicating direct links between the ocean and atmospheric systems, which each record primarily represents, and changes in ice sheet dynamics, recorded by IRD in the Heinrich layers. Following each Heinrich event, there is an abrupt shift to warmer conditions, which (from ice core evidence) apparently took

FIGURE 6.47 Thickness (cm) of Heinrich layers in North Atlantic sediments, based on whole-core magnetic susceptibility data. Cross-hatching denotes major ice sheets; HSt = Hudson Strait, thought to have been the major ice stream supplying material to the North Atlantic from the Laurentide ice sheet. Upper figure: Heinrich Event I (~ 14.3 ka l4C yr B.R; lower figure: Heinrich Event 2 (~21 ka NC yr B.R) (Dowdeswell, et ai, 1995).

FIGURE 6.47 Thickness (cm) of Heinrich layers in North Atlantic sediments, based on whole-core magnetic susceptibility data. Cross-hatching denotes major ice sheets; HSt = Hudson Strait, thought to have been the major ice stream supplying material to the North Atlantic from the Laurentide ice sheet. Upper figure: Heinrich Event I (~ 14.3 ka l4C yr B.R; lower figure: Heinrich Event 2 (~21 ka NC yr B.R) (Dowdeswell, et ai, 1995).

place over just a few decades. The cause of these large-scale changes in the ocean-atmosphere-ice systems is not known. Clearly, they took place on timescales far shorter than forcing related to orbital variations. Indeed, Bond and Lotti (1995) find evidence in high resolution marine sediments for even more episodes of ice-rafting between the major Heinrich events — 13 events from 38,000 to 10,000 B.P. — and these too seem to correspond to low 8lsO in the GRIP ice core (Fig. 6.49). As the detrital material clearly originates from ice-rafting, there must have been quasi-periodic increases in iceberg calving rates into the North Atlantic. High-resolution records show that not only detrital carbonate but also volcanic glass and hematite-

FIGURE 6.48 Comparison of 8lsO, and percentages of the planktonic cold water foram Neogloboquadrtna pachyderma (sinistral) in two ocean sediment cores from the North Atlantic, with the 8lsO record from the GRIP ice core (Summit, Greenland). Dashed lines indicate common features used to match the records, but the timescale shown is based on radiocarbon dates for the past -35,500 yr and estimated ages of Heinrich events (shown as HI to H6) derived in other studies.The ice-core record was then forced to fit the assumed sediment chronology.The records are quite similar, indicating that there were strong links between the ocean-atmosphere-cryosphere system in the North Atlantic during the time period represented here.The lower schematic diagram shows how clusters of millennium-length cycles of 8I80 in the ice-core record (sometimes referred to as Dansgaard-Oeschger cycles) seem to form long-term cooling cycles, which terminate abruptly at irregular intervals; a similar pattern can be seen in the cold-water foram percentage data, especially in core VM23-81 (from -55° N) (Bond etal., 1993).

FIGURE 6.48 Comparison of 8lsO, and percentages of the planktonic cold water foram Neogloboquadrtna pachyderma (sinistral) in two ocean sediment cores from the North Atlantic, with the 8lsO record from the GRIP ice core (Summit, Greenland). Dashed lines indicate common features used to match the records, but the timescale shown is based on radiocarbon dates for the past -35,500 yr and estimated ages of Heinrich events (shown as HI to H6) derived in other studies.The ice-core record was then forced to fit the assumed sediment chronology.The records are quite similar, indicating that there were strong links between the ocean-atmosphere-cryosphere system in the North Atlantic during the time period represented here.The lower schematic diagram shows how clusters of millennium-length cycles of 8I80 in the ice-core record (sometimes referred to as Dansgaard-Oeschger cycles) seem to form long-term cooling cycles, which terminate abruptly at irregular intervals; a similar pattern can be seen in the cold-water foram percentage data, especially in core VM23-81 (from -55° N) (Bond etal., 1993).

Summit ice core(GRIP)

VM23-081

Summit ice core(GRIP)

VM23-081

Oxygen Isotope Grip Last Interglacial

30 I

103 lithic grains/g

0 50 100

FIGURE 6.49 Oxygen isotope record from the GRIP ice core (Summit, Greenland) compared with numbers of lithic grains per gram, and percentage of the planktonic cold water foram Neogloboquadrina pachyderma (sinistral) in sediments from marine core VM23-81 (-55° N in the central North Atlantic). N. pachyderma (s.) is indicative of cold waters (comprising 95% of the fauna in waters with a summer temperature of <5 °C). Major Heinrich events are indicated (HI to H4); sublayers are also shown (a to h).There is a strong correlation between the peaks in lithic concentration and 8lsO in the ice-core record, but less so with the N. pachyderma SST record. Two alternative chronologies for the ice-core record are given; one is based on layer counting to -41,000 (calendar) yr B.P.,the other based on a flow model.The sediment record is largely based on HC dating (Bond and Lotti, 1995).

30 I

103 lithic grains/g

0 50 100

FIGURE 6.49 Oxygen isotope record from the GRIP ice core (Summit, Greenland) compared with numbers of lithic grains per gram, and percentage of the planktonic cold water foram Neogloboquadrina pachyderma (sinistral) in sediments from marine core VM23-81 (-55° N in the central North Atlantic). N. pachyderma (s.) is indicative of cold waters (comprising 95% of the fauna in waters with a summer temperature of <5 °C). Major Heinrich events are indicated (HI to H4); sublayers are also shown (a to h).There is a strong correlation between the peaks in lithic concentration and 8lsO in the ice-core record, but less so with the N. pachyderma SST record. Two alternative chronologies for the ice-core record are given; one is based on layer counting to -41,000 (calendar) yr B.P.,the other based on a flow model.The sediment record is largely based on HC dating (Bond and Lotti, 1995).

coated grains are characteristic of many layers, suggesting more widespread sources of discharge (including Iceland) for the IRD than just the Laurentide ice sheet. Sediment cores from off Norway also show strong correlations between episodes of IRD and 8180 in Greenland ice cores, again with cold periods (low 8lsO in the ice) associated with increased ice-rafting (Fronval et al., 1995). Thus, there appear to have been numerous episodes when there was a massive draw-down of ice in one or more circum-Atlantic ice sheets, resulting in "armadas of icebergs" and entrained basal de-

bris entering the cold waters of the North Atlantic (Broecker, 1994). Surprisingly, these episodes were not brought about by warmer conditions, but occurred when conditions were already cold (Madureira et al., 1997), and much of the North Atlantic was covered by cold polar waters. One possible explanation, suggested by modeling experiments, is that ice sheets may grow to a point where they develop instabilities (basal ice melting), which cause rapid discharge (surges) into marine em-bayments via ice streams and ice shelves (MacAyeal, 1993; Alley and MacAyeal, 1994). Cold, slow-moving ice can freeze rock debris into its base, but during times of destabilization channelized ice streams develop where frictional heating allows ice to slide over a relatively warm and wet bed. Debris-laden icebergs are generated when the surging ice streams enter the marine environment, and this continues until the ice sheet becomes stable once again. This "binge-purge" model could account for the quasi-periodic character of the observed record, though just what climatic conditio^ are needed to bring about destabilization of the ice sheets is not clear. Apparently, there is a fairly delicate balance between conditions that: (a) maintain an ice sheet in a quasi-equilibrium state; (b) cause a periodic collapse, but then allow recovery; and (c) cause irreversible collapse (i.e., complete déglaciation). The interplay of ocean, atmosphere, and ice with the added complexities of eustatic sea-level change and glacio-isostatic adjustments as ice loads changed, allows for many possible scenarios of how the observed changes may have been brought about and provides a fruitful area for future research. One possible consequence of the increased calving rates was that the North Atlantic became flooded with a low salinity melt-water "lid"; this may have acted as a shallow mixed layer, with low thermal inertia, which could have warmed up fairly rapidly, leading to higher air temperatures and the observed 8lsO increase seen in GRIP ice cores (Fairbanks, 1989). This would have been a short-lived episode that came to an end as the meltwater layer became mixed with the deeper ocean. Alternatively, the iceberg flux associated with Heinrich events would presumably have been followed by a period with very little iceberg discharge and a rapid reduction in freshwater flux to the North Atlantic, allowing salinity to increase, and a return to the conveyor "on" mode, with increased advection of warmer water and air masses into the Greenland area (Paillard and Labeyrie, 1994).

One additional complexity to be considered in explaining the abrupt changes seen in marine sediments and ice cores during the last glacial period is that rapid 8lsO shifts in Greenland ice cores are associated with pronounced changes in atmospheric methane (cf. Figs. 5.34 and 6.49). This implicates (or necessitates an explanation involving) tropical and/or high latitude wetlands such that the warmer, high 8lsO episodes following Heinrich events (and other minor IRD events) are somehow associated with periods of CH4 release. Possibly this in turn may have provided some positive feedback due to an enhanced greenhouse effect. Interestingly, there is some evidence that Heinrich events themselves may be a direct consequence of circulation changes in low latitudes brought about by orbital forcing (Mclntyre and Molfino, 1996). Increases in the abundance of the coccolith Florispbaera profunda in the equatorial Atlantic coincide with Heinrich events in the North Atlantic, with a mean period of 8.4 ka. Variations in precession at this frequency cause changes in the strength of the zonal component of the tropical easterlies; when the easterlies diminish in strength (which favors an increase in the abundance of F. profunda) the "reservoir" of warm water within the Caribbean/Gulf of Mexico is no longer restricted and it rapidly "drains out" of these two basins, becoming entrained in the western boundary currents of the North Atlantic. In this way, warm salty waters would have periodically entered the subpolar Atlantic, causing rapid melting of ice sheets and the initiation of Heinrich events. This may help to explain the quasi-periodic nature of Heinrich events that has frequently been noted (Heinrich, 1988; Bond and Lotti, 1995).

6.10.2 Environmental Changes at the End of the Last Glaciation

It is now well-documented that numerous rapid changes in ocean circulation and atmospheric conditions took place throughout the last glacial period. Studies of well-dated high resolution sediments and uplifted coral reefs provide particularly valuable insights into such events at the very end of the last glaciation (Termination 1).

The S180 in benthic forams provides a broad-scale perspective on continental ice volume changes since the last glacial maximum (LGM) around 18,000 (14C yrs) B.P. and many records demonstrate that déglaciation took place in two stages (Terminations la and lb) (Duplessy et al., 1986; Jensen and Veum, 1990). However, because of low sedimentation rates and bioturbation, the details of events during déglaciation are difficult to decipher accurately from the benthic record (Ruddiman, 1987). Fairbanks (1989) was able to address this situation by obtaining a detailed sea-level record from drowned coral reefs off the coast of Barbados (Fig. 6.50). His studies reveal that the maximum sea level lowering due to ice build-up on the continents was 121 ± 5 m at -18,000 (14C yr) B.P. As déglaciation set in, sea level slowly increased by -20 m over the next 5000 yr, followed by a very rapid rise in sea level, centered on 12,000 (14C yr) B.P. At that time (termed Meltwater Pulse IA, mwp-IA) the discharge of water from the continents to the world ocean reached - 14,000 km3 a1 and sea level rose a further -24 m in <1000 yr (see Fig. 6.50). This was followed by a slower rate of sea-level rise, and then a second major meltwater pulse (mwp-IB) centered on 9500 (14C yr) B.P., when sea level rose an additional 28 m in -1500 yr. Thus, more than two thirds of the LGM continental ice had melted by the beginning of the Holocene,23 and almost all of this was discharged to the world ocean (via the North Atlantic) in two short episodes, lasting a total of -2500 yr. Fairbanks argues that the periods of most rapid meltwater flux resulted in a reduction of surface water salinity right across the North Atlantic, as seen in planktic forams from off the coast of Portugal where abrupt decreases in 8180 (Terminations la and lb) appear to be directly linked to the coral reef record of ice volume discharge and sea-level rise (Duplessy et al., 1986; Bard et al., 1989; Fairbanks, 1989).

One of the problems of understanding the exact sequence of events in Late Glacial time is the chronological uncertainty posed by 14C production rate changes at the very time when rapid climatic changes were taking place. Indeed the "14C

23 If the Holocene is defined in calendar years, the 230Th/234U calibration of the 14C timescale (Bard et al., 1990), applied to Fairbanks (1989) sea-level record indicates that sea level had risen 85 m (of the 120 m LGM depression) by the start of the Holocene sensu stricto (see Fig. 6.50).

I ' I i I i I ' I ' M I ' I 1 I ' I i I M ' I ' I ' I

I ' I i I i I ' I ' M I ' I 1 I ' I i I M ' I ' I ' I

0 0

Post a comment