Paleoclimatic Reconstruction From Ice Cores

Ice cores have revolutionized our understanding of Quaternary paleoclimatology by providing high resolution records of many different parameters, recorded simultaneously at each location. Here, we highlight the main results from the northern and southern hemispheres and show how these records are related to each other, and to changes in forcing.

5.4.1 Ice-core Records from Antarctica

A number of long ice-core records are available from Antarctica but the "crown jewel" of Antarctic ice cores is the record from Vostok on the East Antarctic ice plateau (78°28' S, 106°48' E, 3488 m above sea level). This is an important record, not only because it spans a very long interval of time (3350 m, -426 ka) but because it has been recovered from an area where the ice is extremely thick (>3.5 km) and complications due to ice flow and disturbance at the bed are minimal. Furthermore, the relationship between isotopic fractionation and temperature is clear in this region, making a climatic interpretation of the isotopic record fairly straightforward. Thus, Vostok provides the longest well-resolved ice-core record on earth and a yardstick for comparison with other paleoclimatic records (Petit et al., 1997).

Figure 5.17 shows the SlsO record from Vostok over the uppermost 2083 m of the core (Lorius, et al., 1985). Stages A to H designate the major features of the record, with stage A being the Holocene and Stage G being the last interglacial period. Four major cold periods (B, D, F, and H), each with 8lsO of around -62%o, are clearly recognizable. Establishing a precise chronology for this record is a fundamental problem (as with all ice cores) and requires several assumptions, including the original source of the snow (the local ice flow regime) and the accumulation rate. At Vostok sub-ice topographic effects are minor and the most significant factor is the change in accumulation rate over time. Today, accumulation is very low (-2.2 g cm-2 a-1, compared to >50 g at Dye-3 in Greenland) but it was probably even lower during glacial times, as precipitation in Antarctica is closely related to temperature. This relationship has been used to assess precipitation changes by assuming that precipitation is a function of the ratio of the derivative of the saturation vapor pressure (s.v.p.) of water at time Z to the same parameter today. As the slope of the s.v.p. increases exponentially with temperature, so precipitation will

What The Marine Isotope Stages

FIGURE 5.17 8I80 vs depth in the Vostok ice core; climatic stages and their temporal limits are indicated. These are not directly comparable with the marine isotope stages (MIS) 1-6, though stages A-D approximate MIS 1-4, and H approximates MIS 6.Thick line is from continuous sampling, thin line from less detailed analysis (Lorius et a!., 1985). Stages A and G are interglacials.

FIGURE 5.17 8I80 vs depth in the Vostok ice core; climatic stages and their temporal limits are indicated. These are not directly comparable with the marine isotope stages (MIS) 1-6, though stages A-D approximate MIS 1-4, and H approximates MIS 6.Thick line is from continuous sampling, thin line from less detailed analysis (Lorius et a!., 1985). Stages A and G are interglacials.

change as a non-linear function of temperature. Temperature changes are in turn estimated from 8lsO (or 8D) using empirical relationships observed in studies of contemporary snowfall (Jouzel et al., 1983). Using this approach, precipitation rates at Vostok are estimated to have been 50-55% of modern values during glacial times, and, with this estimate, the chronology shown in Fig. 5.17 was established. It should be noted that even small differences in estimates of the "modern" accumulation rate at Vostok are amplified at depth; for example, a 10% difference at the surface produces an uncertainty of 10,000 yr in the chronology at 2000 m. Nevertheless, the proposed time-scale is supported by 10Be data from the same core. The 10Be is a cosmogenic isotope produced by cosmic-ray bombardment of the upper atmosphere. Assuming a constant production rate, any changes in 10Be concentration in snowfall at Vostok would be due to changes in the accumulation rate (Yiou et al., 1985). On this basis, precipitation at the last glacial maximum was -50% of modern values (i.e., only -1.1 g cm"2 a"1). Both of these approaches to estimating paleo-precipitation rates yield surprisingly similar results (Fig. 5.18). Furthermore, using the s.v.p./precipitation rate relationship as the basis for calculating independent chronologies at Vostok and Dome C, a 10Be "spike" is found to coincide (almost) in both cores at around 35,000 yr B.P. (a 3% correction is required in the Dome C s.v.p.-derived chronology), providing confidence that this approach to dating the record has validity (Raisbeck et al., 1987). Additional support comes

Normalised Precipitation

FIGURE 5.18 Normalized precipitation at Vostok, Antarctica with respect to the Holocene mean value (which is I). Upper record is based on the assumption that l0Be concentration in snowfall is a function only of changing accumulation rate (thick line not taking l0Be peaks at -35,000 and 60,000 B.P. into account). Lower record is based on the saturation vapor pressure (s.v.p)-temperature relationship, using 8D to estimate changes in temperature, and assuming precipitation is directly related to s.v.p. (Jouzel et al., 1989a).

FIGURE 5.18 Normalized precipitation at Vostok, Antarctica with respect to the Holocene mean value (which is I). Upper record is based on the assumption that l0Be concentration in snowfall is a function only of changing accumulation rate (thick line not taking l0Be peaks at -35,000 and 60,000 B.P. into account). Lower record is based on the saturation vapor pressure (s.v.p)-temperature relationship, using 8D to estimate changes in temperature, and assuming precipitation is directly related to s.v.p. (Jouzel et al., 1989a).

from the Byrd ice core where accumulation changes can be estimated down-core by continuous acidity measurements (which enable seasonal cycles to be observed) (Jouzel et al., 1989a). These show that accumulation during glacial time was 50% of Holocene levels (as at Dome C and Vostok) and with this taken into account, a 10Be spike is also observed at -35,000 yr B.P. in the Byrd record (Fig. 5.19) (Beer

FIGURE 5.19 The l0Be concentration "spike" observed in Vostok and Byrd ice cores at -35,000 B.P., assuming precipitation was -50% of Holocene levels at LGM (Raisbeck et al., 1992, and Beer et al., 1992).

FIGURE 5.19 The l0Be concentration "spike" observed in Vostok and Byrd ice cores at -35,000 B.P., assuming precipitation was -50% of Holocene levels at LGM (Raisbeck et al., 1992, and Beer et al., 1992).

et al., 1992). Hence through a series of procedures, each based on a somewhat arguable premise, there emerges strong support for the notion that precipitation amount (at least in East Antarctica) is directly related to temperature (as represented by isotopic changes) and that precipitation was much lower in glacial times than in warmer periods. This provides considerable confidence that the chronology shown in Figs. 5.17 and 5.18 is likely to be approximately correct; Lorius et al. (1988) suggest an uncertainty of 10,000-15,000 yr at around 160,000 yr in this chronology.

More recent studies of the Vostok record, including an extension of the record to >400,000 yr B.P., have focused on continuous measurement of 8D (82H) rather than 8lsO (Jouzel et al., 1987b, 1989a, 1989b, 1993a). The 8D changes by 6%o per °C (at the ice surface) in East Antarctica, according to empirical observations and model-derived estimates (Jouzel et al., 1983; Jouzel and Merlivat, 1984). This was discussed in more detail in Section 5.2.3. After correcting for higher 8D levels in water vapor during glacial times (because of isotopic enrichment of the ocean by the heavier isotope of hydrogen) the 8D change from the last glacial to the Holocene represents an increase in surface temperature of ~9°C at Vostok (Jouzel, et al., 1987a). This compares with an independent estimate based on ice crystal growth rate changes with depth, of -11 °C (Petit et al., 1987) (though this approach is disputed by Alley et al., 1988). The isotopic estimates assume no change in ice sheet thickness, though the lower accumulation rates of the last glacial period suggest that the ice sheet elevation may have been lower during stage B than Stage A (Jouzel et al., 1989a). This would make the glacial-Holocene temperature estimate from 8lsO or 8D a minimum estimate.

The 8D values for the last interglacial (Stage G) indicate a period warmer than the Holocene by -2 °C; interstadial Stages C and E were 4-6 °C warmer than the glacial maximum (Stage B). Of particular interest is the "two-step" change in temperature during the last deglaciation, when rapidly rising temperatures were interrupted by a cooling episode lasting -1500 yr. It is estimated that surface temperatures fell by -3-4 °C at Vostok during this "Antarctic Cold Reversal" (based on a peak-to-peak change in 8D of 20%o, for ~25-yr means; Fig. 5.20) (Jouzel et al., 1992; Mayewski et al., 1996). Detailed analysis of the ice from Dome C indicate this cold episode lasted from -13,500 to 11,700 (calendar) yr B.P. and seems to be related in some way to the Younger Dryas oscillation seen in many records from around the world (Wright, 1989; W. Berger, 1990; Peteet, 1992). Unlike the records from Greenland ice cores, this "reversal" is not associated with an increase in continental dust, or with a drop in CH4 or C02 levels; C02 levels appear to have leveled off at this time, and CH4 levels declined slightly later in the cold episode, possibly reflecting tropical aridification and/or cooling (or even freezing) of high latitude peatland (Jouzel et al., 1992). However, levels of chloride at Taylor Dome during this cold episode indicate an increase in the flux of marine salts due to higher wind speeds at that time (Mayewski et al., 1996). Temperatures at Vostok and other locations gradually declined during the Holocene, by -1 °C at the surface (Ciais et al., 1992) a pattern also seen in the Arctic (Koerner and Fisher, 1990; Bradley, 1990).

T I I | I T1-! | I I I | I I r I | -38 12 13 14 15 16

FIGURE 5.20 The 8lsO record (and interpreted changes in atmospheric temperature, above the surface inversion) at four Antarctic sites during the last déglaciation, showing a hiatus or reversal in the 8lsO increase, from ~ 13,500 to ~l 1,700 (calendar) yr B.P. Note that here the absolute chronology of each core is not well known; they are "matched" based on an optimum fit, relative to the Dome C record (Jouzel et al., 1992). Recent work has fixed the Dome C chronology with respect to GISP2 (Mayewski et ai, 1996).

T I I | I T1-! | I I I | I I r I | -38 12 13 14 15 16

FIGURE 5.20 The 8lsO record (and interpreted changes in atmospheric temperature, above the surface inversion) at four Antarctic sites during the last déglaciation, showing a hiatus or reversal in the 8lsO increase, from ~ 13,500 to ~l 1,700 (calendar) yr B.P. Note that here the absolute chronology of each core is not well known; they are "matched" based on an optimum fit, relative to the Dome C record (Jouzel et al., 1992). Recent work has fixed the Dome C chronology with respect to GISP2 (Mayewski et ai, 1996).

Further drilling at Vostok has yielded additional ice down to 3350 m. Comparison with the SPECMAP marine isotope record (see Section 6.3.3) strongly suggests that the ice core record extends to ~426,000 and thus spans the last four interglacial-glacial cycles (Fig. 5.21). Of particular note is the long cold episode from -180-140 ka B.P. when 8D values remained at levels comparable to the Last Glacial Maximum (Petit et al., 1997).

5.4.2 Ice-core Records from Greenland

Four ice cores to bedrock have now been recovered from the Greenland ice sheet: from Camp Century, Dye-3, and two from Summit, the so-called GISP2 (Greenland Ice Sheet Project 2) and GRIP (Greenland Ice Core Project) cores. Three other long

Age (thousands of years)

0 100 200 300 400

FIGURE 5.21 Deuterium record from Vostok, Antarctica, plotted with the SPECMAP 8lsO record of continental ice volume changes (above) (Petit et al., 1997).

FIGURE 5.21 Deuterium record from Vostok, Antarctica, plotted with the SPECMAP 8lsO record of continental ice volume changes (above) (Petit et al., 1997).

records have been obtained from the ice sheet margin (Reeh et al., 1987, 1991, 1993; Johnsen and Dansgaard, 1992). The GISP2 and GRIP cores, in particular, have provided an enormous amount of information about the climatic history of Greenland and of processes that must have operated over a large area of the North Atlantic region, with effects of hemispheric or even global significance. Here, we discuss the long paleoclimatic series from these sites and examine their relationship to other records in the area.

Figure 5.22 shows the 8lsO record from the GRIP ice cores (Dansgaard et al., 1993). The GISP2 and GRIP records are highly correlated down to -2750 m (estimated in the GRIP core to be -103,000 yr ago by means of a flow model) (Grootes et al., 1993). A number of important characteristics of the 8lsO series can be clearly seen. First, the Holocene record was a period of relative stability with a mean 8180 value of -34.9%o at GRIP and -34.7%o at GISP2. Fluctuations are small — on the order of ±l-2%o — and show little correlation in detail between sites, probably due to local differences in accumulation and wind drifting of snow. A slight decline in 8lsO over the course of the Holocene is apparent in both records. Around 8250 yr ago (calendar years) a pronounced episode of low 8lsO values is observed (Fig. 5.22) and this has been seen at several other sites, including ice cores from the Canadian Arctic (Fisher et al., 1995); it also corresponds to an abrupt drop in atmospheric methane levels (Blunier et al., 1995; Alley et al., 1997c).

Depth

1000

1500

-25 -30

^m ,7"

tT~

4-

-40

5

■50

-60

6

■70

■80

3*-

7-

V"

~ 1 Balling IS number

1500 Depth

Denekamp

10 Ii

12 H engelo

-21 Odderade

_T. Brarup

Eem 150 Saale _

Holstein

3000

FIGURE 5.22 The GRIP 8lsO record from Summit, Greenland, plotted linearly with respect to depth. Section A (left) is the Holocene section, showing only minor changes; section B (right) shows the preceding 250 kyr record at the same 8lsO scale. Note the very large and rapid oscillations throughout the pre-Holocene record. Proposed interstadial isotope stages (IS) 1-24 are indicated, together with comparable European pollen stages. Dating was by annual layer counting to 14.5 kyr B.P. and beyond that by an ice-flow model (Dansgaard etai, 1993).

All Greenland ice cores show that dramatically different climatic conditions prevailed in the late Pleistocene, compared to the last 10,000 yr (Dansgaard et al., 1984; Johnsen et al., 1992; Grootes et al., 1993). In contrast to the relative stability of Holocene climate, the preceding -100,000 yr were characterized by rapid changes between two (or more) modes. Dansgaard et al. (1993) recognize 24 interstadial episodes between 12,000 and 110,000 yr B.P. when isotopic values were as high as -37%o at the GRIP site, separated by stadials, with values dropping precipi-

tously to as low as -42 %o (see Fig. 5.22). These abrupt changes can be correlated between cores as far apart as Dye-3 (southern Greenland), Camp Century (northwest Greenland), and Renland (east-central Greenland) so, whatever their cause, the changes were geographically extensive (Fig. 5.23) (Johnsen and Dansgaard, 1992; Johnsen et al., 1992). Indeed, they are well correlated to changes seen in North Atlantic marine sediments (Bond et al., 1992, 1993) that represent large-scale shifts in water masses in that region. This is discussed in more detail in Section 6.10.

Figure 5.22 shows that the changes from low to high 8lsO were rapid, followed by a slower decline to low values once again (Dansgaard et al., 1993). This "sawtooth" characteristic (Dansgaard et al., 1984) is seen most clearly in the detailed studies that have been carried out on the most recent sequence of changes. The 8lsO

Grip Ice Core

contracted scale

FIGURE 5.23 The 8I80 records from four Greenland sites (see Fig. 5.1) showing parallel, large-amplitude changes over very short periods of time. Changes are commonly "saw-tooth" in pattern (see Stage 8, for example) with an abrupt shift to higher 8I80 levels,followed by a slower return to lower values (johnsen et al., I992).

contracted scale

FIGURE 5.23 The 8I80 records from four Greenland sites (see Fig. 5.1) showing parallel, large-amplitude changes over very short periods of time. Changes are commonly "saw-tooth" in pattern (see Stage 8, for example) with an abrupt shift to higher 8I80 levels,followed by a slower return to lower values (johnsen et al., I992).

rose very abruptly (over -10 yr) from the Older Dryas (cold) phase to the Bolling/Allerod warm period, then fell slowly over the next 1700 yr to the very cold Younger Dryas episode (Dansgaard et al., 1989) (Fig. 5.24). This lasted for 1250 ± 70 yr then ended very abruptly again (within a decade, around 11,640 ± 250 yr ago). The transition marked the beginning of Pre-Boreal conditions, which slowly led to higher Holocene 8lsO values. Dansgaard et al. (1989) argue that the Younger Dryas/Pre-Boreal shift in 8lsO of 5 %o at Dye-3 can be interpreted as an increase in mean annual surface temperature of 7 °C, more than half of the total Pleistocene/ Holocene change (estimated as at least -12 °C in southern Greenland). Deuterium excess (d) also increased at the Younger Dryas/Pre-Boreal transition, suggesting the source area of moisture shifted rapidly northward at that time. The exposure of relatively cold seawater closer to the Summit site would provide both a moisture source (for the heavier accumulation of the Pre-Boreal period) and a lower evaporation temperature in the moisture source region, leading to lower d values in snowfall at Summit. Later, as the North Atlantic became warmer, d values slowly increased (Dansgaard et al., 1989).

Pronounced isotopic changes were accompanied by equally dramatic changes in accumulation (Fig. 5.25). Accumulation more than doubled at the transition from the Older Dryas to the Bolling/Allerod, then declined to low values in the Younger Dryas, before doubling within only a few years at the start of the Pre-Boreal period (Alley et al., 1993). This increase in accumulation must have been associated with an increase in temperature, because precipitation amount, 8lsO, and

FIGURE 5.24 The Dye-3, Greenland record at the time of the Younger Dryas episode. Right side of diagram shows in detail the changes at the end of this episode, around 10,700 (calendar yr) B.R 8lsO increased by ~6%o within 50 yr, accompanied by an even more rapid decline in deuterium excess and in dust levels (Dansgaard et al., 1989).

FIGURE 5.24 The Dye-3, Greenland record at the time of the Younger Dryas episode. Right side of diagram shows in detail the changes at the end of this episode, around 10,700 (calendar yr) B.R 8lsO increased by ~6%o within 50 yr, accompanied by an even more rapid decline in deuterium excess and in dust levels (Dansgaard et al., 1989).

Depth (m)

Pre-8«ea< (PB)

Younger Dryds

Bailing,1 Altered (BAJ

Oldest Dryas <OD|

(YD)

J

t

w/<

'r/ti

WAj^jLjjiJtfKih

\

25-yr smoolhetj

\A

Àkll II

F

3-yf change

12J00 14.620

FIGURE 5.25 Accumulation changes at the Summit site, Greenland (from the GRIP ice core) between -17,500 and 9500 calendar yr B.P. The very rapid changes at the transitions between climate stages are shown in detail in the bottom half of the figure. Colder stages were associated with much lower levels of accumulation (Alley et al., 1993).

11.590 12.B6D

12J00 14.620

FIGURE 5.25 Accumulation changes at the Summit site, Greenland (from the GRIP ice core) between -17,500 and 9500 calendar yr B.P. The very rapid changes at the transitions between climate stages are shown in detail in the bottom half of the figure. Colder stages were associated with much lower levels of accumulation (Alley et al., 1993).

temperature are all positively correlated (Clausen et al., 1988; Dahl-Jensen et al., 1993). On this basis, Alley et al. (1993) also estimate that temperature changed by up to 7 °C from the Younger Dryas to the Pre-Boreal period.

Changes in atmospheric dust also occurred, with the colder periods being times when relatively alkaline (Ca++ rich) continental dust accumulated on the ice cap (Mayewski et al., 1993, 1994). This is most clearly seen in the electrical conductivity (ECM) of the ice (Fig. 5.26a) where the cold dry periods are seen as having lower ECM values than the wetter interstadials (Taylor et al., 1993). This is true for the Bolling/Allerod/Younger Dryas episodes as well as earlier stadial/interstadial events (Fig. 5.26b) and provides a vivid picture of how climatic conditions oscillated between different states before 27,000 yr ago, and again in late glacial time. The ECM in the Vostok ice core provides a similar indication of dust levels associated with a changing climate.

The deepest -300 m of both GRIP and GISP2 ice cores poses a dilemma as there is very little correspondence between them (Taylor et al., 1993a; Johnsen et al., 1995). This is surprising in view of the excellent agreement between the two cores (and with other cores) above these levels. The differences appear to be related to disturbances in the ice (at one, or both, core sites) due to deformation at depth. The GRIP site is at the present day ice divide and, therefore, simple vertical strain is the primary factor in thinning each annual layer that accumulated, unless a shift in the position of the ice divide occurred. The GISP2 site is off to the side of the divide

0 0

Responses

  • Jennifer
    What is the marine isotope stages?
    7 years ago

Post a comment