I

FIGURE 5.27 Schematic diagram to illustrate the potential problems caused by flow boudinage on the interpretation of two adjacent ice cores due to the "pinching out" of layers.

period certainly represent some sort of large-scale climatic process, but the differences between GRIP and GISP2 cores at greater depth point to possible disturbances in one or both of the cores. One way to resolve the problem is to look for evidence of inclined or disturbed layering in the core stratigraphy at depth (Meese et al., 1997; Alley et al, 1997b). In the GRIP core, such layering is seen at 2847 m but in the GISP2 site it starts around 2678 m (Grootes et al, 1993). In the GRIP core there is no strong evidence of overturned folding of the ice, though there is a section (from 2900-2954 m) that is quite disturbed; above and below that, however, there appears to be a relatively undisturbed sequence (Johnsen et al., 1995). This suggests that, overall, there may be a longer climatic record at the GRIP site, but the (undisturbed) record may be separated by sections that are uninterpretable. Chappellaz et al. (1997) examined this possibility by comparing CH4 in the lower GRIP and GISP2 ice cores to the Vostok CH4 record, which is undisturbed. This revealed that sections of both cores are very probably undisturbed and of interglacial age, whereas other sections are made up of either older or younger ice. Hence, the seemingly abrupt changes in 8lsO should not be interpreted as representing rapid changes in climate during the Eemian.

One final point worthy of note concerning ice core records from Greenland: remarkable records have been recovered by sampling surface ice along the ice sheet margin, from the equilibrium line to the ice edge (Reeh et al., 1991). The 8lsO from these "horizontal ice cores" is highly correlated with that in deep ice cores (Fig. 5.28) because ice flow has advected ice originally deposited in the accumulation area to the ice sheet margin (see Fig. 5.16). The significance of this is that, potentially, large samples of quite old ice could be obtained by literally mining the ice margin, rather than coring the base of the ice sheet to obtain very small samples (Reeh et al., 1987). Reeh et al. (1993) note that Summit cores yield less than 10 kg of ice per century for studies of ice older than 40,000 yr B.P., and less than 5 kg for ice of interglacial age. Larger samples from the ice margin might permit detailed investigations of, say, dust or pollen content in the distant past.

FIGURE 5.28 Comparison of 8lsO records from Pakitsoq (on the west Greenland ice sheet margin) and Camp Century. Arrows connect the points considered to be synchronous.The dashed lines show the expected change in 8I80, even without a change in climate, due to ice flow to the sample site (from higher elevations, and, therefore, lower Sl80) (modified from Reeh, 1991).

FIGURE 5.28 Comparison of 8lsO records from Pakitsoq (on the west Greenland ice sheet margin) and Camp Century. Arrows connect the points considered to be synchronous.The dashed lines show the expected change in 8I80, even without a change in climate, due to ice flow to the sample site (from higher elevations, and, therefore, lower Sl80) (modified from Reeh, 1991).

5.4.3 Past Atmospheric Composition from Polar Ice Cores

Ice cores are extremely important archives of past atmospheric composition. In particular, they contain records of how radiatively important trace gases — carbon dioxide, methane, and nitrous oxide — have varied both in the recent past, and over longer periods of time (Raynaud et al., 1993). In addition, ice cores provide records of air mass characteristics (seen in total ion glaciochemistry) as well as the history of explosive volcanic eruptions and changes in atmospheric dust content that may have had significant effects on the global energy balance.

Instrumental measurements of radiatively important trace gases ("greenhouse gases") have a relatively brief history, generally providing a perspective on current gas concentrations of less than 40 yr. These measurements reveal dramatic increases in CH4, C02, N20, and industrial chlorofluorocarbons. Over the same period, levels of heavy metals such as lead and vanadium, as well as anthropogenic sulfate and nitrate have also increased dramatically (Oeschger and Siegenthaler, 1988; Ehhalt, 1988; Stauffer and Neftel, 1988; Mayewski et al., 1992). Ice cores enable the short instrumental records of these contaminants to be placed in a longer-term perspective, providing some measure of the background, preindustrial levels that prevailed before global-scale anthropogenic effects became important (Etheridge et al., 1996). Fig. 5.29 shows the concentration of C02 and CH4 in Antarctic ice cores over the last 150-250 yr; in 1995, CH4 concentration reached 220% of its eighteenth century values whereas C02 was at 130% of its preindustrial level. The N20 levels were 110% of what they were 250 yr ago. Collectively, these data unequivocally document the dramatic increases in greenhouse gases over the last 200-300 yr, to levels far higher than anything seen in records spanning the last 220,000 yr (although older records cannot provide the same time resolution of more recent ice cores). The extent to which these changes are responsible for recent changes in global temperature remains controversial (Lindzen, 1993; Karl 1993; Mann et al., 1998) but there is little doubt that if current trends continue, significant changes in global climate will occur.

One of the more important results from the Vostok ice core is the evidence that atmospheric composition has not remained constant over glacial-interglacial cycles. In particular, the concentration of radiatively important trace gases — carbon dioxide, methane, and nitrous oxide — have all changed significantly. There is also evidence that aerosol concentrations changed dramatically and that changes in the global sulphur cycle may have had important consequences for global cloudiness and hence the earth's energy balance. These changes can also be used to link the chronologies of ice core records in both hemispheres because the mixing times of the important trace gases are short (1-2 yr), so that changes observed in one record should be essentially synchronous in both hemispheres. In this section, these issues are examined in more detail.

A fundamental problem in constructing a paleo-record of trace gas concentrations from ice cores is the fact that the air in ice bubbles is always younger than the age of the surrounding ice (Schwander and Stauffer, 1984). This is because as snow is buried by later snowfalls and slowly becomes transformed to firn and ice, the air between the snow crystals remains in contact with the atmosphere until the bubbles or pores of air

FIGURE 5.29 The C02 and CH, levels in Antarctic ice cores (Siple and DE08, see Fig. I). Solid lines are instrumentally recorded values (Raynaud et al, 1993).

become sealed at the firn/ice transition (when density increases to 0.8-0.83 g cm"3). The sealed bubbles thus contain air that is representative of atmospheric conditions long after the time of deposition of the surrounding snow (Fig. 5.30). "Pore close-off" varies with accumulation rate, ranging from -100 yr at high accumulation sites like Dye-3 in Greenland or Siple Station in Antarctica, to as much as 2600 yr at very low accumulation sites in central East Antarctica. Furthermore, this value will have changed over time because accumulation rates were much lower in glacial times and thus the density-depth profile (or time to "pore close-off") will have been considerably longer. At Vostok, the change in accumulation rate from interglacial to glacial time changed the air-ice age difference from -2500 yr to -6000 yr or more (Barnola et al., 1991; Sowers et al., 1992) (Fig. 5.31). Another consideration is that not all pores in a given stratum become sealed at the same time (perhaps closing over a period of -50 yr at Dye-3, but -500 years at Vostok, for example) so the air bubble gas record should be considered a "low pass" filtered record, with each sample of analyzed crushed ice, being representative of gas concentrations over several tens to several hundreds of years. The highest resolution records should therefore be found in high accumulation rate areas where snow is buried quickly and pore close-off is rapid. Such conditions tend to be found in warmer polar environments such as southern Greenland or in the more maritime sections of Antarctica. Unfortunately, this can create an additional

Sintering Process Surface Density (g/cm3)

Sintering Process Surface Density (g/cm3)

FIGURE 5.30 Schematic diagram illustrating how air is trapped in firn and ice during the process of sintering or recrystallization of snow crystals. Depending on the accumulation rate, "pore close off" (complete isolation of air bubbles) may take up to 2500 years; this time will have varied in the past with changes in accumulation rate (see Fig.5.31) (Raynaud, 1992).

FIGURE 5.30 Schematic diagram illustrating how air is trapped in firn and ice during the process of sintering or recrystallization of snow crystals. Depending on the accumulation rate, "pore close off" (complete isolation of air bubbles) may take up to 2500 years; this time will have varied in the past with changes in accumulation rate (see Fig.5.31) (Raynaud, 1992).

FIGURE 5.3 I Difference in age (AT) between the ice and air which it encloses, atVostok, Antarctica, based on a model of the air trapping process. Changes in accumulation rate over time result in variations in AT (Barnola et al., 1991).

FIGURE 5.3 I Difference in age (AT) between the ice and air which it encloses, atVostok, Antarctica, based on a model of the air trapping process. Changes in accumulation rate over time result in variations in AT (Barnola et al., 1991).

problem with some trace gas records if surface melting and refreezing has occurred. This can seriously affect C02 levels in the ice, as appears to have happened in ice cores from Dye-3, southern Greenland; earlier records from that site are now considered to be suspect (Jouzel et al., 1992; Sowers and Bender, 1995).

Figure 5.32 shows the C02 and CH4 records over the last 220,000 yr from Vos-tok, in comparison with the estimated temperature change in the atmosphere (above the surface inversion) derived from 8D, taking into account the air-ice age difference and its changes with depth (Jouzel, et al., 1993b). It is clear that there is a very high correlation between AT and ACH4 (r2*~0.8). During glacial times C02 levels were around 180-190 parts per million by volume (p.p.m.v.) compared to interglacial levels of 270-280 p.p.m.v. Similarly, CH4 levels were around 350^100 parts per billion by volume (p.p.b.v.) in glacial times, versus -650 p.p.b.v. in interglacials. Of particular significance is the phase relationship between C02 levels and AT during the transition from glacial to interglacial climate and back again to glacial times. At the change from Stage H to Stage G (penultimate glacial stage to last interglacial), C02 was essentially in phase with AT, as far as can be determined, given the uncertainty in ice-air age difference as discussed already. Similar in-phase relationships are seen in the Byrd and Dome C records (Raynaud and Barnola, 1985; Neftel et al., 1988). However, in the subsequent shift towards colder conditions (from -130 ka to 115 ka B.P.) C02 levels remained high while AT dropped by an estimated 7 °C (Barnola et al., 1987). This change in temperature occurred before continental ice growth in the northern hemisphere started to influence oceanic 8lsO and sea level (Chappell and Shackleton, 1987). The most rapid decline in C02 occurred from -115-105 ka B.P., when levels fell from -265 to 230 p.p.m.v. This points to some mechanism first

100 150

FIGURE 5.32 The Vostok record of changes In the concentration of carbon dioxide (top) and methane (bottom) with temperature above the surface inversion expressed as differences from present. Temperatures are estimated from changes in 8D. The difference in age between the air and the enclosing ice has been taken into consideration, as have changes in this value, with variations in accumulation rate over time (Jouzel et al., 1993b).

100 150

FIGURE 5.32 The Vostok record of changes In the concentration of carbon dioxide (top) and methane (bottom) with temperature above the surface inversion expressed as differences from present. Temperatures are estimated from changes in 8D. The difference in age between the air and the enclosing ice has been taken into consideration, as have changes in this value, with variations in accumulation rate over time (Jouzel et al., 1993b).

initiating a change in climate and subsequently leading to a situation in which atmospheric C02 levels were drawn down. The most probable mechanism for such a scenario is orbitally driven radiation changes, which brought about changes in the deep ocean circulation. This may have resulted in increased biological activity in areas of upwelling, which then brought about a reduction in atmospheric C02 levels. The radiative consequences of such a reduction would have reinforced any orbitally induced cooling, eventually leading to full glacial conditions. At a later stage, the rapid C02 increase at glacial-interglacial transitions may have been more related to changes in the surface ocean circulation (Barnola et al., 1987).

The long-term record of CH4 is broadly similar to that of C02 in the sense of large glacial-interglacial changes, but some important differences are nevertheless apparent (Fig. 5.32). Whereas C02 levels declined slowly from the last interglacial (Stage G) to Stage B, methane levels following Stage G remained generally low, punctuated by higher levels during interstadial times (Chappellaz et al., 1990; Brook et al., 1996). This difference in the C02 and CH4 records reflects the fact that the primary driving forces for C02 and CH4 are different. Atmospheric C02 levels are largely the result of oceanic changes, whereas there is relatively little CH4 dissolved in the ocean and atmospheric levels are driven by changes in source areas on the continents. In particular, the extent of wetlands in the tropics (but also at high latitudes of the northern hemisphere) is of critical importance to CH4 levels in the atmosphere. This points to the significance of monsoon circulations and their influence on the extent of low latitude wetland areas over glacial-interglacial cycles (Petit-Maire et al., 1991). Considering that CH4 is removed from the atmosphere largely by hydroxyl ion (OH") oxidation in the atmosphere, and that it is likely such a sink was more effective (with more abundant water vapor in the atmosphere) during warmer interglacial times, Raynaud et al. (1988) estimate that the global emissions of CH4 increased by a factor of 2.3 from glacial to interglacial times. This compares with an observed increase (in the ice core record) of 1.8. Such an increase probably resulted from more extensive tropical wetlands and attendant anaerobic bacterial methanogenesis, and from higher rates of bacterial activity in high latitude peatlands during interglacials. Support for this hypothesis comes from detailed CH4 measurements spanning the Holocene period in the GRIP (Summit) ice core from Greenland (Blunier et al., 1995). Methane levels reached a minimum of -590 p.p.b.v. at 5200 years B.P. compared to early and late Holocene levels around 730 p.p.b.v. (Fig. 5.33). This decline in CH4 levels from the early to mid-Holocene corresponds to the well-recognized reduction in area of tropical wetlands over this interval with maximum aridity around 5000-6000 yr B.P. (Street-Perrott, 1993). The subsequent rise in CH4 levels is thought to be due to growth of high latitude peatlands, since tropical regions in general remained relatively dry in the late Holocene.

10,000 12,000

FIGURE 5.33 Methane levels in the GRIP ice core (Summit, Greenland) during the last 12,000 yr. A pronounced drop in 8laO at -8250 (calendar years) B.P. corresponds to a sudden reduction in CH4.The decline in CH4 to -5200 B.P. is related to a reduction in tropical wetlands.The subsequent CH4 increase is ascribed to high-latitude peatland expansion (Blunier et al., 1995).

10,000 12,000

FIGURE 5.33 Methane levels in the GRIP ice core (Summit, Greenland) during the last 12,000 yr. A pronounced drop in 8laO at -8250 (calendar years) B.P. corresponds to a sudden reduction in CH4.The decline in CH4 to -5200 B.P. is related to a reduction in tropical wetlands.The subsequent CH4 increase is ascribed to high-latitude peatland expansion (Blunier et al., 1995).

As noted earlier, the large difference in age between gas content and the age of the enclosing ice in areas of low accumulation like Vostok makes it difficult to compare directly the isotopic and gas records. This problem is minimized at Summit, Greenland where the CH4 and 8lsO records over the period from 8000 to 40,000 yr B.P. provide clear evidence that large CH4 and isotopic shifts have occurred essentially simultaneously (Chappellaz et al., 1993; Brook et al., 1996) (Fig. 5.34). Such rapid changes in the isotopic content of Greenland snow are considered to be linked to changes in North Atlantic themohaline circulation (deepwater production), which may thus provide the connection between tropical wetland extent and high latitude temperature. Alternatively, the higher CH4 levels may reflect de-

10.000

20.000 30.000

FIGURE 5.34 The record of CH, in air bubbles, and SlsO of ice from Summit, Greenland (based on the GRIP ice core).The solid line in the upper figure is the mean concentration and the thin lines represent measurement uncertainty (2cr). Higher CH4 levels are associated with higher 8lsO levels (Chappellaz et al. 1993).

10.000

20.000 30.000

40.000

FIGURE 5.34 The record of CH, in air bubbles, and SlsO of ice from Summit, Greenland (based on the GRIP ice core).The solid line in the upper figure is the mean concentration and the thin lines represent measurement uncertainty (2cr). Higher CH4 levels are associated with higher 8lsO levels (Chappellaz et al. 1993).

gassing of northern peatlands, in unglaciated areas (such as northern Eurasia and Alaska) during the warmer "interstadial" episodes seen in Fig. 5.34. Whatever the cause of these rapid oscillations, they will eventually provide a valuable chronological tool for matching the Antarctic and Greenland ice core records, at least in those areas of high accumulation where high-resolution trace gas data can be obtained.

Long-term changes in nitrous oxide (N20), another greenhouse gas, have also been determined from air bubbles in Antarctic ice — in this case from the Byrd ice core (Leuenberger and Siegenthaler, 1992). As with C02 and CH4, N20 levels were also much lower in glacial times, 30% lower than levels in the Holocene (-190 p.p.b.v. vs -265 p.p.b.v.) (Fig. 5.35). Lower atmospheric N20 concentration probably resulted from a reduction in the production rate in soils during glaciations, when the global terrestrial biomass was significantly reduced (estimated reductions o>

FIGURE 5.35 Concentrations of nitrous oxide and other greenhouse gases in the Byrd station ice core, Antarctica (plotted in terms of estimated gas age) with the 8lsO values.The LGM values of NjO were 30% lower than in the Holocene (Leuenberger and Siegenthaler, 1992).

FIGURE 5.35 Concentrations of nitrous oxide and other greenhouse gases in the Byrd station ice core, Antarctica (plotted in terms of estimated gas age) with the 8lsO values.The LGM values of NjO were 30% lower than in the Holocene (Leuenberger and Siegenthaler, 1992).

range from 22-57% of preindustrial conditions). Interestingly, the main sink for N20 is in the stratosphere, where it reacts with ozone. Hence, stratospheric ozone levels may have been higher during the LGM, leading to lower ultraviolet (UV) radiation levels in the troposphere.

Spectral analysis of the C02 and CH4 records in Vostok ice reveals periodicities that appear to be related to orbital forcing, though the signal is not simple. In the original 160 ka chronology of C02, for example, the strongest periodicity was in the precessional frequency band (-20 ka) but in the extended 220 ka record a strong obliquity (-41 ka) periodicity is apparent, as in the 8D series. The CH4 shows a strong precessional cycle over the last 160 ka but as can be seen in Figure 5.32 this is less significant over the last 220 ka. For reasons as yet unknown, CH4 does not undergo strong variations in the penultimate glacial period and, indeed, there is no evidence of a lag between C02 and AT going from warm conditions around 220 ka to 160 ka B.P., like that observed in the subsequent glacial cycle.

If the 8D or 8180 record from Vostok is considered to be the "product" of various forcing factors, multivariate analysis can be used to try to identify the most important of these "independent" variables (Genthon et al., 1987). Such an analysis reveals that changes in greenhouse gases are of primary importance in explaining the variance of temperature changes in Antarctica over the last glacial-interglacial cycle, though it is certainly likely that orbitally induced changes in solar radiation distribution in some way modulate changes in greenhouse gases such as C02, CH4 and N20. These considerations led Genthon et al. (1987) to conclude that "climatic changes [are] triggered by insolation changes, with the relatively weak orbital forcing being strongly amplified by possibly orbitally induced C02 changes." To this we could certainly add CH4, N20 and other atmospheric constituents (aerosols, DMS, etc.). This is of particular significance in understanding the synchronism of major glaciations in the northern and southern hemispheres, which are difficult to explain from orbital theory alone.

Although the radiative effects of greenhouse gases (and their associated feedbacks) are clearly important for the earth's energy balance, biogenic sulphur production (dimethyl sulphide, DMS, which oxidizes to sulphate in the atmosphere) also plays an important role (Charlson et al., 1987). With higher levels of DMS production, the number of cloud condensation nuclei increase, leading to more extensive stratiform clouds, a higher planetary albedo, and hence (perhaps) lower temperatures (Legrand et al., 1988). Measurements of methane sulphonic acid (an oxidation product of DMS) in Dome C and less specific measurements of non-sea salt sulphate (nss S04 ") in Vostok ice reveal important increases during cold periods (even after taking into account changes in accumulation rate and possible enhanced transportation of nss S04 " to Antarctica at those times) (Saigne and Legrand, 1987; Legrand et al., 1987, 1991). At Vostok, for example, in Stages B, C, and D, nss S04" levels were 27-70% higher than in interglacial Stages A and H, suggesting greatly enhanced productivity of the oceanic biota that produce DMS during the period from -70-18 ka B.P. (Legrand et al., 1991). This probably reflects higher biological activity in upwelling areas of the ocean during colder intervals (Sarnthein et al., 1987).

It is possible to estimate the direct radiative effects of the glacial-interglacial changes in greenhouse gases on global temperature change from radiative-convective models; for C02 this is around 0.5 °C; for CH4 0.08 °C; and for N20, 0.12 °C (Leuenberger and Siegenthaler, 1992). However, the more important issue concerns the associated feedbacks, involving clouds, snow cover, sea ice, etc., which may have resulted from (and amplified) these changes. The overall (equilibrium) temperature change due to doubling of C02, including radiative effects and feedbacks, is referred to as the climate sensitivity and the amplification effect as the net feedback factor (f). General circulation model experiments lead to estimates of f in the range of 1-4 so that the direct radiative effects of a doubling of C02 levels will be increased by a factor of 1 to 4. Using the observed changes in greenhouse gases, dust, non-sea salt sulphate and global ice volume, together with calculated orbitally induced changes in radiation, Lorius et al. (1990) attempted to account for the overall variance in Vostok temperature over the last glacial-interglacial cycle. Their analysis revealed that 50 ± 10% of the variance of the temperature record (-6 °C in the atmosphere above Vostok) is accounted for by greenhouse gas changes. If this figure can be applied to global glacial-interglacial temperature changes (estimated as -4-5 °C; Rind and Peteet, 1985) then it suggests that changes in greenhouse gases were responsible for -2 °C of the global temperature change over the last climatic cycle. Comparing this figure with the calculated direct radiative effect (0.7 °C) suggests a net feedback amplification value (f) of -3. This is in line with model estimates, though at the high end of the general range. Possibly this reflects a higher climate sensitivity in glacial times, when "slow feedbacks" associated with ice extent on land, and semi-permanent sea ice and ice shelves, may have played a stronger role in amplifying the radiative forcing than they do at the present time.

One of the more remarkable features of ice cores from both hemispheres is the dramatic increase in aerosol concentration during the last glacial period. At Vostok, there were three main episodes of increased aluminum concentrations (an index of continental dust, which is primarily made up of aluminosilicates); these peaks are centered on 160 ka, 60 ka, and 20 ka B.P. (i.e., cold Stages H, D, and B) (De Ange-lis et al., 1987; Petit et al., 1990). Taking into account changes in accumulation rates over time, annual dust fluxes have been calculated (Fig. 5.36). These show that dust flux was 15 times higher in Stage B than during the Holocene. This is related to both an increase in mean wind speed (also leading to higher sea salt sodium levels in glacial times; De Angelis et al., 1987) as well as drier conditions in many arid and semiarid areas of the world. Sources of dust in the southern hemisphere are the semiarid areas of Patagonia, and the extensive continental shelves that were exposed during glacial times. Isotopic studies on dust from the LGM in Dome C clearly point to Patagonia as the primary source region (Grousset et al., 1992). Similarly, geochemical analysis of dust in Byrd, Vostock, and Dome C ice cores indicates a mixture of both marine carbonates and clays (mainly illite) from the Patagonia desert areas (Briat et al., 1982; Delmas and Petit, 1994). Studies of the optical properties of LGM dust from Dome C suggest that there may have been important radiative effects on surface temperature due to the increased aerosol load (Royer et

FIGURE 5.36 Dust flux atVostok (x I0"9cm yr1) in relation to the estimated surface temperature change from the present, derived from changes in 8D. Both records have been mathematically smoothed (see Fig. 5.32) (Petit eta/., 1990).

FIGURE 5.36 Dust flux atVostok (x I0"9cm yr1) in relation to the estimated surface temperature change from the present, derived from changes in 8D. Both records have been mathematically smoothed (see Fig. 5.32) (Petit eta/., 1990).

al., 1983). Although subject to considerable uncertainty, they estimate that the temperature effect resulted in a warming of ~2 °C over Antarctica, which would compensate somewhat for the lower levels of atmospheric greenhouse gases at that time (Overpeck et al., 1996).

Glaciochemical analysis of ice cores from Greenland and Antarctica provides a comprehensive perspective on air mass characteristics that can be characterized in terms of the total chemistry in the ice. Thus, Mayewski et al. (1997) recognize that in the GISP2 ice core two major circulation regimes prevailed during the last glacial interglacial cycle. One regime is dominated by a polar/high-latitude air mass (with higher levels of continental dust and marine-derived ions) and a second mid-, low-latitude air mass (with high levels of biogenic nitrate and ammonium ions). Figure 5.37 shows the variations of these two regimes over the last 110,000 yr. As one might expect, this reveals that the abrupt changes in 8180 during the last glacial (Dansgaard-Oeschger oscillations) were associated with pronounced shifts in circulation regimes. During cold events (low 8lsO) a polar/high latitude circulation prevailed whereas during the Holocene and mild interstadials, the mid-, low-latitude circulation pattern was more prevalent (Mayewski et al., 1994). The changes in dominance of circulation regimes seen so prominently during the last glacial period are also identifiable (albeit more subtly) during the Holocene, enabling the principal circulation changes during recent millennia to be identified (O'Brien et al., 1995). Although such changes are defined by the ice-core geochemistry in a remote part of Greenland, there is evidence that they may have significance far beyond this region because of teleconnections linking the atmospheric circulation over long distances (Stager and Mayewski, 1997).

-5

20,000 40,000 60,000 80,000 100,000 Years Ago

20,000 40,000 60,000 80,000 100,000 Years Ago

FIGURE 5.37 Variations in the two main principal components of the glaciochemistry of the GISP2 ice core over the last 110,000 yr (based on covariations in sodium, potassium, ammonium, calcium, magnesium, sulfate, nitrate, and chloride ions).The lower line is interpreted as changes in the relative importance of polar/high latitude circulation regimes and associated air masses, and the upper line as changes in mid-, low-latitude circulation regimes (Mayewski et al., 1997).

Explosive volcanic eruptions may produce large quantities of sulfur and chlorine gases, which are converted to acids in the atmosphere and may be carried long distances from the eruption site (Devine et al., 1984; Rampino and Self, 1984). When these acids are washed out at high latitudes, the resulting acidic snowfall (and dry deposition of acidic particles directly on the ice sheets) produces high levels of electrical conductivity, which appear as "spikes" above natural background levels in ice cores (Hammer, 1977; Hammer et al., 1980). In most cases these acidity spikes result from excess sulfuric acid events (Figs. 5.14 and 5.38). Hammer's original studies showed a remarkable similarity between electrolytic conductivity in the Greenland Crête ice core and Lamb's Dust Veil Index (Lamb, 1970, 1983), indicating that the elevated acidity (above background levels) could be used as an index of volcanic explosivity. Large explosive eruptions are commonly associated with lower temperatures on a hemispheric (or sometimes even global) scale (Bradley, 1988) although circulation anomalies can lead to warmer conditions in some regions (Robock and Mao, 1995).

Volcanically induced conductivity variations have now been extensively investigated in other cores from Greenland, Antarctica and elsewhere, often with ionic analysis to determine the precise chemistry of the acidity spikes (Holdsworth and Peake, 1985; Mayewski et al., 1986; Legrand and Delmas, 1987; Lyons et al., 1990;

5.4.4 Volcanic Eruptions Recorded in Ice Cores

Concentration (nEq l') DVI VEI

Concentration (nEq l') DVI VEI

FIGURE 5.38 Sulfuric acid profile from Dome C, Antarctica over the last ~200-250 yr, compared to Lamb's Dust Veil Index (DVI: Lamb, 1970) and the volcanic explosivity index (VEI) of Newhall and Self (1982).The total sulfate deposition at Dome C (in kg km"2) for major volcanic eruptions is shown by the underlined values (Legrand and Delmas, 1987). Note the two major eruptions in the early nineteenth century, the earlier of which is not seen in the northern hemisphere ice cores (see Fig. 5.14).

FIGURE 5.38 Sulfuric acid profile from Dome C, Antarctica over the last ~200-250 yr, compared to Lamb's Dust Veil Index (DVI: Lamb, 1970) and the volcanic explosivity index (VEI) of Newhall and Self (1982).The total sulfate deposition at Dome C (in kg km"2) for major volcanic eruptions is shown by the underlined values (Legrand and Delmas, 1987). Note the two major eruptions in the early nineteenth century, the earlier of which is not seen in the northern hemisphere ice cores (see Fig. 5.14).

Moore et al., 1991; Delmas et al., 1992; Zielinski et al., 1994). Chemical analysis enables more precise "fingerprinting" of individual eruptions, some of which produce large amounts of HC1 or HF, for example, rather than H2S04 (Symonds et al., 1988; Hammer et al, 1997).

The longest record of S04" concentration in an ice core comes from the GISP2 site in Greenland and spans the last -110,000 yr (Zielinski et al., 1996). This provides a unique record of the volcanic aerosol, which is of most importance climatically. Temporal resolution of the analyses is ~2 yr for the last 11,700 yr, but then declines back in time (3-5 yr-samples back to -14,800, 8-10 to -18,200, 10-15 to -50,000, and up to 50 yr per sample towards the oldest section of the core). This makes it difficult to compare these short-lived events directly because a single acidity spike readily seen in the Holocene would be effectively diluted over the longer periods represented by deeper samples. Nevertheless, there are many episodes when S04 concentrations exceeded those levels observed after recent major eruptions, suggesting that S04 levels may have been extraordinarily high at times in the past, and/or maintained for up to several decades at climatically significant levels. Thus the S04 record from GISP2 provides a very important perspective on volcanic events on the decade to millennial timescale (Zielinski et al., 1996). Several periods stand out as having been particularly active, especially 8-15 ka and 22-35 ka B.P. (Fig. 5.39). It is interesting that these times correspond to the times of major ice growth and decay. This suggests that the greater frequency of eruptions may be a direct consequence of increased crustal stresses associated with glacial loading and unloading, and/or changes in water loading of ocean basins, especially in areas with a thin lithosphere (such as the volcanically active island arcs of the Pacific Rim (Zielinski et al., 1996). Thus, there may be a direct feedback between continental ice growth and explosive volcanism, with orbitally driven ice volume changes driving shorter-term climatic changes associated with the eruptions.

The five largest sulfate anomalies at GISP2 over the last 2000 yr were in 1831 (Babuyan, Philippines), 1815 (Tambora, Indonesia), 1640 (Komataga-Take, Japan), 1600 (Huaynaputina, Peru), and 1259 (possibly El Chichón, Mexico or a near-equatorial source) (Zielinski et al., 1994). Several major eruptions were recorded in cores from Antarctica as well as Greenland, indicating near-equatorial events from which dust and gases spread into both hemispheres (Langway et al., 1988; Palais et al., 1992; Delmas et al., 1992). This points to a difficult problem in assessing the overall eruption size from measurements in an individual ice core. Ice cores from Greenland, for example, are likely to record Icelandic and Alaskan eruptions as larger than equivalent-sized eruptions at lower latitudes, simply because Greenland is located closer to the high latitude source regions (Hammer, 1984). However, even deposition from nearby eruptions will not be dispersed uniformly over the ice sheets, so estimates based on single (~10-15cm diameter!) ice-core samples may be misleading (Clausen and Hammer, 1988). Indeed, some major eruptions were not registered at all in some ice cores (Delmas et al., 1985). Ideally, a suite of cores extending longitudinally along the major mountain ranges of the world is needed to get a more global picture of volcanic aerosol dispersal. However, in many high altitude, low latitude ice cores, deposition of alkaline aerosols neutralizes the volcanic acids and hence eliminates the eruption signal (also a problem during glacial times, when atmospheric dust levels were much higher than in the Holocene). Nevertheless, the collection of many short cores from a wide area of the larger ice sheets, and along polar/alpine transects (e.g., from the South Pole to Ecuador) will eventually enable a better assessment of the spatial pattern of acid deposition to be made, leading to a more reliable record of past explosive volcanism in both the northern and southern hemispheres (Clausen and Hammer, 1988; Mulvaney and Peel, 1987). Cores from Antarctica reveal major eruptions affecting the southern hemisphere that were not recorded in the northern hemisphere, including a second major event shortly before Tambora (possibly around 1809) (Delmas et al., 1992; Cole-Dai et al., 1995).

20,000 40,000 60,000 80,000 100,000 Years ago

ill1

0 20,000 40,000 60,000 80,000 100,000

Years ago

FIGURE 5.39 The GISP2 (Greenland) record of sulfate from volcanic eruptions with the terrestrial dust record, represented by Ca2t(upper diagram).Time is in years before A.D. 2000.The eruptions ofToba, Indonesia (-71,000 yr B.P.), the Icelandic Z2 ash zone (-53,680 years B.P.), and the Phlegraean Fields eruption/Campanian ignimbrite (-34,500 yr B.P.) are indicated.The lower diagram shows the number of eruptions per millennium exceeding an SO, concentration of 74 p.p.b. (dashed line); this is approximately the magnitude of the largest historical eruption (Tambora) recorded at GISP2. Because of the decrease in temporal resolution with depth, the number of individual eruptions is underestimated back in time.The solid line is an estimate of the number of eruptions that may have occurred, if the sampling resolution was the same as in the Holocene throughout the record; thus the lines are coincident for the last -12,000 yr (Zielinski et al., 1997).

One approach that can be used to "scale" the ice-core sulfate (or acidity) record to compensate for long-distance dispersal of volcanic aerosols is to use the concentration of atomic bomb fallout on the Greenland Ice Sheet (from both high and low latitude sources) as a guide to how aerosols, dispersed via the stratosphere, are depleted en route from the source area (Clausen and Hammer, 1988). Using this idea, Zielinski (1995) estimated atmospheric optical depth changes resulting from major explosive eruptions of the last 2100 years. His analysis indicates that the overall impact of volcanic eruptions on atmospheric turbidity was significantly greater in the last 500 years than in the preceding 1600 years. In particular, multiple large eruptions in the interval A.D. 1588-1646 and 1784-1835 probably had a significant cumulative impact on atmospheric optical depth, leading to cooler conditions at those times (Bradley and Jones, 1995).

5.4.5 Correlation of Ice-core Records from Greenland and Antarctica

Because it is not yet possible to date deep ice cores directly by radiometric means (at least not beyond the radiocarbon timescale), establishing a precise chronology for long ice-core records relies on flow models, which are very sensitive to assumptions about past accumulation history and ice dynamics. Thus absolute chronologies are quite uncertain, making comparison of paleoclimatic records from one region to another quite difficult (Reeh, 1991). One means of correlating ice core records over long distances, at least in a relative sense, is to compare those constituents that vary on a global scale and are thus likely to have fluctuations that are essentially simultaneous. One such parameter has already been mentioned: 10Be "spikes," resulting from some short-term increase in production rate (the cause of which is, as yet, unknown) can be seen in ice cores from Byrd, Vostok, and Dome C, Antarctica, and Camp Century, Greenland, providing chronostratigraphic horizons that can be used to correlate these records directly (Raisbeck et al., 1987; Beer et al., 1992). Another approach relies on changes in atmospheric gas composition. As the mixing time of the atmosphere is short (on the order of 1-2 yr), changes in gas content should be essentially synchronous from the Arctic to Antarctica, so the temporal record of gases should be in parallel. Bender et al (1994) have used this fact to good advantage in linking the chronologies of the Vostok and GISP2 ice cores. They use variations of 8lsO in gas bubbles in the ice as the common thread that ties the two records together. The 818Oatm in the atmosphere today is +23.5%o (relative to SMOW), a result of the balance between the fractionation that occurs during photosynthesis and that which occurs during respiration (the "Dole effect"). The 818Osw in the ocean has changed over time due to removal of water relatively enriched in 160 during times of continental ice build-up on the continents. Such changes are recorded in the CaC03 of benthic forams (see Chapter 6). However, if the isotopic content of the ocean changes, the atmospheric oxygen isotope content will undergo parallel changes, because all photosynthetically derived oxygen is affected, directly or indirectly, by the oceanic isotopic composition (i.e., directly, if the 02 is produced by marine biota; indirectly, if 02 is produced by terrestrial biota that are affected by isotopic changes in the hydrological cycle). By extracting 02 from bubbles in the ice, a direct measure of these changes can be obtained (Bender et al., 1985; Sowers et al., 1991, 1993). After taking into account the air-ice age difference at each site, the very similar variations in &18Oatm at Vostok and Summit, Greenland provide a compelling argument for aligning the two records relative to one another, though the absolute chronology remains imprecise (Fig. 5.40). The relatively low frequency changes in §18Oatm (reflecting slow changes in oceanic isotope composition) contrast with the higher frequency changes in 818Oice, which reflect changes in fractionation processes during the formation and delivery of precipitation to each site. During the last glacial period at Summit, very abrupt, large amplitude changes in 818Oice are characteristic of the record. Of the 22 "interstadi-als" recorded in the GISP2 ice core in the interval from 22-105 ka B.P., 8 are also identifiable in the Vostok 8D record. It appears that only the longer episodes (those lasting >2ka in Greenland) are also seen in Antarctica, associated with increases in 8D of >15%o. Because the isotopic changes in Greenland are more frequent and more rapid, Bender et al. (1994) argue that the warming events began in the northern hemisphere and eventually extended to Antarctica if they persisted for a long enough period of time. The most probable mechanism for such a linkage involves changes in heat flux brought about by oceanic circulation changes.

0 0

Post a comment