Synchronous Climate Changes in Antarctica and the North Atlantic

See allHide authors and affiliations

Science  02 Oct 1998:
Vol. 282, Issue 5386, pp. 92-95
DOI: 10.1126/science.282.5386.92


Central Greenland ice cores provide evidence of abrupt changes in climate over the past 100,000 years. Many of these changes have also been identified in sedimentary and geochemical signatures in deep-sea sediment cores from the North Atlantic, confirming the link between millennial-scale climate variability and ocean thermohaline circulation. It is shown here that two of the most prominent North Atlantic events—the rapid warming that marks the end of the last glacial period and the Bølling/Allerød–Younger Dryas oscillation—are also recorded in an ice core from Taylor Dome, in the western Ross Sea sector of Antarctica. This result contrasts with evidence from ice cores in other regions of Antarctica, which show an asynchronous response between the Northern and Southern Hemispheres.

Objective correlation of isotope paleotemperature records from polar ice cores has shown that some climate variations once thought to be synchronous in both hemispheres are in fact out of phase. For example, the Antarctic Cold Reversal (ACR), a period of cooling that appears in many Antarctic stable isotope records (1), has been compared with the Younger Dryas (YD), a prominent feature in Northern Hemisphere records (2). Time series for the cores from Vostok and Byrd Station, Antarctica, correlated to the layer-counted records in central Greenland by measurements of atmospheric trace gas concentrations in trapped air bubbles, show that the ACR occurred at least 1000 years before the YD (3, 4).

Geochemical climate proxies (5, 6) from an ice core at Taylor Dome (77°48′S, 158°43′E, 2374 m above sea level), a near coastal East Antarctic site at the western edge of the Ross Sea (Fig. 1), exhibit large fluctuations during the last glacial-interglacial transition and Holocene that are reminiscent of those in central Greenland. Published records from Taylor Dome, however, use a preliminary time scale (6) that precludes definitive conclusions regarding the timing of rapid climate change events. Here we present a new stable isotope (δD) record (Fig. 2) and a new chronology for the last glacial-interglacial transition in the Taylor Dome core. We use both atmospheric methane (CH4) and the isotopic ratio of molecular oxygen (δ18Oatm) to tie Taylor Dome to the layer-counted chronology of the Greenland Ice Sheet Project 2 (GISP2) (Summit, Greenland) ice core (7). This approach requires calculation of the age difference (Δage) between the ice and the younger gas it encloses. For GISP2, we use the gas-age time scale and Δage values of Brooket al. (8). For Taylor Dome, we obtain a gas chronology by visually matching changes in CH4 and δ18Oatm concentrations with those at GISP2 (Fig. 3). The rapid increases in CH4 before and after the YD provide precise correlation points at 14.6 and 11.6 thousand years before the present (kyr B.P.) (9). The precision of the correlation between 20 and 15 kyr B.P., during which both CH4 and δ18Oatm change relatively slowly, is between 500 and 1500 years.

Figure 1

Map of Antarctica showing locations of Antarctic ice cores.

Figure 2

δD and 10Be concentrations in the Taylor Dome ice core from 0 to 400 m depth (total depth = 554 m), covering the Holocene and last glacial-interglacial transition.

Figure 3

(A) CH4 (○), δ18Oatm (□), and δ15N (•) from trapped air bubbles in the Taylor Dome and GISP2 (+ and ×) cores. (B) Lines show Δage calculated using from 10Be (solid line) and δD (dashed line). Diamonds (⧫) show minimum 10Be Δage constrained by δ15N, assuming a 10-m-thick advective layer at the top of the firn column.

We calculate Δage for Taylor Dome as a function of the effective bubble close-off depth (COD), surface temperature (T) and accumulation rate (), using the empirical Herron-Langway model to describe the firn densification process (10). We assume that the COD occurs at a density ρCOD = 800 ± 10 kg m−3, as determined from nitrogen isotope (δ15N) measurements in firn air (11). Measured δ15N in Taylor Dome ice samples independently constrains Δage by giving a measure of the thickness of the diffusive zone through which gravitational fractionation is manifested. Diffusive layer thicknesses calculated from measured δ15N provide a minimum estimate of the COD and therefore of Δage for given T and (12).

Values for T and are taken as averages over an interval approximating the original thickness of the firn column (13). We assume that T is a linear function of δD, where the slope α = 4.0 ± 1.5‰ °C−1 (14). We calculate from the 10Be concentration (Fig. 2), where we assume that the dry deposition flux is constant and include a term for wet deposition (15). The 10Be method is supported by several observations. First, 10Be concentration and show a strong spatial inverse correlation both locally at Taylor Dome and broadly across the Antarctic continent (16). Second, both empirical and theoretical considerations indicate that the dry deposition flux of10Be at polar latitudes varied little over the last glacial cycle, for averages over time intervals greater than a few decades (17). Third, comparison of 10Be with major ion concentrations in the Taylor Dome core shows a high degree of correlation; variation in accumulation rate produces strong covariance among chemical species, including 10Be and sulfate, which have very different source functions (18). Finally, flow model calculations, based on high-resolution radar profiles and vertical and surface velocity data, provide independent validation of10Be-based estimates of accumulation rates (19). For comparison, we also determine using calculated values for T (from measured δD) by assuming that varies as a linear function of the saturation vapor pressure of water over ice (20). This more commonly used approach, although probably valid for continental sites such as Vostok, is difficult to justify at Taylor Dome, where precipitation is strongly influenced by cyclonic activity (21); relative to the10Be method, it generally overestimates accumulation rates (and therefore underestimates Δage) during cold periods. Values of Δage calculated by the different methods vary by up to ±750 years (Fig. 3). The variance in Δage is greatest in the oldest part of the record (20 to 15 kyr B.P.) but is <300 years during the crucial YD time period and early Holocene and <600 years at 14.6 kyr B.P., at the time of the rapid deglacial warming in the Northern Hemisphere.

We obtain a time scale for Taylor Dome by adding Δage to the gas ages from correlation with GISP2, using the maximum of the estimates shown in Fig. 3. As will become apparent, this approach is the most conservative for comparing Taylor Dome with other ice core records, because it gives the oldest age for a given depth. We estimate the precision of this time scale by propagating uncertainties in ρCOD, T, and (11, 14,15) and adding estimated uncertainties arising from the GISP2 age calculation and the curve-matching technique (22). The resulting δD time series, from 20 to 10 kyr B.P., is compared in Fig. 4 with δD at GISP2. Also shown are δ18O at Byrd and δD at Vostok, both on the Sowers and Bender (3) time scales tied to GISP2 through δ18Oatm. For Byrd, where uncertainties in Δage are small, the time scales of both Sowers and Bender (3) and Blunier et al. (4) are in excellent agreement. For Vostok, uncertainties in Δage are considerably larger; ages from (3) are up to 1200 years greater than those from (4) over the interval from 20 to 10 kyr B.P.

Figure 4

Stable isotope profiles from Taylor Dome, GISP2, Byrd, and Vostok. At the top, the estimated precision in the Taylor Dome age scale is shown [alternative calculations of Δage (Fig. 3) produce younger ages, increasing the contrast with other Antarctic cores]. Boundaries of climate intervals, as defined in the GISP2 record, are shown by dashed vertical lines Hol, Holocene; YD, Younger Dryas; B/A, Bølling/Allerød; LGM, last glacial maximum. ACR is the Antarctic Cold Reversal as defined at Byrd (4).

Figure 4 illustrates three particularly important findings. First, prominent features of the GISP2 record that are absent at Byrd and Vostok appear at Taylor Dome, including generally declining isotope values between 20 and 15 kyr B.P. and near-Holocene isotope values during the Bølling/Allerød (B/A) warm period (14.6 to 12.9 kyr B.P.). Second, the late-glacial cold interval (low δD values) at Taylor Dome, although more subdued than at GISP2, is at least approximately contemporaneous with the Northern Hemisphere YD and definitely lags the ACR. This interval ends with a rapid warming that is synchronous with post-YD warming at GISP2 within a few hundred years. Third, the dramatic warming that marks the end of the last glacial maximum at Taylor Dome lags the onset of gradual warming at Vostok and Byrd by more than 3000 years. In the latter cores, deglacial warming begins before 18 kyr B.P. and continues uninterrupted until the ACR cooling. At GISP2 there is evidence for an initiation of warming as early as 24 kyr B.P., but isotope values generally indicate cold conditions until nearly 14.6 kyr B.P., when rapid deglacial warming occurred. At Taylor Dome, the magnitude of the δD increase during deglacial warming is as large as at GISP2. Uncertainties in both time scales over this interval are considerably larger than for the B/A and YD, but the precision is sufficient to conclude that deglacial warming was synchronous in both cores within 1000 years.

Evidently, climate changes at Taylor Dome during the last glacial-interglacial transition were synchronous or near synchronous with changes in the North Atlantic region. This result has important implications for our understanding of the mechanisms linking climate between the hemispheres. It is generally accepted that abrupt deglacial warming in the Northern Hemisphere was accompanied by the onset of North Atlantic deep water (NADW) formation, promoting northward flow of warm surface waters from the tropics, whereas a circulation pattern marked by reduced NADW formation accounts for cold conditions during the YD interval (23). The Byrd and Vostok records, showing an antiphase relationship between Antarctica and Greenland, have drawn attention to numerical model results in which changes in NADW promote opposing temperature responses in the high latitudes of the Northern and Southern Hemispheres, a consequence of an alternation in the amount of convection or ocean heat convergence (or both) in the two areas (24). The Taylor Dome results, on the other hand, are consistent with earlier arguments that the flow of relatively warm NADW into the Southern Ocean warms circumpolar deep water (CPDW), thereby promoting sea ice melting and atmospheric warming as CPDW upwells along the Antarctic coastal margin (25).

Differences between the isotope-temperature history from Taylor Dome and those from other Antarctic sites are too large to be attributed to dating errors. Rather, the results indicate that the circum-Antarctic climate response to changes in NADW formation and export may not be uniform. We propose that the North Atlantic character of the isotopic record at Taylor Dome, in particular, reflects the relative proximity of this site to the western Ross Sea, an area of active wind-driven convection and ocean-atmosphere heat exchange in today's climate (26). We note that a similarly heterogeneous response to transient reduction of NADW formation and export has been observed in some numerical models (27). For example, in the coupled atmosphere-ocean general circulation model simulations of Schiller et al. (28), near-Antarctic waters of the Southern Ocean (areas of vigorous oceanic convection in control simulations) cool in response to reduced formation and export of NADW, whereas other areas of the Southern Ocean warm as a result of changing patterns of atmospheric circulation and increased ocean heat convergence. Taylor Dome may thus record the direct but localized influence of NADW-borne heat on Antarctic climate (29). Given the substantial difficulty of realistically simulating ocean-atmosphere interactions in general, and the dynamics of the Southern Ocean in particular, it may be some time before the role of NADW in shaping Antarctic climate can be rigorously evaluated. In the meantime, our observations can and should be tested by collection and analysis of additional Antarctic ice cores, especially from near-coastal sites.

  • * To whom correspondence should be addressed. E-mail: steig{at}

  • Present address: Department of Earth and Environmental Science, University of Pennsylvania, Philadelphia, PA 19104, USA.


View Abstract

Navigate This Article