Changes in atmospheric circulation over the past five decades have enhanced the wind-driven inflow of warm ocean water onto the Antarctic continental shelf, where it melts ice shelves from below1,2,3. Atmospheric circulation changes have also caused rapid warming4 over the West Antarctic Ice Sheet, and contributed to declining sea-ice cover in the adjacent Amundsen–Bellingshausen seas5. It is unknown whether these changes are part of a longer-term trend. Here, we use water-isotope (δ18O) data from an array of ice-core records to place recent West Antarctic climate changes in the context of the past two millennia. We find that the δ18O of West Antarctic precipitation has increased significantly in the past 50 years, in parallel with the trend in temperature, and was probably more elevated during the 1990s than at any other time during the past 200 years. However, δ18O anomalies comparable to those of recent decades occur about 1% of the time over the past 2,000 years. General circulation model simulations suggest that recent trends in δ18O and climate in West Antarctica cannot be distinguished from decadal variability that originates in the tropics. We conclude that the uncertain trajectory of tropical climate variability represents a significant source of uncertainty in projections of West Antarctic climate and ice-sheet change.
The West Antarctic Ice Sheet (WAIS), which is grounded largely below sea level, is potentially unstable. Mass loss from the WAIS is contributing to present sea-level rise, owing to the widespread thinning of ice shelves and the acceleration of the large outlet glaciers that drain the ice sheet into the ocean1. Contemporaneous with the loss of mass from the WAIS, air temperatures over the WAIS have increased significantly in the past 50 years4,6,7.
Climate and ice-sheet changes in West Antarctica are closely linked with one another by changes in regional atmospheric circulation8. Observations beneath the floating ice shelf of Pine Island Glacier, a major drainage system for the flow of the WAIS into the Amundsen Sea, show that the primary cause of ice-shelf thinning is the presence of warm Circumpolar Deep Water on the Antarctic continental shelf3. Circumpolar Deep Water inflow onto the continental shelf probably increased between the 1980s and 1990s because of increased wind stress at the shelf edge2,8. The patterns of sea-level pressure and geopotential height anomalies associated with increased westerly wind stress8 also favour reduced sea-ice extent9 and the advection of warm air onto the continent4,10.
It is unknown whether the climate and glaciological changes that have occurred in West Antarctica in recent decades are part of a longer-term trend associated with anthropogenic climate forcing. This question cannot be evaluated with direct observations. West Antarctic temperature and pressure observations begin only in 1957, and reliable satellite observations of Antarctic sea ice date to 1979. Comprehensive observations of glacier dynamics in the most rapidly changing areas were initiated in the 1990s. Borehole temperature data from the WAIS, although confirming the recent rapid rise in temperature, do not resolve decadal-scale variability in the past6.
Here, we use records of the oxygen isotopic composition of precipitation (δ18O, see Methods) from ice cores to assess the significance of recent West Antarctic climate trends, and to place the observed glaciological changes in a longer-term context. Unlike temperature or other conventional climate variables, δ18O is relatively well sampled in West Antarctica and thus provides the best available data for this purpose. We use a network of ice cores (Fig. 1) that includes sixteen new and updated multi-decade to century-length records from across the WAIS, and a new 2000-year-long record from the central ice divide (WAIS Divide).
The use of δ18O in precipitation as a proxy for temperature is well known, and supported by our data. However, we do not use δ18O as a proxy for temperature. Rather, we observe that δ18O in West Antarctica covaries with atmospheric circulation in a manner similar to temperature. Positive anomalies in both near-surface temperature and δ18O of precipitation in West Antarctica arise when anomalous meridional flow in the troposphere advects warmer air from lower latitudes, reducing the net distillation of water vapour that otherwise progressively lowers δ18O values in precipitation11. The δ18O of precipitation in West Antarctica is also increased by reduced sea-ice cover, owing both to an increase in the fraction of locally derived moisture, and an increase in boundary-layer turbulence that produces more effective penetration of local maritime air masses onto the continent12. West Antarctic δ18O thus reflects the same circulation anomalies that have contributed to the trends in temperature, sea ice and the glaciologically significant ocean circulation changes of the past few decades.
Figure 2 shows δ18O averaged over the West Antarctic ice-core records for the past 200 years. A prominent feature is the elevated δ18O of the 1990s and significant upward trend since the 1950s, consistent with the surface temperature trend4,6,7. In nearly all of the individual ice-core records, the highest δ18O values occur after 1985, and the mean decadal maximum occurs during the 1990s (1991–2000). The limited available data indicate lower values during the 2000s. The mean upward trend in δ18O over the past 50 years therefore primarily reflects the anomalous values of the 1990s.
Before the 1990s, there are two other prominent decadal-scale anomalies in West Antarctic δ18O in the past two centuries: the 1830s and the 1940s (1936–1945). We use the population statistics from the ice-core records to assess the level of confidence at which the δ18O values of these decades, averaged over West Antarctica, are distinguishable from the 1990s maximum (Methods). The results (Fig. 2) show that decadal-average δ18O since 1990 is probably more elevated than that of the 1830s and 1940s, but only at low confidence (p∼0.2 and p∼0.3, respectively). It is extremely likely (p<0.05) that the mean δ18O of the decade 1991–2000 was more elevated than during any decade in the previous 40 years, consistent with the evidence for widespread warming and large-scale changes in atmospheric circulation4,10.
Analysis of the long record at WAIS Divide shows that δ18O in West Antarctica is anomalous not only with respect to the past two centuries, but also with respect to the past two millennia. WAIS Divide is well situated to be representative of West Antarctica as a whole: the WAIS Divide δ18O record is equally well correlated with δ18O records from cores east and west of the ice divide11. Decadal-average δ18O values comparable to the 1990s in the WAIS Divide record are reached on only four occasions in the past 1,000 years (Fig. 3). Before 1,000 years ago, modern decadal-average δ18O values are reached more frequently, but these are superimposed on a declining trend attributable to the influence of Milankovitch orbital forcing and ice flow13. Assuming that the decadal variability is independent of orbital forcing, we calculate δ18O anomalies relative to the long-term trend (dashed line in Fig. 3). Anomalies in δ18O similar to those of the 1990s occur just twice in the past 2,000 years; assuming sampling error estimated from the multiple shorter records, comparably elevated δ18O values were reached about 1% of the time. Analysis of trends yields a similar result: trends in δ18O as large as the most recent 50-year trend in the WAIS Divide record is reached ∼ 2% of the time in the past 2,000 years. Hence, the anomalous δ18O values of recent decades in West Antarctica do not seem to be unprecedented, but are near the upper limit of the range of natural variability. A similar conclusion has been reached for the northern Antarctic Peninsula, based on data from an ice core there14.
Recent climate trends in West Antarctica have been shown to result from anomalous convection over the tropical Pacific, propagated to the high latitudes by atmospheric Rossby waves9,10,15. Significant changes have also occurred in the large-scale circumpolar westerlies, expressed as an increase in the Southern Annular Mode (SAM) index, in austral summer16. Changes in the SAM may be reflected in dust and aerosol increases recorded in West Antarctic ice cores17. However, the largest trends in both temperature and in ice-core δ18O in West Antarctica have occurred in the non-summer seasons, during which the influence of tropical forcing has dominated15. Comparisons of West Antarctic δ18O records with tropical sea surface temperature (SST) and rainfall records suggest an important role for the tropical Pacific in influencing West Antarctic decadal δ18O variability throughout the past 200 years18,19. Indeed, the statistics of climate variability in West Antarctic inferred from δ18O are similar to those for tropical Pacific climate. Although the 1990s were the warmest decade of the past century in the tropics, they are distinct from other decades only at moderate confidence20 and proxy SST data indicate that conditions similar to those of the 1990s in the tropical Pacific occur with a frequency of about twice per century over the past 350 years21.
The statistics of the ice-core δ18O records suggest that decadal variability in the tropical Pacific is sufficient to account both for the strong recent trends, and for the magnitude of decadal variability in West Antarctic climate over the past century. This is supported by simulations with a general circulation model (GCM). We used the ECHAM4.6 GCM (ref. 22), which has been shown previously to reproduce the magnitude and spatial pattern of observed atmospheric circulation and temperature change in West Antarctica in response to tropical SST boundary conditions10. When run with its water-isotope module23, ECHAM4.6 reproduces the characteristic spatial covariance relationships between δ18O in West Antarctic precipitation and geopotential height anomalies and SST seen in observations (Supplementary Information). When forced either by global observed SSTs or tropics-only SSTs, ECHAM4.6 captures the magnitude and spatial expression of the prominent 1990s decadal δ18O anomaly (Fig. 4). As in the observations, the simulated δ18O of the 1990s is significantly elevated compared with the past several decades, but is only marginally distinct from other decadal maxima over the past century. Furthermore, whereas there is a significant positive trend in the tropical SST boundary conditions in the past century, there is no significant trend in the resulting δ18O simulated over West Antarctica, consistent with the observations. This occurs because it is the pattern of warming in the tropics, rather than the mean temperature, that produces the Rossby-wave response10,24. The correlation between the forcing (tropical SST) and simulated δ18O time series is only marginally significant (r = 0.22, p = 0.07), whereas the correlation between the central tropical Pacific SSTs and simulated δ18O is highly significant (r = 0.58, p<0.001).
Much attention has been given to the role of anthropogenic radiative forcing on Antarctic climate through the influence of both increasing greenhouse gases and stratospheric ozone decline on the SAM (ref. 16). Our results suggest that in West Antarctica the influence of natural decadal variability originating in the tropical Pacific has masked any anthropogenically forced response. A caveat is that there is a small but robust long-term trend in the pattern of tropical SSTs that may be interpreted as a response to radiative forcing25,26. However, most analyses suggest that observed changes in the key central Pacific and El Niño/Southern Oscillation regions that influence West Antarctic climate cannot be distinguished from stochastically forced variability20,26.
Our results have implications for the response of the West Antarctic Ice Sheet to anthropogenic forcing. How tropical Pacific climate will change in the future is uncertain, with some GCMs projecting a more El Niño-like state and others suggesting a tendency towards greater La Niña-like conditions27. These different possibilities have potentially quite different impacts on the regional circulation affecting West Antarctica, suggesting that tropical climate represents a significant source of uncertainty in West Antarctic climate, and in the contribution of the WAIS to sea level, over the next century. Moreover, some GCMs overestimate the magnitude of variability associated with the SAM (ref. 28)—the dominant response to radiative forcing in most models—and are therefore likely to underestimate the importance of variability associated with changes in the tropical Pacific. Indeed, large-ensemble modelling studies show that internal variability associated with the tropics is a significant source of uncertainty in climate-change projections at high latitudes, especially for the atmospheric circulation29. Projections of the contribution of the WAIS to future sea-level rise that are based on present rates of ice-sheet mass loss30 should be treated with caution.
Ice-core data presented in this manuscript include both previously published and new analyses of ice and firn cores, primarily from the United States’ ITASE (International Trans-Antarctic Expedition). The development of depth-age scales for these cores was conducted using multi-parameter high-resolution chemistry. Water stable-isotope concentrations were measured by mass spectrometry and laser spectroscopy at the University of Washington, and are reported as δ18O deviations in per mil (‰) from Vienna Standard Mean Ocean Water (VSMOW), normalized to VSMOW and SLAP (Standard Light Antarctic Precipitation), where δ18O = R/RVSMOW−1, R is the abundance ratio of 18O/16O in water, δ18OVSMOW = 0.0‰ and δ18OSLAP = −55.5‰. The ice-core δ18O composite records were produced by normalizing the mean over the data overlap period common to all cores. The spatial distribution of decadal-mean δ18O data from West Antarctic cores is indistinguishable from a normal distribution (p<0.05) based on a Lilliefors goodness-of-fit test. We use a one-tailed Student’s t-test of the null hypothesis that a given decade mean comes from the same population as the 1990s (1991–2000) mean. Use of the non-parameteric Wilcoxon rank-sum test yields indistinguishable results. The significance level p of reported correlations accounts for the degrees of freedom reduced by the quotient (1−r1r2)/(1+r1r2), where ri is the lag-1 autocorrelation of time series i. The general circulation modelling results are from experiments with the ECHAM4.6 atmospheric GCM, at a horizontal resolution of T42 (∼2.8° latitude×2.8° longitude) with 19 vertical levels, using observed SST (ref. 26) as a boundary condition. In experiments using only tropical SSTs as the boundary condition, the atmospheric GCM is coupled to a slab ocean in the extratropics.
This work was supported by the National Science Foundation Office of Polar Programs (grant numbers 0537930, 0837988, 0963924 and 1043092 to E.J.S.; 05379853 and 1043167 to J.W.C.W.; 0944730 to S.B.R.; 0230396, 0440817, 0944348 and 0944266 to K.C.T.; 0096305, 9316564, 0096299, 0424589, 0439589, 063740, 063650 and 0837883 to P.A.M.; 0838871 to D.P.S.). NCAR is sponsored by the National Science Foundation. We thank A. Orsi, J. Bautista and J. Flaherty.
About this article
Nature Geoscience (2019)