1Atmospheric and Oceanic Science, Princeton University, Princeton, NJ, USA. 2NOAA/Geophysical Fluid Dynamics Laboratory, Princeton, NJ, USA.
3University Corporation for Atmospheric Research, Boulder, CO, USA. *e-mail: Liping.Zhang@noaa.gov
While Arctic sea ice is rapidly decreasing in association
with increasing surface air temperature1, observations
clearly show an expansion of Southern Ocean (SO) sea-
ice extent (SIE)2 during the satellite era (1979–2012) (Fig. 1a). This
modest increase is consistent with the observed SO cooling trend
(Fig. 1a). The sea surface temperature (SST) and sea-ice concentra-
tion (SIC) trends are not homogeneous in space3,4, with opposing
signs in the Amundsen-Bellingshausen seas versus the Ross and
Weddell seas (Fig. 1b,c). Several mechanisms have been proposed
to explain these trends3–16. A leading idea involves surface wind
changes3–14 driven by a number of factors, including a positive trend
of the Southern Annular Mode in response to stratospheric ozone
depletion, or a deepened Amundsen Sea Low driven by remote
tropical Pacific or North Atlantic SST anomalies. In these studies,
the wind-driven surface heat flux, upper ocean dynamics and sea-
ice drift are key drivers for the observed sea-ice and SST dipoles.
Another explanation involves surface freshening15,16 caused by
anthropogenic warming, possibly via global water cycle amplifica-
tion and/or the melting of Antarctic glaciers and ice sheets. The sur-
face freshening enhances stratification and suppresses convective
mixing with the warmer water at depth, producing cold SST anom-
alies and thus inhibiting the melting of Antarctic sea ice. However,
other studies argue that wind anomalies, induced by ozone deple-
tion, favour an overall decrease rather than increase of Antarctic
sea ice17, and that the surface freshening caused by anthropogenic
warming is not large enough to trigger sea-ice expansion18,19.
In this study we examine the possibility that internal variabil-
ity involving deep ocean convection in the SO could be a major
contributor to the observed trends, probably in concert with other
previously identified factors. To explore this, we use simulations
with a newly developed coupled ocean–atmosphere sea-ice model
(SPEAR_AM2—see Methods for model details). When this model
is driven with estimates of changes in past radiative forcing, the
model simulation does not reproduce the observed SST and SIC
trends around the Antarctic. Instead, the model simulates a steady
warming and Antarctic sea-ice loss (Fig. 1d,e), as is commonly seen
in multimodel mean results of Coupled Model Intercomparison
Project Phase 5 (CMIP5) models20. One possible explanation for
the discrepancy between observations and model projections is that
natural variability may play a large role in the observed trends21,22.
Indeed, the observed sea-ice expansion is within the range of natu-
ral variability in the control run2,21,22.
We provide additional evidence supporting a strong role for
natural variability in the observed trends, probably involving
multidecadal modulation of SO convection and deep-water for-
mation. As shown below, various aspects of the observed changes
are consistent with a multidecadal weakening of Antarctic bottom
water (AABW) formation. In the 1970s, the open ocean Weddell
Polynya23,24 (Supplementary Fig. 1) was first seen by satellite. After
1976, no similar Weddell Polynya was observed until 2016. Ship-
based hydrographic observations show that the AABW has warmed
globally between the 1980s and 2000s25,26. Objectively analysed
ocean data also exhibit a warming temperature trend in the SO
subsurface and a cooling trend in the surface (Supplementary Fig. 2).
Both warming trends are consistent with weakened convection.
These observational results suggest a global-scale slowdown of the
bottom, southern limb of the meridional overturning circulation
(MOC) during 1979–201225. Meanwhile, multidecadal variabil-
ity of SO SST27,28 shown in reanalysis (Fig. 1a) and palaeoclimate
records29,30 highlights the low-frequency character of SO climate
and places the recent Antarctic sea-ice trends into a broader context.
SO internal variability in coupled model
In order to compare the observed changes with an estimate of
natural variability, we first examine SO natural climate variability
from an extended simulation of the SPEAR_AM2 model. We see
that the time series of the AABW cell, related to ocean convection
(Methods and Supplementary Figs. 3–5), has pronounced multi-
decadal to centennial-scale fluctuations that begin after an initial
1,000 year spin-up (Fig. 2a). When convection is strong, the mixed
layer depth (MLD) is largest in the open Weddell Sea, near the
Maud Rise (65° S, 0°) (Fig. 2b). This convection location resembles
the observed 1974–1976 Weddell Polynya (Supplementary Fig. 1).
Some MLD changes are also seen over the Antarctic continental
shelves, such as in the west Ross, Weddell seas and East Antarctic,
but the magnitude is much smaller than that in the open ocean
Natural variability of Southern Ocean convection
as a driver of observed climate trends
LipingZhang 1,2*, ThomasL.Delworth1,2, WilliamCooke2,3 and XiaosongYang2,3
Observed Southern Ocean surface cooling and sea-ice expansion over the past several decades are inconsistent with many his-
torical simulations from climate models. Here we show that natural multidecadal variability involving Southern Ocean convec-
tion may have contributed strongly to the observed temperature and sea-ice trends. These observed trends are consistent with
a particular phase of natural variability of the Southern Ocean as derived from climate model simulations. Ensembles of simula-
tions are conducted starting from differing phases of this variability. The observed spatial pattern of trends is reproduced in
simulations that start from an active phase of Southern Ocean convection. Simulations starting from a neutral phase do not
reproduce the observed changes, similarly to the multimodel mean results of CMIP5 models. The long timescales associated
with this natural variability show potential for skilful decadal prediction.
NATURE CLIMATE CHANGE | VOL 9 | JANUARY 2019 | 59–65 | www.nature.com/natureclimatechange 59
The Nature trademark is a registered trademark of Springer Nature Limited.