LETTERS
PUBLISHED ONLINE: 29 JANUARY 2012 | DOI: 10.1038/NCLIMATE1353
Enhanced warming over the global subtropical
western boundary currents
Lixin Wu
1
*
, Wenju Cai
2
, Liping Zhang
1
, Hisashi Nakamura
3
, Axel Timmermann
4
, Terry Joyce
5
,
Michael J. McPhaden
6
, Michael Alexander
7
, Bo Qiu
4
, Martin Visbeck
8
, Ping Chang
9
and Benjamin Giese
9
Subtropical western boundary currents are warm, fast-flowing
currents that form on the western side of ocean basins.
They carry warm tropical water to the mid-latitudes and
vent large amounts of heat and moisture to the atmosphere
along their paths, affecting atmospheric jet streams and
mid-latitude storms, as well as ocean carbon uptake
1–4
.
The possibility that these highly energetic currents might
change under greenhouse-gas forcing has raised significant
concerns
5–7
, but detecting such changes is challenging owing
to limited observations. Here, using reconstructed sea surface
temperature datasets and century-long ocean and atmosphere
reanalysis products, we find that the post-1900 surface
ocean warming rate over the path of these currents is two
to three times faster than the global mean surface ocean
warming rate. The accelerated warming is associated with a
synchronous poleward shift and/or intensification of global
subtropical western boundary currents in conjunction with
a systematic change in winds over both hemispheres. This
enhanced warming may reduce the ability of the oceans
to absorb anthropogenic carbon dioxide over these regions.
However, uncertainties in detection and attribution of these
warming trends remain, pointing to a need for a long-term
monitoring network of the global western boundary currents
and their extensions.
The increase of carbon dioxide and other greenhouse gases in
the atmosphere has been the major driver of surface warming of
the Earth over the twentieth century, a warming that is projected
to continue in the foreseeable future
8
. The oceans take up both
heat and carbon dioxide, buffering this surface warming. The recent
uptake of heat by the ocean
9
has led to a significant increase in sea
surface temperature (SST) in both the tropics and high latitudes,
with strong warming in the tropical Indian Ocean, and weak warm-
ing over the deep-water-formation region in the high latitudes
10,11
.
Studies so far have focused on tropical warming patterns
8,12–14
owing to their profound influence on atmospheric deep convection
and global atmospheric teleconnections. However, an important
region with some of largest air–sea fluxes of heat, moisture
and carbon dioxide in the world’s oceans is the convergence
zone of extratropical western boundary currents. The poleward
flowing warm/salty subtropical western boundary currents meet
1
Physical Oceanography Laboratory, Ocean University of China, Qingdao 266003, China,
2
CSIRO Marine and Atmosphere Research, Aspendale, Victoria
3195, Australia,
3
Department of Earth, Planetary Science, University of Tokyo, Tokyo 113-8656, Japan,
4
International Pacific Research Center, University of
Hawaii, Honolulu, Hawaii 96822, USA,
5
Physical Oceanography Department, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts 02543,
USA,
6
National Oceanic and Atmospheric Administration/Pacific Marine Environmental Laboratory, Seattle, Washington 98115, USA,
7
National Oceanic
and Atmospheric Administration/Earth System Research Laboratory, Boulder, Colorado 80305, USA,
8
Leibniz-Institut für Meereswissenschaften,
IFM-GEOMAR, D-24105 Kiel, Germany,
9
Department of Oceanography, Texas A&M University, College Station, Texas 77843, USA.
*e-mail: lxwu@ouc.edu.cn.
equatorward flowing cold/fresh counterparts from the subpolar
ocean, separate from the coast and extend eastwards. These
extensions form sharp thermal fronts fueling the mid-latitude
storms that are of vital importance in the climate system. Associated
with wintertime heat loss, carbon is absorbed and sequestrated
into the ocean interior by the subduction and formation of
mode water in these regions. Long air–sea equilibration timescales
also lead to carbon sequestration downstream of the mode
water formation regions.
Here we find a regionally accelerated warming since 1900
over the path of the subtropical western boundary currents in
all the ocean basins that far exceeds the globally averaged surface
ocean warming rate (Fig. 1). The global SST trends since 1900 are
computed based on eight different datasets, including ‘analysed’
SST products that have been optimally interpolated or smoothed in
both time and space (HadISST1, SODA, ERSSTv3b, Kaplanv2), ‘un-
analysed’ SST datasets (HadSST2, Minobe/Maeda SST), and surface
air temperature datasets (MOHMAT43, HadCRUT3; see Methods).
Surface air temperatures (MOHMAT43, HadCRUT3) are indepen-
dently measured, and therefore may serve as cross-validation for
trends derived from SST datasets. In the tropical Pacific, the centre
of action for the interannual El Niño/Southern Oscillation (ENSO)
mode, the SST trend among these datasets differs significantly in
sign and amplitude (Fig. 1), leaving a large uncertainty for the
warming trend of the tropical Pacific
15
. However, in sharp contrast,
all analysed SST datasets (HadISST1, SODA, ERSSTv3b, Kaplanv2)
indicate a conspicuous warming trend over the path of the Pacific,
Atlantic and Indian Ocean subtropical western boundary currents
and their mid-latitude extensions in both hemispheres (Fig. 1a–d).
The magnitude varies regionally and is dataset-dependent, with a
range of 0.8–1.8
◦
C per century, or about two to three times the
rate of the globally averaged SST trend (Fig. 1i). The accelerated
warming trends over the global subtropical western boundary
currents are also evident in the unanalysed SST datasets (HadSST2,
Minobe/Maeda SST; Fig. 1g,h) and surface air temperature datasets
(MOHMAT43, HadCRUT3; Fig. 1e,f). In particular, both Had-
CRUT3 and HadSST2 reveal a nearly uniform warming rate of
approximately 1.5
◦
C per century over these western boundary cur-
rent regions. Note that MOHMAT43, HadCRUT3, and HadSST2
data are independently measured based on different observational
NATURE CLIMATE CHANGE | VOL 2 | MARCH 2012 | www.nature.com/natureclimatechange 161
LETTERS
NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE1353
0° 90° E 180° 90° W0°
60° N
30° N
0°
30° S
60° S
0° 90° E 180° 90° W0°
60° N
30° N
0°
30° S
60° S
0° 90° E 180° 90° W0°
60° N
30° N
0°
30° S
60° S
0° 90° E 180° 90° W0°
60° N
30° N
0°
30° S
60° S
0° 90° E 180
° 90° W0°
60° N
30° N
0°
30° S
60° S
0° 90° E 180° 90° W0°
60° N
30° N
0°
30° S
60° S
0° 90° E 180° 90° W0°
60° N
30° N
0°
30° S
60° S
0° 90° E 180° 90° W0°
60° N
30° N
0°
30° S
60° S
HadISST1 SODA
¬1.5 ¬1.0 ¬0.5 0
°C per century
SST trends (°C per century)
0.5 1.0 1.5
ERSSTv3b Kaplanv2
MOHMAT43 air temperature HadCRUT3 air temperature
abcdefgh
i
GM KC GS EAC BC AC
0
1
2
3
4
ab
cd
ef
g
h
HadSST2 Minobe/Maeda SST
Latitude
Latitude
Latitude
Latitude
Latitude
Latitude
Latitude
Latitude
Longitude Longitude
Longitude
Longitude
Longitude
Longitude
Longitude
Longitude
Figure 1 | Global SST trends over the 1900–2008 period. a–h, SST trends for various datasets. The corresponding global mean SST trend has been
subtracted. White grid boxes denote insufficient data and grey boxes indicate trends that are not statistically significant at the 95% confidence level. i, SST
trends averaged over the mid-latitude extensions of the subtropical western boundary currents in each dataset (labels a–h stand for different datasets).
Error bars denote the 95% confidence interval. Boundary current regions include the Kuroshio Current (KC; 25
◦
E–150
◦
E, 25
◦
N–38
◦
N), the Gulf Stream
(GS; 75
◦
W–45
◦
W, 38
◦
N–48
◦
N), the Eastern Australian Current (EAC; 150
◦
E–165
◦
E, 44
◦
S–34
◦
S), the Brazil Current (BC; 58
◦
W–35
◦
W, 48
◦
S–35
◦
S)
and the Agulhas Current (AC; 25
◦
E–60
◦
E, 45
◦
S–35
◦
S). GM is global mean.
Table 1 | Mean SST warming trends and standard deviations of global and regional subtropical western boundary currents over
different periods (
◦
C per century) for eight different datasets.
Period Global mean Kuroshio Current Gulf Stream East Australian Current Brazil Current Aguhlas Current
1900–2008 0.62± 0.14 1.29± 0.30 1.02 ± 0.37 1.30± 0.23 1.28± 0.15 1.40± 0.40
1900–1949 0.60± 0.18 1.33 ± 0.25 2.31± 0.15 0.90± 0.15 1.13± 0.18 1.71 ± 0.20
1950–2008 0.71 ± 0.22 1.10± 0.31 0.90± 0.30 1.37± 0.23 1.93± 0.28 1.31± 0.42
platforms or instruments
15
, so the consistency between different
datasets suggests a robust process responsible for the accelerated
warming over the global subtropical western boundary currents
and their extensions. The magnitude of the warming trend varies
with dataset but the ensemble mean of these eight datasets is about
1.2
◦
C per century over all the regions (Table 1).
162 NATURE CLIMATE CHANGE | VOL 2 | MARCH 2012 | www.nature.com/natureclimatechange
NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE1353
LETTERS
0.25
0.2
0.15
0.1
0.05
0.02
¬0.02
¬0.02
0
0
Kuroshio Current Extension
32° N33° N34° N35° N36° N37° N
¬0.050
¬0.025
0
0.025
0.050
¬0.050
¬0.025
0
0.025
0.050
¬0.010
¬0.005
0
0.005
0.010
¬0.050
¬0.025
0
0.025
0.050
¬0.050
¬0.025
0
0.025
0.050
0.2
0.15
0.1
0.05
0.02
0.03
¬0.01
¬0.02
0
0
Gulf Stream Extension
35° N36° N37° N38° N39° N40° N41° N
0.11
0.09
0.07
0.05
0.03
0.01
0
0
East Australian Current Extension
35° S34° S33° S32° S
0.09
0.07
0.05
0.07
0.05
0.03
0.01
0
Brazil Current Extension
48° S47° S46° S45° S44° S
0.45
0.35
0.25
0.15
0.05
0.01
Agulhas Current Extension
42° S41° S40° S39° S
32° N33° N34° N35° N36° N37° N
35° N36° N37° N38° N39° N40° N41° N
35° S34° S33° S32° S
48° S47° S46° S45° S44° S
42° S41° S40° S39° S
Kuroshio Current Extension
Gulf Stream Extension
East Australian Current Extension
Brazil Current Extension
Agulhas Current Extension
1,000
800
600
400
200
Depth (m)
1,000
800
600
400
200
Depth (m)
1,000
800
600
400
200
Depth (m)
1,000
800
600
400
200
Depth (m)
1,000
800
600
400
200
Depth (m)
0
0.1
0.2
0.3
0
0.1
0.2
0.3
0
0.05
0.10
¬0.05
0
0.05
0.10
0
0.2
0.4
af
bg
ch
di
ej
Latitude Latitude
Latitude Latitude
Latitude Latitude
Latitude Latitude
Latitude Latitude
Depth-weighted
average eastward
velocity (m s
¬1
)
Depth-weighted
average eastward
velocity (m s
¬1
)
Depth-weighted
average eastward
velocity (m s
¬1
)
Depth-weighted
average eastward
velocity (m s
¬1
)
Depth-weighted
average eastward
velocity (m s
¬1
)
Figure 2 | Trends of eastward velocity across the extensions of the subtropical western boundary currents from SODA (1900–2008).
a–e, Depth–latitudinal profile of mean (contours) and trend (1900–2008, colour). Units for mean and trend are m s
−1
and m s
−1
per century, respectively.
f–j, Latitudinal distribution of depth-weighted average eastward velocity in upper 500-m. Red and blue curves denote mean± trend
∗
N/2 (N is the length of
data used for trend analysis), respectively. Shading indicates the 95% confidence interval of the trend. If the shading around the red and blue curves
overlaps, the trend is not significant. The eastward velocity is zonally averaged over the Kuroshio Current (140
◦
E–180
◦
), the Gulf Stream (290
◦
E–320
◦
E),
the East Australian Current (150
◦
E–160
◦
E), the Brazil Current (295
◦
E–320
◦
E), and the Agulhas Current (21.25
◦
E–40
◦
E), respectively.
The enhanced warming trends suggest a globally synchronous
change of the subtropical western boundary currents over the past
century. One important process that may lead to these enhanced
oceanic warming trends is a poleward shift of the mid-latitude
extensions of these boundary currents
16–18
. However, detecting
human-induced changes in oceanic circulation has been severely
hampered by a lack of direct long-term measurements. To assess the
role of this process in affecting the warming trends, we use the newly
developed century-long oceanic reanalysis product SODA and the
Twentieth Century Atmospheric Reanalysis (20CRv2) product (see
Methods). The evidence of a poleward shift of the subtropical
western boundary currents is seen in the SODA reanalysis. Over the
Kuroshio Extension, for example, the mean current axis, defined as
the latitude of the maximum eastward velocity, is around 34
◦
N
with an eastward velocity of 25 cm s
−1
(Fig. 2a), consistent with
in situ observations
19
. The eastward velocity exhibits a dipolar
trend pattern with an intensification on the northern flank and a
weakening on the southern flank of the mean current axis (Fig. 2a).
This indicates a poleward shift of the Kuroshio Current by about
0.8
◦
± 0.4
◦
over the past century (Fig. 2f) in association with the
warming trend, suggesting a contribution from poleward warm
water advection
17
. A similar dipolar trend pattern is also detected
along the Gulf Stream Extension (Fig. 2b), whose axis undergoes a
poleward shift of 1.4
◦
±0.8
◦
(Fig. 2g). In the Southern Hemisphere,
the poleward axial shift is identified for the mid-latitude extensions
of the Brazil Current (Fig. 2d,i), but less clear for the East Australian
Current (Fig. 2c,h) and the Agulhas Current (Fig. 2e,j). Therefore,
the enhanced warming over the Kuroshio Current, the Gulf Stream
and the Brazil Current may be partly induced by the poleward axial
shift of these western boundary currents.
Owing to the high nonlinearity of the western boundary
currents and insufficient observations of both the winds and ocean
temperatures during the early part of the twentieth century, some
uncertainties remain in the poleward axial shift of the western
boundary currents revealed by SODA. To further validate these
results, we examine changes in the large-scale wind stress over
the twentieth century in 20CRv2. Although the western boundary
currents are not directly forced by the large-scale wind stress, the
boundary between the subtropical and the subpolar gyres, in theory,
is essentially set by the associated mid-latitude zero wind stress curl
NATURE CLIMATE CHANGE | VOL 2 | MARCH 2012 | www.nature.com/natureclimatechange 163
LETTERS
NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE1353
20° N 25° N 30° N 35° N 40° N 45° N 50° N
North Pacific
35.0° N 37.5° N40.0° N42.5° N45.0° N47.5° N50.0° N 52.5° N 5.05° N
North Atlantic
60° S 55° S 50° S 45° S 40° S 35° S
South Pacific
55.0° S 52.5° S 50.0° S 47.5° S 45.0° S 42.5° S 40.0° S
South Atlantic
52° S 50° S 48° S 46° S 44° S 42° S 40° S 38° S
South Indian
¬1.0
¬0.5
0
0.5
1.0
¬1.0
¬0.5
0
0.5
1.0
¬1.0
¬0.5
0
0.5
1.0
¬1.5
¬1.0
¬0.5
0
0.5
1.0
¬1.5
¬1.0
¬0.5
0
0.5
1.0
a
b
c
d
e
Latitude
Latitude
Latitude
Latitude
Latitude
Wind stress curl
(10
¬7
N m
¬3
)
Wind stress curl
(10
¬7
N m
¬3
)
Wind stress curl
(10
¬7
N m
¬3
)
Wind stress curl
(10
¬7
N m
¬3
)
Wind stress curl
(10
¬7
N m
¬3
)
Figure 3 | Trends of zonally averaged wind stress curl over each ocean
basin in 20CRv2 (1900–2008). Red and blue curves denote
mean± trend
∗
N/2 (N is the length of data used for trend analysis),
respectively. Negative (positive) values in the Northern (Southern)
Hemisphere imply equatorward flow in the interior ocean that must be
compensated by poleward flow in the western boundary. Shading indicates
the 95% confidence interval of the trend. If the shading around the red and
blue curves overlaps, the trend is not significant.
line. Therefore, the zero-curl line may be taken as a proxy of the
jet latitude although in reality the axis of the mean jets does not
necessarily follow the zero-curl line exactly. Over the past century,
the zero-curl line averaged zonally across the basin has shifted
poleward by about 3
◦
± 1.4
◦
over the North Pacific (Fig. 3a) and
2.5
◦
±1.0
◦
over the North Atlantic (Fig. 3b). A poleward shift of the
zonally averaged zero-curl line by about 2
◦
± 0.6
◦
is also apparent
over the South Atlantic (Fig. 3d). The poleward shift tendency of the
zero-curl line seems to be consistent with the sign of the poleward
shift in the axis of the Kuroshio Current, the Gulf Stream and
the Brazil Current, but the amplitudes are different. In general,
the poleward shift of these boundary currents is about 1
◦
(Fig. 2),
which is slightly smaller than that of the zero-curl line. Given the
strong meridional temperature gradient in the convergence zones,
the poleward migration of these boundary currents may lead to
a large areal warming through advective changes in SST that are
reinforced by strong air–sea heat exchanges. In the South Pacific
and the South Indian Ocean, the latitude of the zonal extension
of the boundary currents is set by the meridional extent of the
coasts, and thus bears no relationship with the zero wind stress
curl line (Fig. 3c,e). Therefore, the accelerated warming in these two
regions may not be associated with the poleward shift of the western
boundary current axes.
Another important process that may also potentially lead to
these enhanced oceanic warming trends is an intensification of
the western boundary currents, which can transport more warm
tropical water into the cold mid-latitude oceans, augmenting the
warming there. The changes in the amplitudes of the subtropical
western boundary currents may be inferred from the changes in the
surface wind stress curl, although nonlinear vorticity dynamics also
plays a role. In each of the southern subtropical oceanic gyres, the
anticyclonic wind stress curl has intensified over the past century
(Fig. 3c–e), which may accelerate the interior equatorward flow and
thus the poleward western boundary currents. It has been suggested
that intensification of the East Australian Current (Fig. 2c,h), as
manifested in an acceleration of the southern supergyre over the
past decades, is caused by climate change-induced increases in
the westerly wind
5,18
. In the Northern Hemisphere, however, the
wind stress curl seems to have weakened or changed little in the
subtropical North Pacific and the North Atlantic, respectively.
Therefore, the enhanced warming over the western boundary
currents in the Northern Hemisphere may not be associated with
the changes in the amplitudes of the subtropical wind stress curl.
There is a possibility that this connection works well at decadal
timescales, but less so for long-term trends owing to additional
buoyancy forcing effects in a warming climate.
The poleward shift of the zero-curl line and/or intensification
of the subtropical wind stress curls are further supported by
other atmospheric reanalysis products, despite some notable inter-
product discrepancies (see Methods). Analysis of observational data
over the past decades has also revealed a widening of the Hadley
cells
20
and an associated poleward shift of westerly winds in both
hemispheres
21,22
, which has been attributed to increasing green-
house gas concentrations in the atmosphere. These atmospheric
changes over recent decades are largely in the same direction as the
centennial trends identified in our analysis of the surface wind stress
curl. A poleward expansion of the oceanic subtropical gyres in both
hemispheres has been supported by climate model simulations with
increasing greenhouse-gas concentrations
7
.
Although the trends in the twentieth century are consistent
with projections from climate models, an issue arises as to the
extent to which the detected trends are associated with natural
climate variability, particularly at decadal timescales. To address this
issue, we assess the trends in two separate periods: 1900–1949 and
1950–2008. In the early period, enhanced warming can be found
over the Kuroshio Current, the Gulf Stream and the Aguhlas Cur-
rent regions in most of the datasets, whereas the warming over other
subtropical western boundary current regions is marginal (Table 1
and Supplementary Fig. S1). The poleward shift is evident for the
Kuroshio Current and the Brazil Current, but an equatorward shift,
opposite to the long-term trend, is found for the Eastern Australia
Current and the Aguhlas Current (Supplementary Fig. S2). These
results are perhaps not surprising given that greenhouse-gas forcing
in the early period is not strong, and some of the induced changes
do not rise above the background of natural variability. In the
later period, the enhanced warming is seen over all southern ocean
western boundary current regions (Table 1 and Supplementary
Fig. S3), consistent with the intensification trend of the southern
subtropical western boundary currents (Supplementary Fig. S4)
and the wind stress changes (Supplementary Fig. S5). However,
both the Kuroshio Current and the Gulf Stream exhibit a trend
towards an equatorward shift. For the Kuroshio Current, this
equatorward shift is probably attributable to the mid-1970s North
Pacific decadal climate shift
16
. Thus, in the later period, despite
the stronger greenhouse-gas forcing, decadal variability could still
substantially modulate the long-term trends in some ocean basins,
such that the long-term trend can be different from decadal trends
in both amplitude and sign (Supplementary Fig. S6). Nevertheless,
for centennial timescale trends, influences from decadal variability
164 NATURE CLIMATE CHANGE | VOL 2 | MARCH 2012 | www.nature.com/natureclimatechange
NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE1353
LETTERS
diminish, and systematic enhanced warming is detected along the
path of all western boundary currents, making it unlikely to be a
reflection of decadal climate variability.
We conclude that the enhanced warming over the global sub-
tropical western boundary currents in the twentieth century might
be attributable to the poleward shift of their mid-latitude extensions
and/or intensification in their strength. It should be noted that
uncertainties remain large in terms of quantifying, detecting, and
attributing the enhanced warming trends, owing to a lack of long-
term observations of the western boundary currents and to varying
algorithms used in SST reconstructions among reanalysis products.
Some additional mechanisms, for example the enhanced upstream
land warming, may also contribute to the warming over the western
boundary current regions. The estimated errors of the trends
presented here are also probably underestimated because they
do not explicitly take into account uncertainties in observations,
reconstruction algorithms, bias in data assimilations and so on.
Nevertheless, the results highlight the importance of century-long
datasets in detecting climate change signals. To detect future
changes with confidence, a long-term monitoring network of west-
ern boundary current systems that builds on existing programmes
23
is needed, particularly in regions of accelerated warming.
Methods
SST trends. Eight datasets are used for calculating global SST trends: Hadley
Centre Sea Ice and SST version 1 (HadISST1; ref. 24); National Oceanic and
Atmospheric Administration Extended Reconstructed SST version 2 (ERSSTv3b;
ref. 25); Kaplan Extended SST version 2 (Kaplanv2; ref. 26); Simple Ocean Data
Assimilation (SODA) SST product
27
; Hadley Centre SST version 2 (HadSST2;
refs 28,29); night-time marine air temperature (NMAT) from Meteorological
Office Historical Marine Air temperature version 4 (MOHMAT43; ref. 24); and
air temperature from Hadley Centre/Climate Research Unit Temperature version
3 variance-adjusted (HadCRUT3; ref. 30). A detailed description of these datasets
(except SODA) can be found in Deser et al.
15
and cited references. Among these
datasets, MOHMAT43, HadCRUT3, and HadSST2 are independent, and others
may differ in quality-control and bias correction procedures.
The linear trends are calculated from 1900 to 2008 monthly anomalies
using the method of least-squares with statistical significance assessed using a
Student’s t -test. Following Deser et al.
15
, we use a 24-month per decade threshold
as the sampling criterion, but results are not sensitive to this criterion. This
restriction together with the consistency with independently measured marine air
temperatures is used to assess the realism of the SST trends
15
.
Wind stress trends. The Twentieth Century Atmospheric Reanalysis product,
designated as 20CRv2, contains the synoptic-observation-based estimate of
global tropospheric variability spanning from 1871 to 2008 at 6-hourly temporal
resolution and 2
◦
spatial resolution. The product is derived using observations of
synoptic surface pressure and prescribing monthly SST and sea-ice distributions as
boundary conditions for the atmosphere
31
. To assess the trends of the wind stress
in the 20CRv2 atmospheric reanalysis dataset, we conducted an intercomparison
with the National Center for Environmental Prediction and the National
Center for Atmospheric Research (NCEP/NCAR) Reanalysis (1950–2009), the
European Centre for Medium-Range Weather Forecasts (ECMWF) ERA40
Reanalysis (1957–2002), and the newly constructed WASWind (Wave and
Anemometer-based Sea-surface Wind) dataset (1950–2009), in which the bias
in the ship-based measurements of sea surface winds due to an increase in
anemometer height has been corrected
32
. Over the past half century, the wind stress
curl in all these data exhibits an upward trend in the southern subtropical oceans
with the poleward-shifted zero line more pronounced over the Atlantic–Indian
ocean sector (Supplementary Fig. S5). In the northern subtropics, the changes of
the wind stress curl in the North Pacific are not significant in all these data, whereas
in the North Atlantic the wind stress curl is intensified in 20CRv2, NCEP/NCAR
and ERA40 but not in WASWind. The insignificant changes of the wind stress in
the North Pacific may be partly attributable to the opposite effect of the mid-1970s
North Pacific decadal climate shift, which leads to a cooling in the North Pacific
and an equatorward migration of the zero wind stress curl line
16
. Indeed, the
widening of the Hadley cells
20
and associated poleward shift of westerly winds over
the past three decades have been documented from both satellite observations
and model simulations
21,22
. It should be noted the errors shown here are probably
underestimated without explicitly taking into account uncertainties in different
reconstructions. To further assess the robustness of the estimated trends, the
ensemble mean and the standard deviation of trends derived from these four
datasets are calculated. The ensemble mean exhibits an upward trend in global
subtropical wind stress curl, except the subtropical North Pacific. The poleward shift
of the zero-curl line is more evident in the South Atlantic and Indian Ocean sectors.
Oceanic current trends. The oceanic current trends are calculated based on the
century-long SODA reanalysis product
26
, which is based on the Parallel Ocean
Program (POP) ocean model with an average horizontal resolution of 0.4
◦
longitude ×0.25
◦
latitude and with 40 vertical levels. Temperature and salinity
profiles are assimilated and include the recent release of the World Ocean Database
2009 (WOD09). Assimilation is carried out sequentially using a 10-day update cycle
with model error covariances determined from a simulation that does not include
assimilation. The surface boundary conditions are taken from 20CRv2.
To assess whether SODA captures variability of western boundary currents,
we select two sections across the Kuroshio Current and the Gulf Stream where the
transport has been routinely measured. Across the Pollution Nagasaki (PN line)
section of the Kuroshio Current over the East China Sea, repeat hydrographic
surveys have been conducted on a quarterly basis by the Japan Meteorological
Agency since the mid-1950s. For the Gulf Stream, the transport of the Florida
Current between Florida and the Bahamas near 27
◦
N has been collected from
calibration cruises and calibrated cable voltages since the early 1980s. The transports
of the Kuroshio Current across the PN line, which is calculated from geostrophic
balance, and the Florida Current are superimposed over these from SODA
(Supplementary Fig. S7). The agreement is reasonably good, with a correlation of
0.6 and 0.65 (statistically significant at the 95% confidence level) for the Kuroshio
Current and Florida Current during the observed period, respectively.
Received 6 May 2011; accepted 30 November 2011;
published online 29 January 2012
References
1. Minobe, S., Kuwano-Yoshida, A., Komori, N., Xie, S-P. & Small, R. J.
Influence of the Gulf Stream on the troposphere. Nature 452, 206–209 (2008).
2. Kwon, Y-O. et al. Role of Gulf Stream, Kuroshio-Oyashio and their
extensions in large-scale atmosphere–ocean interaction: A review. J. Clim. 23,
3249–3281 (2010).
3. Nakamura, H., Sampe, T., Tanimoto, Y. & Shimpo, A. in Earth’s Climate: The
Ocean-Atmosphere Interaction Vol. 147 (eds Wang, C., Xie, S-P. & Carton, J
.A.) 329–346 (Geophysical Monograph Series, AGU, 2004).
4. Takahahi, T. et al. Climatological mean and decadal change in surface ocean
pCO
2
and net sea–air CO
2
flux over the global oceans. Deep-Sea Res. 56,
554–577 (2009).
5. Roemmich, D. Super spin in the southern seas. Nature 449, 34–35 (2007).
6. Beal, L. M., De Ruijter, W. P. M., Biastoch, A., Zahn, R. & SCOR/WCRP/IAPSO
Working Group 136, On the role of the Aguhlas system in ocean circulation
and Climate. Nature 472, 429–436 (2011).
7. Saenko, O. A., Fyfe, J. C. & England, M. H. On the response of the oceanic
wind-driven circulation to atmospheric CO
2
increase. Clim. Dynam. 25,
415–426 (2005).
8. Meehl, G. et al. The WCRP CMIP3 multimodel dataset: A new era in climate
change research. Bull. Am. Meteorol. Soc. 88, 1383–1394 (2007).
9. Levitus, S. et al. Anthropogenic warming of earth’s climate system. Science 292,
267–270 (2001).
10. Sutton, R. & Hodson, D. Climate response to basin-scale warming and cooling
of the North Atlantic Ocean. J. Clim. 20, 891–907 (2007).
11. Xie, S. P. et al. Global warming pattern formation: Sea surface temperature and
rainfall. J. Clim. 23, 966–986 (2010).
12. Cane, M. A. et al. Twentieth-century sea surface temperature trends. Science
275, 957–960 (1997).
13. Collins, M. & CMIP Modelling Groups, El Niño-or La Niña like climate
change? Clim. Dynam. 24, 89–104 (2005).
14. Vecchi, G. A. & Soden, B. J. Global warming and the weakening of the tropical
circulation. J. Clim. 20, 4316–4340 (2007).
15. Deser, C., Phillips, A. & Alexander, M. Twentieth century tropical sea surface
temperature trends revisited. Geophys. Res. Lett. 37, L10701 (2010).
16. Seager, R., Kushnir, Y., Naik, N. H., Cane, M. A. & Miller, J. Wind-driven
shifts in the latitude of the Kuroshio–Oyashio Extension and generation of SST
anomalies on decadal timescales. J. Clim. 14, 4249–4265 (2001).
17. Marshall, J., Johnson, H. & Goodman, J. A study of the interaction of the North
Atlantic Oscillation with ocean circulation. J. Clim. 14, 1399–1421 (2001).
18. Cai, W. et al. The response of the Southern Annular Mode, the East Australian
Current, and the southern mid-latitude ocean circulation to global warming.
Geophys. Res. Lett. 32, L23706 (2005).
19. Joyce, T. M. & Schmitz, W. J. Zonal velocity structure and transport in the
Kuroshio Extension. J. Phys. Oceanogr. 18, 1484–1491 (1988).
20. Fu, Q., Johanson, C. M., Wallace, J. M. & Reichler, T. Enhanced mid-latitude
tropospheric warming in satellite measurements. Science 312, 1179 (2006).
21. Gillett, N. & Thompson, D. Simulation of recent Southern Hemisphere climate
change. Science 302, 273–275 (2003).
22. Lu, J., Vecchi, G. A. & Reichler, T. Expansion of the Hadley cell under global
warming. Geophys. Res. Lett. 34, L06805 (2007).
23. Cunningham, S. A. et al. Temporal variability of the Atlantic Meridional
Overturning Circulation at 26.5
◦
N. Science 317, 935–938 (2007).
NATURE CLIMATE CHANGE | VOL 2 | MARCH 2012 | www.nature.com/natureclimatechange 165
