Changes in subduction in the South Atlantic Ocean during the 21st century in the CCSM3

Abstract

The Community Climate System Model version 3 is used to analyse changes in water mass subduction rates in the South Atlantic Ocean over the 21st century. The model results are first compared to observations over 1950-2000, and shown to be rather good. The subduction rates do not change significantly over the 21st century, but the densities at which water masses form become significantly lighter. The strong westerly winds in this region do not change much, which suggests small changes to the rate at which the Atlantic sector of the Southern Ocean takes up heat and carbon dioxide over the 21st century.

Full text

Changes in subduction in the South Atlantic Ocean during the
21st century in the CCSM3
Marlos Goes,
1
Ilana Wainer,
1
Peter R. Gent,
2
and Frank O. Bryan
2
Received 21 November 2007; revised 2 February 2008; accepted 14 February 2008; published 18 March 2008.
[1] The Community Climate System Model version 3 is
used to analyse changes in water mass subduction rates in the
South Atlantic Ocean over the 21st century. The model
results are first compared to observations over 1950–2000,
and shown to be rather good. The subduction rates do not
change significantly over the 21st century, but the densities at
which water masses form become significantly lighter. The
strong westerly winds in this region do not change much,
which suggests small changes to the rate at which the Atlantic
sector of the Southern Ocean takes up heat and carbon
dioxide over the 21st century.
Citation: Goes, M., I. Wainer,
P. R. Gent, and F. O. Bryan (2008), Changes in subduction in the
South Atlantic Ocean during the 21st century in the CCSM3,
Geophys. Res. Lett., 35, L06701, doi:10.1029/2007GL032762.
1. Introduction
[2] Water mass formation in the Atlantic sector is an
important component of Southern Ocean ventilation. Com-
ponents of both Subantarctic Mode Water (SAMW) and
Antarctic Intermediate Water (AAIW) are formed in this
basin, see Sloyan and Rintoul [2001] for example, and much
of the Antarctic Bottom Water is formed in the Weddell Sea.
This region is important because it takes up a lot of heat and
CO
2
from the atmosphere. Therefore, an important question
is how water mass formation will c hange ov er the
21st century, especially if formation rates weaken, and the
South Atlantic can’t take up as much heat and CO
2
as it
does at present? There are conflicting published answers to
this question. Russell et al. [2006] analyze future changes in
the Southern Hemisphere from the latest GFDL cl imate
models. They suggest that the ocean may take up more CO
2
in the future, because of a small increase in the strong
westerlies over the high latitude Southern Ocean. Le Quere
et al. [2007] estimate from observations that the Southern
Ocean sink of CO
2
has weakened between 1981 and 2004 at
a faster rate than expected from just the reduced solubility
as CO
2
increases. They attribute this additional weakening
to the observed increase in the Southern Ocean westerlies.
[
3] Possible changes in South Atlantic water mass for-
mation will be analyzed using simulations from the Com-
munity Climate System Model version 3 (CCSM3). The
justification is that SAMW and AAIW are simulated much
better in the ocean components of climate models used in
the IPCC fourth assessment report than in previous reports,
see Sorensen et al. [2001, Figure 2], for example. Very
recently, Sloyan and Kamenkovich [2007] have made a
detailed study of the SAMW and AAIW simulation in eight
of the climate models used in the IPCC fourth assessment
report. The CCSM3 is one of the three best models in
simulating the density structure of the Antarctic Circumpo-
lar Current, the AAIW sa linity minimum in the Indian
Ocean, and the simulation of SAMW and AAIW in general.
Their study is only of present day simulations, and shows
results mostly from the southern Indian and Pacific Oceans.
In this paper, we show results just from the South Atlantic
Ocean, and include projections of future changes over the
21st century.
2. The CCSM3 Model and Integrations Analysed
[4] The CCSM3 is a general circulation climate model that
couples atmosphere, land, ocean, and sea ice components. An
overview and a description of its present day climate simu-
lation using T85 (1.4°) atmosphere and land resolution, and a
nominal 1° ocean and sea ice resolution are given by Collins
et al. [2006]. First, two long control integrations with 1870
values of greenhouse gases were run. Then a total of nine
20th century integrations, that branched from the control runs
at different times, wer e run for 130 years; 1870–1999. These
20th century integrations were all forced using observed
variations of solar input, volcanic and other aerosols, and
observed values of greenhouse gases such as CO
2
.The
21st century integrations continued for 100 years from the
end of the 20th century runs, and spanned years 2000 – 2099.
The runs analyzed in this paper all used the A1B SRES
scenario for the projected future increase in CO
2
, which
increases from 370 ppm in 2000 to 690 ppm in 2100. There
were seven members in this ensemble of 21st century
integrations, and much more detail on all these runs is given
by Meehl et al. [2006, section 2].
3. The 20th Century South Atlantic Simulation
[5] This section will be very brief, and the reader is
referred to the recent Sloyan and Kamenkovich [2007] study
for a more thorough analysis of SAMW and AAIW in
present day simulations from eight climate models incl ud-
ing the CCSM3.
[
6] Figure 1a shows the zonal average salinity in color
and the potential density contour lines in white in the South
Atlantic down to 1.5 km from the Levitus et al. [1998] data.
It shows the salinity minimum that defines AAIW going
from the surface at 50° –55°S, reaching a maximum depth
of about 840 m near 35°S, and then rising to a depth near
680 m equatorward of 20°S. Figure 1b shows the same
fields, but for the ensemble mean of 20th century runs
between 1950–1999. It has a similar subsurface salinity
GEOPHYSICAL RESEARCH LETTERS, VOL. 35, L06701, doi:10.1029/2007GL032762, 2008
Click
Here
for
Full
A
rticl
e
1
Department of Physical Oceanography, University of Sao Paulo, Sao
Paulo, Brazil.
2
National Center for Atmospheric Research, Boulder, Colorado, USA.
Copyright 2008 by the American Geophysical Union.
0094-8276/08/2007GL032762$05.00
L06701 1of5

distribution, with the salinity minimum at the surface near
45° –50°S, reaching a maximum depth of about 830 m near
40°S, and then rising to a depth of about 700 m equatorward
of 30°S. Thus, the salinity is about 0.2 psu fresher than
observations near the surface at 55°S, and 0.1 psu more
saline at depth equatorward of 30°S.
[
7] In observational studies, the location of the salinity
minimum is always taken as the s = 27.2 isopycnal. A very
nice feature of the CCSM3 simulation is that the salinity
minimum below 300 m is also characterized by the same
isopycnal. A quantitative measure of the model error is the
depth of the s = 27.2 isopycnal and the average salinity on
this isopycnal at 30° and 50°S. These values are 730 m and
34.45 psu at 30°S in the model, and 825 m and 34.37 psu in
the Levitus et al. [1998] data. At 50°S, the values are 150 m
and 33.92 psu and 165 m and 34.06 psu in the model and
observations, respectively. These errors imply that the
model temperature is >1° C too cold around 150 m at
50°S, and <1°C too warm around 730 m at 30°S. Overall,
we conclude that the simulation of water mass formati on in
the South Atlantic in CCSM3 is rather good. This gives
confidence that the model will show realistic changes in the
21st century integrations.
4. Water Mass Formation Changes During the
21st Century
[8] We first look at the changes in salinity and potential
density over the 21st century, and their variability across the
ensemble of 21st century runs. Figure 1c shows the zonal
average salinity and potential density in the South Atlantic
for the ensemble mean of 21st century runs between 2080 –
2099. It shows that AAIW becomes fresher, and the
associated isopycnals deepen considerably, over the 21st
century. Contrasting changes in the surface ocean occur
elsewhere: the Weddell Sea freshens significantly, but the
area around 25°S becomes more saline.
[
9] The solid line in Figure 2a shows the ensemble and
zonal average of the s = 27.2 isopycnal depth at 30°S
versus time between 1870–2100, and Figure 2b shows the
same for salinity on the 27.2 isopycnal. The shading shows
the standard deviation across the ensemble. The depth of the
27.2 isopycnal shows variabil ity on the order of 10 m
between 1870–1980, but then begins to deepen monoton-
ically. The deepening starts late in the 20th century, but then
continues strongly throughout the 21st century, so that the
depth reaches 810 m in 2100, compared to the average of
720 m for 1870–1980. Figure 2b shows similar changes
with time for salinity on the 27.2 isopycnal. The average
value, with small variability, is nearly 34.45 psu over 1870–
1980, but then starts to get fresher and this continues over
the 21st century to a value of 34.41 psu at 2100. The
standard deviatio n sho ws tha t both the deepenin g and
freshening of the 27.2 isopycnal are highly significant
compared to the ensemble variability.
[
10] The upper 1 km of the ocean warms throughout the
South Atlantic region, with a near-surface temperature rise
of 1–2°C. However, Figure 2 shows that the 27.2 isopycnal
Figure 1. Plot of zonal average salinity and density contours in the South Atlantic: (a) Levitus et al. [1998] data, (b) CCSM3
ensemble average h1950–1999i, and (c) CCSM3 ensemble average h2080–2099i.
L06701 GOES ET AL.: CHANGES IN SOUTH ATLANTIC SUBDUCTION L06701
2of5

deepens sufficiently that the temperature on it actually goes
down a little.
[
11] The measure of water mass formation rates used is to
diagnose the subduction across the base of the ocean mixed
layer as a function of potential density, see Karstensen and
Quadfasel [2002]. The model mixed layer depth is the
depth where the local buoyancy gradient equals the max-
imum bulk buoyancy gradient calculate d between the
surface and any depth, see Large et al. [1997, p. 2427].
First, the annual maximum depth of the mixed layer is
averaged over a particular decade to calculate the depth, H.
Subduction is then calculated with respect to this depth
under the assump tion that most wate r mass formation
occurs in winter when the mixed layer depth is a maximum.
It is also assumed that using the decadal average value does
not alias the interannual variability. The total subduction is
calculated as,
Sub sðÞ¼
I
sþDs=2
sDs=2
u rH þ w½dA; ð1Þ
where u and w are the decadal averaged horizontal and
vertical velocities at the depth H. The total subduction is an
integral of the local positive subduction over the area
defined by the adjace nt isopycnals s  Ds/2 and s + Ds /2,
where Ds = 0.2.
[
12] Figure 3 shows the transport of subducted waters for
24.9 < s < 27.5 in each decade between 1870 – 2100 from
one of the 20th and 21st century runs. The blue line is the
average value for h1870–1999 i, and the red line the average
value for h2000–2099i. For 24.9 < s < 26.3, the subduction
transports are relatively small, and do not change much over
the 21st century. However, there are signif icant changes for
26.3 < s < 27.1, which is the density range at which SAMW
and AAIW are formed in the CCSM3. The largest changes
are a 3 Sverdr up (Sv) increase at s = 26.8, and a 4 Sv
decrease at s = 27.0, with much smaller increases of <1 Sv
at s = 26.4 and 26.6. The total subduction transport in this
density range is about 15 Sv, which is close to the
observational estimate of Karstensen and Quadfasel
[2002], and the inverse model estimate of Sloyan and
Rintoul [2001]. Total subduction in this density range does
not change significantly over the 21st century.
[
13] Figure 4 shows how the subduction is divided into
the lateral and vertical velocity components at the higher
densities in the range 26.5 < s < 27.3 versus time 1870–
2100 from the same 20th and 21st century run as Figure 3.
The lateral component dominates the vertical component in
all these density classes due to the frontal characteristic of
this region, in agreement with the results of Karstensen and
Quadfasel [2002]. In fact, the vertical component is nega-
tive at s = 26.8 and 27.0. Figure 4 shows that the total
Figure 2. Plot of (a) depth of the s = 27.2 isopycnal, and
(b) salinity on 27.2 at 30° in the South Atlantic versus time
1870–2100. The line is the ensemble mean, and the shading
is the ensemble standard deviation.
Figure 3. Transport of subducted waters for each decade between 1870–2100 versus density 24.9 < s < 27.5 in one 20th
and 21st century run. The blue line shows the average h1870–1999i, and the red line the average h2000–2099i.
L06701 GOES ET AL.: CHANGES IN SOUTH ATLANTIC SUBDUCTION L06701
3of5

subduction rates increase at s = 26.6 and 26.8 over the
21st century, but decrease at s = 27.0 and 27.2. There is
very little subduction at s = 27.2 after about 2040 for this
A1B scenario run. This again shows that the total subduc-
tion rates of SAMW and AAIW do not change much over
the 21st century, but move to a significantly lower density
by about Ds = 0.2.
[
14] Figure 5a shows the decadal maximum mixed layer
depth H, and the density at this depth from the same
21st century run for h2000–2009i, and Figure 5b shows
the same fields for h2090–2099i– h2000 – 2009i. There is
relatively little change in the maximum mixed layer depth
over the 21st century, although the location of the deepest
mixed layer near 40°S moves a little to the south. However,
there is a consistent change to lighter densities throughout
the South Atlantic. The largest change of Ds = 0.5 occurs
between 30° and 40°S where the h2000 – 2009i density is
around s = 26.0. South of 45°S, there is a consistent change
to a ligh ter density by Ds = 0.2 all across the South
Atlantic, which covers the formation region of SAMW
and AAIW. Figure 5b shows how the areas of different
density classes at the base of the maximum mixed layer
change over the 21st century. Additional calculations show
that the changes in subduction over the 21st century at the
higher densiti es shown in Figures 3 and 4 are mostly due to
the changing areas of the density intervals over which the
integral is taken, with the integrand in equation (1) staying
nearly constant.
5. Conclusions
[15] There are four main conclusions from this study:
[
16] a) The CCSM3 simulation in the South Atlantic, and
of AAIW properties in particular, is good enough to give
confidence in the model’s projected changes over the
21st century.
[
17] b) There is natural variability, but small trends, in
the ocean simulation up to 1980. Then changes and trends
appear, which continue throughout the 21st century. These
changes under the A1B scenario forcing are very signif-
icant with respect to both variability within each run, and
the variability across the ensemble of 21st century runs.
[
18] c) In the South Atlantic, the isopycnals become
deeper and the water on them becomes a little fresher and
colder over the 21st century. Isopycnals getting deeper and
salinity decreasing on isopycnals is exactly the interior
ocean signal that Bindoff and McDougall [1994] showed
would result from warming and freshening at the ocean
surface.
[
19] d) Analysis of the subduction across the maximum
mixed layer depth shows that the total SAMW and AAIW
formation rates do not change significantly over the
21st century. However, by 2100 these water masses are
Figure 4. Plot of the lateral and vertical velocity subduction rates for 26.5 < s < 27.3 in the South Atlantic versus time
1870–2100 in one 20th and 21st century run.
L06701 GOES ET AL.: CHANGES IN SOUTH ATLANTIC SUBDUCTION L06701
4of5

formed at significantly lighter densities by Ds = 0.2 than
in 2000. This pattern of change in subduction rates is
consistent across the ensemble of CCSM3 21st century
A1B integrations.
[
20] Thus, the CCSM3 suggests that the uptake of heat
and CO
2
in the South Atlantic over the 21st century will not
be affected by reduced rates of water mass formation. This
conclusion is supported by analyses of water mass trans-
formation rates at the ocean surface and passive tracers done
on the same model runs, which are not shown. Russell et al.
[2006] analyze future changes in the Southern Hemisphere
from the latest GFDL climate models. They suggest that the
ocean may take up more CO
2
in the future, because of a
small increase in the strong westerlies over the high latitude
Southern Ocean. These westerlies increase by only 2–3% in
the South Atlantic sector over the 21st century in the
CCSM3, which is not statistically significant, although the
maximum does move slight ly polewards. Thus, this mech-
anism for increased ocean uptake does not occur in the
CCSM3. Note again that Le Quere et al. [2007] suggest that
increased winds have led to reduced CO
2
uptake over
1981– 2004. Finally, the question of whether the Southern
Ocean continues to take up the same fraction of CO
2
emissions in the future can only be calculated in a model
with a full carbon cycle. Fung et al. [2005] show results
from such a model embedded in an earlier version of this
climate model, CSM1.4. They show that, even if future
ocean circulation changes do not affect the uptake of CO
2
,
whether the ocean continues to take up the same fraction of
emissions depends upon the rate of atmospheric CO
2
increase and the reduced solubility due to rising sea surface
temperatures. Unfortunately, future projections with a car-
bon cycle component have not been performed with the
CCSM3.
References
Bindoff, N. L., and T. J. McDougall (1994), Diagnosing climate change
and ocean ventilation using hydrographic data, J. Phys. Oceanogr., 24,
1137–1152.
Collins, W. D., et al. (2006), The community climate system model version
3(CCSM3), J. Clim., 19, 2122 – 2143.
Fung, I. Y., S. C. Doney, K. Lindsay, and J. John (2005), Evolution of
carbon sinks in a changing climate, Proc. Natl. Acad. Sci. U. S. A., 102,
11,201–11,206.
Karstensen, J., and D. Quadfasel (2002), Formation of Southern Hemi-
sphere thermocline waters: Water mass conversion and subduction,
J. Phys. Oceanogr., 32, 3020 – 3038.
Large, W. G., G. Danabasoglu, S. C. Doney, and J. C. McWilliams (1997),
Sensitivity to surface forcing and boundary layer mixing in a global
ocean model: Annual-mean climatology, J. Phys. Ocea nogr., 27,
2418 – 2447.
Le Quere, C., et al. (2007), Saturation of the Southern Ocean CO
2
sink due
to recent climate change, Science, 316, 1735 – 1738.
Levitus, S., et al. (1998), World Ocean Database 1998, vol. 1, Introduction,
NOAA Atlas NESDIS 18, 346 pp., NOAA, Silver Spring, Md.
Meehl, G. A., et al. (2006), Climate change in the 20th and 21st
centuries and climate change commitment in the CCSM3, J. Clim.,
19, 2597 – 2616.
Russell, J. L., K. W. Dixon, A. Gnanadesikan, R. J. Stouffer, and J. R.
Toggweiler (2006), The Southern Hemisphere westerlies in a warming
world: Propping open the door to the deep ocean, J. Clim., 19,
6382 – 6390.
Sloyan, B. M., and I. V. Kamenkovich (2007), Simulation of Subantarctic
mode and Antarctic intermediate waters in climate models, J. Clim., 20,
5061 – 5080.
Sloyan, B. M., and S. R. Rintoul (2001), Circulation, renewal, and mod-
ification of Antarctic mode and intermediate water, J. Phys. Oceanogr.,
31, 1005 – 1030.
Sorensen, J., J. Ribbe, and G. Shaffer (2001), Antarctic intermediate water
mass formation in ocean general circulation models, J. Phys. Oceanogr.,
31, 3295 – 3311.

F. O. Bryan and P. R. Gent, National Center for Atmospheric Research,
P.O. Box 3000, Boulder, CO 80307-3000, USA. (gent@cgd.ucar.edu)
M. Goes and I. Wainer, Department of Physical Oceanography,
University of Sao Paulo, 05508-900 Sao Paulo, Brazil.
Figure 5. Density, in color, at the annual maximum mixed
layer depth, black contours, in the 21st century run: (a)
h2000–2009i and (b) h2090–2099i–h2000– 2009i.
L06701 GOES ET AL.: CHANGES IN SOUTH ATLANTIC SUBDUCTION L06701
5of5