8/11/2019 2009 - Decadal Increase of Oceanic Carbon Dioxide in Southern Indian Ocean Surface Waters (19912007)
1/13
Decadal increase of oceanic carbon dioxide in Southern Indian Ocean surfacewaters (19912007)
Nicolas Metzl
LOCEAN/IPSL, CNRS, Universite Pierre et Marie Curie, Case 100, 4 Place Jussieu, 75252 Paris Cedex 05, France
a r t i c l e i n f o
Available online 13 December 2008
Keywords:
Carbon dioxide
Southern Ocean
Indian Ocean
Decadal variability
Airsea CO2fluxes
Southern Annular Mode
a b s t r a c t
The decadal variability of the fugacity of carbon dioxide (fCO2) at the sea surface is analyzed for the first
time in the south-western Indian Ocean and corresponding Antarctic sector. This study is based on
seasonal cruises (MINERVE and OISO) conducted onboard the R.S.S. Marion-Dufresneduring the period
19912007. Based on shipboard observations the average annual rate of the atmospheric CO2 was
1.72 ppm/yr, almost equal to the annual growth rate derived from high-quality measurements recorded
at monitoring stations in the Southern Hemisphere. An evaluation based on oceanic observations in the
Southern Indian Ocean (4201S), indicates that oceanicfCO2increased at a rate of 2.11 (70.07)matm/yrfor the period 19912007, i.e. about 0.4 matm/yr faster than in the atmosphere. In order to investigatethe processes that explain the oceanicfCO2variations (and the potential reduction of the ocean carbon
sink), the decadal variability is analyzed in detail in four regions (20351S, 35401S, 40421S and
50551S) for austral summer (DecemberMarch) and winter (JuneAugust). During austral summer, the
fCO2 increase is similar in the four regions (between +2.2 and +2.4 matm/yr). For austral winter thegrowth rate is lower north of 401S (+1.5 to +1.7matm/yr) than at higher latitudes (+2.2 matm/yr). Becausethese regions experienced different warming or cooling, the evolution of temperature normalized fCO2(fCO2
norm) has also been investigated. In the southern subtropical region (35401S), warming occurred in
winter, leading to a small change of fCO2norm (+0.6matm/yr). In this region, anthropogenic CO2 uptake
must be compensated by a reduction of dissolved inorganic carbon (DIC) in surface waters. At latitudes4401S, the observed cooling during winter leads to a rapid increase offCO2
norm (+3.6 to +4.7matm/yr),suggesting that the gradual import of DIC in surface water occurs in addition to anthropogenic CO2. The
contrasting variations observed north and south of 401S are likely related to the high index state of the
Southern Annular Mode (SAM) during the 1990s. The increase of the westerlies at latitudes 4401S could
have enhanced the vertical import of CO2-enriched deep waters in high-latitude surface layers, whereas
the decrease of the wind speed north of 401S would have reduced vertical mixing. Although this analysis
is limited to a relatively short period, 19912007, this is the first time that a link between the SAM and
the decadal reduction of the Southern Ocean carbon sink is suggested from in-situocean carbon dioxide
observations. This offers an encouraging result in the perspective of model validation and under-
standing of the future evolution of the ocean carbon sink and its coupling with climate change.
& 2008 Elsevier Ltd. All rights reserved.
1. Introduction
Anthropogenic emissions of carbon dioxide (CO2) into the
atmosphere from fossil fuel and land use change increased
dramatically from about 5.5PgC/yr (1 Pg 1015g) in 1970 to
8.4 PgC/yr in 2000 (Raupach et al., 2007) up to 9.9 PgC/yr in 2006
(Canadell et al., 2007). This is the consequence of increases in
population size and energy development. About half of these
emissions remain in the atmosphere, leading to significant recent
global warming (IPCC, 2007); the other half is stored in the ocean
and on land, but the partitioning between the ocean andterrestrial carbon sinks are uncertain (Sabine et al., 2003;
Stephens et al., 2007). For the last four decades, many scientists
have developed methods to estimate the global oceanic carbon
uptake. Whatever the method used (ocean observations, ocean
models, atmospheric inversion), the ocean carbon uptake is
estimated to be around 2 PgC/yr (range of 1.7 to 2.8 PgC/yr;
see review inLe Quereand Metzl, 2003). The most recent estimate
based on an international global ocean pCO2 data synthesis
indicates that total ocean CO2 uptake is 1.8 (70.7) PgC/yr for
year 2000 (Takahashi et al., 2009). This value represents about
20% of the anthropogenic emissions of year 2000. For the
anthropocene period (18001994) global ocean carbon inven-
tories derived from in-situ observations suggest that the ocean
ARTICLE IN PRESS
Contents lists available atScienceDirect
journal homepage:www.elsevier.com/locate/dsr2
Deep-Sea Research II
0967-0645/$- see front matter& 2008 Elsevier Ltd. All rights reserved.doi:10.1016/j.dsr2.2008.12.007
Tel.: +33144273394; fax: +33144274993.
E-mail address: [email protected]
Deep-Sea Research II 56 (2009) 607619
http://www.sciencedirect.com/science/journal/dsriihttp://www.elsevier.com/locate/dsr2http://localhost/var/www/apps/conversion/tmp/scratch_2/dx.doi.org/10.1016/j.dsr2.2008.12.007mailto:[email protected]:[email protected]://localhost/var/www/apps/conversion/tmp/scratch_2/dx.doi.org/10.1016/j.dsr2.2008.12.007http://www.elsevier.com/locate/dsr2http://www.sciencedirect.com/science/journal/dsrii8/11/2019 2009 - Decadal Increase of Oceanic Carbon Dioxide in Southern Indian Ocean Surface Waters (19912007)
2/13
absorbed 48% of the emissions over the last 200 years (Sabine
et al., 2004). These authors concluded that the ocean has
constituted the only true net sink for anthropogenic CO2 over
the past 200 years and there are indications that in recent years
the ocean carbon uptake capacity has been reduced (Sabine et al.,
2004;Canadell et al., 2007). How the ocean carbon sink evolved in
the recent period (several decades) and will evolve in the future
(decades to century) are important questions regarding bothclimate change as well as acidification of the oceans and its
impacts on marine ecosystems (Feely et al., 2004).
In this context, observing the long-term change of oceanic
carbon dioxide in surface waters is crucial, not only to better
determine CO2 airsea fluxes at a global scale (Takahashi et al.,
2009), but also to understand how these fluxes will change in the
future under different environmental conditions, including higher
anthropogenic CO2emissions and climate change. The continuous
rise of sea-surface water concentrations of dissolved inorganic
carbon (DIC) and the partial pressure or fugacity of CO2 (pCO2 orfCO2) has been relatively well documented in the North Atlantic
and Pacific Oceans (e.g.,Bates et al., 1996;Bates, 2001;Feely et al.,
2002, 2006; Lefevre et al., 2004; Schuster and Watson, 2007;
Takahashi et al., 2006, 2009). The increase of DIC in sea-surfacewater is generally related to ocean uptake of anthropogenic CO2,
but decadal trends (both positive and negative) also have been
attributed to natural variation or climate change variability,
including evaporation anomalies (Dore et al., 2003), temperature
variation, and water mass transformation (Keeling et al., 2004;
Feely et al., 2006; Corbiere et al., 2007; Takahashi et al., 2009).
Changes in primary productivity also may be responsible for
long-term variations of pCO2, as suggested for the Bering Sea
where pCO2 has decreased for three decades (Takahashi et al.,
2006). However, the impact of the marine biological activity on
CO2 airsea fluxes decadal changes has never been clearly
established.
In the Southern Hemisphere, the long-term evolution of
oceanicpCO2is not well detected, mostly because historical data
are sparse in the remote oceans and the signal-to-noise ratio is
low (Lenton et al., 2006). In addition, during austral summer
when most data are available, the long-term variation of
biogeochemical properties, such as CO2, is often masked by large
spatio-temporal variability (Jabaud-Jan et al., 2004; Inoue and
Ishii, 2005;Breviere et al., 2006). Therefore, the detection of the
decadalpCO2 changes in polar waters requires an analysis over a
very long period (Inoue and Ishii, 2005) and should include winter
observations when the biological activity is low.
In the last 40 years, both greenhouse gas accumulation in the
atmosphere and ozone depletion induced significant thermal
contrast in the Southern Hemisphere (Thompson and Solomon,
2002) and changed the meridional atmospheric pressure gradi-
ents, leading to more positive state of the so-called Southern
Annular Mode (SAM) (Marshall, 2003). The variability of the SAMcan affect wind speeds, heat fluxes, ocean circulation and biology
at mid- and high latitudes (e.g., Lovenduski and Gruber, 2005;Sen
Gupta and England, 2006). Ocean carbon models (Lenton and
Matear, 20 07;Le Quereet al., 2007;Lovenduski et al., 2007;Verdy
et al., 2007) and inversions of atmospheric CO2 observations (Le
Quereet al., 2007) indicate that climate variability in the Southern
Hemisphere may dramatically impact the ocean carbon cycle and
CO2 airsea fluxes in temperate and high latitudes. The link
between surface ocean CO2 and climate variability (SAM and/or
ENSO) also has been recently investigated at regional scale based
on oceanic pCO2 observations conducted south of Tasmania in
19912003 (Borges et al., 2008). Although warming would result
in ocean CO2outgassing anomalies,Borges et al (2008)found that
positive (negative) CO2 airsea fluxes inter-annual anomalies areusually associated with negative (positive) SST anomalies. An
increase of the Southern Ocean carbon sink during warm events
has been occasionally observed and associated with higher
productivity during summer (Jabaud-Jan et al., 2004; Breviere
et al., 2006). How the Southern Ocean carbon sink evolves at
decadal scale has never been directly analyzed from in-situ
observations.
This paper describes for the first time the decadalfCO2changes
in the south-western Indian Ocean (20601S/3090
1E) based on
observations obtained during 19912007. Data were obtained
using consistent instrumentation and processing techniques since
1991. The paper starts with describing the methods and the
atmospheric CO2 trends recorded on board followed by a basin-
wide view of the oceanic fCO2 trends. The analysis is focussed on
four latitudinal bands where oceanic fCO2 variations are likely
driven by different processes in relation to climate changes in the
Southern Hemisphere. Finally, the results in the South-Western
Indian Ocean are compared with decadal fCO2 changes analyzed
in other ocean regions and discussed specifically when comparing
the contrasting patterns observed in the South Indian and South
Pacific oceans.
2. Data collection and atmospheric CO2 trends
Observations of sea-surface and atmospheric fCO2 were
obtained in the Southern Indian Ocean during 19911995
(MINERVE cruises) and 19982007 (OISO cruises) using the same
instrumentation and data processing (Poisson et al., 1993; Metzl
et al., 1995, 1999, 2006). During the cruises (Fig. 1), all conducted
onboard the R.S.S.Marion-Dufresne(IPEV/TAAF), sea-surface water
was continuously pumped and equilibrated with a thin film
type equilibrator, thermostated with surface seawater. After
passing through Peltier cold traps (35 1C), the CO2 in the dried
gas was measured with a non-dispersive infrared analyzer (NDIR,
Siemens Ultramat 5F). During all cruises three standards were
used for calibration with typical low, middle (near atmospheric),
and high CO2 concentrations (corresponding mole fractions
ranges of 250270ppm or mmol CO2 mol1, 350370ppm, and
470490ppm). The standards and atmospheric CO2 were mea-
sured every 67 h. The temperature in the equilibrium cell was
0.21.0 1C warmer than sea-surface temperature (SST) depending
on the location. Since 1991, the fCO2 measurements were all
ARTICLE IN PRESS
-66
-60
-50
-20
-30
-40
20
La Runion
Crozet
Kerguelen
Africa
Antarctica
Tracks of the MINERVE and OISO cruises
in the South-Western Indian Ocean (1991-2007)
Amsterdam
30 40 50 60 70 80 90
Fig. 1. Cruises (MINERVE and OISO) conducted in the south-western Indian Oceanover the period 19912007.
N. Metzl / Deep-Sea Research II 56 (2009) 607619608
8/11/2019 2009 - Decadal Increase of Oceanic Carbon Dioxide in Southern Indian Ocean Surface Waters (19912007)
3/13
corrected to in-situ SST using polynomials given by Copin-
Montegut (1988, 1989). Based on an international at-sea inter-
comparison, the oceanicfCO2data used in this study are accurate
to about 72matm (Koertzinger et al., 2000).In order to investigate the decadal trend of oceanic fCO2, it is
important to evaluate the long-term accuracy of the measure-
ments made on board and associated data-processing used over
more than 15 years. The atmospheric CO2concentrations observedon board were regularly compared with those continuously
monitored at the station established since 1981 on the La
Nouvelle Amsterdam (Gaudry et al., 1983), a French island located
at 371480S771320E in the south-western Indian Ocean. Fig. 2
shows all the atmospheric CO2concentrations measured on board
south of 201S and those recorded at the La Nouvelle Amsterdam
since 1991 (RAMCES data communicated by M. Ramonet, LSCE/
IPSL). The cruises were conducted over a relatively large
latitudinal range (20601S) both near or far from the continents.
This explains why the atmospheric CO2 observations on board
show larger variability compared to the La Nouvelle Amsterdam
record (in this region the seasonality of atmospheric CO2 is very
low, if not the lowest in the world). The differences between the
ship-board measurements and those at the La Nouvelle Amster-dam have been interpreted as real signals, and high concentra-
tions are generally related to air mass trajectories coming from
land especially in the region 201301S influenced by the African
continent and Madagascar. For example, in January 2004 the high
atmospheric CO2 concentrations (4376 ppm, Fig. 2) were re-
corded around 201S401E in the Mozambique Channel. At higher
latitudes, atmospheric CO2 concentrations present much lower
variability (Metzl et al., 2006). Despite these differences, the mean
atmospheric CO2annual growth rate deduced from all individual
ship-board observations, +1.722 (70.004) ppm/yr over the period
19912007, is almost equal to that, +1.701 (70.003) ppm/yr
deduced from the continuous La Nouvelle Amsterdam data. It is
also comparable to the trends deduced from the 2D-Global-View
data set (1.711.73ppm/yr at latitude 20901S) (GLOBALVIEW-
CO2, 2007). The ship-board observations also showed that the
atmospheric CO2 growth rate increased significantly in recent
years from +1.64 (70.01) ppm/yr for 19912000 to +1.94
(70.01) ppm/yr for 20002007. These results derived from cruise
data are coherent with the changes evaluated at the global scale
based on high quality atmospheric observations (Canadell et al.,
2007). During the last OISO cruise in January 2007, the average
atmospheric CO2 in the Southern Indian Ocean was 380.3
(70.7) ppm, i.e. about 100 ppm higher than the preindustrial
value of 280 ppm. The comparisons of atmospheric CO2 trends
described above reflects the long-term quality of the data
obtained on board since 1991. As the same instrument (NDIR)
and its calibration were used for both atmospheric and oceanic
CO2 measurements, one could be confident to interpret the
oceanic fCO2 trends over the period 19912007. For comparingoceanic and atmosphericfCO2trends, a value of +1.7matm/yr willbe used as a reference in the atmosphere (trend estimated after
converting atmosphericxCO2data to fCO2at standard pressure).
3. OceanicfCO2 decadal variations
3.1. The oceanic fCO2 trend in the South Indian Ocean
In order to detect the oceanic fCO2trend, the continuous data
of all cruises conducted between 1991 and 2007 have been first
averaged on 1111latitudelongitude grids (this is the scale that
has been selected for constructing global Surface Ocean CO2Atlas
in the future, SOCAT project as discussed during the SOCOVVmeeting; IOCCP, 2007). In doing so, small-scale variability is
filtered out. In addition, during most cruises the sea-surface
continuous measurements were occasionally maintained at the
same location during several hours or days. Therefore the 1111grid product (hereinafter noted /fCO2S) helps to construct a
uniform weighted data set to analyze the long-term changes.
Because atmospheric pressure varies on seasonal and inter-annual
scales as well as between low and high latitudes, all /fCO2S
values have been normalized at standard pressure of 1013hPa
(hectoPascal 1 atmosphere), and trends of oceanic fCO2 can be
directly compared to the one estimated in the atmosphere
(+1.7matm/yr).To start the trend analysis, all /fCO2S values obtained since
1991 in the South Indian Ocean (20691S, 30901E), for all cruises
and seasons, are represented as a function of time inFig. 3(i.e. the
same way as for atmospheric data,Fig. 2). This overall view leads
to an average increase of ocean fCO2 of +2.11 (70.07)matm/yr(Fig. 3A), higher than in the atmosphere.
Since surface oceanicpCO2measurements started in the 1960s
(Takahashi, 1961; Keeling et al., 1965; Miyake and Sugimura,
1969), it is well known that the range of ocean pCO2is large, with
minimum/maximum values varying between 150 and 550matm inthe open ocean (Takahashi et al., 2002). In the South Indian Ocean,
the values range between 250 and 450 matm (Fig. 3A). The range(maximumminimum) of fCO2 in the South Indian Ocean varies
from 60matm (October 1996) to 140matm (January 2007). Despitethis large variability (spatial, seasonal and interannual), the
observations in the Southern Indian Ocean do show a positive
trend infCO2. Another estimate of the trend can be deduced froma direct comparison of/fCO2S values averaged for the first and
last cruises conducted in the South Indian Ocean. In January
February 1991 the mean of/fCO2S was 336 (718)matm against370 (730)matm in January 2007. This corresponds to an increaseof +34matm over 16 years (or +2.12matm/yr). The trendsestimated from all data and from the difference of two cruises
(austral summer 1991 versus 2007) are almost the same
(+2.1matm/yr). This suggests that since 1991, surface ocean fCO2increased faster than atmospheric CO2 (2.1matm/yr against1.7matm/yr) and consequently the driving force for ocean carbonsink would be reduced by 0.4matm/yr.
The observed CO2 increase at the sea surface should result
from anthropogenic uptake through air-to-sea gas exchange if the
oceanic processes such as thermodynamic, dynamic and biologi-cal activity are in steady state at decadal scale. However, based on
ARTICLE IN PRESS
350
355
360
365
370
375
380
385
1990
Year
xCO2(ppm) Ship Obs. Trend: 1.722 (+0.004) ppm/year
Ams. Station Trend: 1.701 (+0.003) ppm/year
1992 1994 1996 1998 2000 2002 2004 2006 2008
Fig. 2. Atmospheric CO2 concentrations (ppm) observed since 1991 on board in
the South Indian Ocean, south of 201S (open circles, bold line) and continuously
monitored at Nouvelle Amsterdam Island, 381S/771E, (black dots, thin line). Lineartrends and standard deviations calculated for both data sets are indicated.
N. Metzl / Deep-Sea Research II 56 (2009) 607619 609
8/11/2019 2009 - Decadal Increase of Oceanic Carbon Dioxide in Southern Indian Ocean Surface Waters (19912007)
4/13
the same series of cruises, it is observed that sea-surface
temperature (SST) decreased at an average rate of 0.11
(70.03) 1C/yr in the South Indian Ocean. The fCO2 normalized at
a constant temperature (11.76 1C is the mean of all observations)
increased at a rate of 3.47 (70.43)matm/yr (Fig. 3B). The morethan 1 1C/decade cooling makes the thermodynamic effect onfCO2significant and without the cooling the underlying oceanic fCO2increase would be even greater. This result is obtained when
normalizing fCO2 at 20 or 5 1C. Dynamic and/or biological
processes must play a role in order to increase fCO2 in surface
waters and balance the cooling effect.
It is likely that the Southern Indian Ocean, from the subtropics
to the Antarctic zone, experienced regional forcing and circulation
variability at different latitudes and that different processes might
explain the decadalfCO2variations at a regional scale, as has beenshown in the North and Equatorial Pacific Oceans (Inoue et al.,
1995; Feely et al., 2006; Takahashi et al., 2006). In order to
understand the processes that control the fCO2 trends and the
evolution of the ocean carbon sink one has to analyze these data
at a regional scale.
3.2. Oceanic fCO2 regional distributions and changes
The south-western Indian Ocean is characterized by several
dynamic (e.g., frontal systems, gyres) and biogeochemical features
that create large spatial and temporalfCO2variability which have
been described in previous studies (Poisson et al., 1993; Metzl
et al., 1995, 1998, 1999, 2006; Jabaud-Jan et al., 2004). To helpin selecting the regions to investigate the decadal fCO2variations,
a short description of the seasonal distribution is presented here
for two different years. The continuous fCO2 measurements are
presented in Fig. 4 as a function of SST from four cruises
conducted in 1991 and 2000 during the same months (January
February and August).
For the austral summer, the fCO2/SST relationship in the
subtropical and sub-Antarctic zone (1527 1C) presents the same
structure in 1991 and 2000, but is shifted towards higherfCO2in
2000 (Fig. 4A). For example, in warm waters (SST420 1C) the
average fCO2was 370 (713)matm in 2000 against 347 (78)matmin 1991. The 23matm increase in fCO2 over 9 years is larger thanthe atmospheric increase (15matm over the same period). Duringaustral winter (Fig. 4B), the fCO2/SST relations decrease with SST
in both warm and cold waters, and are almost the same in 1991
and 2000 but shifted towards higherfCO2in 2000. In warm waters(SST4141C) the average fCO2was 318 (76)matm in August 2000against 302 (78)matm in August 1991. The increase of oceanicfCO2 based on winter data is 16matm, almost equal to theatmospheric increase.
In the cold waters, south of the sub-Antarctic region around
401S, the fCO2/SST relationship is also well expressed in winter
(Fig. 4B). The data show that fCO2 was higher in 2000 than in
1991. In the temperature range 26 1C, the average fCO2 was 370
(76)matm in August 2000 and 347 (711)matm in August 1991.The average SST was almost the same (3.4 1C in 1991 and 3.5 1C in
2000). Considering that biological activity is low and not changing
dramatically from year to year during austral winter, the fCO2increase of 23matm in the cold waters is likely a signature of
anthropogenic CO2but changes of dynamical processes could alsoexplain this rapid increase of +2.6matm/yr.
ARTICLE IN PRESS
trend = + 2.11 (0.07) atm/yr
250
270
290
310
330
350
370
390
410
430
450
1990
Year
fCO2
(atm)
trend = + 3.47 (0.43) atm/yr
100
200
300
400
500
600
700
1990
Year
fCO2
norm(
atm)
1992 1994 1996 1998 2000 2002 2004 2006 2008
1992 1994 1996 1998 2000 2002 2004 2006 2008
Fig. 3. Evolution of (A) surface water fCO2and (B) temperature normalized fCO2(at SST 11.761C, mean average of all data) based on all measurements conducted in the
South Indian Ocean (south of 201S) during the period 19912007. Each point corresponds to monthly average of the continuous measurement in a 1 111 la/long grid. All
seasons, covering different months are reported in these figures. Without selecting seasons and regions, the data lead to annual increasing rates of +2.11 and +3.47 matm/yrfor fCO2and fCO2
norm, respectively.
N. Metzl / Deep-Sea Research II 56 (2009) 607619610
8/11/2019 2009 - Decadal Increase of Oceanic Carbon Dioxide in Southern Indian Ocean Surface Waters (19912007)
5/13
As opposed to the winter observations, there is no clearfCO2/SST relationship during summer in cold waters at high
latitude (Fig. 4A). This is because the fCO2variability is large andmostly controlled by biological activity, which imprints large fCO2meso-scale gradients in polar waters. In this context, detecting
decadal fCO2 variations could be particularly difficult when
comparing only two cruises in summer. Inoue and Ishii (2005)
reported the same difficulty in the eastern sector of the Southern
Indian Ocean. One has to be very careful when selecting summer
data in order to detect long-term changes.
It is clear that direct cruise to cruise comparisons (1991 versus
2000) cannot lead to firm conclusions regarding the decadal
trends of oceanic fCO2. Indeed, it has been shown that for austral
summer interannual variability of fCO2 is large in the Southern
Indian Ocean (Jabaud-Jan et al., 2004; Inoue and Ishii, 2005;
Breviere et al., 2006; Borges et al., 2008). However, the results
presented inFig. 4suggest that both summer and winter data canbe used to evaluate the long-term trend of oceanic fCO2 over a
large region in the Indian Ocean, from the subtropics to the
Southern Ocean. This is now analyzed in more detail using all
observations in this region in summer and winter since 1991.
3.3. Regional analysis of oceanic fCO2 decadal trends
Based on the spatio-temporal fCO2 distribution, the dynamic
and biogeochemical characteristics of the Southern Indian Ocean
and recent knowledge of large-scale atmospheric and oceanic
changes, the decadalfCO2trends will be analyzed for four regions
(20351S, 35401S, 40421S and 50551S;Fig. 4). The variations of
SST could have a significant effect on fCO2. For this reason, thenormalized fCO2at constant temperature will be also investigated
(fCO2norm). In addition, because the seasonality is large (Fig. 4) and
cruises were not conducted for summer and winter every year, the
fCO2trends will be estimated independently for each season usingthe 1111monthly mean data described in Section 3.1 (/fCO2S,
/fCO2norm
S and /SSTS). Results are presented inFigs. 58for each
region andTable 1gives the values of trends and errors estimated
for SST, fCO2 and fCO2norm.
3.3.1. The subtropical zone, 20351S
The data in the subtropical region (20351S/40801E) have
been selected as oligotrophic warm waters of the western Indian
subtropical gyre, with SST420 1C in summer and SST4141C in
winter. A significant fCO2 increase over time is expected, as
observed in the subtropical zones of the North Atlantic and Pacific
Oceans (e.g.,Takahashi et al., 1983, 2006;Inoue et al., 1995;Bates,
2001). There is some evidence for this in the South Indian gyrewhen comparing the data from years 1991 and 2000 (Section 3.2,
Fig. 4A, B). Over the full period 19912007, the seasonality offCO2is well marked each year with higher fCO2 in austral summer
(Fig. 5B). This is clearly associated with the seasonality of SST
(Fig. 5A). For each season, the oceanic fCO2increase over 16 years
is also observed. The average growth rates are +1.8matm/yr forwinter and +2.2matm/yr for summer (Table 1), i.e. near or abovethe rate of atmospheric CO2 increase. Although the fCO2 appears
relatively stable for some periods (e.g., summer 19982002), the
trends estimated over 16 years suggest that the driving force for
the ocean sink (DfCO2 fCO2sea
fCO2atm) is decreasing. These
observations also suggest a slight decrease of temperature, almost
the same in summer and winter, of 0.03 and 0.04 1C/yr,
respectively. Therefore for both seasons the rate of increase forfCO2
norm is slightly higher than for fCO2 but not statistically
ARTICLE IN PRESS
250
300
350
400
-2
SST (C)
fCO2(atm)
Jan 1991
Jan 2000
20-35S35-40S40-42S50-55S
250
300
350
400
-2
SST (C)
fCO2(atm)
Aug 1991
Aug 2000
20-35S35-40S40-42S50-55S
2 6 10 14 18 22 26 30
2 6 10 14 18 22 26 30
Fig. 4. Continuous oceanfCO2measurements versus SST observed in the south-western Indian Ocean in (A) January and (B) August for years 1991 and 2000. The arrows
indicate SST range in the four zones selected to evaluate and discuss thefCO2 trends at regional scale.
N. Metzl / Deep-Sea Research II 56 (2009) 607619 611
8/11/2019 2009 - Decadal Increase of Oceanic Carbon Dioxide in Southern Indian Ocean Surface Waters (19912007)
6/13
different (Table 1). Although the SST decrease is shown for both
seasons, and could be related to an advective process from the
south (import of high DIC and low alkalinity), the variations are
small and quantification of such small changes is not in hand at
present. In the subtropical Indian gyre, the data suggest that the
increase of fCO2 is mainly explained by progressive invasion of
anthropogenic CO2.
3.3.2. The transitional zone, 35401S
The region between 351S and 401S in the south-western Indian
Ocean represents a transitional zone bordered by the Aghulas
Frontal system and its associated return flow and the sub-
Antarctic front (e.g., Lutjeharms and Valentine, 1984;Belkin and
Gordon,1996). This region separates the oligotrophic warm watersof the subtropics and the more productive waters of the sub-
Antarctic zone and is characterized by large meridional fCO2gradients in all seasons (Fig. 4). In summer, for 15 1CoSSTo20 1C,fCO2 decreases sharply southward (gradients of 100matm arecommon, Fig. 4A). During winter, the SST decreases southward
from 16 to 13 1C andfCO2increases to the south. Gradients offCO2+50matm are regularly observed over the zone in winter, Fig. 4B).The SST presents a clear seasonality (Fig. 6A) but the fCO2seasonality is low (Fig. 6B). As opposed to the subtropical region,
this indicates that SST is not the main factor controlling the
seasonalfCO2 cycle south of 351S. The averaged data also clearly
reveal that fCO2 variability is much larger during summer than
winter. This is observed for all cruises and is probably related to
high primary productivity and shallow mixed-layers in summer,whereas deeper mixing and low biological production in winter
result in homogeneous biogeochemical properties in surface
water (DIC and fCO2). The comparison of continuous measure-
ments between 1991 and 2000 (Fig. 4) suggested a significant
fCO2 increase over 9 years. The full data set also shows a gradualfCO2 increase from 1991 to 2007 of +2.3 matm/yr for summer and+1.5matm/yr for winter (Fig. 6B, Table 1). These are about thesame rates as for the subtropical region. However, because the
data also recorded stable SST in summer (+0.01 1C/yr), but a
warming during winter (+0.07 1C/yr), the increasing rates of
temperature-normalized fCO2 are significantly different for sum-
mer (+2.1matm/yr) and winter (+0.6matm/yr). This suggests thatsince 1991 the oceanic processes that control surfacefCO2(as well
as DIC and perhaps alkalinity) were the same in summer, but
something changed during winter.
To balance the temperature effect that increased winter fCO2one possible scenario is an increase in primary productivity
during the winter months. Indeed, the observed warming in
winter could be related to a decrease of vertical mixing (related to
weaker winds), less import of DIC into the surface water and
increased productivity in illuminated surface waters during
winter (when nutrients are not limiting). Interestingly, based on
ocean-color satellite observations (CZCS 19791983 versus Sea-
WIFS 19982002), significant variations of the Chlorophyll-a
concentrations have been evaluated in the south-western Indian
region at these latitudes during winter but not in other seasons
(Antoine et al., 2005). It is also worth noting that the comparison
of the Chlorophyll-aconcentration in surface water and analysis of
the fCO2 trends are not for the same periods (19791983 vs.
19982002, and 19912007). Unfortunately, satellite color dataare not available for the period 19911997 and the period
ARTICLE IN PRESS
20-35S, 40-80E
280
330
380
430
480
1990
Year
fCO2(atm)
Dec-Feb
Jul-Sep
20-35S, 40-80E
10
15
20
25
30
1990
Year
SST(C)
Dec-Feb
Jul-Sep
20-35S, 40-80E
250
300
350
400
450
500
1990
Year
fCO2
norm(
atm)
Dec-Feb
Jul-Sep
1992 1994 1996 1998 2000 2002 2004 2006 2008
1992 1994 1996 1998 2000 2002 2004 2006 2008
1992 1994 1996 1998 2000 2002 2004 2006 2008
Fig. 5. Evolution of (top) sea-surface temperature (SST), (middle) sea-surface fCO2 and (bottom) temperature normalized fCO2 in the subtropical region of the south-
western Indian Ocean (20351S/40801E). The data obtained during 19912007 are selected only for summer cruises (DecemberFebruary, open circles) and winter cruises
(JulySeptember, filled circles). Linear trends are indicated for summer (dashed line) and winter (black line) and the trends are given in Table 1.
N. Metzl / Deep-Sea Research II 56 (2009) 607619612
8/11/2019 2009 - Decadal Increase of Oceanic Carbon Dioxide in Southern Indian Ocean Surface Waters (19912007)
7/13
19982007 is short to analyze the impact of the primary
productivity changes on decadal fCO2 variations.
3.3.3. The sub-Antarctic zone, 40421S
The sub-Antarctic zone, the region bordered by the subtropical
front and the sub-Antarctic front, is characterized by large spatio-
temporal variability of fCO2, but on average the seasonality is
relatively well identified (Fig. 7). During summer fCO2is low due to
biological activity. In winter fCO2is higher due to deep mixing with
subsurface waters enriched in CO2 (Metzl et al., 1999). In order to
investigate the long-term trends, the data have been selected in a
relatively narrow band, 40421S, wherefCO2reaches a minimum insummer and is higher and rather constant in winter (Fig. 4). In this
region the fCO2 trends estimated for both seasons are almost the
same, +2.2matm/yr, which is faster than in the atmosphere. Asopposed to the regions north of 401S, the observations show a cooling
of surface waters for both seasons, 0.05 1C/yr in summer and
0.15 1C/yr during winter. This leads to a dramatic seasonal contrast
when comparing the decadal trends for fCO2norm (+2.9matm/yr in
summer and +4.7matm/yr in winter). The effect of cooling on fCO2must be offset by an increase of DIC, that could be associated with
deeper mixed-layers and higher winds speeds observed in this region
in recent years (in positive SAM period,Marshall, 2003).
3.3.4. The Polar zone, 50551S
The Polar Front occurs at about 501S in the south-westernIndian Ocean. Summer cruises reached as far south as 561S
(sometimes up to 691S), with winter cruises only reaching 561S on
a few occasions. Therefore, the decadal variations offCO2 will be
estimated over 50551S latitudinal band for summer and over
50521S for winter. Note that if one uses the summer data in the
band 50521S only, results are not dramatically different from
those for 50551S. In addition, the observations obtained in winter
showed that oceanic surface properties (fCO2, DIC, TA, nutrients)
were fairly constant over large distance in the region 50561S
(Metzl et al., 2006). Therefore the trends estimated in the band
50521S in winter are almost certainly representative of the
decadal changes in the POOZ (Permanent Open Ocean Zone). Data
over the Kerguelen Plateau (around 50521S), where strong
blooms create large fCO2 drawdown (Blain et al., 2007; Jouandetet al., 2008), have been filtered out to analyze the long-term
trends. Measurements in coastal waters around Kerguelen Island
are also excluded.
In the Southern Ocean, south of the Polar Front, it is now well
established thatfCO2is higher in winter than in summer (Metzl et
al., 2006;Takahashi et al., 2009). As in the SAZ, this is because the
effect of biological activity in summer, and vertical mixing in
winter dominate the effect of temperature changes on fCO2. The
seasonal amplitude is relatively small (about 20matm, Fig. 8B),much smaller than in the SAZ, and at these high latitudes the
ocean is not far from equilibrium with respect to atmospheric CO2.
The data in this analysis show that the average rate of fCO2increase of +2.4matm/yr in summer and +2.1matm/yr in winter,
are higher than in the atmosphere (+1.7 matm/yr). The rates ofincrease are similar for the SAZ (40421S, where the oceanfCO2is
ARTICLE IN PRESS
35-40S, 30-90E
250
300
350
400
450
fCO2(atm)
Dec-Mar
Jun-Oct
35-40S, 30-90E
10
15
20
25
SST(C)
Dec-Mar
Jun-Oct
35-40S, 30-90E
250
300
350
400
450
1990
Year
fCO2
norm(
atm)
Dec-Mar
Jun-Oct
1992 1994 1996 1998 2000 2002 2004 2006 2008
1990
Year
1992 1994 1996 1998 2000 2002 2004 2006 2008
1990
Year
1992 1994 1996 1998 2000 2002 2004 2006 2008
Fig. 6. Evolution of (top) sea-surface temperature (SST), (middle) sea-surfacefCO2and (bottom) temperature normalizedfCO2in the transitional zone of the south-western
Indian Ocean (35401S/30901E). The data obtained during 19912007 are selected only for summer cruises (DecemberMarch, open circles) and winter cruises
(JuneOctober, filled circles). Linear trends are indicated for summer (dashed line) and winter (black line) and the trends are given in Table 1.
N. Metzl / Deep-Sea Research II 56 (2009) 607619 613
8/11/2019 2009 - Decadal Increase of Oceanic Carbon Dioxide in Southern Indian Ocean Surface Waters (19912007)
8/13
well below equilibrium in summer, Figs. 4 and 7). As in the SAZ,
the SST observations in the POOZ present different trends for
summer and winter (Fig. 8A). A warming of about +0.02 1C/yr is
deduced from the summer data (19912007), whereas winter
data suggest a cooling of0.11C/yr for the period 19912000. This
rapid cooling may be caused by high SST observed in JuneJuly
1993 (Fig. 8A). However, when data for JuneJuly 1993 are filtered
out, the cooling in winter is estimated as 0.06 1C/yr. This leads to
a rate of increase for fCO2norm in winter between +3.6 and
+3.4matm/yr (depending the data selected, with or without1993 observations). Like in the SAZ, the thermal effect should be
balanced by an import of CO2 in surface waters, most likely
through a change in vertical mixing generated by a recent increaseof winds at high latitudes in the Southern Ocean.
4. Discussion
4.1. Evolution of the ocean CO2 sink in the South Indian Ocean
The regional analysis of thefCO2trends estimated in the south-
western Indian Ocean is summarized inFig. 9. For all regions and
seasons south of 201S the oceanic fCO2 increased during
19912007, and the oceanic growth rate is everywhere close to
or larger than in the atmosphere. The four regions analyzed using
summer and winter data lead to an average oceanicfCO2 increase
of 2.10 (70.3)matm/yr. This value is similar to the one derivedwhen using all observations (2.11matm/yr discussed in Section 3.1,
Fig. 3), suggesting this is a robust result. The increase corresponds
to a decadal change of +21 (73)matm/decade. The data obtainedsince 1991 suggest that DfCO2decreased by about 4 matm/decadein the south-western Indian Ocean. For a CO2 sink estimated as
0.4 to 0.7PgC/yr in the South Indian Ocean south of 201S
(Metzl et al. 1995; Takahashi et al., 2002, 2009), the decadal
change ofDfCO2would correspond to a reduction of the ocean CO2sink between 0.08 and 0.14 PgC/yr/decade. This estimate for the
Southern Indian Ocean gives only a measure of the potential
impact of the DfCO2change in terms of the integrated carbon flux
at the basin-wide scale and is similar to a 0.08 PgC/yr/decade
estimate for the Southern Ocean (4451S) for the period
19812004 based on atmospheric data (Le Quere et al., 2007).For an ocean modelling approach,Lenton and Matear (2007)and
Lovenduski et al. (2007) both concluded a reduction of the
Southern Ocean carbon sink would be around 0.1PgC/decade per
unit of change of the SAM. The numbers given here are not
directly comparable (South Indian versus Southern Ocean), but
offer some sense of how a change in observed DfCO2 would
translate in the global carbon budget.
4.2. A link between observed fCO2 decadal trends and the SAM?
In the south-western Indian Ocean, significant warming has
been observed in winter at 35401S, whereas south of 401S the
data indicate winter-time cooling since 1991 (Table 1). These
characteristics promote a strong regional contrast in the annualgrowth rates forfCO2andfCO2
norm in winter (comparison ofFig. 9A
ARTICLE IN PRESS
40-42S/ 30-80E
250
300
350
400
fCO2(atm)
Dec-Mar
Jun-Sep
40-42S/ 30-80E
5
10
15
20
25
1990
Year
SST(C)
Dec-Mar
Jun-Sep
40-42S/ 30-80E
200
250
300
350
400
450
500
fCO2
norm(
atm) Dec-Mar
Jun-Sep
1992 1994 1996 1998 2000 2002 2004 2006 2008
1990
Year
1992 1994 1996 1998 2000 2002 2004 2006 2008
1990
Year
1992 1994 1996 1998 2000 2002 2004 2006 2008
Fig. 7. Evolution of (top) sea-surface temperature (SST), (middle) sea-surfacefCO2and (bottom) temperature normalizedfCO2in the sub-Antarctic zone of the south-western
Indian Ocean (40421S/30801E). The data obtained during 19912007 are selected only for summer cruises (DecemberMarch, open circles) and winter cruises
(JuneSeptember, filled circles). Linear trends are indicated for summer (dashed line) and winter (black line) and the trends are given in Table 1.
N. Metzl / Deep-Sea Research II 56 (2009) 607619614
8/11/2019 2009 - Decadal Increase of Oceanic Carbon Dioxide in Southern Indian Ocean Surface Waters (19912007)
9/13
and B). At 35401S the annual increase of fCO2norm is low,
+0.6matm/yr, whereas south of 401S the growth rate of fCO2norm
reaches 3.64.7matm/yr in winter. In order to balance thetemperature effect, there must be a decrease of CO2 in surface
water north of 401S and an increase of CO2 south of 401S.
The positive trend of the SAM in recent years (especially since
the 1990s, Marshall, 2003) may explain the observed decadal
trends in fCO2 and fCO2norm over 19912007, as it has been
recognized for physical properties and circulation changes.
A recent analysis based on altimetry data (19932005) showed
that in the south-western Indian Ocean (20801E the regioninvestigated here), the sea-level rise decreases south of 401S and
increases north of 401S(Morrow et al., 2008). This is interpreted
by stronger upwelling and cooling south of 401S and warming in
the northern region, as observed in our data set during austral
winter. Morrow et al. (2008) also have investigated the link
between the sea-level anomaly (SLA) and climate index (SAM and
ENSO), and they estimate that SLA was positively correlated with
the SAM during 19932005 in the south-western Indian Ocean
(30901E). In a positive SAM state, stronger westerly winds south
of 401S lead to surface ocean cooling and increase the physical
mixing (e.g., Sen Gupta and England, 2006) and would increase
the import of CO2 in surface waters. The opposite is observednorth of 401S where weaker westerly winds increase SST (increase
ARTICLE IN PRESS
50-55S summer, 50-52S winter, 40-70E
0
2
4
6
8
SST(C)
Dec-Mar
Jun-Aug
50-55S summer, 50-52S winter, 40-70E
300
320
340
360
380
400
420
fCO2(atm)
Dec-Mar
Jun-Aug
50-55S summer, 50-52S winter, 40-70E
300
320
340
360
380
400
420
1990
Year
fCO2
norm(
atm)
Dec-Mar
Jun-Aug
1992 1994 1996 1998 2000 2002 2004 2006 2008
1990
Year
1992 1994 1996 1998 2000 2002 2004 2006 2008
1990
Year
1992 1994 1996 1998 2000 2002 2004 2006 2008
Fig. 8. Evolution of (top) sea-surface temperature (SST), (middle) sea-surface fCO2and (bottom) temperature normalizedfCO2in the Permanent Open Ocean Zone (POOZ)
of the south-western Indian Ocean (50551S/40701E). The data obtained during 19912007 are selected only for summer cruises (DecemberMarch, open circles) and
winter cruises (JuneAugust, filled circles). Linear trends are indicated for summer (dashed line) and winter (black line) and the trends are given in Table 1.For summer datahave been selected in the latitudinal band 50551S and for winter in the band 50521S. See text for explanation of data selection.
Table 1
Annual trends and standard deviations of sea surface temperature (SST), fCO2and temperature normalized fCO2in selected regions of the south-western Indian Ocean, for
summer and winter..
Zone/period SST (1C/yr) fCO2 (matm/yr) fCO2norm (matm/yr) SST norm (1C)
20351S DecemberFebruary 0.03 (0.03) +2.18 (0.18) +2.54 (0.46) 23.90
20351S JulySeptember 0.04 (0.04) +1.75 (0.10) +2.28 (0.58) 18.75
35401S DecemberMarch +0.01 (0.03) +2.35 (0.28) +2.14 (0.30) 18.50
35401S JuneOctober +0.07(0.02) +1.52 (0.15) +0.62 (0.45) 14.30
40421S DecemberMarch 0.05 (0.04) +2.24 (0.32) +2.92 (0.76) 14.95
40421S JuneSeptember 0.15 (0.06) +2.18 (0.22) +4.72 (1.03) 11.09
50551S DecemberMarch +0.02 (0.01) +2.39 (0.16) +2.04 (0.26) 3.72
50521S JuneAugust 0.10 (0.03) +2.10 (0.26) +3.62 (0.43) 2.85
Based on observations conducted during 19912007 in all regions expect for 50521S in JuneAugust (period 19912000). Last column indicates SST used to normalizefCO2for each region and season.
N. Metzl / Deep-Sea Research II 56 (2009) 607619 615
8/11/2019 2009 - Decadal Increase of Oceanic Carbon Dioxide in Southern Indian Ocean Surface Waters (19912007)
10/13
SLR) and reduce the mixed-layer depth. Consequently during
positive SAM state, less DIC would be imported in surface layers
north of 401S. At large scale, this scenario has been proposed by
several authors, based on ocean carbon models, to interpret the
inter-annual to decadal CO2 airsea flux variability (Lenton and
Matear, 20 07;Le Quereet al., 2007;Lovenduski et al., 2007;Verdy
et al., 2007). The data presented in this study support the
modelled predictions although the analysis only focuses on the
south-western Indian Ocean.
4.3. Comparison with other ocean sectors
Compared to other oceanic sectors, the average growth rate
estimated in the south-western Indian Ocean for the period
19912007 is generally higher than those reported in other oceans
(Table 2, here quoting several studies based on multi-year
observations). The decadal variability of fCO2, or any other sea-
surface property, also depends on the period investigated. For
example, in the Equatorial Pacific, Feely et al. (2006)note that the
growth rate of ocean fCO2 was faster in recent decades (Table 2).
This clearly indicates that one could observe significant differences
in the decadal oceanic fCO2 growth rate depending on the
investigated period and region. It is also important to note that
few studies investigated the oceans in the Southern Hemisphere.Inoue and Ishii (2005) analyzed the long-term observations
of pCO2 in the Southern Ocean from a series of four cruises
conducted south of Tasmania (at 1401601E) in austral summer of
years 1969, 1984, 1995 and 2002. They deducedpCO2growth rates
that vary between +10 (75) and +15 (74)matm/decade in the SAZand Polar Front zone. The interannual variability could be large in
this sector, especially during austral summer (Breviere et al.,
2006) and could create difficulties to detect the trends. A recent
data synthesis obtained in the south-western Pacific (20551S,
1401801E) indicates that oceanic pCO2increased at a rate of +14.4
(73.0)matm/decade during 19842006 (Takahashi et al., 2009).
ARTICLE IN PRESS
0.0
0.5
1.0
1.5
2.0
2.5
3.0
AnnualRatefCO2s(atm/yr) 20-35S 35-40S 40-42S 50-55S
20-35S 35-40S 40-42S 50-55S
0.0
1.0
2.0
3.0
4.0
5.0
6.0
AnnualRatefCO2n
orm(
atm/yr)
Fig. 9. Annual mean trends of (A) sea-surface fCO2 and (B) temperature normal-
ized fCO2 in 4 regions of the south-western Indian Ocean. The open bars indicate
the growth rates estimated for summer and black bars for winter. Standard errors
associated to each trends are also indicated. In each figure, the dashed line
indicates the atmospheric CO2annual growth rate. Note that the larger scale in (B).
Table 2
Decadal trends of sea surface fCO2 observed in various regions of the oceans..
Region Period fCO2growth rate (matm/decade) Reference
Northern and tropical regions
North Atlantic
Subpolar Gyre 19932003 +18 to +28 Corbiere et al. (2007)
Subtropical 19832003 +17 (73) Bates (2007)
North Atlantic 4151N 19722006 +18 (74) Takahashi et al. (2009)
North Pacific 19702004 +12 (74.8) Takahashi et al. (2006)
West. North Pacific 19841993 +12 (79) Inoue et al. (1995)
Equatorial Pacific 19902003 +20 (72) Ishii et al. (2004)
Central Equ. Pacific 19751990 11 (710) Feely et al. (2006)
Central Equ. Pacific 19902005 +17 (77) Feely et al. (2006)
Southern Hemisphere
South-western Pacific
(451501S, SAZ) 19692004 +10 (75) Inoue and Ishii (2005)
(501551S, PFZ) 19692004 +15 (74) Inoue and Ishii (2005)
South-western Pacific (201551S) 19842004 +14.4 (73.0) Takahashi et al. (2009)
South-western Indian (201S421S) 19912007 +20.4 (73.3) This study, average summer/winter
Southern Ocean South of the Polar Front
Pacific 1401701E
Summer 19692004 +18 (72) Inoue and Ishii (2005)
Circumpolar
SST o6.5 1C winter 19862006 +21 (74) Takahashi et al. (2009)
Indian 30901E
Summer (50551S) 19912007 +23.9 (71.6) This study
Winter (50521S) 19912000 +21.0 (72.6) This study
This table presents some examples of the results based on regular monitoring and/or data synthesis.
N. Metzl / Deep-Sea Research II 56 (2009) 607619616
8/11/2019 2009 - Decadal Increase of Oceanic Carbon Dioxide in Southern Indian Ocean Surface Waters (19912007)
11/13
The oceanic fCO2 increase of +20.4 (73.3)matm/decade in thesouth-western Indian Ocean (this study, 30901E/20421S) is faster
than values reported for the south-western Pacific. Interestingly, it
has been observed that since 1993, sea-level rise (SLR) decreased in
the south-western Indian Antarctic sector (stronger upwelling and
cooling), whereas the SLR increased in the south-western Pacific
(Morrow et al., 2008). The difference of decadal trend offCO2(as for
the SLA) could be explained by the contrasting response of theSouth Indian and South Pacific Oceans to the SAM and ENSO
variability (Morrow et al., 2008).
At higher latitudes, in the cold waters of the Antarctic Ocean to
the south of the Polar front, the trend in the south-western Indian
sector south was higher (+24 (72)matm/decade) in summer thanthe pCO2 trend of +18 (72)matm/decade reported by Inoue andIshii (2005). As for the SAZ, this could be explained by the regional
response of the ocean to the atmospheric variability, or because
the period investigated are different.
Takahashi et al. (2009) investigated the decadal change of
oceanic pCO2 in the circumpolar waters at high latitudes, using
only winter observations. For SST below 6.5 1C they evaluated that
the oceanic pCO2 increased at a rate of +21 (74)matm/decade
during the period 19862006. This is the same value as that forwinter data south of 501S in the south-western Indian Ocean since
1991, +21.0 (72.6)matm/decade. Takahashi et al. (2009) includedalmost all the winter data analyzed in this paper as well as several
other data sets, but they used another approach to detect the
trends (by evaluating trends for different SST ranges using all data
in circumpolar waters). This comparison is thus encouraging and
demonstrates that the oceanic pCO2 (or fCO2) increased faster
than in the atmosphere at high latitudes in several regions of the
Southern Ocean for the past two decades.
5. Conclusions
The decadal trends of oceanic fCO2have been investigated for
the first time in the south-western Indian Ocean based on
observations conducted during 19912007. The shipboard data
show an annual atmospheric CO2increase of 1.72 ppm/yr over the
period 19912007 with a more rapid increase in recent years,
+1.64 (70.01) ppm/yr for 19912000 to +1.94 (70.01) ppm/yr for
20002007. This is coherent with the analysis of atmospheric CO2at a global scale (Canadell et al., 2007).
In the ocean, the observations show a significant fCO2 increase
in all sectors of the south-western Indian Ocean, in warm and cold
waters. The mean rate is +2.11 (0.07)matm/yr, suggesting thatocean fCO2increased at about the same rate or faster than in the
atmosphere. For a constant gas transfer velocity, this would imply
an ocean carbon sink reduction of about 0.08 and 0.14PgC/yr/
decade in the South Indian Ocean.
At a regional scale, the fCO2 increase is detected for bothsummer and winter. It varies between 1.5 and 2.4 matm/yr, i.e.almost always equal or higher than in the atmosphere. The
regional analysis also shows that the annual increase of fCO2 is
rather similar in different seasons and regions (Fig. 9A), but that
the rates of normalized temperature fCO2norm present contrasting
values, north and south of 401S and especially during austral
winter (Fig. 9B). To balance the thermal effect, more CO2(DIC) has
to be imported in surface waters south of 401S, whereas CO2concentrations have to be reduced at 35401S (reduction of
vertical mixing and/or increased productivity in winter). This
meridional contrast could be explained by the change in climate
over the Southern Hemisphere associated to the upward trend of
the SAM index. This would be coherent with a recent scenario
based on modelling and suggestions that variability in theSouthern Hemisphere climate forcing (e.g., the SAM) impacts on
ocean circulation, by increasing the supply of DIC from the deep
ocean at high latitudes and reducing the oceanic carbon sink for
atmospheric CO2(Lenton and Matear, 2007;Le Quereet al, 2007;
Lovenduski et al., 2007, Verdy et al., 2007). This is the first time
that such scenario is suggested from in-situ data.
This study also shows that the growth rate of oceanic fCO2 is
higher in the subtropical and sub-Antarctic zones of the south-
western Indian Ocean (+20matm/decade) than in the south-western Pacific (+10 to +15matm/decade, Inoue and Ishii, 2005;Takahashi et al., 2009). The difference may be explained by
regional oceanic response to the SAM and ENSO since the 1990s,
as evaluated from altimetry data (Morrow et al., 2008), and
suggests that the decadal changes of the ocean carbon sink
for atmospheric CO2 may not be homogeneous around the
circumpolar region.
This work offers new findings, but clearly calls for a more
comprehensive analysis of the oceanic CO2 in the Indian and
Southern Oceans that should be conducted in the frame of the
international collaborative synthesis (IOCCP, 2007). As the data
are now available to evaluate the decadal changes of the ocean
carbon cycle, a strong interaction with the modelling community
is expected, not only to validate ocean models, but to helpunderstanding why the oceanic CO2 sink is changing and
presumably decreasing. For this issue, an important analysis
should be performed not only withfCO2data (as in this paper) but
by adding regular observations of DIC and alkalinity in surface
waters and at depth. The same is true for other oceanic regions
(e.g., the North Atlantic) where it has been shown that the oceanic
CO2sink has decreased over the past decade (Corbiere et al., 2007;
Schuster and Watson, 2007; Schuster et al., 2009). As the
reduction or saturation of the oceanic carbon sink has been
identified by several observational studies in the Northern and
Southern Oceans, this has major implications and provides a
strong motivation to pursue long-term oceanic CO2 observations
to better understand the global carbon budget, its evolution and
its coupling with climate change.
Acknowledgements
The long-term OISO observational program is supported by
three French Institutes INSU (Institut National des Sciences de
lUnivers), IPSL (Institut Pierre-Simon Laplace) and IPEV (Institut
Paul-Emile Victor). Warm thanks to the captains and crews of the
R.S.S.Marion-Dufresne, and many colleagues at the laboratories in
Paris, LPCM (Laboratoire de Physique et Chimie Marines,
19911996), LBCM (Laboratoire de Biogeochimie et Chimie
Marines, 19972004) and LOCEAN (Laboratoire dOceanographie
et du Climat: Experimentation et Approches Numeriques). This
work is part of the national program LEFE/Cyber/FlamenCO2, acomponent of SOLAS-France. Support from the European Inte-
grated Project CARBOOCEAN Contract 511176 (GOCE) is also
acknowledged. Un Grand merci to Andrew Lenton (LOCEAN/
IPSL) for his questions and comments in a previous version of this
manuscript, to Dorothee Bakker (UEA, Norwich) for a very helpful
review and Bronte Tilbrook (CSIRO, Hobart), co-editor, for
important comments on the revised manuscript.
References
Antoine, D., Morel, A., Gordon, H.R., Banzon, V.F., Evans, R.H., 2005. Bridging oceancolor observations of the 1980s and 2000s in search of long term trends.
Journal o f Geophysical Research 110 (C06009).Bates, N.R., 2001. Interannual changes of oceanic CO2 and biogeochemical
properties in the Western North Atlantic subtropical gyre. Deep-Sea ResearchII 48 (89), 15071528.
ARTICLE IN PRESS
N. Metzl / Deep-Sea Research II 56 (2009) 607619 617
8/11/2019 2009 - Decadal Increase of Oceanic Carbon Dioxide in Southern Indian Ocean Surface Waters (19912007)
12/13
Bates, N.R., 2007. Interannual variability of the oceanic CO2sink in the subtropicalgyre of the North Atlantic Ocean over the last two decades. Journal ofGeophysical Research 112.
Bates, N.R., Michaels, A.F., Knap, A.H., 1996. Seasonal and interannual variability ofoceanic carbon dioxide species at the US JGOFS Bermuda Atlantic Time-seriesStudy (BATS) site. Deep-Sea Research II 43, 347383.
Belkin, I., Gordon, A., 1996. Southern Ocean fronts from the Greenwich meridian toTasmania. Journal of Geophysical Research 101 (C2), 36753696.
Blain, S., and 46 co-authors, 2007. Effect of natural iron fertilisation on carbon
sequestration in the Southern Ocean. Nature.Borges, A.V., Tilbrook, B., Metzl, N., Lenton, A., Delille, B., 2008. Inter-annualvariability of the carbon dioxide oceanic sink south of Tasmania. Biogeos-ciences 5, 141155.
Breviere, E., Metzl, N., Poisson, A., Tilbrook, B., 2006. Changes of oceanic CO2sink inthe Eastern Indian sector of the Southern Ocean. Tellus 58B, 438446.
Canadell, J.G., Le Quere, C., Raupach, M.R., Field, C.B., Buitehuis, E.T., Ciais, P.,Conway, T.J., Gillett, N.P., Houghton, R.A., Marland, G., 2007. Contributions toaccelerating atmospheric CO2growth from economic activity, carbon intensity,and efficiency of natural sinks. Proceedings of the National Academy ofSciences 5.
Copin-Montegut, C., 1988. A new formula for the effect of temperature on thepartial pressure of CO
2in seawater. Marine Chemistry 25, 2937.
Copin-Montegut, C., 1989. A new formula for the effect of temperature on thepartial pressure of CO
2in seawaterCorrigendum. Marine Chemistry 27,
143144.Corbiere, A., Metzl, N., Reverdin, G., Brunet, C., Takahashi, T., 2007. Interannual and
decadal variability of the oceanic carbon sink in the North Atlantic subpolargyre. Tellus B 59 (2), 168179.
Dore, J.E., Lukas, R., Sadler, D.W., Karl, D.M., 2003. Climate-driven changes to theatmospheric CO2 sink in the subtropical North Pacific Ocean. Nature 424,754757.
Feely, R., Boutin, J., Cosca, C., Dandonneau, Y., Etcheto, J., Inoue, H., Ishii, M., LeQuere, C., Mackey, D., McPhaden, M., Metzl, N., Poisson, A., Wanninkhov, R.,2002. Seasonal and interannual variability of CO2 in the Equatorial Pacific.Deep Sea Research II 49 (1314), 24432469.
Feely, R.A., Sabine, C., Lee, K., Berelson, W., Kleypas, J., Fabry, V.J., Millero, F.J., 2004.Impact of anthropogenic CO2on the CaCO3system in the oceans. Science 305,362366.
Feely, R.A., Takahashi, T., Wanninkhof, R., McPhaden, M.J., Cosca, C.E., Sutherland,S.C., Carr, M.-E., 2006. Decadal variability of the airsea CO2 fluxes in theequatorial Pacific Ocean. Journal of Geophysical Research 111, C07S03.
Gaudry, A., Ascensio, J.M., Lambert, G., 1983. Preliminary study of CO2variations atAmsterdam Island (Territoire des Terres Australes et Antarctiques Francaises).
Journal of Geophysical Research 88, 13231329.GLOBALVIEW-CO2, 2007. Cooperative Atmospheric Data Integration Project
Carbon Dioxide. CD-ROM, NOAA ESLRL, Boulder, CO [Also available on Internetvia anonymous FTP toftp.cmdl.noaa.gov,Path: ccg/co2/GLOBALVIEW].
Inoue, H.Y., Ishii, M., 2005. Variations and trends of CO2in the surface seawater inthe Southern Ocean south of Australia between 1969 and 2002. Tellus B 57 (1),5869.
Inoue, H.Y., Matsueda, H., Ishii, H., Fushimi, K., Hirota, M., Asanuma, I., Takasugi, Y.,1995. Long-term trend of the partial pressure of carbon dioxide (pCO2)in surface waters of the western North Pacific, 19841993. Tellus 47B,391413.
IOCCP, 2007. Surface Ocean CO2variability and vulnerabilities workshop, UNESCO,Paris, France, April 1114, 2007, IOCCP Report, No. 7. Available Online at/http://www.ioc.unesco.org/ioccp/pCO2_2007.htmS.
IPCC, 2007. Climate change 2007: the physical science basis. In: Solomon, S., Quin,D., Manning, M., Chen, Z., Marquis, M., Averyt, K.B., Tignor, M., Miller, H.L.(Eds.), Contribution of Working Group I of the Fourth Assessment Report of theIntergovernmental Panel on Climate Change. Cambridge University Press,Cambridge, United Kingdom and New York, NY, USA, p. 996pp.
Ishii, M., Saito, S., Tokieda, T., Kawano, T., Matsumoto, K., Inoue, H.Y., 2004.Variability of surface layer CO2 parameters in the western and centralequatorial Pacific. In: Shiyomi, et al. (Eds.), Global Environmental Change in
the Ocean and on Land. Terrapub, Tokyo, pp. 5994.Jabaud-Jan, A., Metzl, N., Brunet, C., P oisson, A., Schauer, B., 2004. Variability of the
carbon dioxide system in the Southern Indian Ocean (20601S): the impact of awarm anomaly in austral summer 1998. Global Biogeochemical Cycles 18 (1),GB1042.
Jouandet, M.-P., Blain, S., Metzl, N., Brunet, C., Trull, T., Obernosterer, I., 2008.A seasonal carbon budget for a naturally iron fertilized bloom (Kerguelen I.Southern Ocean). Deep-Sea Research II 55, 856867.
Keeling, C.D., Rakestraw, N.W., Waterman, L.S., 1965. Carbon dioxide in surfacewaters of the Pacific Ocean, 1: measurements of the distribution. Journal ofGeophysical Research 70 (N24), 60876097.
Keeling, C.D., Brix, H., Gruber, N., 2004. Seasonal and long-term dynamics of theupper ocean carbon cycle at station ALOHA near Hawaii. Global Biogeochem-ical Cycles 18.
Koertzinger, A., Mintrop, L., Wallace, D.W.R., Johnson, K.M., Neill, C., et al., 2000.The international at-sea intercomparison of fCO2 systems during the R/VMeteor Cruise 36/1 in the North Atlantic Ocean. Marine Chemistry 2 (24),171192.
Lefevre, N., Watson, A.J., Olsen, A., Rios, A.F., Perez, F.F., Johannessen, T., 2004.A decrease in the sink for atmospheric CO2 in the North Atlantic. GeophysicalResearch Letters 31 (L07306).
Lenton, A., Matear, R.J., 2007. The role of the Southern Annular Mode (SAM) inSouthern Ocean CO2 uptake. Global Biogeochemical Cycles 5.
Lenton, A., Matear, R., Tilbrook, B., 2006. Observational strategy to constrain theannual airsea flux of CO2 in the Southern Ocean. Global BiogeochemicalCycles 20.
Le Quere, C., Metzl, N., 2003. Natural processes regulating the oceanic uptake ofCO2. In: Field, C.B., Raupach, M.R. (Eds.), The Global Carbon Cycle: IntegratingHumans, Climate, and the Natural World, SCOPE 62. Island Press, Washington,DC, pp. 243255.
Le Quere, C., Rodenbeck, C., Buitenhuis, E.T., Conway, T.J., Langenfelds, R., Gomez,A., Labuschagne, C., Ramonet, M., Nakazawa, T., Metzl, N., Gillett, N., Heimann,M., 2007. Saturation of the Southern Ocean CO2 sink due to recent climatechange. Science 316, 1735.
Lovenduski, N.S., Gruber, N., 2005. Impact of the Southern Annular Mode onSouthern Ocean circulation and biology. Geophysical Research Letters 32 (11),L11603.
Lovenduski, N.S., Gruber, N., Doney, S.C., Lima, I.D., 2007. Enhanced CO2outgassingin the Southern Ocean from a positive phase of the Southern Annular Mode.Global Biogeochemical Cycles 21, GB2026.
Lutjeharms, J.R.E., Valentine, H.R., 1984. Southern Ocean thermal fronts south ofAfrica. Deep Sea Research 31, 14611476.
Marshall, G.J., 2003. Trends in the Southern Annular Mode from observations andreanalyses. Journal of Climate 16, 41344143.
Metzl, N., Poisson, A., Louanchi, F., Brunet, C., Schauer, B., Bres, B., 1995. Spatio-temporal distributions of airsea fluxes of CO2 in the Indian and AntarcticOceans: a first step. Tellus 47B, 5669.
Metzl, N., Louanchi, F., Poisson, A., 1998. Seasonal and interannual variations of seasurface carbon dioxide in the subtropical Indian Ocean. Marine Chemistry 60,
131146.Metzl, N., Tilbrook, B., Poisson, A., 1999. The annualfCO2cycle and the airsea CO2
flux in the sub-Antarctic Ocean. Tellus 51B (4), 849861.Metzl, N., Brunet, C., Jabaud-Jan, A., Poisson, A., Schauer, B., 2006. Summer and
winter airsea CO2 fluxes in the Southern Ocean. Deep Sea Research I 53,15481563.
Miyake, Y., Sugimura, Y., 1969. Carbon dioxide in the surface water and theatmosphere in the Pacific, the Indian and the Antarctic Ocean areas. Records ofOceanographic Works in Japan 10 (N1), 2833.
Morrow, R., Valladeau, G., Sallee, J., 2008. Observed subsurface signature ofSouthern Ocean decadal a level rise. Progress in Oceanography 77 (4),351366.
Poisson, A., Metzl, N., Brunet, C., Schauer, B., Bres, B., Ruiz-Pino, D., Louanchi, F.,1993. Variability of sources and sinks of CO2 and in the western Indian andSouthern Oceans during the year 1991. Journal of Geophysical Research 98(C12), 22,75922,778.
Raupach, M.R., Marland, G., Ciais, P., Le Quere, C., Canadell, J.G., Kleppe, G., Field,C.B., 2007. Global and regional drivers of accelerating CO2 emissions.Proceedings of the National Acadamy Sciences, USA 104, 1028810293.
Sabine, C.L., Heimann, M., Artaxo, P., Bakker, D., Chen, C.-T.A., Field, C.B., Gruber, N.,Le Quere, C., Prinn, R.G., Richey, J.E., Romero-Lankao, P., Sathaye, J., Valentini, R.,2003. Current status and past trends of the carbon cycle. In: Field, C.B.,Raupach, M.R. (Eds.), The Global Carbon Cycle: Integrating Humans, Climate,and the Natural World. SCOPE 62, Island Press, Washington, DC, pp. 1744.
Sabine, C.L., Feely, R.A., Gruber, N., Key, R.M., Lee, K., Bullister, J.L., Wanninkhof, R.,Wong, C.S., Wallace, D.W.R., Tilbrook, B., Millero, F.J., Peng, T.-H., Kozyr, A., Ono,T., Ros, A.F., 2004. The oceanic sink for anthropogenic CO2. Science 305,367371.
Schuster, U., Watson, A.J.W., 2007. A variable and decreasing sink for atmosphericCO2in the North Atlantic. Journal of Geophysical Research 112, C11006.
Schuster, U., Watson, A.J., Bates, N., Corbiere, A., Gonzalez-Davila, M., Metzl, N.,Pierrot, D., Santana-Casiano, M., 2009. Trends in North Atlantic sea surface
fCO2 from 1990 to 2006. Deep-Sea Research II, this issue [doi:10.1016/j.dsr2.2008.12.011].
Sen Gupta, A., England, M.H., 2006. Coupled oceanatmosphereice response tovariations inthe Southern Annular Mode. Journal of Climate 19 (18), 44574486.
Stephens, B.B., Gurney, K.R., Tans, P.P., Sweeney, C., Peters, W., et al., 2007. Weak
northern and strong tropical land carbon uptake from vertical profiles ofatmospheric CO2. Science 316, 1732.
Takahashi, T., 1961. Carbon dioxide in the atmosphere and in the Atlantic Oceanwater. Journal of Geophysical Research 66, 477494.
Takahashi, T., Chipman, D., Volk, T., 1983. Geographical, seasonal, and secularvariations of the partial pressure of CO2 in the surface waters of the NorthAtlantic Ocean. The result of the Atlantic TTO Program. In: Proceedings: CarbonDioxide, Science and Consensus, CONF 820970, pp. 125145, US DOE,Washington, DC.
Takahashi, T., Sutherland, S.C., Sweeney, C., Poisson, A., Metzl, N., Tilbrook, B.,Bates, N., Wanninkhof, R., Feely, R.A., Sabine, C., Olafsson, J., Nojiri, Y., 2002.Global seaair CO2 flux based on climatological Surface Ocean pCO2, andseasonal biological and temperature effect. Deep Sea Research II 49 (910),16011622 Revised airsea fluxes calculations from this article are available at/www.ldeo.columbia.edu/CO2S.
Takahashi, T., Sutherland, S.C., Feely, R.A., Wanninkhof, R., 2006. Decadal change ofthe surface water pCO2 in the North Pacific: a synthesis of 35 years ofobservations. Journal of Geophysical Research 111, C07S05.
Takahashi, T., Sutherland, S.C., Wanninkhof, R., Sweeney, C., Feely, R.A., Chipman,D.W., Hales, B., Friederich, G., Chavez, F., Sabine, C., Watson, A., Bakker, D.C.E.,Schuster, U., Metzl, N., Yoshikawa-Inoue, H., Ishii, M., Midorikawa, T., Nojiri, Y.,
ARTICLE IN PRESS
N. Metzl / Deep-Sea Research II 56 (2009) 607619618
http://ftp.cmdl.noaa.gov/http://ftp.cmdl.noaa.gov/http://www.ioc.unesco.org/ioccp/pCO2_2007.htmhttp://localhost/var/www/apps/conversion/tmp/scratch_2/dx.doi.org/10.1016/j.dsr2.2008.12.011http://localhost/var/www/apps/conversion/tmp/scratch_2/dx.doi.org/10.1016/j.dsr2.2008.12.011http://ww.ldeo.columbia.edu/CO2http://ww.ldeo.columbia.edu/CO2http://ww.ldeo.columbia.edu/CO2http://ww.ldeo.columbia.edu/CO2http://ww.ldeo.columbia.edu/CO2http://localhost/var/www/apps/conversion/tmp/scratch_2/dx.doi.org/10.1016/j.dsr2.2008.12.011http://localhost/var/www/apps/conversion/tmp/scratch_2/dx.doi.org/10.1016/j.dsr2.2008.12.011http://www.ioc.unesco.org/ioccp/pCO2_2007.htmhttp://ftp.cmdl.noaa.gov/8/11/2019 2009 - Decadal Increase of Oceanic Carbon Dioxide in Southern Indian Ocean Surface Waters (19912007)
13/13