Вы находитесь на странице: 1из 13

Click

Here

GLOBAL BIOGEOCHEMICAL CYCLES, VOL. 20, GB3020, doi:10.1029/2005GB002491, 2006

for

Full
Article

Ocean carbon cycling in the Indian Ocean:


1. Spatiotemporal variability of inorganic carbon and
air-sea CO2 gas exchange
Nicholas R. Bates,1 A. Christine Pequignet,1 and Christopher L. Sabine2
Received 18 February 2005; revised 27 October 2005; accepted 17 February 2006; published 9 September 2006.

[1] The spatiotemporal variability of upper ocean inorganic carbon parameters and air-

sea CO2 exchange in the Indian Ocean was examined using inorganic carbon data
collected as part of the World Ocean Circulation Experiment (WOCE) cruises in 1995.
Multiple linear regression methods were used to interpolate and extrapolate the
temporally and geographically limited inorganic carbon data set to the entire Indian
Ocean basin using other climatological hydrographic and biogeochemical data. The
spatiotemporal distributions of total carbon dioxide (TCO2), alkalinity, and seawater
pCO2 were evaluated for the Indian Ocean and regions of interest including the Arabian
Sea, Bay of Bengal, and 10N35S zones. The Indian Ocean was a net source of CO2 to
the atmosphere, and a net sea-to-air CO2 flux of +237 132 Tg C yr1 (+0.24 Pg C yr1)
was estimated. Regionally, the Arabian Sea, Bay of Bengal, and 10N10S zones
were perennial sources of CO2 to the atmosphere. In the 10S35S zone, the CO2 sink
or source status of the surface ocean shifts seasonally, although the region is a net oceanic
sink of atmospheric CO2.
Citation: Bates, N. R., A. C. Pequignet, and C. L. Sabine (2006), Ocean carbon cycling in the Indian Ocean: 1. Spatiotemporal
variability of inorganic carbon and air-sea CO2 gas exchange, Global Biogeochem. Cycles, 20, GB3020,
doi:10.1029/2005GB002491.

1. Introduction
[2] Determining the past, present and future fate of
anthropogenic carbon dioxide (CO2) requires detailed
knowledge of the exchange between and transformations
of carbon within the mobile reservoirs of the atmosphere,
terrestrial biosphere and ocean. This entails improvements
in knowledge about ocean carbon sources and sinks, their
geographic and temporal variability, and their modulation
by biological and physical processes.
[3] The Indian Ocean, influenced by seasonal monsoonal
forcing, is an important component of the global ocean
system and a modulator of heat and salinity transport, and
the biogeochemical cycling of carbon and nutrient elements
(e.g., nitrogen, phosphorus, silicon). Biological productivity
in the Indian Ocean, for example, accounts for 15 20% of
global ocean productivity [e.g., Chavez and Barber, 1987;
Behrenfield and Falkowski, 1997], with large variation
observed in the timing and spatial distribution of new and
export production.
[4] The Indian Ocean has been the focus of previous
studies that evaluated inorganic carbon cycling and rates of
air-sea CO2 gas exchange. The southwestern Indian Ocean,
1

Bermuda Biological Station For Research, Inc., Ferry Reach, Bermuda.


Pacific Marine Environmental Laboratory, NOAA, Seattle, Washington, USA.
2

Copyright 2006 by the American Geophysical Union.


0886-6236/06/2005GB002491$12.00

south of 35S, has been the focus of several survey and


time series (e.g., Kerguelen time series) investigations [e.g.,
Metzl et al., 1991; Poisson et al., 1993; Metzl et al., 1995,
1998]. Similarly, the Arabian Sea and Bay of Bengal has
been investigated during the U.S. Joint Global Ocean Flux
Study (JGOFS) process cruises of 1995 [Millero et al.,
1998a; Goyet et al., 1998] and Indian JGOFS programs, in
particular [e.g., George et al., 1994; Kumar et al., 1996;
Sarma et al., 1998, 2003; Sarma, 2003, 2004]. For the
entire Indian Ocean basin, synthesis of surface partial
pressure of CO2 (pCO2) data collected primarily on World
Ocean Circulation Experiment (WOCE) cruises, and the
National Oceanographic and Atmospheric Administration
(NOAA) Ocean Atmosphere Carbon Exchange Study
(OACES) Indian Ocean cruises have led to improved
knowledge of the variability of pCO2 and air-sea CO2
fluxes [Louanchi et al., 1996; Sabine et al., 2000], while
others have focused on the distribution of anthropogenic
CO2 in the ocean interior [Sabine et al., 1999; Sabine and
Feely, 2001; Coatanoan et al., 2001]. Analysis of the
variability of ocean sinks and sources of CO2 in the Indian
Ocean, however, has been somewhat limited by coarse
spatial resolution employed in previous studies (2  2
[Louanchi et al., 1996; Sabine et al., 2000], 4  5
[Takahashi et al., 1993, 2002]).
[5] The present work examines the seasonal and spatial
variability of upper ocean inorganic carbon and the exchange of CO2 between ocean and atmosphere in the Indian
Ocean. As pointed out by a number of authors [e.g.,

GB3020

1 of 13

GB3020

BATES ET AL.: INDIAN OCEAN CARBON CYCLE, 1

Takahashi et al., 2002; Sabine et al., 2000], previous studies


of the region have been somewhat limited by spatially and
temporally insufficient data coverage. Here multiple linear
regression (MLR) approaches were used to interpolate and
extrapolate inorganic carbon data, collected as part of the
WOCE CO2 survey in 1995, to the entire Indian Ocean
basin. Similar MLR approaches have been widely used in
recent studies of inorganic carbon cycling [e.g., Goyet and
Davis, 1997; Goyet et al., 2000; Lee, 2001; Lee et al.,
2002], inorganic carbon data quality issues [e.g., Lamb et
al., 2002]; calcium carbonate dissolution [Chung et al.,
2003; Feely et al., 2002], anthropogenic CO2 inventories
[Sabine et al., 1999; 2002a, 2002b; Sabine and Feely,
2001], and global pCO2 climatotologies [Takahashi et al.,
2002]. Here total carbon dioxide (TCO2), alkalinity (TA)
and seawater pCO2 data set were extrapolated to the basin
scale (Indian Ocean; 35S!15N) using a spatial resolution of 1  1, vertical differentiation into 14 layers in the
upper 500 m, and temporal resolution of 1 month in order to
evaluate the spatiotemporal distribution of hydrographic
and seawater CO2 properties. The seasonal and annual rates
of air-sea CO2 exchange were then quantified to determine
the regional and basin sources and sinks of CO2. In a
companion paper [Bates et al., 2006], the analyses presented herein are used to quantify the annual and spatialtemporal pattern of new production (or net community
production, NCP) and export production, and examine the
net metabolism (heterotrophy versus autotrophy) for the
Indian Ocean.

2. Indian Ocean Overview


[6] The Indian Ocean is dominated by monsoon forcing
and intermonsoon transitions. During the Northeast Monsoon (NEM; December February), the Northeast Monsoon Current (NMC) brings waters from east to west across
the north Indian Ocean and into the Arabian Sea (Figure 1)
[e.g., Wyrtki, 1973; Weller et al., 1998; Schott and
McCreary, 2001]. At the equator, the South Equatorial
Counter Current (SECC) returns water to the east, while
south of the equator (5S 20S), the South Equatorial
Current (SEC) transports water from east to west. The
coolest surface temperatures (24 1C) are typically
observed in this period, particularly in the Arabian Sea
[e.g., Weller et al., 1998; Morrison et al., 1998]. During
the Spring Intermonsoon (SIM; March May) transition,
the westward flow of NMC declines and stops prior to the
Southwest Monsoon (SWM; June September). A broad
swath of warm surface temperatures (>28C) occupies the
Indian Ocean north of 10S (Figure 1), with temperatures
reaching their seasonal maxima in the Arabian Sea (>28C
[see Weller et al., 1998; Morrison et al., 1998]) (see also
auxiliary materials, Figures S2 and S31). In the southern
Indian Ocean south of 30S, the surface layer cools during
the austral fall period. During the SWM period, there is
intense coastal upwelling off the Arabian Peninsula with the
Somali Jet (SJ) [Schott and McCreary, 2001]. At the
1
Auxiliary materials are available in the HTML. doi:10.1029/
2005GB002491.

GB3020

Figure 1. Major features of the surface circulation in the


Indian Ocean [after Schott and McCreary, 2001]. Surface
currents include: EICC, East Indian Coastal Current; NMW,
Northeast Monsoon Current; SC, Somali Current; SEC,
South Equatorial Current; SECC, South Equatorial Counter
Current; SJ, Somali Jet, STF, Subtropical Front; SWMC,
Southwest Monsoon Current.

equator, the Somali Current (SC) forms part of a strong


anticyclonic gyre, while the South West Monsoon Current
(SWMC) transports waters to the east, north of the equator
(Figure 1). Notable hydrographic features include the presence of cooler surface temperatures (<26C), lower salinities (<35.5 salinity) and higher nitrate in waters off the
coasts of Somalia and the Arabian Peninsula, associated
with coastal upwelling. In the Bay of Bengal, salinities are
low (<30 salinity), reflecting the enhanced freshwater inputs
to the Bay [e.g., George et al., 1994]. In the southern Indian
Ocean, south of 30S, surface temperatures reach their
seasonal minima. The Fall Intermonsoon (FIM; or Northeast
Intermonsoon) during the September November period
marks the transition back to the NEM. The coastal upwelling declines and stops, and transports of water associated
with the eastward SECC (0) and westward NMC (5
10N) resume.

3. Data Sources and Analysis


3.1. Oceanographic Data Sets
[7] During the U.S. World Ocean Circulation Experiment
(WOCE) hydrographic survey of the Indian Ocean between
December 1994 and January 1996, inorganic carbon parameters were measured on nine cruises (see auxiliary materials
Figure S1) as part of a cooperative effort with the Joint
Global Ocean Flux Study (JGOFS) program. The CO2 data
used for this study were obtained from the Carbon Dioxide
Information and Analysis Center (CDIAC). Cruise data,
metadata information (including chief scientist, CO2 analysts, and CO2 analytical protocols), and ASCII documentation files were obtained at: http://cdiac.esd.ornl.gov/
oceans/woceindian.html (see auxiliary material for more
information).

2 of 13

BATES ET AL.: INDIAN OCEAN CARBON CYCLE, 1

GB3020

GB3020

Table 1. Multiple Linear Regression Equations for the Best Fit Between Observed Inorganic Carbon Data and Modeled Inorganic
Carbon Data Computed From Other Hydrographic Parameters for the Upper 0 150 ma
Season

Mean

Optimal Multiparameter Fit

Dec Feb
March May
June Aug
Sep Nov

5.13
5.66
5.99
4.40

700
311
703
681

722.27
743.93
661.38
779.56

+
+
+
+

(6.473.T)
(7.587.T)
(6.058.T)
(5.331.T)

Total Carbon Dioxide (TCO2 0 150 m)


+ (40.348.S) + (1.513.NO3) + (0.064.dep) + (0.009.lat) + (0.464.AOU) + (59.299.PO4)
+ (40.506.S) + (0.743.NO3) + (0.107.dep) + (0.008.lat) + (0.374.AOU) + (57.45.PO4)
+ (41.361.S) + (1.393.NO3) + (0.014.dep) + (0.044.lat) + (0.073.AOU) + (68.737.PO4)
+ (37.712.S) + (0.262.NO3) + (0.028.dep) + (0.034.lat) + (0.159.AOU) + (79.882.PO4)

Dec Feb
March May
June Aug
Sep Nov

6.43
4.99
4.48
4.78

659
299
733
672

735.73
416.23
338.55
508.56

+
+
+
+

(2.308.T)
(1.552.T)
(1.481.T)
(1.239.T)

+
+
+
+

Alkalinity (TA) 0 150 m


(46.136.S) + (0.568.NO3) + (0.023.dep) + (0.032.lat) + (0.022.AOU) + (4.455.PO4)
(54.673.S) + (1.179.NO3) + (0.013.dep) + (0.004.lat) + (0.059.AOU) + (12.954.PO4)
(57.046.S) + (0.550.NO3) + (0.008.dep) + (0.114.lat) + (0.051.AOU) + (8.116.PO4)
(51.923.S) + (1.041.NO3) + (0.007.dep) + (0.081.lat) + (0.117.AOU) + (3.152.PO4)

a
The best multiparameter fits for TCO2 and alkalinity were generated as a function of temperature (T), salinity (S), apparent oxygen utilization (AOU),
nitrate (NO3), phosphate (PO4), depth, latitude and season. RMS values are mmoles kg1. N equals number of data. For example, a salinity error of 0.1
would increase (decrease) TCO2 and TA by 5.5 and 6.5 mmoles kg1, respectively. This would either increase (decrease) seawater pCO2 by 1 matm.

[8] In addition, data from the US JGOFS Arabian Sea


Expedition [e.g., Smith et al., 1998; Millero et al., 1998a],
and NOAA-OACES CO2 repeat hydrographic transect
survey (I8N) in the Indian Ocean in 1995, were not
incorporated into the primary WOCE CO2 data set for
interpolation and extrapolation purposes, but rather used
as independent validation of the modeled DIC and TA data.
Climatological hydrographic and biogeochemical data (i.e.,
T, S, NO3, PO4), with a spatial resolution of 1  1,
vertical resolution of 14 layers in the upper 500 m, and
temporal resolution of 1 month, were also used in this
analysis (please see auxiliary material).
3.2. Interpolation and Extrapolation of
Inorganic Carbon Data
[9] Determining the time and space scale of variability of
the carbon cycle requires interpolation of discrete water
column data to a basin scale. As stated earlier, inorganic
carbon parameters (i.e., TCO2 and TA) were sampled on
nine cruises across the Indian Ocean. Although the survey
cruises were broadly distributed, large swaths of the region
were not sampled in time or space (Figure 1). However, in
the Indian Ocean there is an abundance of other climatological and hydrographic data (e.g., temperature, salinity,
inorganic nutrients). Thus TCO2 and TA can be interpolated
and extrapolated as a function of other water mass properties, such as potential temperature and salinity.
[10] In previous studies, interpolation of TCO2 and TA
distributions using the MLR approach has an uncertainty of
5 15 mmoles kg1 when applied to data below the mixed
layer [e.g., Goyet and Davis, 1997; Sabine et al., 1999;
Goyet et al., 2000; Sabine and Feely, 2001; Coatanoan et
al., 2001]. In the mixed layer, the interpolation of TCO2 in
particular, has a greater uncertainty due to seasonal variability. However, Lee et al. [2000, 2002], applied MLR
analyses (albeit using a much more limited data set than
used here) to interpolate DIC from temperature, salinity and
nitrate data, and extrapolated to produce 4  4 maps of
global TCO2 and TA (including the Indian Ocean).
[11] Here different interpolation schemes were investigated
using data available for all cruises, including: T, potential
temperature (q), S, DO, density, apparent oxygen utilization

(AOU), nitrate (NO3), phosphate (PO4), sample depth, latitude and longitude. AOU was computed from bottle DO, T,
and S data sets. Various combinations of parameters were
examined in order to improve the quality of the fit and
reduce the residual errors between the measured and synthetic data. Initially, carbon parameters were fit simply as a
function of q, salinity and AOU, similar to the approach of
Goyet et al. [1999, 2000].
TCO2 a1 a2 q a3 AOU a4 S;

where a1 to a4 are constants. However, the uncertainties


of the interpolation for TCO2 and TA were relatively high
(>10 mmoles kg1) for the upper ocean. Various combinations of data were evaluated and the quality of the MLR fit
was determined by the RMS error, comparison of cruise and
synthetic data and examination of the spatial pattern of the
residuals. The optimal interpolation (with the lowest
associated uncertainty) for TCO2 and TA in the upper
ocean (0 150 m) was a function of several properties,
TCO2 a1 a2 T a3 AOU a4 S a5 NO3 a6 PO4
a7 latitude a8 sample depth

TCO2 b1 b2 T b3 AOU b4 S b5 NO3 b6 PO4


b7 latitude b8 sample depth;

where a and b are constants (Table 1). The optimal


interpolation for TCO2 and TA in the water column (150mbottom) is reported in auxiliary materials (auxiliary Table 1
in Text S1). The best fits for the upper ocean were generated
with the following considerations in mind.
[12] First, the MLR approach was initially applied to
Indian Ocean TCO2 and TA data from the entire water
column. However, the best fits for the upper ocean occurred
if the MLR approach was used for data from the 0- to 150-m
layer only rather than the entire water column. Thus MLR
equations are reported for 0- to 150-m layer and deeper
>150 m layer (see auxiliary Table 1 in auxiliary Text S1).
The error analysis outlined in section 2.3 is focused on the
0- to 150-m layer data only.

3 of 13

GB3020

BATES ET AL.: INDIAN OCEAN CARBON CYCLE, 1

[13] Second, TCO2 and TA data from regions where the


bottom depth was less than 200 m were excluded from the
MLR analysis. This step effectively excludes the continental
shelf of the Indian Ocean from the analysis. Recent reviews
indicate that there is much greater spatiotemporal variability
of the CO2 in the coastal ocean compared to the open ocean
[Borges, 2005; Borges et al., 2005; Ducklow and McAllister,
2005]. Given the paucity of CO2 data collected in the
coastal ocean of the Indian Ocean during the WOCE
program, there were insufficient data to characterize the
spatiotemporal variability of inorganic carbon and air-sea
CO2 exchange within the complex coastal zone of the
Indian Ocean.
[14] Third, this study was limited to the Indian Ocean
region north of 35S. Within the region of 35S 42S,
WOCE CO2 survey data were insufficient to adequately
characterize the seasonal and temporal variability of inorganic carbon in the region of the Subtropical Front (STF).
[15] Fourth, the smallest interpolation errors were produced if the data were grouped into respective monsoon and
intermonsoon seasons relevant to the Indian Ocean, i.e.,
December to February (NEM); March to May (SIM); June
to August (SWM); and September to November (FIM).
Greater differences between the observed and MLR modeled data occurred if the MLR approach was applied to the
full year of TCO2 and TA data. Similarly, greater errors
occurred when the MLR approach was applied to TCO2 and
TA data grouped into each month. This reflected a paucity
of data in different months in different regions of the Indian
Ocean.
[16] Fifth, AOU values rather than DO, improved the
MLR fit, while inclusion of PO4 and depth improved the fit
the least. Dissolved oxygen concentrations are influenced
by physics, biology and air-sea O2 exchange. As AOU is a
partial function of the dominant physical forcing (i.e.,
temperature and salinity), and since AOU (rather than
DO) exhibits a stronger covariance with TCO2 in response
to a variety of biological processes, the inclusion of AOU
significantly improved the MLR fit. Both nitrate and phosphate were included in the MLR approach, reflecting the
variability of N:P elemental stoichiometry in the upper
ocean in response to processes such as nitrogen fixation
and denitrification in parts of the Indian Ocean. This
contrasts to the deeper water column, where the elemental
stoichiometric ratios of N and P have a very restricted range
[e.g. Redfield et al., 1963; Anderson and Sarmiento, 1994].
[17] Last, the inclusion of latitude in the interpolation
improved the fit, but longitude did not (reflecting the
stronger meridional gradients compared to zonal gradients).
In situ temperature (T) was used rather than potential
temperature (q) since there was negligible difference
between T and q in the 0- to 150-m layer. The sample
depth of the observed TCO2 and TA data was also
included in the MLR approach, since this helped to
improve the MLR fits. The inclusion of sample depth
probably reflects the contribution of sinking and dissolved
organic matter remineralized to DIC vertically through the
water column.
[18] The optimal interpolation of surface layer (0 150 m)
TCO2 had an uncertainty of 5.0 6.5 mmoles kg1 for the

GB3020

four monsoonal/intermonsoonal seasons (Table 1; see auxiliary material Figure S1). These error estimates were small
compared to the large range of TCO2 (300 mmoles kg1)
values observed in the upper 150 m during the survey
cruises. It should also be noted that owing to the relatively
short period over which TCO2 samples were collected in the
Indian Ocean, no correction was applied to account for
secular increase in seawater TCO2 due to uptake of anthropogenic CO2 from the atmosphere. The optimal interpolation of surface layer (0 150 m) TA had an uncertainty of
4.0 5.5 mmoles kg1 for the four seasons (Table 1, see auxiliary material Figure S2). Again, this error was small compared to the range of TA (150 mmoles kg1) in the upper
150 m observed during the survey cruises. In the deeper
depths (i.e., >150 m), TCO2 had an uncertainty of 6.6
8.6 mmoles kg1 for the four monsoonal/intermonsoonal
seasons (Table 1). Additionally, in the deeper depths (i.e.,
>150 m), TA had an uncertainty of 3.7 6.1 mmoles kg1
for the four monsoonal/intermonsoonal seasons (Table 1).
3.3. Extrapolation to the Indian Ocean
[19] In the extrapolation step, optimal MLR equations
(Table 1) were applied to gridded, 1  1 climatological
data of hydrographic parameters each month (i.e., T, S,
AOU, NO3, PO4) to produce basin-wide maps of carbon
parameters (e.g., DIC and TA). In addition, basin-wide
distributions of seawater partial pressure of CO2 (pCO2)
were calculated each month from the 1  1 monthly grids
of DIC, TA, T and S (see auxiliary materials Figures S2 S4
and S10 S14). Seawater pCO2 was calculated from DIC
and TA using the program of Lewis and Wallace [1998] (see
auxiliary materials for more information).
3.4. Sensitivity Tests, Errors, and Caveats
[20] The interpolation of DIC and TA in the upper 0
150 m from hydrographic and biogeochemical properties
(equations (2) and (3)) had a relatively small error (4
6 mmoles kg1; Table 1). However, there may be additional
systematic uncertainties associated with extrapolating data
outside of the defined interpolation range, particularly in the
regions of the Indian Ocean where there is limited data in
time and space. This problem is inherent to other studies
using the MLR approach including: carbon cycle and data
studies [e.g., Goyet and Davis, 1997; Sabine et al., 1999;
Goyet et al., 2000; Lee, 2001; Sabine and Feely, 2001; Lee
et al., 2002; Takahashi et al., 2002; Chung et al., 2003;
Feely et al., 2002]. In order to get some quantitative sense
of the potential errors associated with extrapolation, two
error analyses were attempted.
[21] First, modeled DIC and TA (generated from the
primary CO2 data sources) were compared with observational data collected on the I8N NOAA OACES repeat
section and during the JGOFS program in the Arabian Sea.
As stated earlier, these secondary data sets were not included
with the WOCE CO2 data in the MLR interpolation and
thus represent independent data sets with which to compare.
The MLR regressions (for the 0- to 150-m layer) were used
to determine model DIC and TA data for each 1  1 grid
in the Indian Ocean. The DIC and TA data (determined
from climatological hydrographic data and at depths equiv-

4 of 13

GB3020

BATES ET AL.: INDIAN OCEAN CARBON CYCLE, 1

alent to the observed data) were compared with observed


DIC and TA data from each CTD sample depth collected on
the I8N and Arabian Sea cruises. For the I8N NOAA
OACES repeat section, the mean difference between model
and observed TCO2 and TA data was small (11.3 and
4.1 mmoles kg1, respectively; see auxiliary materials
Table 2 in Text S1 and Figure S3). This repeat section was
conducted in a different season than the primary WOCE CO2
survey cruises that were used in the multiparameter analysis.
In the Arabian Sea, modeled DIC and TA data were
compared to observed data collected from four different
Arabian Sea cruises, along the separate north and south
repeat transects (see Smith et al. [1998] for CTD station
locations). Along the south transect, for example, the mean
difference between calculated and observed DIC and TA was
8 10 mmoles kg1 (see auxiliary materials Table 2 in
Text S1 and Figure S5). For TA, there was a large difference
(19 mmoles kg1) between model and observed data for the
TN43 cruise. This indicates that the MLR approach may not
have captured all the processes that influence alkalinity
during the NE Monsoon period (e.g., upwelling of alkalinity,
calcification due to pelagic calcifying plankton). In summary,
these comparisons suggest that the extrapolation errors
(9.0 11.3 mmoles kg 1 for DIC; auxiliary materials
Table 2 in Text S1) were typically larger than the interpolation errors (4.4 6.0 mmoles kg1; Table 1). However, the
differences between model and observed data may partly
relate to differences imparted by the hydrographic data sets
used (i.e., climatological hydrography versus observed
WOCE CTD hydrography), and scale incompatibility of
the comparison (1  1 model data compared to observed
data from 1 or more CTD station within a 1  1 grid).
[22] Second, a comparison of observed and modeled
seawater pCO2 was attempted for the entire Indian Ocean.
To facilitate comparison, model seawater pCO2 data were
calculated from model TCO2 and TA data for each 1  1
grid every month, and compared to seawater pCO2 data
calculated from observed TCO2 and TA data at colocated
and contemporaneous WOCE CTD stations in the Indian
Ocean. This approach tests whether there are systematic
errors in the TCO2 or TA data that present themselves in the
model pCO2 data. For the 0- to 150-m layer, the mean
difference between model and observed seawater pCO2 was
small (1 matm; see auxiliary material Table 3 in Text S1
and Figure S6), although the standard deviation was quite
variable (12 21 matm). This comparison indicates that the
model pCO2 data simulate the observed spatiotemporal
variability pCO2 quite well [Sabine et al., 2000]. If there
are underlying systematic differences between model and
observed TCO2 and TA data, these differences do not
manifest themselves in the model pCO2 data.
[23] Unfortunately, there are no other CO2 data sets that
can be entrained to test whether there are systematic differences between model and observed CO2 data outside the
time and space frame used to generate the MLR analysis.
Systematic errors in the model TCO2 and TA data could not
be clearly identified with the available observational data
sets. The error analysis conducted here indicates that the
errors associated with interpolation and extrapolation are
10 mmoles for model TCO2 and TA data.

GB3020

3.5. Quantifying the Rates of Air-Sea CO2


Gas Exchange (Cgasex)
[24] Basin-wide estimates for the rate of air-sea flux of
CO2 (Cgasex) were estimated from seawater pCO2 and
atmospheric pCO2, and wind data (see auxiliary material
for more information). The flux of CO2 (F) across the airsea interface is typically determined from the bulk formula,
Ft k s DpCO2 t

where Ft is the CO2 flux averaged over a certain time period


(t), k is the gas transfer velocity, s is the solubility of CO2 in
seawater, and DpCO2 is the difference between seawater
and atmospheric pCO2, respectively. The DpCO2, or air-sea
CO2 disequilibrium, sets the direction of CO2 gas exchange
while k and the magnitude of DpCO2 determines the rate of
air-sea CO2 transfer.
[25] Wind speed is currently the most robust parameter
available to determine air-sea CO2 exchange and several gas
transfer velocity-wind speed relationships are frequently used
[Liss and Merlivat, 1986; Wanninkhof, 1992; Wanninkhof and
McGillis, 1999]. Here a quadratic dependency between wind
speed and k is used [Wanninkhof, 1992],
k 0:39 U210 Sc=6600:5 ;

where U10 is wind speed corrected to 10 m, and Sc is the


Schmidt number for CO2.
[26] The flux of CO2 due to gas exchange was computed
using equation (5) and monthly wind speed data, facilitating
comparison with the air-sea CO2 fluxes determined by
Louanchi et al. [1996], Louanchi and Najjar [2000] and
Sabine et al. [2000]. The F (equation 4) or Cgasex term was
quantified every month for each 1 by 1 box in the Indian
Ocean (see auxiliary material for more information). The
errors for both air-sea CO2 flux terms were determined
assuming a systematic 5 matm uncertainty for the seawater
pCO2 values.

4. Results and Discussion


4.1. Spatiotemporal Distributions of CO2 in the Upper
Ocean of the Indian Ocean
[27] The spatiotemporal distributions of TCO2, TA, and
seawater pCO2 were determined monthly but presented here
(Figures 2 4) as averages for the four different monsoon/
intermonsoon seasons. The Indian Ocean, particularly the
northern sector and Arabian Sea, is influenced by the
seasonal transition between the NE and SW monsoon. This
physical forcing provides the context for interpreting the
spatiotemporal variability of inorganic and organic carbon
in the Indian Ocean.
[28] The spatiotemporal distribution of TA was similar to
salinity. This finding is not unexpected since TA is typically a
conservative function of salinity, particularly in the subtropical gyres [e.g., Millero et al., 1998b]. The range of alkalinity
across the Indian Ocean was >200 mmoles kg1, with the
highest salinity and TA found in the Arabian Sea (>36
salinity; >2300 mmoles kg1 TA) and in the region between
the SEC and STF (>35 salinity; >2300 mmoles kg1 TA)

5 of 13

GB3020

BATES ET AL.: INDIAN OCEAN CARBON CYCLE, 1

GB3020

Figure 2. Spatial distribution of surface alkalinity (mmoles kg1) in the Indian Ocean each season.
(a) December to February (NEM period). (b) March to May (SIM period). (c) June to August (SWM
period). (d) September to November (FIM period).

(Figure 2). The lowest salinity and TAvalues were found in the
equatorial region (34 35 salinity; <2300 mmoles kg1 TA)
and Bay of Bengal (<30 salinity; <2200 mmoles kg1 TA).
The smallest seasonal change in TA (i.e., DTA) of less than
20 mmoles kg1 occurred in three main regions: (1) zonal
band between 15S and 40S, (2) zonal band at the
equator, and (3) in the northwest Arabian Sea (Figure 5b).
There are modest seasonal changes of TA within the zonal
band located approximately at the latitude of the South
Equatorial Current (i.e., 5S 15S). The highest seasonal
TA changes were observed southwest of the Indian continent
(up to 60 mmoles kg1) and in the Bay of Bengal (>100
200 mmoles kg1). The large seasonal changes in salinity
and TA in the Bay of Bengal reflect seasonally variable
riverine inputs to the ocean during monsoonal transitions.
During the SWM and FIM, TA values are at their seasonal
minimum (Figure 2), presumably owing to freshening of this
region [George et al., 1994].
[29] The spatiotemporal distribution of TCO2 in the surface
layer was somewhat similar to alkalinity, with high TCO2
values occurring in the Arabian Sea (>2000 mmoles kg1)

and low TCO2 values found between the equator and


15S (<2000 mmoles kg1), and in the Bay of Bengal
(<1900 mmoles kg1) (Figure 3). However, unlike alkalinity, TCO2 increased southward from the latitudes of the SEC
(15S 20S) to the STF (2000 to >2100 mmoles kg1). In
the zonal band between 30S and 45S, TCO2 values reach a
seasonal maximum during the SW monsoon and Fall Intermonsoon periods (July December, Figure 3). This feature is
also apparent for salinity normalized TCO2 (nTCO2), which
is corrected to a salinity of 35 (see auxiliary materials Figure
S12). This region undergoes seasonal changes in TCO2 (i.e.,
DTCO2) of 20 60 mmoles kg1 (Figure 5c), and has the
largest seasonal temperature (Figure 5e) and mixed layer
depth (see auxiliary materials Figure S10f) changes in the
Indian Ocean.
[30] There are several notable features to the spatial and
temporal distribution of surface pCO2 in the Indian Ocean.
From June to March, surface pCO2 values are generally
higher than the atmosphere north of 15S 20S. In the
zonal band between 30S and 45S (Figure 4), pCO2 values
greater than 400 matm were generally observed across the
Arabian Sea into the tropical regions of the Indian Ocean.

6 of 13

GB3020

BATES ET AL.: INDIAN OCEAN CARBON CYCLE, 1

GB3020

Figure 3. Spatial distribution of surface TCO2 (mmoles kg1) in the Indian Ocean each season.
(a) December to February (NEM period). (b) March to May (SIM period). (c) June to August (SWM
period). (d) September to November (FIM period).
During the SIM (April June), pCO2 values greater than
400 matm are localized to the Arabian Sea, and pCO2 values
are close to or below equilibrium values across much of the
Indian Ocean. South of 15S 25S, surface pCO2 values
were close to equilibrium values or below during all
seasons. The highest seasonal variability of surface pCO2
was observed south of 30S (>100 matm), within the Bay of
Bengal (>100 matm) and localized within the Arabian Sea,
close to the Oman coast (Figure 4d).
[31] Although considerable complexity is evident in the
spatiotemporal variability of TCO2, alkalinity and pCO2,
simple generalizations can be made (with exceptions in the
Arabian Sea and Bay of Bengal). In the 10S 35S region,
the annual amplitude of surface temperature and seawater
pCO2 was 6 2C and 65 15 matm, respectively
(Figure 5). The thermodynamic effect of temperature on
pCO2 is 4.21% pCO2 change C1 (15 17 matm C1).
Similar to the subtropical gyres of the North Pacific and
North Atlantic Oceans [Bates et al., 1996], the seasonal
influence of temperature on seawater pCO2 (i.e., 6  16 =
90 matm) was offset by seasonal influence on pCO2 by

TCO2 changes. About one third of the annual change of


TCO2 (35 10 mmoles kg1) was due to salinity changes
(i.e., annual change of nTCO2 was 255 mmoles kg1; see
auxiliary materials Figure S12). The remaining annual
nTCO2 change was primarily caused by primary production
(two thirds [Bates et al., 2006]) and sea-to-air CO2 gas
exchange (one third), respectively.
[32] In the 10N10S region, the annual amplitude of
surface temperature and seawater pCO2 was 2 1C and
40 10 matm, respectively (Figure 5). Here the annual
amplitude of pCO2 was mostly caused by temperature
variability.
4.2. Indian Ocean Air-Sea CO2 Fluxes
[33] Air-sea CO2 flux rates (Figure 6) were calculated
from monthly maps of DpCO2 (Figure 7) and wind speed
(not shown). From these monthly air-sea CO2 flux maps,
annual rates were computed for the entire Indian Ocean and
regions of interest (e.g., Arabian Sea) (Table 2). The net
annual sea-to-air CO2 flux for the entire Indian Ocean
(35S 20N) was estimated at +237 132 Tg C yr1 (or

7 of 13

GB3020

BATES ET AL.: INDIAN OCEAN CARBON CYCLE, 1

GB3020

Figure 4. Spatiotemporal distribution of surface seawater pCO2 (matm). (a) December to February
(NEM period) surface pCO2. (b) March to May (SIM period) surface pCO2. (c) June to August (SWM
period) surface pCO2. (d) September to November (FIM period) surface pCO2.
+0.24 0.13 Pg C yr1) using the quadratic wind speed-k
dependency (equation (5)) of Wanninkhof [1992]. Mean
monthly winds speeds in the Indian Ocean typically ranged
between 6 and 11 m s1, with an annual mean of 7.8 m s1.
At this range of wind speeds, the cubic wind speed-k
dependency of Wanninkhof and McGillis [1999] predicts
lower rates of air-sea CO2 exchange [e.g., Bates and
Merlivat, 2001]. If the cubic wind speed-k dependency is
used, a net annual sea-to-air CO2 flux for the entire Indian
Ocean (35S 20N) was estimated at +190 106 Tg C yr1
(or +0.19 0.11 Pg C yr1), 80% of the value predicted by
the quadratic equation.
[34] Previous studies have reported a wide range of airsea CO2 fluxes for the Indian Ocean. Louanchi et al.
[1996], using 2  2 spatial resolution for maps of
seawater pCO2, estimated a smaller net annual sea-to-air
flux of +22 Tg C yr1 for the 35S 20N zone. The major
differences between the annual CO2 flux estimates of this
study and the Louanchi et al. [1996] study appear to be
associated with higher sea-to-air CO2 fluxes in the 20S
20N region (estimated in this study). In a recent synthesis
of surface seawater pCO2 measurements from the Indian

Ocean during the WOCE survey in 1995, Sabine et al.


[2000] estimated a net annual sea-to-air CO2 flux of
+150 Tg C yr1 (or +0.150 Pg C yr1) for the 36S
20N zone of the Indian Ocean. In a recent study of
anthropogenic CO2 in the Indian Ocean, Hall et al.
[2004] suggested that the rate of sea-to-air CO2 flux in
the Indian Ocean was 250 Tg C yr1. Both the Sabine et
al. [2000] and Hall et al. [2004] estimates of annual sea-toair CO2 flux were similar to the estimates reported in this
study. No estimates of error were given for the Louanchi et
al. [1996] or Sabine et al. [2000] air-sea CO2 flux rates.
4.3. Arabian Sea
[35] In the Arabian Sea, TCO2 ranged from approximately
2000 to 2050 mmoles kg1 (Figure 3). The seasonal DTCO2
changes were relatively modest (<40 mmoles kg1) over
much of the Arabian Sea (Figure 5c). The exception was
higher DTCO2 and DnTCO2 changes (>80 mmoles kg1)
close to the NE Arabian Peninsula off Oman. During the
SW monsoon period, high TCO2 and nTCO2 was observed
off the coast (Figure 4) and associated with coastal upwelling. On the US JGOFS Arabian Sea expeditions, surface

8 of 13

GB3020

BATES ET AL.: INDIAN OCEAN CARBON CYCLE, 1

GB3020

Figure 5. Seasonal variability of hydrographic properties, TA, TCO2 and pCO2 (mmoles kg1) in the
surface layer of the Indian Ocean. (a) Salinity (DS). (b) Alkalinity (DTA) (mmoles kg1). (c) Surface TCO2
(or DTCO2) (mmoles kg1). (d) Seawater pCO2 (matm). (e) Temperature (C). (f) mixed layer depth (m). The
seasonal range was calculated for each parameter in every 1  1 box. Mixed layer depth was calculated
using a 0.1 sigma theta density change criterion.
temperatures were 4 6C cooler during the SWM than the
proceeding Spring Intermonsoon period [Morrison et al.,
1998; Weller et al., 1998; Gundersen et al., 1998]. The
spatiotemporal distribution of nTCO2 in the Arabian Sea
were similar to the observations of Millero et al. [1998a],
collected on the US JGOFS Arabian Sea expeditions
(although the US JGOFS Arabian Sea TCO2 was not
included in the MLR fits, the modeled nTCO2 agrees within

10 15 mmoles kg1). Both the Arabian Sea data and the


modeled data show high nTCO2 (>2050 mmoles kg1) off
the coast of Oman during the SWM period. Over the rest of
the Arabian Sea, the seasonal range of nTCO2 was 1925
1975 mmoles kg1, with the lowest values (1925
1940 mmoles kg1) observed during the Spring Intermonsoon (SIM) period [Millero et al., 1998a, Figure 8].

9 of 13

GB3020

BATES ET AL.: INDIAN OCEAN CARBON CYCLE, 1

GB3020

Figure 6. Monthly distribution of air-sea CO2 flux (mmoles CO2 m2) in the Indian Ocean. Positive
values denote sea-to-air CO2 flux, whereas negative values denote air-to-sea CO2 flux. (top, left to right)
January to April; (middle, left to right) May to August; (bottom, left to right) September to December.
[36] The large seasonal TCO2 (and TA) changes southwest of the Indian continent, reflect a seasonal increase in
salinity during the SWM and FIM. It appears that the SW
Monsoon Current (SWMC; Figure 1) transports higher
salinity (and high TCO2 and TA) waters from the northern
Arabian Sea toward the equator. Such seasonal advection of
high-salinity water to the southeast from the Arabian Sea
was also indicated during the US JGOFS Arabian Sea
expeditions [e.g., Morrison et al., 1998].
[37] The geographic and temporal variability observed
compared well with other observed pCO2 data from basin
[Louanchi et al., 1996; Sabine et al., 2000] and regional
studies [Goyet et al., 1998]. For example, high (>400 matm)
but localized pCO2 values were observed in the Arabian
Sea off Pakistan during the NEM period in the model data
(Figure 4a) and observed data [Sabine et al., 2000]. The
lowest pCO2 values found during the NEM period occurred
close to the southwest coast of India, a feature also observed
by Sarma [2003] who ascribed this feature to low-salinity
water inflow from the Bay of Bengal. Similarly, high
(>400 matm) but localized pCO2 values were observed in
the Arabian Sea off Oman during the SIM in both model data
and observed data [Goyet et al., 1998; Sabine et al., 2000]. As

pointed out by others [Goyet et al., 1998; Sarma et al., 1998;


Sabine et al., 2000; Sarma, 2004], this reflects upwelling of
pCO2 rich waters close to the coast of Oman. During the
SWM, seawater pCO2 values decreased to a seasonal minimum (350 400 matm) throughout much of the Arabian Sea
(higher near to the coast), although the region remained
supersaturated with respect to the atmosphere.
[38] In the Arabian Sea, DpCO2 values were generally
positive each month (Figure 6) and this region was a
perennial source of CO2 to the atmosphere (Figure 7). In
the Arabian Sea north of 10N, fluxes ranged from +2 to
+9 Tg C month1, with a net annual sea-to-air flux of +64
30 Tg C yr1 (Table 2). For comparison, Sarma et al.
[1998] estimated an annual flux of +47 Tg C yr1 for the
Arabian Sea from limited pCO2 observations. More recently,
Sarma [2003] estimated an annual net air-sea CO2 flux of
90 Tg C yr1 north of 10N using a model approach. In both
studies, no estimates of error were given for the air-sea CO2
flux rates, but it is expected that there is a similar level of
uncertainty as reported herein. Seasonally, the highest fluxes
occurred during both monsoon (i.e., SWM and NEM)
periods presumably associated with higher wind speeds
prevalent in this region. The lowest sea-to-air flux period

10 of 13

GB3020

BATES ET AL.: INDIAN OCEAN CARBON CYCLE, 1

GB3020

Figure 7. Monthly distribution of DpCO2 (matm; pCO2sea- pCO2atm) in the Indian Ocean. (top, left to
right) January to April; (middle, left to right) May to August; (bottom, left to right) September to
December.

occurred during the Spring Intermonsoon, a feature observed


in several other studies [George et al., 1994; Louanchi et al.,
1996; Sabine et al., 2000].
4.4. Bay of Bengal
[39] In the Bay of Bengal, there was a general northward
decrease in TCO2 (Figure 3), with the strongest gradients
present during the SWM and FIM periods. George et al.
[1994] observed similar gradients of TCO 2 (1920
1800 mmoles kg1) between 10N and 17N in March
April 1991. The seasonal DTCO2 and DnTCO2 changes
were greater than 100 mmoles kg1 across most of the Bay
of Bengal (Figure 5). In the Bay of Bengal, large variability
of surface pCO2 was observed but there are very few data to
compare directly [George et al., 1994]. From January to
September, pCO2 values are close to equilibrium or below,
similar to a very limited calculated pCO2 data set collected
in March April 1991 [George et al., 1994]. In contrast,
pCO2 values were seasonally higher (>400 matm) during the
October December period. This feature coincides with the
seasonally low salinities and high salinity normalized TCO2
values observed in the Bay of Bengal. Presumably, high

river (and carbon) output after the SWM elevates seawater


pCO2, thereby switching the Bay of Bengal from neutral/
sink status to a source of CO2 status.
[40] In the Bay of Bengal, fluxes ranged from 1 to
+2 Tg C month1 and the net annual sea-to-air CO2 flux
was +13 6 Tg C yr1 (Table 2). The highest sea-to-air
CO2 fluxes occurred during the SWM and FIM period,
when DpCO2 values were higher (i.e., seawater pCO2
higher). River discharge into the Bay of Bengal primarily
occurs during this period (e.g., June to September

Table 2. Annual Sea-to-Sea CO2 Flux for the Indian Ocean and
Regions of Interesta
Area

Air-Sea CO2 Flux, Tg C yr1

Indian Ocean
Arabian Sea
Bay of Bengal
10N 10S
10S 20S
20S 35S

+237
+64
+13
+180
+110
130

a
Ocean sources of CO2 to the atmosphere are given as positive values.
Ocean sinks for atmospheric CO2 are given as negative values.

11 of 13

GB3020

BATES ET AL.: INDIAN OCEAN CARBON CYCLE, 1

[Subramanian and Ittekkot, 1991]), and the higher sea-to-air


CO2 fluxes is presumably associated with enhanced transport of riverine TCO2 and remineralization of riverine
dissolved organic carbon within the Bay of Bengal.
4.5. 10N 35S Region of the Indian Ocean
[41] In the region of 10N 10S, seawater pCO2 values
were also higher than atmospheric values (Figure 7; DpCO2
values were positive) and the area was a perennial source of
CO2 to the atmosphere. Sea-to-air CO2 fluxes ranged from
+8 22 Tg C month1 and the net annual sea-to-air CO2
flux was estimated at +180 98 Tg C yr1. As a comparison, Louanchi et al. [1996] estimated a flux of +169 Tg C
for the 10S20N zone. The highest CO2 fluxes were
observed in the FIM/NEM periods (October to February)
associated with highest seawater pCO2 values (and highest
positive DpCO2 values) across this region. A secondary
peak in CO2 flux occurred in May and June at the onset of
the SWM (i.e., when wind speeds increase).
[42] Compared to the equatorial region, Arabian Sea and
Bay of Bengal, different seasonal patterns of air-sea CO2
fluxes were observed in the 10S 20S and 20S 35S
zones. During the NEM to SIM period (December
May), DpCO2 values were positive and these regions
(i.e., 10S 35S) were sources of CO2 to the atmosphere
(Figure 7). However, during the SWM to FIM period (i.e.,
austral winter and spring), DpCO2 values become negative
and these regions switch to oceanic sinks for atmospheric
CO2. In the 10S 20S zone, the reversal from CO2 source
to sink status occurs from July to September, and it was
estimated that the annual sea-to-air flux CO2 was 110
59 Tg C yr1 (Table 2). In a previous study, Louanchi et al.
[1996] estimated a zero flux for this region (i.e., 10S
20S).
[43] In the 20S 35S zones, air-to-sea fluxes were highest (40 24 Tg C month1) during the SWM/FIM
transition period. In contrast to other regions of the Indian
Ocean, this region is a net annual sink for atmospheric CO2 of
130 72 Tg C yr1 (Table 2). In a previous study, Louanchi
et al. [1996] estimated a CO2 flux of 147 Tg C yr1 for
the 20S 35S region of the Indian Ocean.

5. Conclusions
[44] Spatiotemporal variability of ocean carbon cycling
and air-sea CO2 exchange in the Indian Ocean are examined
using inorganic carbon data collected as part of the WOCE
and OACES cruises in 1995. The spatiotemporal distributions of TCO2, alkalinity and seawater pCO2 were estimated
for the Indian Ocean and regions of interest including
the Arabian Sea, Bay of Bengal and 10N 35S zones.
The range of alkalinity across the Indian Ocean was
200 mmoles kg1. The highest seasonal variability was
observed in the Bay of Bengal (>100 200 mmoles kg1) and
in the Arabian Sea (up to 60 mmoles kg1). In other regions,
seasonality of alkalinity was small (<20 mmoles kg1)
and linked to salinity variability. Similarly, the largest
seasonal ranges of TCO2 and pCO2 were observed in the
Bay of Bengal, presumably associated with seasonal river
discharge from the Himalayas. Over much of the Indian

GB3020

Ocean, seasonal variability of TCO2 and pCO2 was


<50 mmoles kg1 and <100 matm, respectively.
[45] A net sea-to-air CO2 flux of +237 132 Tg C yr1
(237  1012 g C yr1) was determined for the Indian Ocean.
The Arabian Sea, Bay of Bengal, and 10N10S zones
were perennial sources of CO2 to the atmosphere, although
the Bay of Bengal became a sink for CO2 in December. In
the 10S 35S zone, the ocean is a source of CO2 to the
atmosphere during the NEM and SIM period. During the
SWM to FIM period, the region shifts to a net oceanic sink
of atmospheric CO2.

[46] Acknowledgments. We would like to thank the Chief Scientists,


scientific personnel, and crews who collected TCO2 and alkalinity data for
all the WOCE Survey, JGOFS Arabian Sea Expedition, and NOAA
OACES cruises in the Indian Ocean. We are especially indebted to Doug
Wallace, Ken Johnson, Chris Winn, Catherine Goyet, Frank Millero, Bob
Key, Dick Feely, and Rik Wanninkhof for all their efforts collecting and
analyzing inorganic carbon parameters. Andrew Dickson is also thanked for
his long-standing efforts in producing certified reference materials (CRMs)
for TCO2. Burke Hales is thanked for his comprehensive review of an
earlier draft. Three reviewers are thanked for their constructive criticisms.
This research was funded by National Science Foundation grant OCE0096089. This is PMEL contribution 2589.

References
Anderson, L. A., and J. L. Sarmiento (1994), Redfield ratios of remineralization determined by nutrient data analysis, Global Biogeochem. Cycles,
8(1), 65 80.
Bates, N. R., and L. Merlivat (2001), The influence of short-term wind
variability on air-sea CO2 exchange, Geophys. Res. Lett., 28(17),
3281 3284.
Bates, N. R., A. F. Michaels, and A. H. Knap (1996), Seasonal and interannual variability of the oceanic carbon dioxide system at the U.S.
JGOFS Bermuda Atlantic Time-series Site, Deep Sea Res., Part II, 43,
347 383.
Bates, N. R., A. C. Pequignet, and C. L. Sabine (2006), Ocean carbon
cycling in the Indian Ocean: 2. Estimates of net community production,
Global Biogeochem. Cycles, 20, GB3021, doi:10.1029/2005GB002942.
Behrenfield, M. J., and P. G. Falkowski (1997), Photosynthetic rates derived from satellite-based chlorophyll concentrations, Limnol. Oceanogr.,
42, 1 20.
Borges, A. V. (2005), Do we have enough pieces of the jigsaw to integrate
CO2 fluxes in the coastal ocean?, Estuaries, 28(1), 3 27.
Borges, A. V., B. Delille, and M. Frankignoulle (2005), Budgeting sinks
and sources of CO2 in the coastal ocean: Diversity of ecosystems counts,
Geophys. Res. Lett., 32, L14601, doi:10.1029/2005GL023053.
Chavez, F. P., and R. T. Barber (1987), An estimate of new production in
the equatorial Pacific, Deep Sea Res., 34, 1229 1243.
Chung, S. N., K. Lee, R. A. Feely, C. L. Sabine, F. J. Millero, R. Wanninkhof,
J. L. Bullister, R. M. Key, and T.-H. Peng (2003), Calcium carbonate
budget in the Atlantic Ocean based on water column inorganic carbon
chemistry, Global Biogeochem. Cycles, 17(4), 1093, doi:10.1029/
2002GB002001.
Coatanoan, C., C. Goyet, N. Gruber, C. L. Sabine, and M. Warner (2001),
Comparison of two approaches to quantify anthropogenic CO2 in the
ocean: Results from the northern Indian Ocean, Global Biogeochem.
Cycles, 15(1), 11 25.
Ducklow, H. W., and S. L. McAllister (2005), Biogeochemistry of carbon
dioxide in the coastal oceans, in The Sea, vol. 13, The Global Coastal
Ocean-Multiscale Interdisciplinary Processes, edited by A. R. Robinson
and K. Brink, pp. 269 315, John Wiley, Hoboken, N. J.
Feely, R. A., et al. (2002), In situ calcium carbonate dissolution in the
Pacific Ocean, Global Biogeochem. Cycles, 16(4), 1144, doi:10.1029/
2002GB001866.
George, M. D., M. D. Kumar, S. W. A. Naqvi, S. Banerjee, S. Narvekar,
P. V. deSousa, and D. A. Jayakumar (1994), A study of the carbon
dioxide system in the northern Indian Ocean during premonsoon, Mar.
Chem., 47, 243 254.
Goyet, C., and D. L. Davis (1997), Estimation of TCO2 concentration
throughout the water column, Deep Sea Res., Part I, 44, 859 877.
Goyet, C., F. J. Millero, D. W. OSullivan, G. Eisheid, S. J. McCue, and
R. G. J. Bellerby (1998), Temporal variations of pCO2 in surface seawater of the Arabian Sea in 1995, Deep Sea Res., Part I, 45, 609 623.

12 of 13

GB3020

BATES ET AL.: INDIAN OCEAN CARBON CYCLE, 1

Goyet, C., C. Coatanoan, G. Eisheid, T. Amaoka, K. Okuda, and R. Healy


(1999), Spatial variation of total CO2 and total alkalinity in the northern
Indian Ocean: A novel approach for the quantification of anthropogenic
CO2 in seawater, J. Mar. Res., 57, 135 163.
Goyet, C., R. Healy, and J. P. Ryan (2000), Global distribution of total
inorganic carbon and total alkalinity below the deepest winter mixed
layer depths, ORNL/CDIAC-127, Carbon Dioxide Inf. Anal. Cent., Oak
Ridge Natl. Lab., U.S. Dep. of Energy, Oak Ridge, Tenn.
Gundersen, J. S., W. D. Gardner, M. J. Richardson, and I. D. Walsh
(1998), Effect of monsoons on the seasonal and spatial distributions of
POC and chlorophyll in the Arabian Sea, Deep Sea Res., Part II, 45,
2103 2132.
Hall, T. M., D. W. Waugh, T. W. N. Haine, P. E. Robbins, and S. Khatiwala
(2004), Estimates of anthropogenic carbon in the Indian Ocean with
allowance for mixing and time-varying air-sea CO2 disequilibrium, Global Biogeochem. Cycles, 18, GB1031, doi:10.1029/2003GB002120.
Kumar, D., S. W. A. Naqvi, M. D. George, and D. A. Jayakumar (1996), A
sink for atmospheric carbon dioxide in the northeast Indian Ocean,
J. Geophys. Res., 101, 18,121 18,125.
Lamb, M. F., et al. (2002), Consistency and synthesis of Pacific Ocean CO2
survey data, Deep Sea Res., Part II, 49, 21 58.
Lee, K. (2001), Global net community production estimated from the annual cycle of surface water total dissolved inorganic carbon, Limnol.
Oceanogr., 46, 1287 1297.
Lee, K., R. Wanninkhof, R. A. Feely, F. J. Millero, and T.-H. Peng
(2000), Global relationships of total inorganic carbon with temperature
and nitrate in surface seawater, Global Biogeochem. Cycles, 14(3),
979 994.
Lee, K., D. M. Karl, R. Wanninkhof, and J.-Z. Zhang (2002), Global
estimates of net carbon production in the nitrate-depleted tropical and
subtropical oceans, Geophys. Res. Lett., 29(19), 1907, doi:10.1029/
2001GL014198.
Lewis, E., and D. W. R. Wallace (1998), Program developed for CO2
system calculations, ORNL/CDIAC-105, Carbon Dioxide Inf. Anal.
Cent., Oak Ridge Natl. Lab., U.S. Dep. of Energy, Oak Ridge, Tenn.
Liss, P. S., and L. Merlivat (1986), Air-sea gas exchange rates: Introduction
and synthesis, in The Role of Air-Sea Exchange in Geochemical Cycling,
NATO ASI Ser. C: Math. Phys. Sci., vol. 185, edited by P. Buat-Menard,
pp. 113 128, Springer, New York.
Louanchi, F., and R. G. Najjar (2000), A global monthly climatology of
phosphate, nitrate and silicate in the upper ocean: Spring-summer production and shallow remineralization, Global Biogeochem. Cycles, 14(3),
957 978.
Louanchi, F., N. Metzl, and A. P. Poisson (1996), Modelling the monthly
sea surface pCO2 fields in the Indian Ocean, Mar. Chem., 55, 265 279.
Metzl, N., C. Beauverger, C. Brunet, C. Goyet, and A. P. Poisson (1991),
Surface water carbon dioxide in the southwest Indian Ocean sector of the
Southern Ocean: A highly variable CO2 source/sink region in summer,
Mar. Chem., 35, 85 95.
Metzl, N., A. P. Poisson, F. Louanchi, C. Brunet, C. Schauer, and B. Bres
(1995), Spatio-temporal distributions of air-sea fluxes in the Indian and
Antarctic oceans: A first step, Tellus, Ser. B, 47, 56 69.
Metzl, N., F. Louanchi, and A. P. Poisson (1998), Seasonal and interannual
variations of sea surface carbon dioxide in the subtropical Indian Ocean,
Mar. Chem., 60, 131 146.
Millero, F. J., E. A. Degler, D. W. OSullivan, C. Goyet, and G. Eischeid
(1998a), The carbon dioxide system in the Arabian Sea, Deep Sea Res.,
Part II, 45, 2225 2252.
Millero, F. J., K. Lee, and M. Roche (1998b), Distribution of alkalinity in
the surface waters of the major oceans, Mar. Chem., 60, 111 130.
Morrison, J. M., L. A. Codispoti, S. Gaurin, B. Jones, V. Manghnani, and
Z. Zheng (1998), Seasonal variation of hydrographic and nutrient fields
during the US JGOFS Arabian Sea Process Study, Deep Sea Res., Part II,
45, 2053 2101.
Poisson, A. P., N. Metzl, C. Brunet, B. Schauer, B. Bres, and D. Ruiz-Pino
(1993), Variability of sources and sinks of CO2 in the Western Indian and
Southern Oceans during the year 1991, J. Geophys. Res., 98, 22,759
22,778.

GB3020

Redfield, A. C., B. H. Ketchum, and F. A. Richard (1963), The influence of


organisms on the composition of seawater, in The Sea, edited by M. N.
Hill, pp. 27 77, John Wiley, Hoboken, N. J.
Sabine, C. L., and R. A. Feely (2001), Comparison of recent Indian Ocean
anthropogenic CO2 estimates with a historical approach, Global Biogeochem. Cycles, 15(1), 31 42.
Sabine, C. L., R. M. Key, K. M. Johnson, F. J. Millero, A. P. Poisson, J. L.
Sarmiento, D. W. R. Wallace, and C. D. Winn (1999), Anthropogenic
CO2 inventory of the Indian Ocean, Global Biogeochem. Cycles, 13(1),
179 198.
Sabine, C. L., R. Wanninkhof, R. M. Key, C. Goyet, and F. J. Millero
(2000), Seasonal CO2 fluxes in the tropical and subtropical Indian Ocean,
Mar. Chem., 72, 33 53.
Sabine, C. L., R. M. Key, R. A. Feely, and D. Greeley (2002a), Inorganic
carbon in the Indian Ocean: Distribution and dissolution processes, Global Biogeochem. Cycles, 16(4), 1067, doi:10.1029/2002GB001869.
Sabine, C. L., R. A. Feely, R. M. Key, J. L. Bullister, F. J. Millero, K. Lee,
T.-H. Peng, B. Tilbrook, T. Ono, and C. S. Wong (2002b), Distribution of
anthropogenic CO2 in the Pacific Ocean, Global Biogeochem. Cycles,
16(4), 1083, doi:10.1029/2001GB001639.
Sarma, V. V. S. S. (2003), Monthly variability in surface pCO2 and net airsea CO2 flux in the Arabian Sea, J. Geophys. Res., 108(C8), 3255,
doi:10.1029/2001JC001062.
Sarma, V. V. S. S. (2004), Net plankton community production in the
Arabian Sea based on O2 mass balance model, Global Biogeochem.
Cycles, 18, GB4001, doi:10.1029/2003GB002198.
Sarma, V. V. S. S., D. Kumar, and M. D. George (1998), The central and
eastern Arabian Sea as a perennial source of atmospheric carbon dioxide,
Tellus, Ser. B, 50, 179 184.
Sarma, V. V. S. S., et al. (2003), Carbon budget in the eastern and central
Arabian Sea: An Indian JGOFS synthesis, Global Biogeochem. Cycles,
17(4), 1102, doi:10.1029/2002GB001978.
Schott, F., and J. P. McCreary (2001), The monsoonal circulation of the
Indian Ocean, Prog. Oceanogr., 51, 1 123.
Smith, S. L., L. A. Codispoti, J. M. Morrison, and R. T. Barber (1998), The
1994 1996 Arabian Sea Expedition: An integrated, interdisciplinary investigation of the response of the northwestern Indian Ocean to monsoonal forcing, Deep Sea Res., Part II, 45, 1905 1916.
Subramanian, V., and V. Ittekkot (1991), Carbon transport by the Himalayan
rivers, in Biogeochemistry of Major World Rivers, edited by E. T. Degens,
S. Kempe, and J. E. Richey, pp. 157 168, John Wiley, Hoboken, N. J.
Takahashi, T., J. Olafsson, J. G. Goddard, D. W. Chipman, and S. C.
Sutherland (1993), Seasonal variation of CO2 and nutrients in the highlatitude surface oceans: A comparative study, Global Biogeochem.
Cycles, 7(4), 843 878.
Takahashi, T., et al. (2002), Biological and temperature effects on seasonal
changes of pCO2 in global ocean surface waters, Deep Sea Res., Part II,
49, 1601 1622.
Wanninkhof, R. (1992), Relationship between wind speed and gas exchange over the ocean, J. Geophys. Res., 97, 7373 7382.
Wanninkhof, R., and W. R. McGillis (1999), A cubic relationship between
air-sea CO2 exchange and wind speed, Geophys. Res. Lett., 26(13),
1889 1892.
Weller, R. A., M. F. Baumgartner, S. A. Josey, A. S. Fischer, and J. S.
Kindle (1998), Atmospheric forcing in the Arabian Sea during 1994
1995: Observations and comparisons with climatology and models, Deep
Sea Res., Part II, 45, 1961 1999.
Wyrtki, K. (1973), Physical oceanography of the Indian Ocean, in The
Biology of the Indian Ocean, edited by B. Zeitschel and S. A. Gerlach,
pp. 18 36, CRC Press, Boca Raton, Fla.


N. R. Bates and A. C. Pequignet, Bermuda Biological Station For


Research, Inc., 17 Biological Station Lane, Ferry Reach, Bermuda, GE01.
(nick.bates@bbsr.edu)
C. L. Sabine, Pacific Marine Environmental Laboratory, NOAA, 7600
Sand Point Way NE, Seattle, WA 98115-0070, USA.

13 of 13

Вам также может понравиться