River runoff, on the other hand, is important for
the Atlantic freshwater balance and carries significant concentrations of TCO 2 . It is possible that
riverine inputs of carbon to the North Atlantic are
not correctly accounted for with the salinity normalization used by Broecker and Peng (1992) and
Keeling and Peng (1995).
Consideration of the net mass transport, including the freshwater transport divergence, is critical
for comparing transports at different sections
(including transport through Bering Strait) and for
assessing the effects of runoff on the carbon budget. An early review of Atlantic carbon transports
(Fig. 9 in Sarmiento et al., 1995, see also Section
6.3.4.3) erroneously compared some observationbased transport estimates, which were based on a
zero net transport assumption, with the net carbon
transport through Bering Strait. As a result the net
divergence of the carbon transport between Bering
Strait and 18°S was underestimated by as much as
0.75 PgC yr
91 (Holfort et al., 1998).
In the three WOCE-era studies discussed here
the carbon transports across each section and
through Bering Strait can be directly compared.
Most of these studies utilized sections covering the
entire Atlantic from east to west. (The AR7E section of Stoll et al. (1996) was unable to sample the
East Greenland Current due to ice-cover and did
not cover transport routes via the Labrador Sea
and English Channel.) All of these studies benefited from relatively high-resolution carbon data
along the sections.
Present-day carbon transports are composed of
a natural (or preindustrial) component and a component representing the transport of Excess CO 2 .
Both long-term storage (of Excess CO 2 ) and net
air–sea exchange can balance the contemporary
divergence of carbon transport. Riverine inputs
and transport divergence of dissolved organic carbon (DOC) can also play a role in balancing
regional budgets (Rintoul and Wunsch, 1991;
Sarmiento et al., 1995; Stoll et al., 1996; Holfort
et al., 1998). The CO 2 Survey was planned at a
time when a major controversy concerning the
measurement of DOC in seawater had broken out.
Consequently, few DOC measurements were made
along WOCE sections (except, significantly, as
part of the NOAA-OACES programme) and the
DOC transport divergence can usually not be
calculated. The contribution of this divergence to
regional carbon budgets, while significant, is likely
to be relatively small compared with carbon divergences driven by air–sea exchanges and river
inputs (cf. Holfort et al., 1998; Sarmiento et al.,
1995). Recent sections collected in the North
Atlantic have included DOC measurements so that
its contribution will be estimable in future studies
(D. Hansell, personal communication).
The Excess CO 2 transport can be estimated
directly by using the preformed CO 2 techniques to
calculate the Excess CO 2 (C excess ) distribution
along a section and substituting C excess for C in
equation (6.3.3). Subtraction of this Excess CO 2
transport from the total contemporary transport
gives an estimate of the preindustrial or natural transport (assuming ocean circulation is
unchanged since 1750). For the preindustrial era it
can be assumed that storage of inorganic carbon
was negligible. This makes it possible to estimate
the regional magnitude of air–sea and riverine carbon inputs for the preindustrial era. When compared with present-day estimates of riverine inputs
and net air–sea flux based on contemporary surface pCO 2 data, the anthropogenic perturbation of
the net air–sea flux can then be assessed on a
regional basis.
Needless to say, there are many uncertainties
in these approaches, some of which are difficult
to quantify. Holfort et al. (1998) identified, not
surprisingly, the Ekman transport (where this is
significant) and the specification of the barotropic
velocity distribution as significant sources of
uncertainty. Uncertainty in specification of the
deep, subthermocline circulation may have a larger
effect on the uncertainty of the carbon transport
than on heat tranport, especially in the Atlantic
Ocean. This is because significant lateral and vertical gradients in dissolved carbon concentrations
are found at all depths, whereas deep ocean temperatures are relatively uniform spatially below the
thermocline. Better constraints for the specification of the barotropic velocity distribution through
inverse modelling must therefore be developed.
Dissolved inorganic carbon concentrations do
vary seasonally in the oceans’ upper layers due to
the formation and degradation of organic material
and the air–sea exchange of CO 2 . Measurements or
estimates of the variability of dissolved and particulate as well as inorganic forms of carbon in winter
and summer along a section are required to address
this, but such data are rarely available. The possibility of correlation of seasonal concentration
6.3 Storage and Transport of Excess CO 2 in the Oceans
515
Wallace
the Atlantic freshwater balance and carries significant concentrations of TCO 2 . It is possible that
riverine inputs of carbon to the North Atlantic are
not correctly accounted for with the salinity normalization used by Broecker and Peng (1992) and
Keeling and Peng (1995).
Consideration of the net mass transport, including the freshwater transport divergence, is critical
for comparing transports at different sections
(including transport through Bering Strait) and for
assessing the effects of runoff on the carbon budget. An early review of Atlantic carbon transports
(Fig. 9 in Sarmiento et al., 1995, see also Section
6.3.4.3) erroneously compared some observationbased transport estimates, which were based on a
zero net transport assumption, with the net carbon
transport through Bering Strait. As a result the net
divergence of the carbon transport between Bering
Strait and 18°S was underestimated by as much as
0.75 PgC yr
91 (Holfort et al., 1998).
In the three WOCE-era studies discussed here
the carbon transports across each section and
through Bering Strait can be directly compared.
Most of these studies utilized sections covering the
entire Atlantic from east to west. (The AR7E section of Stoll et al. (1996) was unable to sample the
East Greenland Current due to ice-cover and did
not cover transport routes via the Labrador Sea
and English Channel.) All of these studies benefited from relatively high-resolution carbon data
along the sections.
Present-day carbon transports are composed of
a natural (or preindustrial) component and a component representing the transport of Excess CO 2 .
Both long-term storage (of Excess CO 2 ) and net
air–sea exchange can balance the contemporary
divergence of carbon transport. Riverine inputs
and transport divergence of dissolved organic carbon (DOC) can also play a role in balancing
regional budgets (Rintoul and Wunsch, 1991;
Sarmiento et al., 1995; Stoll et al., 1996; Holfort
et al., 1998). The CO 2 Survey was planned at a
time when a major controversy concerning the
measurement of DOC in seawater had broken out.
Consequently, few DOC measurements were made
along WOCE sections (except, significantly, as
part of the NOAA-OACES programme) and the
DOC transport divergence can usually not be
calculated. The contribution of this divergence to
regional carbon budgets, while significant, is likely
to be relatively small compared with carbon divergences driven by air–sea exchanges and river
inputs (cf. Holfort et al., 1998; Sarmiento et al.,
1995). Recent sections collected in the North
Atlantic have included DOC measurements so that
its contribution will be estimable in future studies
(D. Hansell, personal communication).
The Excess CO 2 transport can be estimated
directly by using the preformed CO 2 techniques to
calculate the Excess CO 2 (C excess ) distribution
along a section and substituting C excess for C in
equation (6.3.3). Subtraction of this Excess CO 2
transport from the total contemporary transport
gives an estimate of the preindustrial or natural transport (assuming ocean circulation is
unchanged since 1750). For the preindustrial era it
can be assumed that storage of inorganic carbon
was negligible. This makes it possible to estimate
the regional magnitude of air–sea and riverine carbon inputs for the preindustrial era. When compared with present-day estimates of riverine inputs
and net air–sea flux based on contemporary surface pCO 2 data, the anthropogenic perturbation of
the net air–sea flux can then be assessed on a
regional basis.
Needless to say, there are many uncertainties
in these approaches, some of which are difficult
to quantify. Holfort et al. (1998) identified, not
surprisingly, the Ekman transport (where this is
significant) and the specification of the barotropic
velocity distribution as significant sources of
uncertainty. Uncertainty in specification of the
deep, subthermocline circulation may have a larger
effect on the uncertainty of the carbon transport
than on heat tranport, especially in the Atlantic
Ocean. This is because significant lateral and vertical gradients in dissolved carbon concentrations
are found at all depths, whereas deep ocean temperatures are relatively uniform spatially below the
thermocline. Better constraints for the specification of the barotropic velocity distribution through
inverse modelling must therefore be developed.
Dissolved inorganic carbon concentrations do
vary seasonally in the oceans’ upper layers due to
the formation and degradation of organic material
and the air–sea exchange of CO 2 . Measurements or
estimates of the variability of dissolved and particulate as well as inorganic forms of carbon in winter
and summer along a section are required to address
this, but such data are rarely available. The possibility of correlation of seasonal concentration
6.3 Storage and Transport of Excess CO 2 in the Oceans
515
Wallace