¥ AOU. Of the samples left available for comparison after these geographically and salinity-based
selections, only the 2–4 interpolated values with
the closest agreement of AOU between GEOSECS
and P2 were finally compared.
Even after this restrictive data-selection procedure, significant differences in salinity and AOU
remained between the GEOSECS and P2 values
for any given isopycnal. These were corrected for
by comparing salinity-normalized values of TCO 2
and by correcting the TCO 2 values to a common
value of AOU using a stoichiometric coefficient
⌬CO 2 : ⌬O 2 of 90.768. Uncertainties associated
with this stoichiometric coefficient are minimized
because the AOU difference between the two
cruises was always :17 mol kg
91 as a result of
the data selection.
Systematic TCO 2 differences of the order
13<8 mol kg
91 remained between GEOSECS
and P2 values in waters deeper than :27.5. It
was assumed that Excess CO 2 had not penetrated
this deep and that these offsets represented analytical errors in the GEOSECS data that were corrected for on a station-by-station basis. Results
from the comparison are illustrated as profiles in
Fig. 6.3.11. A clear temporal increase is seen down
to depths of the order of 750 m, with the nearsurface increase averaging about 22 mol kg
91
over the 20-year interval between the two cruises.
This near-surface rate of increase is similar to the
expected increase (17–19 mol kg
91
) if surface
waters had maintained a constant relative equilibrium with the increasing atmospheric pCO 2 .
Peng et al. (1998) reported a similar analysis
for the Indian Ocean WOCE Section I8NR along
80°E (sampled in 1995) based on comparison with
GEOSECS data collected in 1978. An assumed
invariance of deep water (92000 m) properties was
also used to correct the GEOSECS TCO 2 data
by a constant offset and values were then compared
on isopycnals within the main thermocline (300–
1000 m). No data-selection procedure was used,
and the comparison was based primarily on station
position (latitude). Only small GEOSECS-WOCE
differences were observed in TCO 2 -related properties such as AOU and alkalinity on the chosen
isopycnals, and the TCO 2 differences between the
cruises were adjusted for these differences using stoichiometric corrections. WOCE-era TCO 2 concentrations were found to be significantly higher than
the corrected GEOSECS values on all isopycnal surfaces examined down to :27.2. The temporal
increase was considerably greater between 20°S and
10°S (6–25 mol kg
91
) than in the equatorial zone
(almost undetectable at 1–5 mol kg
91
).
Summary
These studies have shown that decadal time-scale
increases in TCO 2 due to Excess CO 2 uptake are
indeed detectable above the natural TCO 2 variability using direct measurements. Because of their
reduced dependence on assumptions and models
such measurement-based comparisons will ultimately become a powerful and reliable test of
models’ ability to assess Excess CO 2 storage. At
present the need to correct historical data sets for
analytical inaccuracies of the pre-WOCE data sets
limits the use of such techniques to locations
where deep-water Excess CO 2 build-up can safely
be assumed to be negligible. If the improved data
accuracy and precision attained during the CO 2
Survey can be maintained, future comparisons
with the WOCE-era CO 2 Survey data set will be
much more robust and can be applied more
generally, as they will no longer require analytical
offsets to be estimated and applied based on comparisons with deep-water values.
6.3.4.3 Within-ocean transport of
inorganic carbon
Background
The meridional transport of carbon carried by the
ocean circulation is of significance for global
carbon cycle research for two main reasons:
1 The ‘natural’ preindustrial transport of carbon
between the hemispheres is a key constraint for
the modelling and interpretation of atmospheric
pCO 2 meridional gradients (Section 6.3.2.4).
2 The within-ocean transport divergence of Excess
CO 2 can be compared with regional estimates of
Excess CO 2 storage to infer regional distributions of the air–sea uptake of Excess CO 2 . As
previously discussed (Section 6.3.2.1), Excess
CO 2 uptake cannot be directly measured but is
modelled using carbon cycle GCMs. Inferences
concerning sensitivity of future uptake to climate
change may be sensitive to how this uptake is
modelled (Caldeira and Duffy, 2000). It will
therefore be useful to have a means to validate
models’ depiction of this process.
6.3 Storage and Transport of Excess CO 2 in the Oceans
513
Wallace
selections, only the 2–4 interpolated values with
the closest agreement of AOU between GEOSECS
and P2 were finally compared.
Even after this restrictive data-selection procedure, significant differences in salinity and AOU
remained between the GEOSECS and P2 values
for any given isopycnal. These were corrected for
by comparing salinity-normalized values of TCO 2
and by correcting the TCO 2 values to a common
value of AOU using a stoichiometric coefficient
⌬CO 2 : ⌬O 2 of 90.768. Uncertainties associated
with this stoichiometric coefficient are minimized
because the AOU difference between the two
cruises was always :17 mol kg
91 as a result of
the data selection.
Systematic TCO 2 differences of the order
13<8 mol kg
91 remained between GEOSECS
and P2 values in waters deeper than :27.5. It
was assumed that Excess CO 2 had not penetrated
this deep and that these offsets represented analytical errors in the GEOSECS data that were corrected for on a station-by-station basis. Results
from the comparison are illustrated as profiles in
Fig. 6.3.11. A clear temporal increase is seen down
to depths of the order of 750 m, with the nearsurface increase averaging about 22 mol kg
91
over the 20-year interval between the two cruises.
This near-surface rate of increase is similar to the
expected increase (17–19 mol kg
91
) if surface
waters had maintained a constant relative equilibrium with the increasing atmospheric pCO 2 .
Peng et al. (1998) reported a similar analysis
for the Indian Ocean WOCE Section I8NR along
80°E (sampled in 1995) based on comparison with
GEOSECS data collected in 1978. An assumed
invariance of deep water (92000 m) properties was
also used to correct the GEOSECS TCO 2 data
by a constant offset and values were then compared
on isopycnals within the main thermocline (300–
1000 m). No data-selection procedure was used,
and the comparison was based primarily on station
position (latitude). Only small GEOSECS-WOCE
differences were observed in TCO 2 -related properties such as AOU and alkalinity on the chosen
isopycnals, and the TCO 2 differences between the
cruises were adjusted for these differences using stoichiometric corrections. WOCE-era TCO 2 concentrations were found to be significantly higher than
the corrected GEOSECS values on all isopycnal surfaces examined down to :27.2. The temporal
increase was considerably greater between 20°S and
10°S (6–25 mol kg
91
) than in the equatorial zone
(almost undetectable at 1–5 mol kg
91
).
Summary
These studies have shown that decadal time-scale
increases in TCO 2 due to Excess CO 2 uptake are
indeed detectable above the natural TCO 2 variability using direct measurements. Because of their
reduced dependence on assumptions and models
such measurement-based comparisons will ultimately become a powerful and reliable test of
models’ ability to assess Excess CO 2 storage. At
present the need to correct historical data sets for
analytical inaccuracies of the pre-WOCE data sets
limits the use of such techniques to locations
where deep-water Excess CO 2 build-up can safely
be assumed to be negligible. If the improved data
accuracy and precision attained during the CO 2
Survey can be maintained, future comparisons
with the WOCE-era CO 2 Survey data set will be
much more robust and can be applied more
generally, as they will no longer require analytical
offsets to be estimated and applied based on comparisons with deep-water values.
6.3.4.3 Within-ocean transport of
inorganic carbon
Background
The meridional transport of carbon carried by the
ocean circulation is of significance for global
carbon cycle research for two main reasons:
1 The ‘natural’ preindustrial transport of carbon
between the hemispheres is a key constraint for
the modelling and interpretation of atmospheric
pCO 2 meridional gradients (Section 6.3.2.4).
2 The within-ocean transport divergence of Excess
CO 2 can be compared with regional estimates of
Excess CO 2 storage to infer regional distributions of the air–sea uptake of Excess CO 2 . As
previously discussed (Section 6.3.2.1), Excess
CO 2 uptake cannot be directly measured but is
modelled using carbon cycle GCMs. Inferences
concerning sensitivity of future uptake to climate
change may be sensitive to how this uptake is
modelled (Caldeira and Duffy, 2000). It will
therefore be useful to have a means to validate
models’ depiction of this process.
6.3 Storage and Transport of Excess CO 2 in the Oceans
513
Wallace