estimate of their errors – those shown in Fig. 6.2.9
are based on the scale arguments discussed earlier
and so are rather conservative.
6.2.8 Summary
Direct estimates of oceanic freshwater transport
can be determined to useful accuracy from the
analysis of ocean sections. Most significant is that,
in contrast to values based on integrating surface
fluxes, the error in the direct ocean estimates are
largely independent of the size of the area of ocean
enclosed. While they can only shed light on the
long-term mean flux at locations where sections
have been made, they provide estimates of large
areal integrals that are not reliably gained from
any other source. It is the combination of these
estimates with those that do resolve both the true
spatial and temporal variability (e.g. satellite data
sets) that holds much potential.
Until surface flux products or atmospheric
models improve, the direct estimates are likely to
remain the most accurate of large areal averages of
the water flux between the ocean and the atmosphere. When synthesized globally, direct oceanic
freshwater transports can be used to assess the
performance of atmospheric and coupled models.
Estimates available to date show more agreement
with atmospheric model output than with direct
estimates of atmospheric moisture fluxes.
Presently, direct oceanic freshwater fluxes are
not as well reported or analysed as are the companion heat fluxes. While most estimates are fairly
consonant with each other and error estimates
made here based on simple scaling arguments, others are quite anomalous (such as Dobroliubov,
1997). Tracking down the source of these differences requires a detailed breakdown of the mechanisms making up these fluxes.
In respect of freshwater flux estimates, the full
potential of the WOCE data set is yet to be realized.
We can anticipate global, better-documented and
more accurate freshwater transport estimates to be
produced. In addition there may be great potential
in the idea of ‘tuning’ surface flux products using
direct ocean estimates to remove flux biases (e.g.
Isemer et al., 1989; da Silva et al., 1994). This may
lead to flux products that are accurate enough to
force ocean climate models directly with confidence
and thus remove the need to use the unphysical
surface relaxation boundary condition. This in turn
will open the door for the much more meaningful
use of salinity as a model diagnostic.
The limited number of direct oceanic freshwater
transport estimates presently available points to
the need to reassess our view of basin-wide balances and interbasin exchanges. Already, the idea
that the Atlantic Ocean is highly evaporative and
must import large amounts of oceanic fresh water
must be revised: the new estimates suggest nearzero gain down to 40°S (Saunders and King,
1995b). We can also deduce a larger freshwater
exchange than previously estimated between the
Southern Ocean and the basins to the north via the
injection of Antarctic Intermediate Waters into the
southern hemisphere subtropical gyres. It is noteworthy that this horizontal vЈ SЈ flux effected by
the subtropical gyres is not accounted for in simple
box models of the global thermohaline circulation
(e.g. Broecker et al., 1990). Such models must fold
this upper ocean transport into the deep water
component of the global thermohaline circulation,
confusing the role of freshwater transport as a
control on the circulation (Rahmstorf, 1996).
More careful and detailed studies of how the
ocean transports fresh water are required to pick
apart the relative role of the shallow wind-driven
gyres relative to that of the deep circulation in balancing the surface forcing. The data collected during WOCE will make such analyses possible.
Acknowledgements
Thanks is given to the following: Dr Hils from
the Global Runoff Data Center for providing the
Center’s runoff data; Sergie Dobroliubov for providing his results; Steve Covey for access to the
AMIP numbers; Todd Mitchell for advice and access
to some of the new climatologies; Siobhan O’Farrel
and Tony Hirst for use of the CSIRO coupled
model output; Stefan Rhamstorf for bringing some
of the new modelling work to my attention; Jim
Mansbridge for help with data processing; and the
International WOCE Office for sponsoring my
attendance at the WOCE Conference in Halifax.
This article was supported by the Australian Climate Change Research Project funded by Environment Australia.
SECTION 6 LARGE-SCALE OCEAN TRANSPORTS
488
are based on the scale arguments discussed earlier
and so are rather conservative.
6.2.8 Summary
Direct estimates of oceanic freshwater transport
can be determined to useful accuracy from the
analysis of ocean sections. Most significant is that,
in contrast to values based on integrating surface
fluxes, the error in the direct ocean estimates are
largely independent of the size of the area of ocean
enclosed. While they can only shed light on the
long-term mean flux at locations where sections
have been made, they provide estimates of large
areal integrals that are not reliably gained from
any other source. It is the combination of these
estimates with those that do resolve both the true
spatial and temporal variability (e.g. satellite data
sets) that holds much potential.
Until surface flux products or atmospheric
models improve, the direct estimates are likely to
remain the most accurate of large areal averages of
the water flux between the ocean and the atmosphere. When synthesized globally, direct oceanic
freshwater transports can be used to assess the
performance of atmospheric and coupled models.
Estimates available to date show more agreement
with atmospheric model output than with direct
estimates of atmospheric moisture fluxes.
Presently, direct oceanic freshwater fluxes are
not as well reported or analysed as are the companion heat fluxes. While most estimates are fairly
consonant with each other and error estimates
made here based on simple scaling arguments, others are quite anomalous (such as Dobroliubov,
1997). Tracking down the source of these differences requires a detailed breakdown of the mechanisms making up these fluxes.
In respect of freshwater flux estimates, the full
potential of the WOCE data set is yet to be realized.
We can anticipate global, better-documented and
more accurate freshwater transport estimates to be
produced. In addition there may be great potential
in the idea of ‘tuning’ surface flux products using
direct ocean estimates to remove flux biases (e.g.
Isemer et al., 1989; da Silva et al., 1994). This may
lead to flux products that are accurate enough to
force ocean climate models directly with confidence
and thus remove the need to use the unphysical
surface relaxation boundary condition. This in turn
will open the door for the much more meaningful
use of salinity as a model diagnostic.
The limited number of direct oceanic freshwater
transport estimates presently available points to
the need to reassess our view of basin-wide balances and interbasin exchanges. Already, the idea
that the Atlantic Ocean is highly evaporative and
must import large amounts of oceanic fresh water
must be revised: the new estimates suggest nearzero gain down to 40°S (Saunders and King,
1995b). We can also deduce a larger freshwater
exchange than previously estimated between the
Southern Ocean and the basins to the north via the
injection of Antarctic Intermediate Waters into the
southern hemisphere subtropical gyres. It is noteworthy that this horizontal vЈ SЈ flux effected by
the subtropical gyres is not accounted for in simple
box models of the global thermohaline circulation
(e.g. Broecker et al., 1990). Such models must fold
this upper ocean transport into the deep water
component of the global thermohaline circulation,
confusing the role of freshwater transport as a
control on the circulation (Rahmstorf, 1996).
More careful and detailed studies of how the
ocean transports fresh water are required to pick
apart the relative role of the shallow wind-driven
gyres relative to that of the deep circulation in balancing the surface forcing. The data collected during WOCE will make such analyses possible.
Acknowledgements
Thanks is given to the following: Dr Hils from
the Global Runoff Data Center for providing the
Center’s runoff data; Sergie Dobroliubov for providing his results; Steve Covey for access to the
AMIP numbers; Todd Mitchell for advice and access
to some of the new climatologies; Siobhan O’Farrel
and Tony Hirst for use of the CSIRO coupled
model output; Stefan Rhamstorf for bringing some
of the new modelling work to my attention; Jim
Mansbridge for help with data processing; and the
International WOCE Office for sponsoring my
attendance at the WOCE Conference in Halifax.
This article was supported by the Australian Climate Change Research Project funded by Environment Australia.
SECTION 6 LARGE-SCALE OCEAN TRANSPORTS
488