Notable differences exist in the amounts of water gained or lost by each ocean. There is an excess of precipitation in the North Pacific, especially in the eastern tropical Pacific, and a dominance of evaporation in the Atlantic. The Pacific-Atlantic difference is thought to be maintained by water vapor transport across Central America [ Zaucker and Broecker, 1992] and the lack of any similar transport into the Atlantic from Africa. [ Schmitt, et al. 1989] also point out that the Atlantic is relatively narrow, and thus under the influence of continental air over a greater fraction of its area. The western tropical Pacific also gains water vapor exported from the Indian Ocean. As a result, the North Pacific is notably fresher than the other oceans, especially the North Atlantic.
The differences in surface water fluxes also mean that there is a
necessity for interbasin transport in the ocean. This could be
accomplished by a number of different paths, given the multiply- connected
nature of the ocean basins. Baumgartner and Reichel
[1975] constructed a scheme which assumed zero water transport across the
Atlantic equator, and integrated their surface flux estimates relative to
that point. However, Wijffels, et al. [1992] showed that
this arbitrary assumption could be replaced by direct oceanographic
transport measurements in Bering Strait. The scheme for ocean water
transports they developed differs dramatically from the earlier work
(Figure 4),
which had confused the flux
convergence in the Arctic basin with a flux. Indeed, they find that most
of the North Pacific water gain (nearly a Sverdrup), is transported
through Bering Strait. An assumption of zero water flux across the
Pacific Equator appears to be nearer the truth.
The interbasin water transport is accompanied by a salt transport. Since
there is no significant atmospheric pathway for salt, the salt flux
through coast to coast sections within a given sub-section of the
interconnected oceans must be the same. That is, the northward salt flux
across 24 N in the Pacific is equal to that through Bering Strait and
the southward transport across any zonal section in the Atlantic. This
constraint allows us to calculate the freshwater flux divergence between
sections. Such direct ocean estimates can be used to evaluate
integrations of surface fluxes [ Schmitt and Wijffels,
1993]. When complemented by western boundary current measurements and
Ekman transport (upper-ocean, wind-driven flow) estimates, the salinity and
geostrophic velocity fields (orthogonal to the pressure distribution) from
a hydrographic section can be used to calculate the salt flux. As noted
by [ Schmitt and Wijffels 1993], the salt flux can be
subtracted from the mass flux to give the net (fresh) water flux
divergence between sections in the same basin (or channel). Two examples
are given there, one for the freshwater gain between Bering Strait and
24 N in the Atlantic, the other for the gain between 24 N in the
Pacific and 24 N in the Atlantic. Estimates of the flow through
Bering Strait [ Coachman and Aagaard, 1988] and 24 N
data analyzed by Hall and Bryden [1982] allow a
calculation of the freshwater input between the sections. This gain is
estimated to be about 0.1 Sv. In contrast, the climatologies sum to about
zero water gain in the Arctic and Atlantic north of 24 N. This would
indicate the necessity to increase high latitude precipitation or runoff
estimates (or decrease evaporation) in the climatologies, though the
discrepancy is not substantial compared to the uncertainties which must
apply to both types of data (Figure 5)
. Also
shown is the recent freshwater flux estimate of Friedrichs
and Hall [1994] for 11 N. This latitude is at an extrema in water
transport, and also indicates that the climatologies have insufficient
water gain to the north. However, because of the uncertainties associated
with stronger, and more variable, Ekman and eddy fluxes at 11 N, the
error bars are larger than at 24 N. Field studies in the dynamically
interesting tropics will be very valuable for improving error estimates
there.
A more extensive area of ocean can be evaluated by considering sections at 24 N in both the Atlantic and Pacific. Since the salt flux in these two sections must be equal (though opposite in sign), by virtue of the Bering Strait through-flow, it is possible to calculate the total ocean freshwater gain north of 24 N. Only the section data of Hall and Bryden [1982] (Atlantic) and Bryden, et al., [1991] (Pacific) are necessary for this calculation, the precise magnitude of the Bering Strait through-flow is not critical. The result is a net southward transport of water by the ocean of 0.31 Sv. Combined with southward meridional river flows of about 10% of this value, the ocean/river return flow is in reasonable agreement with atmospheric water transport at this latitude [ Peixoto and Oort, 1983]. This is an interesting result, as there has been speculation that the ``missing petawatt'' [ Bryden, 1993] resulting from the discrepancies among the meridional heat flux estimates due to the Earth's radiation budget, atmospheric transport and ocean transport, might be found in unresolved latent heat transport in the atmosphere. However, since the return flow would be in the ocean, the data do not support an increase of more than double the presently estimated meridional transport of water at 24 N, which would be required for the extra petawatt [ Schmitt and Wijffels, 1993].
The more complete set of zonal lines to be collected during the World
Ocean Circulation Experiment
(WOCE) program will permit even more
points of comparison between the surface estimates and ocean flux
divergences. Indeed, preliminary results of J. Toole [
personal communication, 1994] indicate that the mid-latitude Southern
hemisphere oceans carry half the heat flux and twice the water flux of the
Northern hemisphere oceans. This striking asymmetry suggests that the
atmosphere/ocean system has multiple ways of satisfying the global
radiation budget (i.e., an enhanced latent heat flux in the
atmosphere/ocean system can make up for reduced sensible heat transport in
the ocean). It is quite possible that the enhanced water flux over the
southern ocean helps to decrease the heat-flux-carrying thermohaline
circulation there by freshening the surface waters and limiting deep
convection. In a stronger CO
greenhouse climate it is hypothesized
that the hydrologic cycle will intensify. Could it intensify
sufficiently to limit the thermohaline circulation in the North Atlantic?
This question is of obvious first order importance, yet we will have
trouble addressing it because our knowledge of precipitation and
evaporation over the ocean is so rudimentary. The difficulties of making
precipitation measurements at sea, the paucity of data for calculating
evaporation, and lingering uncertainty about the bulk formula combine to
make the net air--sea water exchange a very poorly known quantity. As
can be seen in Figure 5, there are significant discrepancies between
available climatologies for the North Atlantic which serve to illustrate
the problem. When summed over the North Atlantic the
Schmitt et al., [1989), numbers yield 10
m
/s more evaporation
than the Baumgartner and Reichel (1975) summation. This is over half the
flow of the Amazon. Even greater discrepancies arise when more recent
climatologies are considered for the evaporation field. For instance,
Isemer and Hasse, [1987] have revised the Bunker
estimates for the North Atlantic. While they adjust the exchange
coefficients in the bulk formulae downward, their revision of the Beaufort
wind scale leads to stronger trade winds and increased evaporation in the
tropics and subtropics. Their estimate of 20 to 40 cm/year more
evaporation south of 40 N suggests that an additional freshwater loss
of nearly 4 
10
m
/s may occur over the North Atlantic.
Thus, current estimates of net E-P over the basin differ by 2.5 times the
flow of the Amazon (1.9 
10
m
/s). Since the North
Atlantic is a small, relatively well sampled basin, we can only infer that
the uncertainties for the other ocean basins are even larger.
Zauker and Broecker [1993] find that atmospheric general circulation
models at present also do a poor job of representing the vapor transport
out of the Atlantic basin.