As discussed in Section 1, oceanographic flux divergences can be used to constrain the net water exchange between ocean, atmosphere and land. Some aspects of making such estimates are discussed here. Depending on the volume in question, river flows can be an important part of the water budget. These are generally monitored to sufficient accuracy and are available from the Global Runoff Data Center in Koblenz, Germany. However, some rivers, such as the Amazon, are gauged well upstream of the mouth, and the discharge to the ocean may not be adequately known.
The latitudes of most interest for the global water cycle are
distinct from those that best define the meridional heat flux. Ocean
water fluxes reach extrema at around 10 and 45 N and S (Figure 2),
reflecting evaporation in mid-latitudes and precipitation at high and
low latitudes (Figure 3). In contrast, heat flux is maximal at
25--30 N and S. If we suppose that freshwater flux divergence
could be estimated to about 
m
/s from
hydrographic data at 45 N and 10 N in the Atlantic,
then this would correspond to an error in E-P of about 12 cm/yr,
significantly less than the differences between present
climatologies. With few zonal ocean sections having been occupied,
and none on a seasonal basis, it is difficult to estimate the error
of such calculations due to unsampled seasonal and interannual
variability. Friedrichs and Hall [1993] estimate the error at
11 N to be twice that at 24 N in the Atlantic, based
on only one section. The ``direct'' method of flux calculation
appears to yield an accurate estimate of mean heat flux, which, in
the Atlantic, is largely carried by the baroclinic overturning cell.
Estimates of heat flux at 24 N from section data taken
3 different times over 35 years are found to be identical [ Hall and
Bryden, 1981; Roemmich and Wunsch, 1985;
Parilla et al., 1994].
Recent model runs [ Boning and Herrmann, 1994] indicate that the North
Atlantic response to wind stress variations is mainly barotropic, and
thus does not substantially change the density field. While other
oceans (and latitudes) may be more sensitive to seasonal variability,
these results do suggest that further development of zonal
hydrographic sections would be an important contribution toward
understanding the global water and heat cycles.
The WOCE Hydrographic Programme will improve the global coverage of sections, however, long time series of repeat sections would be a very valuable follow-on to the first order picture developed from the WOCE data set. Occupation of sections with seasonal resolution may be justified at some latitudes (10 N), while others with a weak seasonal cycle (24 N) might be occupied every few years. In the Atlantic a zonal section can be completed in less than 3 weeks. As a example, we note the 27 occupations of a section at 165 E, from 20 S to 10 N, over a 7-year period by the French, U. S., and Chinese during TOGA. With the number of countries with strong oceanographic programs being quite high around the Atlantic, it should be relatively easy to focus resources on repeat occupations of several zonal sections in that ocean. The resulting improvement in ocean heat and freshwater flux estimates would provide valuable constraints on surface fluxes derived from satellites, climatologies, and weather forecasting models. They would also provide insight into the mechanisms of freshwater transport that is presently lacking.
Strait monitoring will also be necessary to define the water budget for
certain regions. Bering Strait is the main mass flux conduit for water
returning from the Pacific to the Atlantic, and an important source of
buoyancy for the Arctic Ocean. Coachman and Aagaard
[1988] estimate a standard deviation in interannual variability in Bering
Strait transport of about 
10
m
/s. This seems large
enough to have climatic significance Shaffer and Bendtsen,
1994], especially for the Arctic ice budget. A more complete monitoring
program, including salinity measurements, would be a very valuable climate
indicator for the high latitude fresh water budget.
The two layer flow at Gibraltar is the result of a loss of water to evaporation in the Mediterranean basin that represents one of the larger terms in the North Atlantic mass budget. Fram Strait carries freshwater and ice from the Arctic to key deep convection areas of the North Atlantic. It is clear that time series measurements of transports in these straits would be very valuable in deciphering oceanic climate response to variations in freshwater forcing. Thus, establishment and long term maintenance of transport monitoring programs in these straits should be an immediate priority for new ocean monitoring programs. For the Mediterranean, the constraint on the water budget provides a link with the heat budget through the latent heat term [ Garrett et al., 1992]. Improved Gibraltar flux measurements could be instrumental in deciphering the ongoing controversy about exchange coefficients and radiation terms in the surface heat fluxes.