The exchange of water between ocean and atmosphere has consequences for ocean flows beyond the requirement of mass conservation. The simplest (yet largely neglected) response is that governed by vorticity conservation requirements which derive from the conservation of angular momentum on our rotating planet. Goldsbrough[1933] first showed how surface mass fluxes drive steady flows in the ocean, in what may be viewed as a precursor to the well-known Sverdrup relation. That is, it is readily shown [ Huang and Schmitt, 1993] that the steady vorticity balance in the ocean interior is given by:
where f is the Coriolis parameter, 
its meridional gradient,
the wind stress and V the meridional velocity. Goldsbrough (who
considered only the (E-P) term above) suggested that a
subtropical-gyre-like circulation would result from an interior
precipitation regime and an evaporative western boundary region. This is
of course quite unrealistic; Stommel [1957] suggested
that the interior Goldsbrough circulation could be closed by adding
western boundary currents. The complete gyres can be refered to as the
Goldsbrough-Stommel circulation.
Huang and Schmitt [1993] have presented a preliminary calculation of
the Goldsbrough-Stommel circulation for the world ocean
(Figure 6).
While E-P is smaller than Ekman
convergence ( 1 m/yr vs. 20 m/yr), the flows generated are generally
opposite to the wind driven currents. In the North Atlantic, a western
boundary current is required which reaches 2 Sverdrups southward at 35
N, in direct opposition to the Gulf Stream. Other basins have similarly
strong boundary currents. We suggest that these adverse flows can affect
the separation latitude of the western boundary currents. For the Gulf
Stream, the effect may be as large as 75 km. As there are strong
horizontal temperature gradients associated with the western boundary
currents, any small shifts in separation latitude could have large impact
on SST distributions in the western portions of the subtropical gyres.