404 BOUNDARIES OF THE SEA 



based on the assumptions that the friction, caused by vertical 

 shear, and the Coriolis force together balance the pressure gradient, 

 and that all motions disappear at great depths. The resulting 

 equation predicts a northward transport proportional to the wind 

 stress curl. Starting from a coastline where the normal transport 

 is zero, one can from the continuity equation derive also the zonal 

 transport all over the sea. The computation fails, however, when 

 a second coastline is met, and the model obviously breaks down, 

 at least locally. For theoretical reasons developed by several 

 investigators, it seems obvious that the breakdown will take place 

 at the western edges of the ocean basins. Accepting this idea, and 

 placing boundary currents at these edges so as to fulfill all con- 

 tinuity requirements, one would find for the entire ocean system 

 the circulation pattern given in Fig. 1. The wind stress data used 

 are obtained from Hidaka and the Scripps Institution of Ocean- 

 ography. The computation gives, of course, no information about 

 the structure of the boundary currents. Detailed studies of 

 conditions at the western edges have been made by Stommel, 

 Munk, Charney, and others. 



The map (Fig. 1) gives transport values that are of a correct 

 order of magnitude, and the streamline pattern is in reasonable 

 agreement with observations. Thus we may ha\e some confidence 

 in the simple Sverdrup model. Nevertheless, there is little hope 

 that we could be satisfied with the Sverdrup model after we obtain 

 sufficient, accurate data on wind stress and ocean transport to 

 test the model critically. In fact, there are some obvious weaknesses 

 of the Sverdrup model. To begin with, the model gives no infor- 

 mation about how the water of the boundary currents separates 

 at the western edges and then rejoins the Sverdrup circulation. 

 At the southern tips of the continents the boundary currents are 

 forced out into the sea, balancing the pure Sverdrup flow north- 

 ward into the basin. This water must be taken care of by some 

 mechanism other than the Sverdrup transport. 



A second difficulty is the effect of convective sinking of water 

 with winter cooling. This is a nonlinear process that may involve 

 large quantities of water, say in the North Atlantic, and is not 

 considered in the linear Sverdrup model. Other such nonlinear 



