and in computations of transport. Generally, the depth of zero velocity will not be a level 

 surface, but will vary from place to place. In some early attempts to define this variable- 

 depth reference surface, it was placed at the depth of minimum dissolved oxygen. How- 

 ever, it is difficult to accept the assumption that the minimum of dissolved oxygen is a 

 consequence of no horizontal advection. Further, velocities obtained using this reference 

 surface were unrealistic. 



Defant calculated the depth of the reference surface by a method which is implicit 

 in the dynamic height observations. He used his method to construct a chart of the 

 depth of the reference surface in the Atlantic Ocean. A plot was made of pressure (or 

 depth) versus the difference in dynamic heights for each pair of adjacent stations. The 

 plotted curve represents the velocity component normal to the section. Any part of the 

 plotted curve that is parallel to the pressure coordinate represents a depth interval of 

 constant velocity. Defant reasoned that the velocity is more likely to be zero in this layer 

 of no shear stress than at any other level. In constructing his chart of the depth of this 

 reference surface, he made the additional assumption that the horizontal flow parallel 

 to the dynamic height section is also zero at this level. Although this method of selecting 

 a reference surface is subjective, its use has resulted in consistently realistic velocity 

 fields. This is essentially the method used by Russian investigators working with 1958- 

 1959 Vityaz data from the North Pacific Ocean (Chekotillo, 1961). 



Stommel (1956) described a method for determining the depth of no meridional motion 

 by equating the divergence of the Ekman transport with the divergence of the geostrophic 

 transport. Using the concept introduced by Fofonoff (1962) that the total transport can 

 be separated into a barotropic mode, a baroclinic mode, and an Ekman transport, Stom- 

 mel's method can be expressed as a relationship between these three modes: 



pmV b = ^(V-V B -Vg) 



where the Vs represent mass transport per unit width, u& is the velocity at the bottom, 

 h is the total depth, p m is the mean density of the water column, and the subscripts e and g 

 denote Ekman and baroclinic modes, respectively. Stommel did not consider the effects 

 of bottom topography on the divergence of the barotropic mode of transport. 



Murty and Rattray (1962), Fofonoff (1961), Favorite (1961), and Dodimead, Favorite, 

 and Hirano (1962) have compared baroclinic transports with wind-driven transports. 

 The method used by Murty and Rattray, in which the baroclinic transport is compared 

 with the wind-driven geostrophic transport to obtain a velocity near the bottom, is used 

 herein. The method is developed in appendix I and a numerical example is given in ap- 

 pendix II. 



Russian workers have determined deep currents by similar methods. Koshlyakov 

 (1961) reported computations in the North Pacific Ocean based on arguments similar to 

 Stommel's, but included the effects of variations in depth by means of an expression 

 similar to: 



pmVb' 



B(fh) 



dy 



In this expression the new terms t sx and t. w are wind stresses at the surface, (3 is the rate 

 of change of the Coriolis parameter with latitude, and / is the Coriolis parameter. This 

 expression is valid in areas where the product of the zonal component of velocity at the 

 bottom and the zonal component of the bottom slope is much smaller than the product 

 of the meridional components of the same variables. 



