Ai,i-ini 15 



variation depends on the choice of the two parameters, density- 

 difference and thickness of top layer. 



By varying the two parameters we can get good qualita- 

 tive agreement with observations of the shape' of the thermocline. 

 In Fig. 9y a cross-section of the computed thermocline is shown 

 for four pairs of values of the parameters* Because of the 

 rather vague definition of the actual thermocline, we cannot 

 state specifically the extent of quantitative agreement between 

 our computed results and the observed values. Consider, 

 howGVGr, the curve in Fig, 9 with a depth of the top layer of 

 200 meters and a density difference of 0,0025. For that curve 

 the results disagree by a factor of three when compared to some 

 of the measurements of the thermocline off Chesapeake Bay [10]» 



The two-layer non-steady problem constitutes an attempt 

 to drop the assumption made in the one-layer problem that the 

 velocities vanish at some great depth. As a consequence the 

 problem becomes much more complicated and it is necessary to 

 introduce some other simplifying assumptions, viz* , to neglect 

 the shear forces at the bottom and at the thermocline. This 

 may have far-reaching effects. These simplifications notv;ith- 

 standing, we were unable to obtain a solution, A brief descrip- 

 tion of our attempts at such a solution followsn 



First, the equations are non-dimensionalized as in the 

 one-layer case. The integrated continuity equation for the top 

 layer now contains the time derivative of the magnitude of the 

 deviation of the thermocline from its equilibrium position. 

 Since this term is very large, the perturbation method used in 



