The rather widely scattered distribution of points in Figure 28 does 

 not permit definite conclusions. However, the normal curve for the At 

 interval of 11° F (Figure 19) can "be used to predict the weak thermocline 

 depth, if there is evidence of surface temperature increase. Convergence 

 and divergence effects probably enter the problem of predicting the weak 

 thermocline depth and could partly account for the wide scattering of the 

 points in Figure 28. Extent of decay in the lower part of the mixed layer 

 depends largely on the strength of flow in the upper part of the thermo- 

 cline . 



Variation of Thermocline Depth With Latitude 



Theoretical studies of thermocline depth have indicated a considerable 

 increase of mixed-layer thickness with decreasing latitude, if wind con- 

 ditions are the same. No empirical evidence to support such an increase 

 can be found, and such conclusions are not substantiated by this study. 

 Data from the North Atlantic at various locations between 35° and 56 N 

 do not indicate measurable differences of mixed-layer thickness. Perma- 

 nent convergence areas in subtropical zones may often be misleading; the 

 rather thick mixed layer in such areas cannot be attributed to mechanical 

 mixing. Strong negative salinity gradients in the thermocline often per- 

 mit deeper mixing in these areas because of reduced stability. Tnis fact 

 may be one reason for the fictitious influence of latitude. If some small 

 latitude effect is present, it may have been lost among several errors 

 involved. For the time being, however, its existence and magnitude remain 

 uncertain, and this prediction method can be considered valid for any 

 latitude . 



PRACTICAL PREDICTION OF THE THERMOCLINE DEPTH 



Verification 



The thermocline depth predicted by this method is the mean depth of 

 the interface between the mixed layer and the thermocline. Consequently, 

 verification of the prediction cannot be made with an individual BT obser- 

 vation, but must be determined from several (preferably 6 or more) BT 

 observations taken over a period of about one day, because the amplitude 

 of oscillation of the interface may be quite large. An individual BT is 

 also a rather poor source for information concerning the temperature gra- 

 dient in the thermocline and temperature distribution in the upper layers. 

 Indeed, prediction of mean thermocline depth will often be more accurate 

 and reliable than interpretation of the mean thermocline depth from a 

 single BT observation. 



Means of Prediction 



Table 5 lists values of wave parameters and minimum fetch and duration 

 for various wind speeds for fully developed sea. Only t) values are re- 

 quired if the curves in Figures 15 through 22 are used. This 77 value is 

 on the extreme left scale in the corresponding At graph. From the hori- 

 zontal intersection of the 77 value with the proper k(ii) curve (normal, 



59 



