convergent, or divergent), proceed vertically to the intersection with the 

 corresponding dashed curve, h(k). Horizontally from this intersection, the 

 thermocline depth (h) is determined on the second scale at the left of the 

 graph. 



For application of this prediction method in non fully developed sea, 

 B.1/0 and T max are taken from graphs for non fully developed sea in H.O. 

 Pub. 603, and 17 is computed in the usual way (17 = HT). 



Determination of the Stability Index 



The - stability index, At, must be determined and corrected for the 

 salinity gradient, if possible, before a prediction can be made. Improved 

 accuracy of At can be obtained by averaging temperature differences between 

 the surface and the ^00-foot level from k to 6 BT's covering a 2U-hour 

 period. Such opportunities seldom exist. In fact, predictions would be 

 less difficult if frequent BT's were made. Usually, the best situation 

 will include BT's taken up to several days earlier in a nearby area. 



The most likely sources of such observations are "Selected BT Charts" 

 of the North Atlantic which are transmitted daily by facsimile from the 

 Hydrographic Office. These charts are compiled from BT messages from vari- 

 ous ships and may contain as many as 22 observations distributed throughout 

 the Worth Atlantic. Observation points other than weather ship locations 

 may change from day to day. Reference to recent BT charts may provide 

 satisfactory observations for the area under consideration. 



At computed from several BT's, even though they might have been taken 

 a few days apart, is preferable because of the short range oscillations 

 of temperature at U00 feet due to internal waves. In the absence of advec- 

 tion, surface temperatures do not change significantly in periods of less 

 than 5 to 10 days. Mean temperature variation at 400 feet would be even 

 less; however, instantaneous temperature measurements may vary considerably 

 from the mean. When observations are not available for the immediate area, 

 the stability index can be evaluated by interpolation from 2 or more BT's 

 proximal to the prediction area, provided that neither was taken in an 

 area of permanent currents. 



Statistical computations of temperatures at the surface and at 400 

 feet and of At for limited areas, perhaps 5 degrees square, would be of 

 some help. However, data for such complete coverage are not available. 

 Figures 30, 31* and 32 represent monthly frequency distributions of At at 

 stations CHARLIE, DELTA, and ECHO. Table 6 contains ranges, means, and 

 standard deviations of temperatures and At at the surface and 400 feet for 

 these stations. 



Station DELTA is located in the North Atlantic Current; therefore, 

 statistical results for this station are quite different from those for 

 stations CHARLIE and ECHO. The mean of all standard deviations for DELTA 

 is 3^ percent larger than the mean of all standard deviations for CHARLIE 

 and 37 percent larger than the mean of all standard deviations for ECHO. 



61 



