depth of the selected isotherms in the two 30-hour samples were sometimes as 

 great as the variability of the entire sample . 



Although we do not fully understand the causes, by means of statistical 

 time-series analysis, we are able to describe certain features of the tem- 

 perature variability associated with internal waves . 



The relatively complete 6-minute thermistor chain series of 5,332 obser- 

 vations extended from 0918 September 6, 1963 to 1624 September 28, 1963, a 

 period of 22 days, 7 hours. These data have been analyzed to determine 

 quantitatively the characteristics of both the low and high frequency inter- 

 nal waves which can be seen in Figures 4 and 5. Time-series computer analy- 

 sis programs, developed at Scripps Institution of Oceanography by Bullard, 

 Oglebay, Munk and Miller (1964), were used. 



In Figure 7 are presented power spectra for the 20°, 14° and 12.5° iso- 

 therms whose T-D relation is shown in Figure 6. The spectra for the 17° and 

 15.5° isotherms are essentially the same as that of the 20° isotherm and are 

 omitted from Figure 7 . High frequencies have been filtered from these spec- 

 tra, the cutoff frequency being .5 CPH . The significant low frequency peaks 

 occur at .08 CPH (the semi-diurnal tidal period) and at .055 CPH (the 18.6 

 hour inertial period at this latitude). For 90% confidence limits, the 

 ratio of computed to true values for these spectra fall between .49 and 

 1.85 



The ratio of the power in the semi-diurnal and inertial peaks to the 

 total power in these spectra (Fig 7) and the amplitude of these peaks are 

 given in Table 3. Amplitudes and ratios were summed over the peak and the 

 two adjacent frequencies . 



In the case of the 14° isotherm, the peak inertial frequency has shifted 

 to the right to .06 CPH. There is no peak at any of the inertial frequen- 

 cies on the 12.5° spectrum. There is, however, a small "blip" at .07 CPH on 

 the shoulder of the semi-diurnal peak. When the three spectra of Figure 7 

 are considered together, there appears to be a shift in the inertial peak 

 toward the high frequencies in the semi-diurnal peak. This shift may be an 

 artifact caused by a diminution in the height of the inertial wave component 

 as shown in Table 3 by the amplitude summed over the inertial frequencies 

 for the 12.5° isotherm. 



Defant (1940) reported a change in phase but no diminution of strength 

 of inertial components of currents with depth in his discussion of the 

 ALTAIR anchor station data in the Gulf Stream area north of the Azores 

 (44.6°N, 34.0°W, 16-20 June 1938). His interpretation of the ALTAIR data 

 was that the change in phase with d«pth indicated that the inertial oscil- 

 lations were coupled with semi-diurnal internal waves. The FLIP data, on 

 the other hand, show a diminution of amplitude of the inertial component but 

 no systematic phase shift in the internal waves within FLIP'S depth range, 

 although, occasionally, both low and high frequency internal waves in the 

 14° and 12.5° series (located near the bottom of the thermocline) are out of 



416 



