frequencies, since the amplitudes are squared. The smoothed 

 power spectrum values were obtained as follows: 



1 I" X=n-1 ^ ■ 



U(h)=~ \R(o) + Y, R ( X ) t 1 + cos— ) cos- 



n L x=i n n . 



where h - 0, 1, 2, 3 . . . . n index number of frequency (actual 



frequencies are given by h/(2h t) 

 cycles/min, At = 1/2 min), and 



X = 0, 1, 2, 3 .... n is the lag number 



The power spectrum {/(ft) was converted to units of vari- 

 ance per cycle per minute by multiplying by 2nAt where At is the 

 time interval between depth samples and is equal to 0. 5 minute. 



The results of the computation are shown in figures 17 

 through 21. The power spectrum of the 9°C-isotherm depth on 

 the offshore leg is plotted in figure 17. The importance of the 

 power spectrum lies in the peaks in the curve which indicate fre- 

 quencies (or periods) in the original data which may have been 

 obscured by "background noise. " Significant is the fact that this 

 power spectrum has a large number of peaks ranging in periods 

 from 3. 2 to 13. 5 minutes which corresponds to 0. 3 and 1. 3 miles, 

 but none of these is especially dominant. The greatest power is 

 in the low frequencies which show no peaks. There are, however, 

 several high frequency peaks but there is no dominant one. 



Similar computations of power spectrum of the 9°C-isotherm 

 depth (fig. 18) on the alongshore leg show a number of peaks with 

 periods of 3. to 21. 3 minutes (0. 3 to 2. 1 miles). 



The power spectrum for 9°C of the onshore traverse (fig. 19) 

 shows only a low frequency peak at 14. 7 minutes (1. 5 miles) and 

 quite weak higher frequency peaks. In comparison, the offshore 

 9° spectrum has higher power in all frequencies (by a factor of 10 

 or more) than that for the alongshore or onshore examples. 



A similar comparison can be made between the alongshore 

 14°C isotherm power spectrum and that for the onshore 16 °C iso- 

 therm. The power spectrum for the 14°C isotherm in the 



29 



