Y = 



aa+bb +ab-ab 



12 12 12 2 1 



112 2 



1/2 



(13) 



2. Results 



Individual Fourier coefficients were determined using the complex 

 fast Fourier transform on continuous current records equal to or longer 

 than 28.4 days (4,096 data points). Each record was divided into 

 sections containing 2,048 data points for both velocity components, and 

 each section was separately transformed. No special treatment was applied 

 to periods of wery low or zero current speeds. Frequency bands were 

 selected to place diurnal, inertia!, and semidiurnal frequencies near the 

 centers of separate frequency bands for most of the current records. To 

 compress the analysis results at frequencies greater than 0.15 cph, a band- 

 averaging technique was chosen to make the spectral estimates almost 

 equally spaced when plotted versus the logarithm of the frequency. To 

 obtain the analysis results for each current record, band averages were 

 computed over tne corresponding bands within each section of the record. 



Total energy spectra are shown in figures 3 through 10. Energy spectra 

 of currents are not shown for array A, because the record lengths are less 

 than 28.4 days. Table 4 provides the record lengths and degrees of freedom 

 which accompany the spectra. The 90-percent confidence bands shown on 

 figures 3 through 10 apply to frequencies less than 0.15 cph, the region of 

 the primary contributions to the measured currents. Each spectrum is char- 

 acterized by peaks at the local inertial frequency and at the semidiurnal 

 frequency which are indicated on the frequency scale by i and s, respectivel. 

 For arrays B, D, E, F, and G, the local inertial periods are shorter than 

 the diurnal period, and diurnal internal wave motion cannot occur. At these 

 arrays, the energy spectra show essentially no diurnal energy (marked by d) 

 above a noisy background. This finding is particularly evident for the 

 uppermost records from each location. For arrays C, H, and I, the local 

 inertial periods are also shorter than the diurnal period but are too near 

 the diurnal period to allow an accurate separation using the techniques 

 which have been described. 



The depth variations of energy densities for three frequency bands are 

 shown in figure 11. The diurnal energy is within the band centered at 

 0.038 cph, and the semidiurnal energy is within the band centered at 0.082 

 cph. For arrays B, D, E, and F the inertial energy is within the band 

 centered at 0.053 cph. For arrays C, H, and I, the inertial energy is 

 within the band centered at 0.038 cph. The energy densities computed for 

 array G are not included in figure 11, because spectral analysis results 

 at only three depths may not adequately specify the energy density profile. 

 Also presented in figure 11 is the depth variation of the Brunt-Vaisala 

 frequency in the vicinity of the arrays. The plotted Brunt-Vaisal^ frequency 



