- 124 



The "graininess" of the absolute gauge signal is dus to 

 resolution limitations of the transducer and digitizer. 



It is reasonable to conclude from inspection of the raw data, 

 then, that the dPl records should not be included in the spectral 



analyses, and that less faith should be placed in dPA because of its 



tendency to drift (as is exhibited by its changing mean). If these 



two channels do not provide valid data, it is not possible to 



generate the water surface curvature terms as described in Chapter IV 

 with any confidence. 



3. Record Length 



In beginning the conversion of the data to the frequency 

 domain, it was discovered that the DEC-10 computer system could not 

 handle the 17 minute long f our-samples-per-second data records, (4096 

 points) . So the data size was halved by averaging two adjacent 

 points in the time series. The Fast Fourier Transform, then, 

 operated upon a 17 minute record with an effective sampling rate of 

 two Hertz. (The corresponding Nyquist frequency was still safely 

 above the two rad/sec cut-off frequency used in the analysis 

 program.) The number of data points in the time series was halved 

 after the initial scan for bad points and calculation of the record 

 mean. 



