date and time, followed by a variable number of records necessary to obtain 20 

 minutes of data from all sensors at a sample rate of four values per second. 

 Each record contained 384 20-bit integer words (i.e., binary format). Each 

 integer word represented the computer units corresponding to the instantaneous 

 voltage output of the sensor. The above sequence of records was repeated for 

 each recording interval until the data tape was filled. 



The 20-bit word size is unusual but necessary because CERC processes the 

 data on a CDC6600 machine which has a 60-bit word size. CERC has the software 

 to convert the data tapes to an ASCII format. 



d . Data Analysis and Summarization Procedures. The CERC procedure for 

 analyzing and summarizing digital wave data is based on a fast Fourier trans- 

 form (FFT) spectral analysis procedure. The final results are also subjected 

 to an editorial review and quality control before public distribution (Harris, 

 1974; Thompson, 1977). The computer analysis routine uses 4,096 data points 

 (1,024 seconds of data sampled four times per second) for each data record 

 processed. The program first edits the digital data record, checking for 

 nonnumeric characters, jumps, and spikes (i.e., deviations greater than 2.5 

 and 5 standard deviations from the mean, respectively). If more than five bad 

 data points are found in a row or more than 2.5 percent of the digital values 

 are determined to be bad, the record is considered unsuitable for analysis and 

 rejected. For a few bad data points, the routine will linearly interpolate 

 between the erroneous values. If the record is determined to be suitable for 

 analysis, the distribution function of the sea-surface elevations and the 

 first five moments are computed. The variance (the second moment) and skew- 

 ness (the third moment) are checked to determine if full analysis of the data 

 record is warranted. Records with very low variance values and excessively 

 skewed distribution functions are not fully analyzed. After it is determined 

 that the record justifies full analysis, a cosine bell data window is applied 

 to increase the resolution for the energy spectrum of the record; use of the 

 data window is discussed by Harris (1974). After application of the data 

 window, the program computes the variance spectrum (energy spectrum) using the 

 FFT procedure. 



Significant wave height and peak spectral (or significant) period values 

 conveniently characterize the wave conditions contained in the data record and 

 are more conducive to statistical summarization than the more complete but 

 complex description provided by the spectrum. Although significant wave 

 height is defined as the average height of the highest one-third of the waves 

 in a record, experimental results and calculations based on the Rayleigh 

 distribution function show that the significant height is approximately equal 

 to four times the standard deviation of the wave record (U.S. Army, Corps of 

 Engineers, Coastal Engineering Research Center, 1977). The peak spectral wave 

 period (also referred to as the significant or peak period) for each digital 

 record is defined as that period associated with the maximum energy density in 

 the spectrum (Thompson, 1977). 



After 1 month of data has been analyzed, the significant wave height and 

 peak period values are segregated by gage and tabulated for an editorial 

 review. The editor checks for such things as unreasonable distribution of 

 the sea-surface elevations; clipping of the crest or troughs; inconsistencies 

 between successive observations; large trends in the 17-mlnute, 4-second data 

 record; and discontinuities in the data. After the data are edited, two-page 

 monthly sximmaries of significant height and peak period are generated for 

 inclusion in monthly reports. 



16 



