data on a CDC 6600 machine with a 60-bit word size; when necessary, CERC con- 

 verted the data tapes to an ASCII format. 

 Analysis/Summarization procedures 



42. The CERC procedure for analyzing and summarizing digital wave data 

 was based on a Fast-Fourier Transform (FFT) spectral analysis procedure. The 

 final results were subjected to human editing and quality control before pub- 

 lic distribution (Thompson 1977; Harris 1974). The computer analysis routine 

 used 4,096 data points (1,024 sec of data sampled four times per sec) for each 

 data file processed. 



43. The program computed an initial distribution of the data along with 

 the first 5 moments of the distribution function and then edited the digital 

 data file, checking for data points out of the 0- to 5-V range, "jumps," and 

 "spikes." A jump is defined as a data value in excess of 2.5 standard devia- 

 tions from the previous data point, while a spike is a data point 5 standard 

 deviations or more from the mean. If fewer than 5 jumps, spikes, or points 



out of range, in a row, were found, the program linearly interpolated between ac- 

 ceptable data and replaced the erroneous data points. If either more than 5 

 in a row or a total of 100 bad data points for the file were found, the pro- 

 gram (a) stopped interpolation and further editing, (b) analyzed the data, and 

 (c) printed a flag indicating there was a problem with this data file. A 

 variance of less than 0.001 sq m indicated that the waves were calm; therefore 

 the record was not analyzed. 



44. After editing, the distribution function and first 5 moments of the 

 sea surface elevations were again computed. A cosine bell data window was 

 applied to increase the resolution for the energy spectrum of the file (use of 

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

 window, the program computed the variance spectrum (energy spectrum) using an 

 FFT procedure. 



45. The symbols H (defined as four times the standard deviation of 



m 



the sea surface elevation) and T (defined as that period associated with 

 the maximum energy density in the spectrum (Thompson 1977)) provided a con- 

 venient way to characterize the wave conditions contained in the data file; 

 they were more conducive to statistical summarization than the more complete, 

 but complex, description provided by the spectrum. 



46. After the data files were analyzed, the results were eliminated 

 for files that were flagged as bad or that appeared inconsistent with 



17 



