and all mean spectra in a row represent the same significant height 

 interval. Thus, the effect of increasing height on spectra with similar 

 peak periods and the effect of increasing peak period on spectra with 

 similar heights can be easily followed. A small niomber of plotted mean 

 spectra are not included in Appendix B because the spectra (a) usually 

 represented low or moderate wave heights with relatively few cases or 

 high wave heights with only one case for which the spectrum is shown in 

 Appendix A, or (b) were high-energy cases for Pt. Mugu which had obviously 

 been overconpensated for attenuation of pressure with depth. 



Mean spectra for low wave heights and peak periods less than 5 seconds 

 along the Atlantic coast show evidence of secondary peaks at low frequency 

 in many cases. The low- frequency peaks are broad and flat, are generally 

 located near 0.1 hertz, and are likely to represent low-energy swell in 

 the Atlantic Ocean. Corresponding broad, low- frequency secondary peaks 

 are also evident in mean spectra for Pt. Mugu on the Pacific coast. These 

 peaks are not evident in mean spectra for the gulf and Great Lakes gages, 

 although the Presque Isle pressure-gage spectra show a sharp secondary 

 peak which is consistently centered on 0.038 hertz. This low-frequency 

 peak at Presque Isle may not be physically meaningful. For all locations, 

 the differences between mean spectra are smaller than the differences 

 between the individual spectra used to compute the means. 



Mean spectra from the ocean gages for peak periods longer than about 

 10 seconds show evidence of secondary peaks at high frequency, especially 

 for high wave heights. For example, several selected mean spectra for 

 high wave, long-period groups are plotted in Figure 10 in dimensionless 

 form. Energy density was made dimensionless for each mean spectrum by 

 dividing by the total energy in the spectrum and converting to percent; 

 frequency was made dimensionless by dividing by the peak frequency. 



The figure clearly shows that the main secondary peak in each mean 

 spectrum is at twice the peak frequency. The spectrum for one cnoidal 

 wave (see Fig. 2) is also shown to clearly resemble the field spectra. 

 A section of the Lake Worth pen-and-ink strip-chart record taken at 

 nearly the same time as the Lake Worth spectrum in Figure 10 is shown 

 in Figure 11. The profiles of the larger waves closely resemble the 

 cnoidal wave profile. Many of the high-frequency secondary peaks in 

 mean spectra for high, long-period waves are probably indicative of 

 cnoidal-type wave profiles. Spectra for high, long-period waves in 

 which secondary peaks at about two or three times the peak frequency 

 represent independent wave trains are expected to be rare. 



The plots in Appendix B also show interesting variations with wave 

 height in mean spectra with short peak periods. In some cases, the 

 mean spectra for high waves exhibit secondary peaks at low frequency. 

 The emergence of the low- frequency secondary peaks for Nags Head is 

 shown by dimensionless spectra plotted in Figure 12. The figure also 

 indicates that the largest secondary peak is at one-half the frequency 

 of the dominant frequency. A section of pen-and-ink strip-chart record 

 corresponding to two high wave, short-period cases is shown in Figure 13. 



33 



