high frequency signal using the same filtering procedures. For the 

 high-frequency case this was done to remove the influence of the large 

 low- frequency spikes in the spectra. A high-pass filter with a cutoff 

 frequency of 0.1 hertz was used. 



For a closer look at the low-frequency information in the anchor 

 cable force data, a new time series was generated from the original re- 

 cord by sampling every eighth data point. To reduce aliasing of the 

 higher frequency energy in the original signal, each record was low-pass 

 filtered prior to this sampling using the filtering techniques previous- 

 ly outlined with a cutoff frequency of 0.2 hertz. The sampling of every 

 eighth point of the original time series gives a sampling interval of 

 4 seconds, a Nyquist frequency of 0.125 hertz and a record length of 

 256 points or 1,024 seconds. Five raw spectral points were averaged 

 together to give the final smoothed spectral estimates. This results 

 in a frequency resolution of 0.0049 hertz with 10 degrees of freedom per 

 spectral estimate. 



A total of 95 records was recorded at the site from 1330 hours on 

 30 December to 3 May 1975. There were no known equipment failures or 

 breakdowns except for one of the load cells going off scale at low tide 

 on the first tape (FH 7, NW load cell channel 3). A complete summary of 

 these events is given in Appendix G. Also, Figure 25 gives the relative 

 locations of the individual transducers. 



The wind direction in all cases is referred to the long leg, which 

 has a north-south compass bearing (magnetic declination in this area is 

 23° east). There are two wind-direction windows of interest. For the 

 long leg, the directions are approximately 50° to 95°; for the short leg, 

 130° to 160° (Figs. 24 and 25). 



Two storm events were chosen for presentation and further analysis. 

 These events cover records FH 7-6 through FH 7-12 and FH 11-8 through 

 FH 11-14 (Apps . G and H) . They were chosen because of their directions 

 relative to the short and long legs, respectively, and because of their 

 duration and magnitude. Both events lasted for over 7 hours with maxi- 

 mum windspeeds in excess of 35 miles per hour, with all the mean wind 

 direction within or close to the desired wind-direction windows. Appen- 

 dix H gives the pertinent wave spectra and transmission curves for the 

 above two events . 



The average overall response or transmission curves for the events 

 within each wind-direction window and for all the recorded data, are 

 given in Figure 27. These plots were obtained by averaging the square root 

 of the ratio of the transmitted to the incident wave spectras for the 

 records indicated f,OT each curve. Therefore, they have the same frequen-, 

 cy resolution of 0.0195 hertz. 



A puzzling feature in all the transmission response curves calculat- 

 ed from field data is the rise at lower frequency to a value near one and 

 then dropping off again. This can partially be attributed to a lack of 



58 



