estimates for all bands 0.01 hertz wide between approximately 30 and 3 

 seconds were obtained for the 10 arrays. The results displayed discrep- 

 ancies of the order of 20" for those bands with central periods above 10 

 seconds and of 180° for those with shorter central periods. It had been 

 expected that the array would yield direction to better than 20° and for 

 periods between 25 and 7 seconds. 



To isolate problems associated with the calculations, the propagation 

 of narrow-banded wave trains across the array was simulated in a computer. 

 The computational model was applied to the simulated observations using 

 the maximum frequency resolution available from spectral computations 

 based on 20-minute records. It was found that the directional results 

 obtained with this model are highly dependent on the spectral width, 

 both in frequency and direction, of the wave train involved and on the 

 relationship between wavelength at the site and gage separations. The 

 assigned directions were recovered within 1° for 16-second waves when the 

 frequencies of the spectral components in the wave train differed by 0.003 

 hertz or more, and the directions were spread within a 5° arc. This fre- 

 quency separation results in a minimum difference in periods of 0.7 second 

 for waves with periods near 16 seconds. 



Application of the same analyses to field wave pressure records with 

 standard deviations above 0.61 meter (2 feet) resulted in an average dis- 

 crepancy of 20° among computed directions for narrow-banded wave trains 

 with periods longer than 10 seconds. Larger discrepancies resulted for 

 shorter periods. ' Thus, accuracies no better than 20° can be expected for 

 wave directions resulting from three-gage arrays. 



1. The System . 



A minimum of three gages is required for a unique determination of 

 wave direction by most proposed models. Since these models make a few 

 assumptions about the nature of ocean waves which have not been estab- 

 lished, some redundance was thought to be necessary which would require 

 a minimum of four gages. However, it was agreed that a five- gage array 

 would increase the probability of redundance in the ocean environment. 

 An array was designed at CERC by Leon E. Borgman, statistician- engineer, 

 while on sabbatical leave from the University of California, Berkeley, 

 when the experiment was being planned. He investigated the directional 

 resolving power of several array geometries and concluded that the pat- 

 tern shown in Figure 3 would be the most suitable for the conditions to 

 be expected at Pt. Mugu (Borgman and Panicker, 1970). 



The array was installed off Pt. Mugu, approximately 80.47 kilometers 

 (50 miles) northwest of Los Angeles (Fig. 4), in about 9.14 meters (30 

 feet) of water, 0.76 meter (2.5 feet) from the bottom. The gages in the 

 array are pressure transducers developed mostly at CERC (Williams, 1969). 

 The heart of the system is a Fairchild pressure transducer which is potted 

 inside a 2-inch Plexiglas tube (Fig. 5) (Peacock, 1974). A plastic tube 

 filled with silicone oil transmits the pressure from the seawater to the 

 pressure transducer. The silicone oil is separated from seawater by 



