The periods assigned to the sinusoids giving rise to simulated obser- 

 vations 1, 2, 5, and 6, differed by exactly three spectral periods (see 

 Table 3). The spectra for these four observations (Figs. D-1, D-2, D-5, 

 and D-6) exliibit three maxima separated by two minima. The energy at 

 these two minima may be interpreted as due to spillover since no sinu- 

 soids were combined with the corresponding periods. 



The high-resolution spectra from field wave observations are dis- 

 cussed later; however, the use of the minima in these spectra to esti- 

 mate spillover and noise is discussed here. The average energy at the 

 minima between 25 and 7 seconds in the high- resolution spectra of the 

 field observations was used as a measure of spillover and noise. Only 

 spectral periods displaying an energy content at least twice this "back- 

 ground" energy were interpreted as possibly arising from physical wave 

 components in the wave field. 



The average directions resulting from the 10 three-gage arrays are 

 given in Table 4 (last column). Only results of computations at the 

 spectral periods closest to the assigned periods are shown. Tlie table 

 shows that for wave trains 1, 2, and 3 the computed directions for the 

 10 arrays agree with the input directions to within 1°. 



The directional results for wave train 4 are correct only for spec- 

 tral period 62, The main difference between this train and wave train 2 

 is the narrower spectral width. Wave train 2 gave the correct directions 

 for all spectral periods; wave train 4 did not„ 



For simulated wave trains 5 to 8, the average directions seem meaning- 

 less. To determine whether these poor results were due to a programing 

 deficiency, another eight sets of simulated records were generated, inter- 

 changing periods and directions. The computer output for the simulated 

 observations is in Appendix D. This appendix and Table 5 should be 

 referenced in the following discussion of additional simulated wave 

 trains. 



The directional results for the sinusoids with periods clustered 

 around 16 seconds were of the same quality regardless of the assigned 

 direction. However, the directional results for the 8-second sinusoids 

 indicate that the capability to sense the correct direction for these 

 shorter waves depends on the orientation of the three-gage array rela- 

 tive to the direction of propagation of the incoming wave. The result- 

 ing directions, which differed by less than 87° from the assigned direc- 

 tions, are given in Table 5. The top of each column in the table shows 

 the shape and orientation of the array. These results are not surprising 

 since the effective gage separation for the different gage pairs varies 

 with orientation relative to direction of wave propagation. Table 5 

 also shows that the more nearly equilateral arrays have wider direction 

 discemability. The design considerations for the array indicated an 

 effectiveness for wave periods between 25 and 7 seconds (Borgman and 

 Panicker, 1970). 



25 



