introduce much error. This will not be the case for the shorter waves 

 and orientation of the array becomes important. 



Because of refraction, the curvature of a wave train changes, perhaps 

 only slightly, while propagating over the array. This change will intro- 

 duce differences in the direction at each gage and differences in the 

 direction sensed by different gage pairs, causing undetermined additional 

 errors in the computation of direction. To determine the magnitude of 

 these errors, two additional sets of simulated wave records were gene- 

 rated. The periods of the sinusoids combined were those for simulated 

 observation 3; the directions assigned were spread within a 10° arc for 

 the first set and 20° for the second. The last two computer outputs in 

 Appendix D show that spreads of the order of 16° and 32°, respectively, 

 resulted in computed directions. 



A stationary condition in time is usually assumed when developing 

 wave directional models. Indications are that this is not strictly 

 applicable at all times. 



The three factors discussed above are sufficient to account for the 

 inaccuracies encountered in the computations. 



5. Conclusions . 



The results of directional computations, for both simulated and field 

 wave data records, indicate three-gage arrays have some capabilities to 

 determine wave direction under certain conditions. These capabilities 

 depend on : 



(a) The dimension of the array and the water depth at the 

 site which place a lower limit on the wave period for which 

 possibly accurate directions may be computed. 



(b) The orientation of the array for the shorter periods. 



(c) The nature of the wave field; directional results for 

 wave trains with a narrow frequency distribution or where the 

 computed directions differ little at the adjacent spectral 

 periods might be meaningful. 



For wave trains with narrow frequency and directional width, and 

 period above 10 seconds, the three- gage arrays at Pt. Mugu yield direc- 

 tions to an estimated accuracy of 20°. 



At the Pt. Mugu site, 16-second waves may approach the coastline at 

 angles of 22° or less from the normal. The directional information pro- 

 vided by the array adds little to this and seems hardly cost effective. 



