Table 6. Average spread in computed directions for 280 wave trains identified 

 in the high-resolution spectra of 44 field wave observations. 



Spectral width 



Avg. total spread 



Avg. spread in 



Cases 





in direction for 



direction for 







all periods 



period of train 





(hz) 



(°) 



(°) 



(No.) 



0.001 



21.8 



21.8 



96 



0.002 



30.9 



22.9 



58 



0.003 



33.6 



20.2 



41 



0.004 



41.5 



20.4 



22 



0.005 



38.7 



19.8 



21 



0.006 



43.8 



19.8 



16 



0.007 



48.2 



17.0 



5 



0.008 



84.5 



17.6 



8 



0.009 



53.0 



15.3 



6 



0.010 



49.3 



13.7 



3 



0.011 



38.0 



18.0 



1 



0.012 



81.0 



15.0 



1 



0.013 



48.5 



15.2 



2 



(a) The phase differences be known accurately or that the 

 probable error in their computed values be known. 



fb) The wave crests over the array site be long and 

 straight; thus, the waves must not have undergone appreciable 

 refraction. 



(c) The sea surface be stationary in time for the dura- 

 tion of the record and in space over the span of the array. 



The mathematical exercise in Appendix D indicates that the analysis 

 yields accurate phase differences only for strictly monochromatic condi- 

 tions. When this is not the case, no accurate estimate of the error 

 involved in the computation of direction can be given. This inability 

 is inherent to the computational procedure and cannot be resolved. 



Waves with periods over 8 seconds have been and are undergoing 

 refraction at the site of the array. Therefore, the wave crests are 

 not exactly straight. For the longer waves, with wavelengths at the 

 array site several times the gage separations, the curvature will not 



30 



