incipient motion. Actual damage should be measured by before and after sound- 

 ings on a fine grid, but some judgment must be made as to the additional dam- 

 age that might have occurred in prototype from armor unit breakage. 



105. Wave transmission characteristics of the apparent optimum cross 

 section must also be verified. Tests of these design conditions simulating 

 the fully described forecast conditions at the site as accurately as possible 

 should be performed for the functional performance criteria and a number of 

 more extreme conditions. Again, these extremes should relate to the F(Hg , 

 T ) derived in Step 1. At least three H . (H ■) points should be measured 

 in order to verify or refine the economic loss function derived in Step 7. 

 Operational techniques should include efforts to accurately model reflection 

 and wave transmission by both overtopping and permeation. Transmitted waves 

 should be measured as time series comparable to time series measured of inci- 

 dent waves. Coherence and cross-correlation analyses should be performed for 

 the incident and transmitted time series along with computation of more common 

 spectral parameters. Individual runs of 100 or more waves are recommended for 

 the wave transmission tests in keeping with the widely accepted assumption of 

 stationarity in natural sea states. 



106. The measured f.D(H/Hj) data and H^(H^) data should be applied 

 in Steps 6 through 9 for the apparent optimum cross section. All its associ- 

 ated costs should then be adjusted according to the revised expected damages 

 and economic losses with the breakwater in place. Model tests often make sig- 

 nificant refinements to a design cross section obvious, and any such refine- 

 ments should be incorporated. Drastic changes to the original apparent opti- 

 mum cross section may require similar changes to be made to all the alterna- 

 tives and for Steps 5 through 9 to be repeated for these cross sections as 

 well. If the original apparent optimum is still indicated as the optimum 

 cross section, then no further model testing will be necessary. A new appar- 

 ent optimum should have its ^D(H/H.) and Hj^(H-) functions verified in as 

 thorough a manner as the first. I 



