when the wave attenuation was high, the transmitted wave had multiple crests, 

 implying that the mattresses filled with water are a highly nonlinear mecha- 

 nism. When the breakwater was only partially effective, the waves trans- 

 mitted past the breakwater were only partially deformed. The effect of wave 

 diffraction in the mattress floating breakwater was significant (phenomena 

 which were precluded in the two-dimensional tests). The three-dimensional 

 laboratory tests indicated the water-filled floating mattress should have a 

 width of about twice the wavelength to be an effective wave attenuator. In 

 the effective range, the ratio of transmitted wave height-to-incident wave 

 height was approximately equal to the ratio of wavelength-to-breakwater 

 length. The mattress breakwater hovering in the top one-third of the water 

 was nearly as effective as the breakwater extending to the bottom of the 

 basin. 



Breakwater devices should be tested in the irregular waves of an ocean 

 environment. Water-filled mattresses were placed in a test installation in 

 San Francisco Bay where the waves are both wind- and ship-generated waves, 

 although somewhat smaller than the waves of the open Pacific Ocean, and were 

 tested by the same investigators. The field test site, located directly 

 across the bay from the bay entrance, was usually relatively calm in the morn- 

 ings; onshore winds in the afternoons generated waves with 1- to 2-second 

 periods and 0.5- to 1.5-foot heights. 



Diffraction was found to be as important in these field tests as had been 

 determined in the three-dimensional laboratory experiments. However, the two- 

 dimensional wave spectra characteristics of wind-generated waves appeared to 

 be more significant. Wind-generated waves have components that are radiated 

 with a considerable spread of directions, and have a two-dimensional energy 

 spectra. It was obvious that wave components were arriving at an angle in the 

 lee of the breakwater. The wave height sensor in the lee of the breakwater 

 was placed near enough to the structure to minimize these effects. Multi- 

 crested records were due to oscillations of the mattresses, not to either 

 diffraction or angular components of the two-dimensional wave spectra. 



Wind-wave tests on San Francisco Bay showed that if the significant wave- 

 length was used in data reduction and analysis, the coefficient of transmis- 

 sion, H t /Hj, was only about one-half the value obtained in the laboratory 

 for the same relative breakwater width, L/W. This was attributed to the 

 relative instability of the wind waves compared with the laboratory waves, 

 making the water-filled - floating mattress breakwater more effective in wind 

 waves than in swell. Computations based on the component wave period asso- 

 ciated with the maximum energy density did not correlate well with the labor- 

 atory results. There was no linear coherence between the incident waves and 

 the transmitted waves, indicating the breakwater acts as a nonlinear process. 

 To provide an observational format consistent with other data displays, Jones 

 (1974) reanalyzed Wiegel, Shen, and Wright's (1960) data, and displayed the 

 transmission coefficients, C t , as a function of relative water depth, L/d. 

 The asymptotic leveling of the curve with increasing L/d suggests a better 

 performance than most other barriers for very long waves (L/d < 8). However, 

 both the length and the submergence of the bags were relatively large. Jones' 

 (1974) reanalyzed data are presented in Figure 178. 



238 



