The results for a(6) also coaform reasooably well to the model pce- 

 dlctioa. The values, plotted below and labelled "Intercept", represent 

 the amplitude (In meters) of the Fourier component at a wavelength of 1 

 kilometer. The mean intercept is 1.68 meters, indicating a relatively 

 rough topography, and corresponding to the "isotropic" term u in the 

 model. The anlsotropy of the surface is obvious from the large ampli- 

 tude (0.51 meters) of the model sinusoid (the v term). The maximum 

 value of the model occurs at an azimuth of 6^ « 115". This corresponds 

 to the normal to the linear trend (which was measured as 25° in the 

 full-coverage chart) as expected. These values fully parameterize the 

 effect of surface anlsotropy on the frequency domain description. 



Although the functional models for b(6) and a(6) appear to be of 

 the proper form, there are some obvious variations in the measured par- 

 ameters from the model. In order to give an Intuitive impression for 

 the degree that the generalized model departs from the actual measured 

 spectrum. Figure 6-7 illustrates a "worst case" example in which the 

 modelled a(6) differs from the observed a(6) by ~0.5 s. The selected 

 profile is from 6 - 115° (see Figure 6-6). Plotted with the measured 

 spectrum is the regression line derived from the functional model, 

 rather than the least-square fit usually shown. The model-derived spec- 

 trum provides an excellent representation of the spectrum and appears to 

 fall well within the estimation noise of the spectrum, even in this 

 "worst case" example. 



The second study area is shown in Figure 6-8. This data set repre- 

 sents a totally different style of sea-floor topography due to the geo- 

 logic environinent, which is controlled by sedimentological processes 

 rather than the tectonic setting of the Gorda Rise. The broad trend of 



