6-5. Amplitude estimates appear as Irregular contours. The amplitude 

 values were log transformed before plotting, and these values are plot- 

 ted at Integer Increments. Contours are drawn each .5 order of magni- 

 tude of amplitude. The data set, and therefore its amplitude spectrum. 

 Is oriented with the columns parallel to longitude and rows parallel to 

 latitude. 



Plotted with the spectrum In heavy lines is the two-dimensional 

 spectrum as modelled from one-dlmenslonal profiles by the method 

 described In the previous section. Because the grldded data base used 

 In the analysis was spaced evenly in latitude and longitude, the spec- 

 trum as a function of spatial frequency Is necessarily distorted. High 

 frequency noise associated with the east-west oriented track lines 

 appears as a smearing of the contours In the vertical and horizontal. 



The simple model spectrum explains most of the variance in ampli- 

 tude. The llneation of the topography with a strike of 6 » 25" can be 

 easily seen as an elongation of the contours in the cross-strike (6 » 

 115") direction. The degree of anisotropy (v) term in the n»del deter- 

 mines the elongation of the contours. The model appears to overestimate 

 the amplitudes in the high frequencies slightly, which would indicate a 

 slightly lower slope (b) value than that derived by the described 

 method. An improved fit results if the value b <■ ~1*5, the value 

 derived for the ridge crest in Chapter 5, is used. Overall, however, 

 the match of the true two-dimensional spectrum with the model is quite 

 good. It is questionable whether the additional detail present in the 

 true spectrum represents a valuable signal or simply additional noise. 



There are several advantages to the model proposed in this study 

 (derived with respect to azimuth) over the two-dimensional FFT method 



105 



