For example, quasi column noise whose effect would disappear at plus or 

 minus ten lags in the vertical direction on the covariance surface could produce 

 the negative areas in the spectra found on the horizontal aods. 



However the attemipts were in general unsatisfactory as the different types 

 of corrections propagated very oddly from, one system to another. No notable 

 success was achieved by these attempts. 



Detailed analysis of leveled spot height data 



The analysis of the data had reached an impasse. After a number of con- 

 ferences with JLeo Tick and Prof. Max Woodbury, Prof. Woodbury suggested that 

 the original data be studied to see if they could be corrected. Such a procedure 

 would involve re-comiputation of the results, but the use of the Logistics computer 

 at George Washington University was assured, and the problem was deemed so 

 important that the added effort to obtain a satisfactory solution should be made. 



The ridge along the verticeil axes of the covariance surfaces suggested 

 Sonne source of error in the vertical direction of the stereo data. Figures 

 11.4 and 11.5 and the leveled spot height values as tabulated were then studied 

 very carefully to see if any discrepancies could be found, 



I la fig. Hv4, it had been noted that the diagonally oriented wave crest- 



wave trough pattern on the left side and in the center of the figure changed to a 

 vertical orientation on the right hand edge of the pattern. Very strong verti- 

 cally oriented crests are especially pronounced in the lower right corner. This 



variation had been thought to be a possible perfectly natural variation in the data, 



but now this assumption was checked. 



173 



