- 34 — 
bability: (14,65. -o) fori x and y as + 100 m andufora zag #50 mw 
(for.1.:2.50 O00 in x as * 23 m, in y as. *t J6 m, and imas 
*.5.m) cannot be reached. 
Contrary to the results for infrared imagery the results: for 
radar imagery might still further be improved by polynomials, 
or better, fourier transformations in order to reach mapping 
specifications for | : 250 O00 in x and y (as was: proved in 
a computation carried out by Dowideit at the Technical Uni- 
versity of Hannover). For the determination of z to mapping 
specifications it becomes clear that the use of a radar inter- 
ferometer becomes a necessity. 
The results supplement quite well the single image SLAR-geo- 
metry tests carried out by Bosman, Clerici, Eckhart and 
Kubik [25] and by Leberl [24] over flat terrain in Holland. 
Table 6 shows, as a comparison to table 5 Leberl's test values 
obtained for four radar strips over the North-East-Polder: 
Table 6 
Geometrical Test of Single image Radar Imagery over Holland 
by Leberl 
  
  
  
  
  
  
  
  
  
  
  
single images (strip 2) Ax Ay 
standard error (RMS) m 
orthogonal transformation only 266 184 
(comparable to table 5) 
- single images 
arithmetic mean between two 
strips 62 76 
(comparable due to flat ter- 
rain to stereo images - op- 
posite directions in table 5) 
polynomial fit 51 16 
graphical interpolation 
(Strinz) 52 35 
  
- 35 - 
  
S 4 
ui 
j , 
5 > 0 H | oo 
o