Te — 
  
A simplified approach to the formation of a 
stereo model assumes again, similar to Equation (9), 
straight and level parallel flights (Derenyi, 1975a, 
b; Leberl, 1975e): 
gu (x, 0, h) So (0, B, h) 
so that 
*, = (x + x 3.4.2 
a. (T uesri^ - n^3/2n (19) 
2 -h- iot B) 2); + (pl ya) 34/2 
When ground control points are available, then 
there are again the same two possibilities of in- 
terpolative or parametric use of these points as 
in the single image approach: either the model de- 
formations in control points are used to interpo- 
late corrections in radargrammetric points, or the 
slant ranges and times of imaging are corrected 
using Equations (11), (12). 
6.2 Mapping Methods - Stereo Configurations: 
  
The normal case of stereo-radargrammetry con- 
sists of two parallel flight lines on the same 
side of the imaged area ("same-side stereo"). Other 
configurations, e.g. parallel flight lines on oppo- 
site sides of the imaged area ("opposite-side"), 
or at right angles ("cross-wise'" - See Graham, 1975b) 
can create great difficulties in visually perceiv- 
ing a stereoscopic model , to the point that such 
configurations cannot be employed. 
This exhausts the possibilities for synthetic 
aperture radar. For real aperture radar, there 
are still a number of possible stereo configura- 
tions, which can be generated along a single flight- 
line, for example imaging with convergent scanning 
planes (imaging with ¢-tilted antennas, Leberl, 
  
  
  
    
  
   
   
  
   
  
  
  
  
  
  
   
  
  
    
    
  
  
    
  
   
  
  
  
   
  
   
  
  
  
  
  
  
  
  
  
  
   
  
  
    
    
planes; or the combination of one k-swung vertical 
scanning plane with a scanning cone (7$ 0; see Carl- 
Son, 1973; Bair and Carlson, 1974,1975). The latter 
configuration particularly is claimed to provide for 
a satisfactory visual stereo-effect. 
6.3 Mapping Accuracy: 
Theoretical studies into the accuracy of a 
side-looking radar stereo model have been under- 
taken by Leberl (1972b, 1976b). Expressions for 
the deformation of a synthetic aperture stereo 
model were derived by Leberl (1976b). It is shown 
that for errors pertaining to image ('), one ob- 
tains: 
  
Ax H tanQ' H 
=> + ——— b —— ' 
ax. 2 2757. 25. 
1 1 
Ay Se MS ses Ar deno) 
Yo l-cotQ! tanQ" tan&£?'-tan€" sin(Q'-Q") 
1 
AZ. sin " Ar! 
  
1 
AY. 
cotQ'-cotQ" 1-tanf cot" + sin(Q'-Q") 
Equations (20) show clearly, that the model co- 
ordinate in flight direction (x ) is deformed main- 
ly due to errors of the velocity vector ( Aÿ , AZ ) 
and that errors in across track direction (y ) an 
in height (z ) are only the results of erronéous 
sensor positions ( Ay, Az.) and of an error of 
range r. 
Figure 4 gives a.graphical representation of Ay 
and Az due to Ar' and for specific stereo arrange- 
ments.' The change-over from a same-side to oppo- 
site side configuration represents a singularity 
of the error curve It is obvious that small 
elevation angles Q', Q lead to small Az, errors, 
but that they inversely create larger ay? errors. 
The opposite is true for large elevation àngles. 
Experimental stereo analyses have been per- 
formed by a number of authors. An overview of the 
  
   
1972a); imaging with two X -swung vertical scanning results is given as Table 4. In some cases the 
© = 15° —-—0'= 30° — Qm 45" 
50 T 50 T 
Ar! = 10m Air! = 10m 
SAME 
25|- SIDE 25|- 
SAME [ 
SIDE OPP 
SIDE | 
tags 
SAME SAME 1. 
SIDE Shp SIDE OPP 
DE S. SIDE 
| N 
NI 
S. | ered 
NN À 
N. | 
N 
| 
  
  
  
  
45 
  
  
  
  
  
-45 0 
Figure 4 
Examples of error curves for stereo radargrammetry