ation 
  
       
Imax (8°) Imax (0°) 
Imin (a) 
  
  
| 
| | 
| | 
| | 
| | 
| | 
x(o') x(o'*) 
BRIGHT BRIGHT 
Figure 4: Edge migration as a result of changing look angles. 
For Mt. Shasta and for Gordón la Graza, SIR-B image pairs were also the source for digital elevation 
models. These have been presented at earlier occasions (see reference 8). The use of the radar 
derived DEMs is for secondary radar images as discussed in the next chapter. 
3. SECONDARY RADAR IMAGE PRODUCTS 
Domik et al. (1986) illustrate a set of SAR images that are derived from the primary correlator output. 
Secondary image products are the result of digital image processing. Table 4 lists various types of 
such products. 
Primary interest is of course on the rectified radar image; a new terminology refers to this also as 
"geo-coded" images. Rectification is achieved using the stereo-derived DEM, and it may concern the 
geometric resampling of the primary images as well as the radiometric correction for slope effects. 
Figure 6 illustrates a set of raw primary images, and their geometric as well as geometrically and 
radiometrically rectified versions. The application of these and other techniques have been discussed 
in the papers by Domik et al. (1986) and Cimino et al. (1986). It is evident that rectified multi-date 
and multi-angle SAR images lend themselves to multi-temporal and incidence-angle-signature 
analyses. 
4. CONCLUSIONS 
SIR-B provided for the first time a variety of SAR images at differing look angles that cover areas for 
stereoscopic analysis. The results lead to the conclusions that 
* planimetric and height accuracies are of the same order of magnitude; 
% errors are less dependent on intersection angles than previously expected from theoretical 
predictions; 
* the errors are of the order of magnitude of several pixels or range resolution diameters; 
* only the steep-looking opposite-side stereo-pair of a flat area (Illinois) produced height 
accuracies near the theoretical prediction. 
339