SOURCE ACCURACY CONTROL RESOLUTION ANTENNA RADAR 
(Meters) per (Ground -) System Code 
100 km Meters ; 
along across height Stabi- Type 
lc lc lo lized 
Gracie 1970 12.2 7.7 13.2 35.0 17 yes synth.  AN-APQ 102 Opp. Side 
68.0 138.0 240.0 Opp. Side 
Konecny 1972 12 yes real Westinghouse = 
130.0 428.0 1548.0 Same Side 
DBA- 26.8 21.9 16.7 Opp. Side 
Systems 1974 1.2 3 yes synth.  AN-ASQ142 
29.5 25.6 19.7 Same Side 
Goodyear 1974 93.0 3.3 12 yes synth. GEMS 1000 Same Side 
33.0 Opp. Side 
Derenyi 1975 12 yes synth. GEMS 1000 ee 
177.0 Same Side 
Leberl 1975 | 173.0 510.0 109.0 0.3 30-150 no synth. Apollo 17 Same SideoseSnall € - 
Base - Satellite 
Table 4 
Mapping accuracy achieved by 
accuracies quoted are optimistic, particularly if 
they concern opposite-side stereo configurations. 
Opposite-side stereoscopic viewing may be mainly 
possible in the case of fairly flat terrain or only 
isolated mountains surrounded by flat terrain. In 
other cases, lay-over, shadowing, and general dif- 
ferences in the contents of overlapping image pairs 
may perhaps not permit opposite-side stereo meas- 
urements to be taken, although geometrically, the 
opposite-side stereo arrangement is certainly 
superior. 
The two results marked by an asterisk (*) de- 
serve special comments: Goodyear (1974) measured 
a single profile of 6 km length and used one con- 
trol point for "absolute orientation." The results 
are therefore not really comparable to those of 
other authors (control density: I point. per 
3 x 3 x T1 30 km® corresponds to —3 points per 
100 km@. The other comment pertains to the re- 
sults of Leberl (1975e) using lunar orbital radar: 
here, a 2 m radar wavelength, very steep look-angles 
and not very well identifiable surface features 
(craters) were employed. As a result, the accuracy 
achieved had to be somewhat less than in studies 
with airborne radar. The lunar mapping study provided 
evidence that the stereo-radar computation according 
to Equ. (16) is superior to Equ. (18). 
6.5 Contouring: 
contouring from radar stereo models has been 
reported on two occasions: Norvelle (1972) demon- 
strated the use of the analytical plotter AS-11-A 
to directly plot contour lines from a deformed 
radar model.  Leberl (1975e) produced a radar con- 
tour plot of a lunar feature, however not by 
directly tracing the countour lines, but by first 
acquiring a digital height model, from which the 
contours were interpolated numerically. 
7. MAPPING FROM BLOCKS OF OVERLAPPING IMAGERY 
7.1 General: 
Investigations into the three-dimensional ad- 
justment of a radar block have been reported by 
DBA-Systems (1974) and Dowideit (1975). A review 
of methods of sequential and simultaneous adjust- 
stereo-radargrammetry 
ment of original radar strips and of independent 
stereo models has been compiled by Leberl (1976b). 
However, these studies mainly concern the mathe- 
matical and statistical formulation of a solution 
to the problem. Actual computations concern them- 
selves in the case of Dowideit (1975) with only a 
single simulated stereo pair, and in the case of 
DBA-Systems (1974) with a triplet of actual radar 
images. The modesty of the past efforts in this 
field could in part be caused by a certain lack of 
prospective block adjustment applications. 
Applications exist, however, for the prepara- 
tion of base maps for radar mosaicking of large 
numbers of radar strips. Consequently, a number of 
results are available on the accuracy of purely 
planimetric adjustments of radar strips (Leberl 
1975c, 1975d; Leberl, Jensen et al., 1976) and on 
the mere mosaicking of radar images (Lewis and 
MacDonald,1970; Berlin, 1971; v. Roessel and 
de Godoy, 1974). 
7.2 Planimetric Adjustment of Overlapping 
Radar Strips: 
Computation of ground coordinates from a block 
of overlapping side-looking radar images in a pro- 
cess similar to photogrammetric strip adjustment 
was attempted first by Bosman et al., (1972b) but 
with/a block of only two radar strips. The first 
adjustment of an extensive block of overlapping 
radar strips to a set of control points had to be 
in the framework of an operational radar map- 
project -- PRORADAM of Colombia (Leberl,1975c). 
was followed by a numerical simulation experi- 
ment -- as an afterthought to the actual operational 
task (Leberl, 1975d). Most recently an exceptional 
block of radar imagery of a well-mapped area in the 
U.S.A. (W. Virginia) required an operational adjust- 
ment, but also permitted an experiment to be carried 
out to study the performance of several methods 
(Leberl, Jensen et al., 1976). 
The numerical simulation study referred to 
above (Leberl, 1975d) did show the following: if 
a radar block is formed by sequential connection 
of the original radar strips (or stereomodels) 
whereby spline functions are used to describe image 
deformations, and if then an external block adjust- 
ment, according to Schut (1970), is carried out 
made 
ping 
This