49 - 
The standard deviation of a computed fictitious satellite position is less than ^ 0. 3 [A 
which corresponds, with a focal length of 300 mm, to t 0. 2 seconds of arc. If combined with 
the aforementioned standard deviation of the orientation at the center of the plate, which is a - 
bout + 0. "3, the accuracy of the final direction to the satellite should be better than ± 0. "5. 
Practical tests, based on multi-camera observations at the same station, indicate at present 
an accuracy of t 0. M 7 or better. The slight degradation in the accuracy is caused by some in 
sufficiently resolved bias errors. The mathematical model simulating the photogrammetric da 
ta acquisition process is at present over-economized and must be somewhat relaxed by addi 
tional sophistication in the simulation of the distortion model. More important, there still ex 
ists a small but significant bias error in the coordinate measurements between star and satel 
lite images. Automation of the image measuring process by electronic image sensing is expec 
ted to be helpful. 
From the results so far obtained, it appears justified to conclude that the photogram - 
metric satellite triangulation method, especially if an optimized lens system with approximate 
ly 450 mm focal length is considered, will give the direction to a satellite to within t 0. 4 to 
+ 0. 5 seconds of arc. The fact that such a direction has an absolute meaning with respect to the 
right ascension-declination system, allows the conclusion that such a result is promising if 
compared with the accuracy of + 0. "2 to t 0. "3 which is today obtained in first order triangula 
tion for a relative direction. 
Closing this consideration of photogrammetric satellite triangulation, it can be predic 
ted that this method will not only prove useful for a world-wide triangulation scheme, but will 
eventually provide the necessary accuracy for increasing the geometric fidelity within indivi 
dual geodetic datums. 
Extensive numerical analysis on various possible schemes for the application of pho - 
togrammetric satellite triangulation is presently in progress for the purpose of studying the 
problem of error propagation and for establishing an optimized field operational procedure. 
Some Aspects of Satellite-Borne Photogrammetry 
Rather detailed information could be given for the application of photogrammetry in 
connection with the tracking of satellites. A discussion of the application of photogrammetry 
using satellite-borne photogrammetric data acquisition systems, however, must be conducted 
in generalities and must be considered conjectural, because of the lack of actual raw data. 
One of the basic problems to be considered is the use of satellite-borne photography 
for establishing geodetic control, especially in support of the compilation of topographic maps. 
The corresponding techniques are the same as those recently under development in computa - 
tional photogrammetry for photogrammetric control extension. Continental reference nets es - 
tablished by photogrammetric, stellar triangulation, as outlined before (compare flow diagram 
on page 14), with an average distance between control points of about 300 km and a positional 
accuracy of ^ 3 meters, are assumed to be existence. These stations are assumed to be mar 
ked for photogrammetric recording. 
Considering the lowest feasible height for a satellite orbit as 200 to 250 km, the trou 
blesome problems of extended photogrammetric aerial triangulation are avoided, because the 
coverage of a single photograph corresponds to the spacing of the control points. When trea 
ting a few photographs which, in practice, must be combined, the error propagation is further 
reduced by certain orbital constraints, two important ones being the consistency of the inclina 
tion of the orbit and the smoothness of short orbital arcs. 
Further geometrical information becomes available if the intervals between exposures 
are measured with high accuracy (10 sec). The use of additional star or sun photography (as 
important as the latter is for conventional strip and block triangulation) appears, under these 
conditions, of questionable value, considering the unavoidable degradation in absolute accuracy 
when the orientation of the stellar camera must be referenced to the camera taking the topo -