nated satelli-
ystem, which
and Geode -
system it is
record.
irget in terms
un as an illu-
I and Echo
of a world-
First, the
disturbing
patial direc-
rical strength
dellite should
lg from unfa-
possible
Lise varia -
plan for a
102 sides.
, a nominal
sr practical
meed by
ons, the
jor world
neighbor -
shed in
in jud -
angulation.
are inter-
s Cata -
910) a
advan -
errors,
at the
the who-
* contai-
than the
racy as
on of star
ng south-
mean
faint,
epart from
geometric
by 1975
these two systems will have deviated from each other by as much as 0. "9 in right ascension
at-60° declination. Precision geodetic satellite triangulation depends on the removal of such
discrepancies, emphasizing the need for intensive series of fundamental astronomical obser -
vational programs [10].
Entirely in accordance with any other photogrammetric triangulation procedure, the
collinearity condition is the basic element of measurement in stellar triangulation. In other
words, any system of stellar triangulation can be conceived as a multitude of individual rays,
where the direction of each of these rays is determined from a corresponding photogramme -
trie record.
Therefore, it is first of interest to study the accuracy with which such a direction can
be obtained, in order to follow ultimately with the problem of determining the accuracy of the
final triangulation as the result of propagating the errors of the individual rays into the confi
guration of the spatial triangulation.
The problem of how accurately an individual ray can be interpolated into the star back
ground can be studied by considering two independent error-sources :
a - The accuracy with which the spatial orientation of the photogrammetric record can be
established from the star photography, and
b - the accuracy with which the image of the satellite can be measured.
The accuracy of the determination of the orientation of the photogrammetric record is
obtained from the inverse matrix of the normal equation system associated with the analytical
treatment of the single photogrammetric camera. Experience has shown that, despite previous
ly established camera calibrations, it is necessary to incorporate, in addition to the six geome
trical parameters (three rotations and three translations), three to seven parameters, depen -
ding on the quality of the photogrammetric camera, describing the distortion characteristics
of the specific photogrammetric record at the moment of exposure. In addition, it is obviously
necessary to provide means for correcting both the plate measurements and the star catalog
values in accordance with assigned weighting factors.
The following diagram shows the number of stars which must be measured on an 18x18
cm photography, obtained with f = 300 mm, as a function of the number of parameters carried
in the solution, in order to obtain a certain overall orientation accuracy. It is important to note
that not only the absolute number of parameters, but also the need for additional star imagery
increases with the size of the area of the photogrammetric record used for the recording of the
satellite information. If, for example, 2/3 of the record is to be used for recording satellite
images, and 9 parameters are considered necessary for describing the mathematical model of
the photogrammetric camera, it is necessary to measure 15 0 approximately evenly distributed
star images in order to obtain a minimum orientation accuracy of t 0. 5 seconds of arc, if the
mean error of unit weight for a single coordinate measurement is + 3 microns.
The coordinate measuring process obviously is one of the key operations during the da
ta evaluation. The problem begins with the need for a calibration standard in the form of a cali
brated grid with which the comparator calibration can be performed. Quite a few papers have
been published in the past few years which are concerned, in part, v/ith the measuring of plate
coordinates for precision photogrammetric triangulation. In these publications it has become
customary to claim a coordinate measuring accuracy of better than one micron. It is of inte -
rest to consider such statements in light of the fact that at present the national institutes for ca
libration - to name a few, the Bureau of Standards in the U. S. A. ; the Physikalische Technis -
che Bundesanstalt in Braunschweig, Germany ; the Eidgenössische Amt für Mass und Gewicht
in Bern, Switzerland ; - do not accept the task of calibrating a line grid of 200 x 200 mm to an
accuracy of 1 1 H- . As a matter of fact, these institutes apparently have to apply their full ca -
pabilities in order to calibrate the positions of a sequence of lines on a 200 mm glass scale to
an accuracy of t 0. 5 p. .
It was necessary, therefore, to develop a special measuring procedure in order to ob
tain the coordinates of a selected number of grid intersections (25 points arranged in a 5 cm
- 45 -