ders, 
cated 
hods. 
uares 
matic 
del. 
sition 
nown 
lerms 
ional 
each 
uring 
t for 
eters 
com- 
rning 
sure- 
r and 
IS are 
| the 
tion. 
of the 
netric 
mera 
ure is 
ratio 
is an 
ble 1 
large 
  
vard 
  
Oo Wn Oo 
  
vhich 
on of 
(bay 
11 be 
vn in 
n can 
ther 
or by 
ensor 
By the simultaneous adjustment of two crossing strips 
recorded from two different orbits further improvement can 
be achieved, especially in the common area covered by both 
strips. Intersection angles of ground tracks of two orbits can 
range from very small angles at 28.5? northern or southern 
latitude up to 57° at the equator. Simulations have been 
performed using the three intersection angles 5°, 30° and 
35°. 
All points located within the first or the last baselength of a 
strip are imaged by two sensors only (2-ray points). Within 
the overlapping area of two crossing strips the points can be 
projected into a maximum of six sensor lines (6-ray points). 
Figure 3 shows the different numbers of possible image 
rays, contributing to the determination of the corresponding 
object point. 
  
NER 
E Ces 
  
  
  
Fig. 3: Numbers of possible image rays, contributing to the 
point determination 
3.1.2 Image coordinates of conjugate points and ground 
control points (GCP) The digitally recorded MOMS-02 
imagery will be available on computer compatible tapes 
(CCT) and thus will directly be accessible for the further 
digital processing. After radiometric correction a large 
number of image coordinates of conjugate points can 
automatically be derived by digital image matching 
techniques. Involving least squares adjustment a precision 
up to 1/10 of the pixel size can be achieved. How far the 
three times higher ground resolution of the HR channel 
additionally improves the accuracy has still to be 
investigated. 
The determination of image coordinates of GCP implies the 
identification of the points in the image. Up to now this task 
must be carried out interactively using e. g. a softcopy 
photogrammetric workstation. 
For the simulations the image coordinates of 11905 object 
points with a constant height of 0.0 m were computed, 
assuming a straight forward flight path (¢,=w,=x,=0°). The 
points are arranged in a grid within an 36 x 468 km? area. 
The distance between two points is chosen to be 9 km 
across track and 0.2 km along track. The x-axis of the object 
coordinate system is defined parallel to the direction of 
flight. All image coordinates were introduced into the 
adjustment as uncorrelated observations with equal standard 
deviations of 2 um (1/5 pixel size). 
3.1.3 Ground control information In principle, ground 
control information is necessary to define the datum, i.e. for 
positioning, orientation and scale of the photogrammetric 
461 
model in a selected reference frame, and to improve the 
accuracy of the point determination. Due to the low orbital 
inclination only regions close to the equator can be imaged 
by the MOMS-02 camera. In most of these countries the 
number and accuracy of GCP is rather limited. 
In the simulations two kinds of control information are 
used: 
» 4 XYZ GCP, error-free; 
» 125 XY GCP and a Digital Terrain Model (DTM) with 
standard deviations of 25 m (GCP) and 50 m (DTM), 
respectively. 
In the first case, the four XYZ GCP are arranged at the 
corners of the 3-ray area. Their coordinates are treated as 
error-free. By means of measurements with the Global 
Positioning System (GPS), that is available all over the 
world, it is possible to determine these GCP economically at 
an accuracy level of a few centimetres. 
In the second case, a lot of XY GCP and a DTM, which 
might be derived from existing maps via digitization, are 
employed. From topographic maps in the scale of 1:50.000 
characteristic points, e. g. cross-roads, can be digitized with 
a planimetric accuracy of about 20 to 30 m. The generation 
of DTM using contour lines is presented in (Aumann and 
Ebner, 1992); from the mentioned topographic maps 
1:50.000 a height accuracy of at least 50 m can be achieved. 
The mathematical model for using DTM information in a 
bundle block adjustment is described in (Ebner and Strunz, 
1988). 
3.1.4 Observations of the exterior orientation parameters 
The derived position and attitude data from navigation 
systems and orbit models, mentioned in chapter 2.3, are 
introduced with the following standard deviations: 
position parameters attitude parameters 
  
» 0 m (error-free) 0 mgrad (error-free) 
» 1m 5 mgrad 
» 2m 10 mgrad 
> © (no observation) œ (no observation) 
According to the current knowledge the position data can 
be obtained more precisely than the attitude data. For 
instance, an attitude accuracy of 5 mgrad is equivalent to a 
position accuracy of 25 m considering the orbit height of 
296 km. In our case, an accuracy of about 1 to 2 m for the 
position and 5 to 10 mgrad for the attitude is expected. In 
addition, the two extreme cases - error-free and no 
observations - are applied comparatively. 
Furthermore, the observed position and attitude data have 
low absolute but high relative accuracy with respect to the 
reference coordinate system because of unknown systematic 
errors, which are modeled by the twelve offset and drift 
parameters as described in chapter 2.3. 
3.1.5 Distances between the orientation images (DOT) 
The suitable DOI has to be adapted to the temporal 
variations of the position and attitude of the shuttle. The 
aim is to derive such a DOI, that the exterior orientation 
parameters are optimally approximated by a mathematical 
model, i.e. a 3rd order Lagrange polynomial in this case. It 
must be considered that a long distance between the 
orientation images leads to interpolation errors and 
deteriorates the accuracy. In future, the program CLIC