2. MATHEMATICAL MODEL FOR NON-PHOTOGRAMMETRIC INFORMATION 
2.1. Functional model 
When the original geodetic, navigational or satellite 
surveying measurements are introduced into a photogram- 
metric block adjustment, one has to take into account to 
what extent the data are influenced by the definition of 
the measuring system, comparing it to the desired system 
of the point determination. For example most of the infor- 
mation about geometric forms of objects can be formulated 
in any system. On the other hand e.g. slope distances 
depend on the scale factor of the measuring instrument, 
horizontal angles on the definition of the horizontal 
plane or the plumb-line, or satellite measurements in 
general on the definition of the satellite system. 
Based on the original data of a measuring system normally 
positions of points are estimated. When introducing these 
coordinates in a block adjustment one has to take care 
that point coordinates, which are given in a system with 
another definition of the datum and/or with certain syste- 
matic effects, are formulated in an extended functional 
model. This means that transformation or trend parameters 
should be included; possibly, if available, also with a 
priori information about these parameters. 
In a similar way the functional models can be formulated 
for the positions and orientations of the camera stations. 
Here it may sometimes be convenient to introduce strip- 
invariant trend parameters, presuming they are signifi- 
cant, for certain types of data. This nas proved to'be 
very effective for the processing of statoscope data and 
it seems to be in principle also a proper formulation for 
the introduction of data e.g. from inertial surveying 
systems. 
An important question, which arises especially with the 
introduction of relatively "new data", is, whether one 
should use the often very distorted raw data from a meas- 
uring system or whether it is more convenient to use the 
results of a separate data processing, e.g. coordinates or 
coordinate differences of points. Often the more practica- 
ble way seems to be the second one (e.g. Anderson /2/ for 
Doppler measurements), provided that the derived coordi- 
nates are introduced rigorously with their full covariance 
matrix. 
2.2. Stochastic model 
Although the situation is similar for general non-photo- 
grammetric information, the following remarks apply to the 
case when coordinates of points are given. 
Introducing control points with their stochastic proper- 
233