Coming from five data bases, two main data streams can be distinguished. The first describes the generation 
of a Corrected Product, using only geographic coordinate informations from the satellite auxiliary data and, 
if required and available, Ground Control Points. When the GCPs are detected in the radar image by automatic 
algorithms, standard rectification and resampling procedures will generate the Corrected Product. 
The second data stream uses GCPs to compute the internal distortions of the SAR image especially in azimuth 
with respect to slant range distance. Coming from exact located DEM points, the slant range is computed 
involving the distortions found above and thus giving a reference to a point in the slant range input image, 
Therefore pixel coordinates will be found even in regions of typical radar distortions.Notice, that the process 
of generating Corrected and Precision products will enhance the GCP library, thus contributing "bootstrap- 
ping"-information for following radar images. Map Products will be generated by mosaicking methods from 
the Precision Product. 
To meet the requirements of the Geocoding System, approaches and algorithms are modified for restituting the 
radar geometric distortions and producing level precision and map products. (More detailed information can 
be found in [Domik, 1984; Ehlers, 1983 and Maier, 1985].) 
To cope with the huge amount of GCP and DEM data, a GCP library and a DEM data base will be installed, 
both able to accept data from different sources and in different formats. The design of these data bases fulfills 
the requirements of SAR data Geocoding, but will also serve as a data pool for mapping applications using 
data of other satellites. 
The GCP data base will consist of small image chips containing features such as water boundaries, road 
crossings and other outstanding points on the earth surface, which can be found in satellite images. The effort 
to fill the data base with suitable Ground Control Point chips will be supported by automatic pattern recognition 
algorithms, which could be used to extract features from Landsat MSS/TM and SPOT images. In addition to 
that, the GCP data base can also be filled with chips from SAR images processed earlier and topographic map 
information. A special problem will arise from the fundamental difference between optical satellite data and 
radar images. Therefore algorithms are now under investigation, to increase the correlation between image 
chips of different sources. 
The implementation of a GCP data base will be a dynamic process, which will start before ERS-1 is in orbit. 
The data needs maintenance and update and efforts will be done to cover all regions of interest with Ground 
Control Point chips. 
The second necessary information, to rectify radar images and to restitute radar specific distortions like lay- 
over, foreshortening and shadow, is the digital elevation data of the terrain, which will be stored in a DEM 
data base. The storage of DEM data needs no sophisticated data format design, if a regular grid of values is 
used and the location is referenced with geographical coordinates. For example, the area of the Federal 
Republic of Germany is covered by approx. 500 Megabyte of 16 bit elevation data using a grid of 30 m x 
30 m. Transformation and resampling to other geographic coordinate systems can be performed with standard 
map and image transformation algorithms. 
The severe problem is the availability of DEM data in the required accuracy. Even for countries covered with 
small scale maps, digital elevation information is available only for selected regions. Huge data sets of terrain 
information of the Federal Republic of Germany will be compiled for governmental planning purposes until 
1987. Besides, few data is existent for Middle Europe. All of them come with different format and accuracy 
and need thorough compilation before it can contribute to a DEM data base. 
However, the use of DEM data to be derived from the new generation of satellites such as SPOT is being 
investigated as a serious alternative. 
378 
  
The 
at d 
cess 
Pro 
faci 
Dig 
sets 
suci 
For 
reqi 
The 
wit 
erri 
The 
for 
stor 
Im: 
fori 
pro 
It i 
24( 
mo 
via 
nec 
Ar 
of 
dist 
Ca 
ma