only empirical functions, like arbitrary polynomial 
equations. For the future a great improvement in this 
field is anticipated. Radarmosaics and Radar 
blockadjustment can bridge areas with lack in ground 
controlpoints. If this gap extends about one strip 
width, polynomial equations used for an image to 
image registration, should be of first order, due to 
error propagation. In future on board GPS promises 
to register the flight path with acceptable accuracy, 
which allows to use a more realistic formulation of 
the flight behaviour in the radar blockadjustemt. Also 
inflight GPS can replace ground control to a great 
extend. 
In conclusion can be stated, sofar only the 
photogrammetric frame camera fulfills the 
requirements for topographic maps in scaleranges of 
about 1:50 000. Therefore a special space mission 
aimed at cartographic interests still is required to 
meet the worldwide demand for maps. Another 
question has to answered with regard to the type of 
map products for the future(e.g., orthophotos at 
regular intervals instead of conventional linemap and 
their revision). 
3. ESTIMATES FOR SENSOR PERFORMANCE 
Sofar the alternatives to meet medium and small 
scale mapping needs have been discussed. The 
following shall be seen as a contribution to answer 
the question, how far acquired images, showing a 
particular pixelsize can actually be used for 
topographic mapping. To answer this question, 
several estimates have been given in literature. 
DOYLE, 1975 , uses the formula:suited mapscale=2 
times Pixelsize (in meter) times 10 power 4 to 
estimate the suitability of digital sensors for mapping. 
According to this for topographic maps of the scale 
1:50000 a ground pixelsize of about 2.5 m is 
necessary. 
At the Institute for Photogrammetry and Engeneering 
Surveys of the University of Hannover three types of 
imagery have been used for interpretation with 
respect to the content of medium scale topographic 
map: 
1) A stereo pair taken with a RMK 15/23 at the scale 
1:50 000 of the City of Hannover(base-height ratio — 
0.6) representing an urban area; for interpretation 
results see figure 2. 
2) A stereopair taken over an area near Villefranche- 
sur-Cher with a RMK 30/23 at the scale 1:35 000 
(base-height ratio = 0.3), representing a rural area 
under Western European conditions; for 
interpretation results see figure 3. 
3) A stereo pair of a part of the island of 
Sylt(German North Sea Cost) with a RMK 8.5/23 at 
the scale 1:125000 (base-height ratio =1.1), 
representing an uncultivated area with dunes; for 
interpretation results see figure 4. 
This imagery has been digitized at pixel sizes 2,5 
micron, 50 micron, 100 micron and resampled to 200 
micron, 300 micron, 400 micron and 800 micron. The 
resulting images have been interpreted monoscopic 
and stereoscopic, to derive mapfeatures contained in 
1:25000, 1:50000 and 1:100000 maps. 
The following topographic features have been 
extracted: roads and paths, buildings, 
vegetation(forest, trees and shrubs), creeks and 
ditches and topographic forms. The mapping 
potential of this images with respect to these features 
has been judged as high, good, media, low or 
blank(meaning lack of particular features) see figures 
2 to 4. According to the expectations, the 
interpretation of the original images is satisfactory for 
all cases. There is one slight exception, for building 
details such as ledges, etc., cannot be detected in the 
1:125 000 scale image. However this is not 
significant for the required map content of 1:50000. 
In order to depict such building details, pixel sizes 
between of at least 2 and 5 m are required. 
The most significant result of the interpretation test 
is, that stereo-observation is by far superior to 
monoscopic observation. The advantage of stereo 
nearly increases proportional with the base height 
ratio. The comparison between the different areas 
photographed clearly shows, that the interpretability 
of houses, roads, land forms, creeks and ditches also 
varies with the type of the terrain, since urban areas, 
rural areas and uncultivated areas generelly contain 
different types of housing, roads etc, even though 
they are of rthe same topographic category of 
features. 
Concerening the topographic requirements, the most 
critical features are the buildings. To map them 
monoscopically a pixel size of at least 3 m is 
required. Under stereo observation a pixelsize of 
about 6 m might suffice. This is the requirement for 
urban areas to map at 1:50 000. For the mapping of 
roads and vegetation this requirement may slightly be 
relaxed. 
The results of the interpretation of digitzed images at 
25 micron were identical to those in the originals. 
Therefore they are not included in figure 2 to 4. This 
just proves, the observer's eye cannot resolve better 
than about 5 to 10 Ip/mm. 
Interpretation under stereo observation was possible 
for 50 micron and 100 micron pixelsizes. The stereo 
effect also appaers for images showing pixelsizes of 
200 micron and 400 micron, but does not aid the 
interpretation any more and does not permit 
contouring. 
It is interesting, the stereo effect is still unexpe cted 
high, if images of different pixel sizes are 
454