International Archives of Photogrammetry and Remote Sensing. Vol. XXXII Part 7C2, UNISPACE III, Vienna. 1999 
21 
I5PR5 
UNISPACE III - ISPRS Workshop on 
“Resource Mapping from Space” 
9:00 am -12:00 pm, 22 July 1999, VIC Room B 
Vienna, Austria 
ISPRS 
Such images may be used for the creation of planimetric image 
maps at the scale of 1:10 000, but the lack of a large number of 
ground control points to differentially rectify the image maps on 
account of image defonnation to planimetric map accuracy 7 
standards at that scale renders the expected planimetric accuracy 
more to the 1:25 000 level. 
Moreover, the height determination from images generally 
depends on the height-base ratio of the imagery flown: 
, h h 
a, =± er 
b c 
px 
with 
= 
h 
b 
c = 
camera objective 
or, 
point error in elevation 
orbital height 
orbital base 
principal distance (focal length) of the 
"px 
parallax measurement error in the order of 
the point positioning error op of about 10 
(im in a photographic image, 
corresponding more or less to the pixel 
size on the ground with the image scale 
used. 
Due to the very long focal length of the KVR camera of 1 m the 
stereoscopic overlap conditions of that imagery will not permit 
a smaller and more favorable height-base ratio than 10, 
rendering the expected height accuracy of less than ± 20 m. 
The present KVR 1000 high resolution images are therefore 
suitable for planimetric map updates 1:10 000 or 1:25 000, but 
not for digital elevation model measurements. 
Cartographic satellites, which permit a better height 
determination, even if they do not reach the same detectability 
are the French panchromatic Spot system with 10 m pixels and 
the Indian IRS-1 C system with 5.8 m pixels. Due to their 
capability to incline the sensor by a mirror in cross-track 
direction a favourable height-base ratio of up to 1 may be 
achieved from subsequent orbits. This, however, is often a 
handicap due to changing cloud cover, which severely limits 
stereoscopic coverage for a time period in which the radiometry 
of the ground lias not changed. 
The best test result achieved thus far with Spot Pan and IRS-1C 
stereo imagery are in the order of ± 5 to 10 m in elevation and ± 
3 to 5 m in position, making these sensors suitable for 1:50 000 
to 1:100 000 mapping in mountainous areas. 
Another approach has been provided by the inflight stereo 
capability 7 of the German Stereo-MOMS system. It consists of a 
triple line scanner looking forward, vertically down and aft, 
which has been flown for 10 days on the U.S.-German Space 
Shuttle mission D2 in 1993 and which since 1996 operates on 
the MIR-Space Station’s Priroda module with interruptions. 
On MIR the vertical sensor yields 5 m panchromatic ground 
pixels and/or 15 m multispectral ground pixels, and the fore and 
aft sensors give 15 m panchromatic ground pixels. Figure 9 
show's such an image over the German city of Augsburg, in 
which the vertical panchromatic 5 m pixels have been fused 
with the vertical multispectral 15 m pixels. 
Figure 10 shows an oblique view of three processed fore, down 
and aft panchromatic stereo images after the generation of a 
D.E.M. and the resulting orthoimage over the area of Barcelona, 
Spain. 
The advantage of in-track stereo sensing is that all images to 
create a DTM, orthophotos and oblique views are taken at the 
same time. 
The cartographic requirements for these images show, that 
planimetric requirements can be met for the 1:25 000 scale, that 
elevations can be determined with ± 5 to 10 pm accuracy and 
that the detectability is suitable for 1:50 000 mapping.