€ Mw
- VJ Co gl
AM NC XM "WE pe nn)
(p N WV WV 2 = UD 9 oe vM
NA ce
e»
de UW
Besides mosaicked image data records there exist other
results such as those from the antenna Hlistening® to the
emitted energy from the planet's surface producing
temperature maps; or echoes from a radar-altimeter,
producing a coarse topographic relief and measures of
reflectivity; and finally the gravity observations producing
a detailed gravity map (see table 2).
3 MANAGING MAGELLAN'S IMAGES
At Institute for Computer Graphics of the Technical
University in Graz, the European Magellan Data Node
EMDN has been installed as part of NASA's PDS.
Related to the responsibility for supporting Magellan users
in Europe, several activities have been started.
3.1 A Venusian Atlas
Since the size of the Magellan data set is to big to
distribute it entirely to anyone interested, efforts are
undertaken to grant convenient access to the data via an
Internet-based image data catalogue. The inspection of the
Venusian surface for features or areas of interest so far
was done by visual inspection of paper printouts of the
digital data. To implement a similiar functionality with the
Internet-based retrieval system, a digital map is provided
which can be used to explore the Venusian surface at 8
different levels of resolution from 225 m/pixel to about 30
km/pixel. The Venusian Atlas map data were derived from
the C1-MIDR (see Tablel) data set. To reduce the size of
the resulting image data pyramid from about 13 Gbytes
down to 500 Mbytes, lossy compression was applied,
therefore it can not be used for any further processing. It's
main role is to provide interactive access to the Magellan
images.
3.2 Interactive Access to Magellan Images
The Interactive Venus atlas can be overlaid with any
information available about existing images and other
data. One can zoom in and out the digital surface map and
define areas of interest using the mouse. The search for the
image coverage of a given area becomes a matter of
seconds. A color coding scheme helps to identify areas
with stereoscopic coverage, where same side images at
different inclination angles do exist. Any of the images
found during this process can be marked for ordering or
downloading depending on their size and the network
bandwith available. The system is designed in a way that
alows to add any new images or other data like DEMs to
the database as they become available.
3.3 Distributed Data and Software
Software and data for Magellan may well be distributed
over multiple sites (Rehatschek, 1996; in print; Walcher
and Rehatschek, 1995). Access to software and the images
should be open to remote locations. Such concepts are
currently being tested for PDS / EMDN, but also apply to
terrestrial remote sensing, as it is done by EOSDIS in the
US, or CEO in Europe.
1) MST is available from Vexcel Corporation by contacting
kelly vexcel.com.
2) P. Chodas, oral communication.
493
4 DIGITAL ELEVATION DATA
4.1 Stereo Matching and the MST
The mission originally did not plan to acquire overlapping
stereoscopic radar images at different incindence angles
from the same side. It was only in the ,extendend
mission", when the so-called Cycle 2 images were
collected at opposite viewing direction (1991), that it was
shown in an eight orbit imaging sequence that useful
stereo observations could be made (Leberl et al., 1992).
This led to the initiative (as an „on the fly decision'^ of a
third imaging cycle for obtaining a stereo coverage in
combination with Cycle 1 data. Such, about a third of the
planet was covered when the radar sytsem ceased to
function. While the stereo coverage exists, funds are
insufficient to actually systematically extract stereo
information. Funding was available, however, to develop
software to support individual researchers in the
production of digital (Leberl, 1993; MST, the Magellan
Stereo Tolkit)"
A number of stereo experiments were performed at
NASA's Jet Propulsion Laboratory, at the Center for
Astrogeology of the US Geological Survey, at Vexcel
Corporation and elsewhere to verify the ability of
extracting a digital elevation model with predictable and
useful accuracy. Leberl et al. (1992) and others report that
the accuracy may be in the range of better than 1 pixel for
image matching, resulting in an elevation accuracy of
about 2 pixels or 150 meters. This is an optimistic
assessment that assumes that the ephemeris of the
spacecraft is known at great accuracy. In fact the
ephemeris can have large errors of up to 10 - 15 km when
considering image pairs taken several months apart.
Unfortunately, this is most likely the case when data from
different acquisition cycles have to be combined for stereo
restitution.
Figure 2: Stereo-derived DEM of an area at 2°S, 74°E.
Area covered is 75 km x 45 km.
Figure 2 respresents a digital elevation model of an area
obtained close to the equator with differences in altitude
of up to 5000m. In that case an improved ephemeris was
computed for a limited set of orbits. The Jet Propulsion
Laboratory has shown that such improvement is possible
within about + 30 m”. However, the ephemerides of the
entire Mission will still have to be processed with that
technique. The result in Figure 2 is a demonstration of a
prototype suite of stereo processing algorithms that have
been implemented in MST in a preliminary fashion.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996
