STEREO-MAPPING OF PLANET VENUS FROM MAGELLAN SAR IMAGES: A STATUS REPORT
F. Leberl, K. Maurice
VEXCEL Corporation, Boulder, Colorado, USA 80301
ABSTRACT
We report on the current status of stereo-mapping of the surface of planet Venus from side-looking radar images created in the
Magellan-mission of NASA. We find excellent vertical exaggeration that is more pronounced than in aerial photography.
Measurements appear repeatable within * 30 m and accurate to within € 100 m.
1. INTRODUCTION
NASA's Magellan mission to map the planet Venus is the most
ambitious planetary imaging project ever. Venus has a surface
equal in size to Earth, but has no oceans. Coverage of the
planet's surface therefore requires three times the images one
would need on Earth. At a resolution of 75 m per pixel,
Magellan exceeds the resolution of the original Landsat-MSS
sensor and has collected more data than all previous planetary
missions combined. Table 1 summarizes key facts about
Magellan. The data of interest to map the surface are:
* the synthetic aperture radar images (pixel sizes of 75
m);
* the altimetric measurements spaced at about 10 km to
20 km;
* the radiometric observations of emmissivity of the
surface with pixel sizes of 25 km.
The mission offers unique challenges to radargrammetry: more
than 9046 of the surface have been imaged, and nearly half of
the surface was imaged twice or more. No ground control
exists. About 20% of the images present a same-side stereo
coverage; about 45% of the surface is covered by opposite-side
stereo data.
Date of launch from Kennedy 4 May 1989
Space Center, Florida
Beginning of systematic 15 September 1990
radar imaging
Completion of initial coverage 15 May 1991
(all 360°, Cycle 1)
Completion of second coverage 14 January 1992
(Cycle 2)
Orbit inclination (but images 85°
pole!)
Spacecraft altitude above surface 294 km
at periapsis
Spacecraft altitude above surface 8458 km
at apoapsis
Radar image pixel size 75mby75m
Radar image range resolution 88 m (0.59usec)
Radar image azimuth resolution 120 m
at 5 to 17 looks per pixel
Look angle off-nadir
Radar wavelength
Angular width of image at
antenna, at periapsis and 80°N
See Figure 1
12.6 cm
2° and 0.5°
Tab le 1: Key facts about Magellan with relevance to stereo-
radargrammetry.
While a considerable number of technical papers has been
authored about the subject of stereo-radargrammetry, very little
has been exercised on large, real-life data sets. Therefore, use
of Magellan radar images to develop a detailed topographic
relief requires pioneering efforts.
795
We report in this paper about the current state of affairs in
extracting Digital Elevation Models (DEMs) from overlapping
radar images of the surface of Venus. This effort is supervised
by a Stereo Analysis Work Group of the NASA Mission’s
governing Project Science Group with the active participation
of engineers and scientists at the Jet Propulsion Laboratory, US
Geological Survey, VEXCEL Corporation, MIT and others.
The mission consumed in excess of $700 million to build the
system, get it into orbit around Venus, and receive and store
the data. This resulted in satisfying the nominal goals of the
Mission, namely to develop a single image coverage of 70% of
the planet. It was only in an extended mission that the second
coverage was created that supports the stereo work, at a cost of
about $45 million. In contrast, nearly no funding has become
available to process the stereo images into useful DEMs.
2. DATA COVERAGE
With a planetary radius of 6,051 km, the surface of Venus
covers 4.6 * 10? sq. km. These are imaged into 1,650 images
extending from pole to pole with 220,000 pixel along track,
350 pixels across track. These Full Resolution Basic Image
Data Records (F-BIDRs) are subject to radargrammetric
processing. Images are taken from an elliptical orbit from an
altitude above the poles of about 2,000 km and at periapsis of
about 295 km. As the side looking radar (SAR) images are
being taken the sensor looked East (Cycle 1). A second
coverage of the planet was taken with the sensor looking West
(in Cycle 2). About 45% of the planet is covered by both East-
and West- looking data. The missing 55% are a result of
occasional obstacles to imaging, e.g. if the sun is behind Venus
and Earth, or if the radar malfunctions.
A third coverage created images again by looking East, but at a
different angle (Cycle 3). About 20% of the surface will be
covered by September, 1992, with two sets of overlapping
images both looking East and representing a same-side stereo
data set. Given that on Earth only 1/3 of the surface is land,
the Magellan stereo data cover a surface equal to 6096 of the
Earth's land masses.
The elliptical spacecraft orbit forced the imaging radar to look
at the surface with varying angles. Over the poles the
spacecraft is 2,000 km above the surface, looking down steeply
at look-angles of 7* to 11* off nadir. At the 300 km perapsis,
the look-angles can be kept much larger at 43° to 55°. Each
imaging cycle therefore has a specific desired look angle
profile (DLAP). Figure 1 presents Magellan’s look angle
profiles.
The ellipticity of the orbit and differences in look angles result
in the fact that an individual Magellan image (an F-BIDR)
traverses numerous other images from a different Cycle:
1 image of Cycle 1 East Looking (EL) can traverse more
than 100 images of Cycle 2, West-looking (WL).
1 image of Cycle 1 (LL) can traverse 35 images of Cycle 3
(EL).
1 image of Cycle 2 (WL) can traverse more than 100 images
of Cycle 3 (EL).
