„A
r (MOLA)
uires high
us studies.
egistration
cneralized
points and
stment, we
consistent
resolution
for precise
As the first
out in the
| the same
the MOC
y to bea
Next, we
on quality
credible
o included
odels of
presented.
A
>.
netry data,
for MGS
und points
icient and
on model
TA
rom three
ma, Gusev
ches, final
es for past
conditions
). Among
chosen as
f the MER
chosen as
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004
The data include MOC NA images, MOLA profile and MGS
trajectory data. MOC is a linear pushbroom scanner taking one
line of an image at a time (Albee et al., 2001). The NA camera
with 2048 detectors and 3.5 m focal length acquires high-
resolution images with 1.4 meter/pixel at nadir. Stereopair of
high-resolution images from the NA camera are used for this
study. Table 1 illustrates the properties of high-resolution MOC
stereo pair images based on the three study sites. The stereo
geometry of MOC is across track configuration with one small
emission angle for one nadir image and one large emission
angle for the other off-nadir image. Line exposure time is a
quite important property for the processing of linear pushbroom
images. Images are acquired from an acquisition time at the rate
of the line exposure time. The image acquisition time in Table 1
shows all images are taken between March and May 2001. Line
exposure times and ground space distances (GSD) are different
for every image as illustrated in Table 1. Ground space distance
indicates the ground distance per pixel, and varies from 3.3
meter/pixel to 5.5 meter/pixel depending on the image.
Table 1. Properties of MOC stereo images
Site Name Eos Chasma Gusev Crater Isidis Planitia
Image E02 E04 E02 E02 E02 E02
Name 02855 | 01275 | 00665 | 01453 | 01301 | 02016
Emission [046 11797 |-02 224: 1.3391 92
Angle (?)
Acquisition 54
Date (2001) Mar.31 [May 18| Mar. 8 (Mar. 17|Mar.15 |Mar. 23
Exposure |, 8078 | 15052 | 1.4462 | 1.4462 | 0.9642 | 1.4462
Time (ms)
File Size | 9856 | 7424 | 10112 | 8960 | 7680 | 7680
(H*W) *672 | *1024 | *1024 | *1024 | *1024 | *1024
CD. usb hf age Ga Fa
(m/pixel)
MOLA is designed to understand global three-dimensional
topography and atmosphere around Mars using laser signals
(Smith et al, 2001). If MOLA data and MOC images are
obtained at the same time, the MOLA profiles are called
simultaneous MOLA profiles. Thus, one MOC image has one
linear-pattern MOLA profile and this study uses the
simultaneous MOLA profiles of each image. Among several
standard MOLA data products, this study is based on Precision
Experiment Data Record (PEDR) data generated using
precision orbit data. PEDR data consists of areocentric
longitude and latitude, range, planetary radius, topographic
height, and ephemeris time. Figure | shows the MOLA ranging
principle in measuring distance between MGS and a footprint
of laser signal on the surface. The range is calculated from the
time-of-flight of laser pulses and the vacuum speed of light
(Abshire et al, 1999). In Figure 1, planetary radius, R_MGS and
R_areoid indicate the distances from the center of Mars to the
surface, MGS and areoid respectively. Areoid is the reference
surface on Mars. Topographic height can be calculated using
geometry in Figure 1 (Abshire et al, 1999). The ephemeris time
IS the time instant that a laser signal is shot. Along with the
information provided in PEDR, 3-D ground coordinates, X, Y
and Z, of MOLA profiles in Mars body-fixed system, IAU 2000
reference system, can be derived from areocentric longitude
and latitude (Shan et al. 2004).
823
MGS
MOLA
range
R MGS
Topographi
height
Planetary
Radius
R_areoid
Surface
Areoid
Mars center
Figure |. MOLA range, topographic height and planetary radius
3. CONSISTENCY OF MOLA AND MOC
REGISTRATION
The property of linear pushbroom images and collinearity
equations are used for the calculation of MOC image
coordinates of MOLA points. The registration of MOLA, which
is a precedent step of this research, is reported in (Yoon and
Shan, 2003). For the convenience of readers, this process is
briefly summarized. First, MOC exterior orientations are
extracted at a constant time interval from SPICE (Spacecraft,
Planet, Instrument, C matrix (rotation) and Event) that is a
library provided by NAIF NASA. Using SPICE, binary
navigation data as it is called a kernel can be accessed by time.
Time-dependent exterior orientation of each MOC scan line is
modeled by a second order polynomial (Shan et al, 2004).
Secondly, image coordinates of corresponding MOLA profiles
are calculated using the collinearity equations with the exterior
orientation from the sensor model and the ground coordinates
of MOLA profiles.
The result of the above calculation shows MOLA profiles are
registered into different positions on the two stereopair images.
It was previously reported that the registration shifts are around
325 meters mainly in the flight direction in all three study sites.
The results are shown in Figure 2 and Figure 3 to compare with
bundle adjustment results. To precisely correct this mis-
registration and obtain accurate point determination, a bundle
adjustment is developed and implemented.
4. BUNDLE ADJUSTMENT
The combined adjustment integrates primarily MOLA profiles,
MOC image orientation data, and tie points collected on MOC
stereo images. Various types of measurements and their a priori
standard deviations are introduced in the bundle adjustment.
Image coordinates of MOLA and tie points, MOLA ranges,
MOLA ground coordinates, MOC exterior orientation are
considered as measurements in the bundle adjustment. Image
coordinates of MOLA footprints initially result from the
previous registration procedure, while image coordinates of tie
points are manually and automatically measured on stereo
