International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part Bl. Istanbul 2004
1.2 HRS' Geometry
HRS is fully integrated on board SPOTS and ancillary data
provided with each image of an HRS stereopair has got same
characteristics as SPOTS High Geometric Resolution principal
mission ancillary data.
All data necessary for a physical geometric modelling of an
image are given in a DIMAP format file named
METADATA.DIM :
- dating is made through two models : correspondence
between board time (BT) and universal time (UT) is given
by Doris ultra stable oscillator and correspondence
between image line and board time is given by onboard
oscillator. Image centre line time, line period and each
ephemeris and attitude sample time are deduced from these
models and given in universal time ;
- satellite position and velocity is given every 10 s by
DORIS DIODE on board orbitography determination. The
best orbit arc is fit on DIODE data thanks to TRIODE
software and result in ephemeris points given every 30 s. ;
- Satellite yaw, pitch and roll orientation angles are
computed out of quaternions delivered by the on board
stellar location unit (ULS) and given every 125 ms (8 Hz) ;
- pair of viewing angles psiX, psiY are calibrated on ground
.. and given for each detector of the CCD array.
Details about how to proceed with the geometric modelling of
an image out of its METADATA.DIM are given in [Spot
Image 2002].
With those, HRS location performance is expected to be better
than 50 m rms (specification). Before launch expectations even
showed that a 25 m rms performance could be reached.
This performance level, much better than for the previous
SPOT satellites, is mainly due to the onboard stellar location
unit. This system uses both star tracker and gyroscopes data to
compute absolute orientation of the satellite. The result is a
good low frequency accuracy thanks to the star tracker, and a
good high frequency accuracy thanks to the gyroscopes.
1.3 Geometric calibration of HRS performed during the
quality image commissioning phase
The image quality commissioning phase was divided into two
parts. The first two months were dedicated to first calibrations
and characterisation in order to ensure that SPOTS was
fulfilling its specifications and ready for commercial
exploitation. During the next 4 months, image quality
commissioning went on, with a fine characterisation, and
improvement of the first calibration processed.
HRS' geometric calibration during this phase has consisted in :
- measuring orientation bias of the two cameras on SPOTS
platform thanks to ground control points (outer
orientation) ;
- calibration of the optic distortion and relative orientation
of the two cameras by comparison of HRS images with a
precise aerial reference (inner orientation).
This process has resulted in updates of the list of viewing
directions psiX, psiY given for each detector (See [Bouillon,
2002], [Breton, 2002] and [Gachet, 2004] for more details about
geometric calibrations and performance measurements carried
out during the commissioning phase).
At this point, the goal for HRS was to meet the location
specification of 50 m rms and it was attained. Still, HRS
challenge is to provide a worldwide database of Digital Terrain
Models and Orthoimages, as automatic as possible, called
Reference3D. This database is produced from HRS acquisition
with no use of ground control points and a horizontal location
specified as 16 m for 90% of the points (see [Airault, 2003] and
[Bernard, 2004]).
So, at the end of the commissioning phase, some steps remained
to get location results in accordance with pre flight expectations
and it was also necessary to check that Reference3D
specifications were held.
2. PRINCIPLES FOR OPTIMISATION OF HRS
». LOCATION PERFORMANCE
After the end of the commissioning phase, some possible
improvements of HRS location performance were already
identified. In this section, we will describe method and means
used to complete the work performed during the commissioning
phase in order to get the best possible location performance for
HRS.
2.1 Adopted working method
A working group involving French Space Agency (CNES),
French Mapping Agency (IGN), SPOTS satellite integrator
(ASTRIUM) and star tracker developer (SODERN) has been
created. The method held back was to carry on an analysis of
every potential error source affecting HRS' location
performance at three levels of interpretation :
- elementary sources of errors, such as attitude restitution's
accuracy, structure deformations or geometric calibration
remainders ;
- System location performance : in this context, each image,
even of an HRS stereopair, is considered independently and
the location performance is computed separately for each
camera ;
- location performance obtained with the added value of tie
points between the two images of an HRS stereopair
(stereoscopic location performance) and of several HRS
stereopairs considered in a bundle block adjustment process
(with only tie points between the different strips, no ground
control points are used).
Each level of interpretation allows the detection of different
phenomena likely to have some impact on the location
performance. Each phenomenon has been analysed to
understand its source and exact impact. Potential improvements
have been examined, but only in terms of on board modification
of existing threshold or parameter, or of ground algorithms.
Finally, improvements considered as necessary have been
carried out and validated, as well as monitoring activities
necessary to uphold the performance.