UCSD/SIO Final Report
Advanced Computing Technology Applications to
SAR Interferometry and Imaging Science
Professor David T. Sandwell, dsandwell@ucsd.edu
Professor J-Bernard Minster, jbminster@ucsd.edu
Contents:
-
Objectives
-
Publications
-
Small-Scale
Deformation Associated with the Landers Earthquake Mapped by SAR Interferometry,
JGR, v. 103, p. 27001-27016, 1998.
-
Phase
Gradient Approach to Stacking Interferograms, JGR, v. 103, p. 30183-30204,
1998.
-
Near-Real-Time
Radar Interferometry of the Mw 7.1 Hector Mine Earthquake - in press,
GRL,May 2000.
-
Topographic
Recovery from Stacked ERS interferometry - in press, JGR, May 2000.
-
X-Band
Downlink
-
Near-Real-Time Data Acquisition and Processing
Objectives - Monitoring
Strain Buildup and Earthquake Displacements in California
The long-range objectives of our research are to understand
the kinematics and dynamics of active plate boundaries through precise
geodetic and geophysical measurements. Because the continental lithosphere
has a more complex layered rheology than the oceanic lithosphere, strain
across the major plate boundaries such as subduction zones and transform
faults is more diffuse on the continents than it is in the oceans. For
example, the strain along the 800 km long Eltanin transform fault in the
South Pacific is confined to a 30 km-wide zone along two major faults.
In contrast, the strain along the 800 km long San Andreas transform system
occurs over a zone at least 200 km-wide containing numerous parallel faults.
Thus the continental tectonics are more complicated and the location and
timing of earthquake ruptures are less predictable. Earthquake damage and
loss of life is a major concern in Southern California (e.g., January 13,
1994 Northridge earthquake). While timely earthquake prediction may elude
us for the foreseeable future, a better understanding of the physics of
the earthquake cycle may aid in earthquake preparation and seismic hazard
assessment. As has already been demonstrated by other groups, the southwestern
US is an ideal location to use interferometric SAR methods to measure ground
displacements. These applications have so far been confined to coseismic
and postseismic motions. We propose to use SAR interferometry to also measure
interseismic motions by interpolating between the existing array of permanent
GPS receivers (PGGA) coordinated by the Southern California Earthquake
Center with NASA and other funding. Being able to measure the complete
deformation field in space and time would be immensely valuable in understanding
earthquake hazard; because of our involvement in the Southern California
Earthquake Center we would be able to apply these results directly to this
problem.
Near Real Time Interferometry
Direct Readout from ERS-2:
The SIO X-band ground station has been in operation since
June 1, 1999. Funding from this proposal paid for a spectrum analyzer
as well as some initial data purchases. About 3.0 Tbytes of raw SAR
data from ERS-2 have been collected and about 1.3 bytes of tectonically
useful data have been archived. The system can operate without human
contact although we have hired an undergraduate student to monitor the
operations. Currently there is no external funding for operating
the facility although SeaSpace Co. has a commitment to keep things running.
The operations are largely automatic:
Acquisition of raw ERS-2 data
-
Fresh orbital elements for relevant satellites are gathered
each day at 17:00 GMT from the Naval Space Command under an MOU with
SIO.
-
TeraScan software directs the 5-m dish to track every ERS-2
overflight.
-
X-band telemetry is monitored, digitized, and written to
a local raid-disk array whenever the radar is turned on.
-
When the pass is complete, the raw telemetry is frame synchronized,
converted to ESA/DPAF format, and copied to a 64 Gbyte raid disk for intermediate
storage.
-
Relevant files are written to the IGPP mass storage area
and deleted after two days.
-
(Note there is still a problem with antenna pointing accuracy;
sometimes data are lost when the satellite is overhead. We are currently
working with Scientific Atlanta engineers to solve this problem.)
Ancillary Data -Two types of ancillary data
are required for the construction of interferograms in real time as well
as for the maintenance of the SAR data archive.
-
Accurate Time ? Each raw ERS-2 SAR echo is time-tagged using
the onboard clock. These relative times are converted to absolute
time (~1 millisecond accuracy) using PATN files provided by ESA on a daily
basis.
-
Accurate Orbits ? Three types of accurate orbits, all provided
by Delft University [Scharroo et al., 1998] are used (http://deos.lr.tudelft.nl/ers/precorbs/status.html).
Precise orbits (~0.05 m accuracy) with about a 6 month delay are available
for the entire ERS-1 and ERS-2 missions. Fast delivery orbits (<
0.10-m accuracy) with about a 5-day delay are used for near-real time analysis.
Real-time orbital predictions are available for real-time interferometry.
All three types of orbits are generated to support the NOAA real-time radar
altimetry program for ERS-2.
Near Real-Time Interferometry -Data are stored files
of typically 0.5 million echoes along a single track (~ 5 Gbyte).
To create an interferogram between a newly-acquired SAR image and a stack
of archive images, the start time and stop time of the master image in
the stack are used along with the timing information and the precise orbits
to identify the relevant records in the newly acquired pass. Interferometric
baselines are also computed. The relevant data are then extracted
from the pass file, the image is focussed (either at SIO or JPL), and the
image is aligned at the sub-pixel level to the master. Finally interferograms
are constructed between the new data and any one of the members of the
stack. With ordinary workstations, an interferogram can be ready
about 3 hours after the data are downlinked but this time is substantially
reduced by using the interferometric code developed under this proposal.