92
subsurface phenomena and compares them with the ship based measurements. Finally, in
Section 5, some conclusions about the utility of the system are presented.
2. EXPERIMENTAL SETUP
2.1 CCRS InSAR
The structure and operation of the CCRS C-band SAR has been described in detail
previously [Livingstone et al., 1987, Livingstone et. al., 1995]. In 1991, this system was
modified to include an across-track interferometric capability by adding a microstrip
antenna on the right side of the aircraft. In 1993, the InSAR system was further enhanced
to allow along track interferometry by dividing the microstrip antenna into two separate
sub-antennas with phase centres separated by d = 0.46 meters along the aircraft. The
main C-band antenna located underneath the aircraft transmits horizontally polarized
pulses which are received and recorded separately from each of the microstrip antennas.
The microstrip antennas have asymmetric azimuth weighting functions which results in
each antenna having a main lobe squint of approximately 1.1° in opposite directions. To
compensate for this, the antennas are mounted with a dihedral having an included angle of
approximately 2°. This results in two-way transmit-receive sidelobes of better than -40 dB.
The sub-antennas have a 3 dB beam width of 6°, however, which can give rise to azimuth
ambiguities using the normal ratio of the pulse repetition frequency to aircraft ground
speed (PRF/V = 2.5672 m' 1 ). This is why the along track InSAR is run at double speed
with PRF/V = 5.1345 m' 1 .
Both the transmit and receive antennas are steered in azimuth using data from a LTN-92
inertial navigation system sensor head located directly above the main antenna, combined
with a real-time clutter lock which ensures that the antenna boresights remain steered to
zero doppler. All other motion compensation is performed during post-flight processing
using data from the LTN-92 and from an Ashtech GPS receiver. Normally the GPS data is
differentially processed with data from a second receiver at a known ground point. The
antenna phase centres can then be located with submeter precision. Phase calibration is
always necessary to achieve accurate estimates of radial velocity. Any variation in the
electrical path lengths associated with the two sub-antennas, or the two C-band receivers
in the radar system will change the differential phase and corrupt the radial velocity
estimation. More importantly, however, any error in the measurement of the positions of
the antenna phase centres will lead to a phase calibration problem. Although the antenna
positions are well known with respect to the aircraft and the absolute position of the
imaging plane (the zero doppler plane) can be derived from the GPS data, the absolute
attitude of the aircraft is subject to uncertainty because of drift in the laser ring gyros used
in the LTN-92 unit. The only way to guarantee a reliable phase calibration is to find a
large flat object with zero radial motion close to the object being imaged. Fortunately,
coastal land serves this purpose if it is included in the radar image.
After the radar image from each along track antenna has been processed to a complex
SAR image, an interferogram is produced by multiplying one image by the complex