93
conjugate of the other. The magnitude of each pixel in this interferogram is a measure of
the target pixel’s radar backscatter and its phase, O, is a measure of the target’s radial
velocity. The reason for this is that the doppler processed image from the second antenna
represents the same scene as shown in the image from the first antenna but at a time which
is later by the period taken for the aircraft (and thus the phase centre of the antenna) to
travel the distance between the two antennas, t-dlV. In that time a target with radial
velocity, v, will have moved a distance vt parallel to the zero doppler direction and thus
its phase will have changed by 4 nvtlk, where A is the wavelength of the radar (5.656
cm in the case of the CCRS along track InSAR). The phase of the interferogram can thus
be directly translated to radial velocity using v -VX O / 2dn , where the reduction by a
factor of two is due to the fact that the same transmit antenna is being used for both
InSAR antennas and so the effective difference between the transmit/receive zero doppler
positions is only d / 2. The necessity of exactly registering both images prior to combining
them in an interferogram adds some complexity, but is not a problem provided the physical
separation and electronic path lengths associated with each antenna are known. If the
target becomes decorrelated over the time interval, t , however, there will be a problem
because the two images will not be phase coherent and the interferogram will become
meaningless. This is not a problem for the CCRS InSAR when viewing typical ocean
targets, since t - dIV is typically around 5 ms whereas the ocean decorrelation time at
C-band is expected to be between 50 and 100 ms.
The first test of the CCRS along track InSAR took place on January 12, 1994 and
demonstrated that the radial velocities determined by the InSAR system were accurate to
within approximately 5 cm/s even though the InSAR antennas were stuck at their steering
limits due to high cross winds [A. Laurence Gray et al., 1994].
2.2 CHS SURVEY SHIP
The Canadian Hydrographic Service of the Department of Fisheries and Oceans provided
the survey ship NSC Frederick G. Creed to take part in the June, 1994 experiment. It was
equipped with a SIMRAD EM 1000 multibeam sonar, which had 60 separate beams, each
of 3.3° by 3.3°. These beams were spaced 2.5° apart in an angular sector covering +75° to
-75° from the vertical so that as the ship moved they could be used in a pushbroom
imaging mode. The insonification frequency was 95 kHz using a 0.2 ms CW pulse.
In standard survey mode the ship would cruise at 12-16 knots, covering a swath of a few
hundred meters on the ocean bottom. The size of the swath varied because the bottom
footprint of each sonar beam varied from 2-5 m at the depths where the surveys were
being conducted. A typical survey of the Scots Bay dune field required approximately
twelve hours. Since this corresponds to a full tidal cycle in which the water level could
change by as much as 10 m, it was necessary to reference the bottom topography to
depths below the WGS-84 ellipsoid rather than below the water surface. This was
accomplished using differentially processed GPS data from a receiver on board the ship,
which also allowed the x-y coordinates of the bottom topographical features to be
converted to an accurate UTM map.