2. Baseline design
The length of baseline should be
optimized with S/N, imaging looks and
geometrical design difficulty. We know that
for the two antenna interferometric system
(with a horizontal baseline), the correlation
between the echoes received by two antennae
decreases as the baseline increases. When the
baseline equals BC , the correlation becomes
zero
[8].
Where Ry is the ground range resolution.
Baseline length can be optimized
according to the relation between the phase
uncertainty ( e
f ) and signal to noise ratio
(SNR) and number of looks (N
L)
[9]:
Where
g is decorrelation coefficient.
In practice, the baseline length B is
chosen about 0.1 to 0.2 times of B
C trading off
the sensitivity of phase measurement and the
engineering complexity in realizing the
interferometric system. In our design, the
length of baseline B is three meters.
3. Observation angle
Fig.3 The average return response of a
rectangular pulse with nadir looking
Fig.4 The average return response of a
rectangular pulse with off-nadir 2° looking
The observation angle should be
designed compatible with the accuracy of
significant wave height, range resolution and
the length of baseline. In ocean observation,
when the antenna is nadir looking, the echo
model can be described using Brown’s Model.
If the antenna is off-nadir looking for a small
angle, the Brown’s Model is still appropriate
to use after making some modifications. From
equation (1), one can see that, the larger the
off nadir looking angle, the longer the length
of baseline is needed under the same slope
range, however, the higher ground range
resolution and the wider swath can be
achieved.
Above all, the off nadir angle is better in
the range of several degrees. In our design, 2
° off nadir angle is used.
Fig. 3 indicates the average return
response of a rectangular pulse with nadir
looking and Fig. 4 indicates he average return
response of a rectangular pulse with off nadir 2
°
looking. Comparing these two figures, one
can see, the return shapes are quite different.
In the off-nadir case, the discrimination
between echo shapes at small SWH is not as
clear as that of nadir case.
4. PRF design
PRF is very important for constraining
the azimuthal and range ambiguity, not
loosing Doppler information, and lowering
down the data rate.
In our system design, the observation
angle is very small, and the swath is also very
narrow compared with conventional SAR
system, so the range ambiguity is not serious.
We can select PRF simply according to the
following equation:
fd<PRF<c/2(Rmax-Rmin) (4)
Where R
min and R
max are the near slope range
and far slope range.
5. Signal bandwidth
The wider the transmitted signal
bandwidth, the higher the accuracy in range
measurement, but the narrower the tracking
scope. Therefor according to the different
surface characteristics of ocean, oceanic ice
and land, the bandwidth should be different.
In CIALT, three bandwidths is used, they are
320MHz, 80MHz and 20MHz respectively.
These three chirps are generated digitally, so
the change of bandwidth according to
different surface is easy to realize.
6. Compression in range direction
Full-deramp technique and matched
filtering technique can be used to complete
range compression, but they are fit to different
cases. In CIALT, high speed A/D and matched
filter is involved.
7. Compression in azimuth direction
Focused or unfocused processing should
be selected according to the resolution
requirement, complexity of system, and the
operational height of platform. In CIALT,
unfocused processing is enough to meet the
design goal.
8. Tracking Algorithm
In order to be able applied to ocean,
oceanic ice and land altogether, model-free
tracker should be designed. Tracking
algorithm based on offset center of gravity
(OCOG) can meet the requirement of model-free
tracker. In this algorithm, the height error
is linear across the entire range window and it
is independent of pulse shape.
In OCOG algorithm, there are only two
parameters to be estimated from echo profile,
i.e. pulse amplitude, A, and the pulse width W.
According to them, errors signals which
control height and gain loops are generated.
When this error become large and the echo
will be outside the tracking window, so the
altitude resolution will be degraded. This
results in the broadening of tracking window
to capture the echo again.
