Geoscience Reference
In-Depth Information
Fig. 13. Autocorrelation spectral density (ASD) of a radar return, top: magnitude and
bottom: phase. Clutter is well defined with its peak at zero and flat phase (red). Based on
this phase five coefficients are replaced with interpolated values resulting in 14.4 dB of
suppression (defined as the ratio of total S+C power to remaining power). Interpolated
powers are indicated by the dotted line; dash line represents linear phase;
v
a
=
27 m s
-1
. Data
obtained with the phased array radar (NWRT). (Figure courtesy of Sebastian Torres).
The choice of window function
d
(
m
) is very important because its sidelobes limit the amount
of power that can be filtered. The clutter power is computed from the sum of
V
(
m
) to obtain
the clutter to noise ratio (CNR). Then the CNR is compared with the peak to first sidelobe
level (PS
w
) ratio of four windows (w=rectangular, von Hann, Blackman, and Blackman-
Nuttall) and the window whose PS
w
exceeds the CNR by the smallest amount is chosen.
That way the leakage of the clutter signal away from zero will be below the noise level,
while the notch width will be smaller than the one for the other windows satisfying the
condition PS
w
>CNR.
Data windows spread the phase of clutter's
S
1
(
k
) either side of zero (
k
=0) Doppler (Fig. 13).
Recognition of the flat phase identifies clutter's presence. Doppler index at which the phase
begins to depart from zero (according to a set of criteria) defines the clutter filter width. In
the mean the autocorrelation spectral density of noise has linear phase as seen in Fig.13 but
semi coherent signals have flattened phases in the vicinity of their mean Doppler shifts.
Panels in Fig. 14 demonstrate qualitatively performance of this clutter mitigation technique
and the caption highlights results.
