Geoscience Reference
In-Depth Information
call Clutter Phase Alignment (CPA) defined as CPA=| sum V k |/sum| V k | , where V k is the
complex voltage (I + j Q) from a fixed clutter location at consecutive times (spaced by the
PRT) indicated by time index k and the sum is over the total number of pulses in the dwell
time. Local standard deviation (termed texture) of reflectivity factor Z i in range ( i index
indicates adjacent values in range) and changes in sign of the differences Z i+1 - Z i are also
used; the frequency of change in reflectivity gradient along range is obtained from this
difference and it defines the spin variable. The CPA, texture, and spin are combined in a
fuzzy classification scheme to identify locations where clutter filter should be applied.
Fig. 8. Doppler spectrum of simulated weather signal (red) and clutter (blue). Interpolated
(filtered) Gaussian part and estimated noise level are shown. The v a =32 m s -1 . (Figure as in
Torres et al., 2004c).
The GMAP filter and censoring (Free & Patel, 2007) is applied to surveillance and Doppler
scans. In the “batch” mode the number of samples is insufficient for spectral processing
hence the average voltage (i.e., DC) from the samples spaced by the long PRT is removed.
The system also employs strong point clutter (typically caused by aircraft) removal along
radials. It is done on each spectral moment independently by comparing the sample power
with two adjacent values either side of it. If the value is outside prescribed criteria it is
replaced by interpolation of neighboring values.
3.4 Computation of spectral moments
In computations of Z and σ v receiver noise powers are subtracted from the returned powers.
Thus, the receiver noise power is estimated at the end of each volume scan at high elevation
angle. The noise depends on the elevation angle because contributions from ground
radiation and air constituents are larger if the beam is closer to the ground. To account for
the increase the noise is extrapolated to lower elevations using empirical relations.
The reflectivity factor is obtained by summing the pulse powers, subtracting the noise
power, and using the radar equation (Doviak & Zrnic, 2006). At the lowest few elevations Z
is computed from the long PRT (surveillance scan). At mid elevations (“batch mode”) the
Search WWH ::




Custom Search