Vaisala RVP900, Руководство пользователя

Страница: 48 / 484

Table 9 Autocorrelation Mode
Mode Description
Pulse Pair Mode Filtering for clutter removal may be performed in the time domain or
frequency domain.
Traditional IIR type clutter filters are available in the time domain.
However, the frequency domain filter is more adaptable.
Clutter filtering can be optionally performed in the frequency domain,
and then inverse FFT or DFT may be performed to return to time domain
after clutter removal. Autocorrelations are then computed in the time
domain.
DFT/FFT Mode Filtering for clutter is performed in the frequency domain using both
fixed width filters and the Gaussian Model Adaptive Processing (GMAP)
technique. Autocorrelations are computed from the inverse transform.
Random Phase Filtering for clutter and second trip echo is performed in the frequency
domain by adaptive algorithms. Autocorrelations are computed from the
inverse transform.
3.3.9.1 RVP900 Pulse Pair Time Domain Processing
Pulse pair processing is done by direct calculation of the autocorrelation.
Prior to pulse pair processing, the input
I and Q values are filtered for clutter using a time
domain notch filter, frequency domain fixed, or variable width filters. IIR filters of various
selectable widths are available for either 40 dB or 50 dB stop band attenuation. The filtered
I/Q values are processed to obtain the autocorrelation lags R0, R1, and R2. The unfiltered
power is also calculated (T0). The autocorrelations are sent to the range averaging and
moment extraction steps.
3.3.9.2 RVP900 DFT/FFT Processing
The DFT/FFT mode allows clutter cancellation to be performed in the frequency domain.
DFT is used in general, with FFTs used if the requested sample size is a power of two.
The following windows provide the best match of window width to the spectrum dynamic
range:
• Rectangular
• Hamming
• Blackman
• Exact Blackman
• Von Han
After the FFT step, clutter cancellation is done with the options of using
GMAP
, a selectable
fixed width filter that interpolates across the noise or any overlapped weather, or an
adaptive filter which automatically determines the optimal width. This technique preserves
overlapped weather as compared to time domain notch filters, which always attenuate
overlapped weather to some extent, depending on the spectrum width.
After clutter cancellation,
R0 , R1 , and R2 are computed by inverse transform and these are
used for moment estimation.
RVP900 User Guide M211322EN-J
46