4.2-O A new intermittent clutter filtering algorithm for Radar Wind Profiler

The standard RWP signal processing usually assumes, that the received voltage signal at the receiver output S(t) is the sum realization of the atmospheric scattering process I(t) and the receiver noise N(t). Both can be regarded as independent stationary Gaussian random processes. All terms are complex-valued base-band voltages that are obtained after demodulation at the receiver output. Stationarity can be assumed over typical dwell-times (quasi-stationarity). Under these assumptions, the stochastic process S(t) is completely described by its time-independent second-order properties. The natural way to process RWP data is therefore based on a non-parametric estimation of the power spectrum (Doppler spectrum) using a discrete Fourier transform of the (usually coherently integrated) raw signal over a fixed time interval. In reality, however, there is often a third component contributing to the total signal S(t), namely clutter: