A Novel Regression Band-Stop Filter and Application to NEXRAD SZ Phase Coded Data

Filtering of short-length time series is typically accomplished with FIR, IIR or spectral domain filters. Recently, global regression filtering has been shown to be effective in eliminating radar ground clutter. The novel filter presented here shows how a high-pass regression filter can also be used as an arbitrary frequency band-stop filter. For applications where band-stop filtering is required for short length time series, this novel filter has several significant advantages over the more traditional filtering approaches. For example, FIR and IIR filters require special attention at the beginning and ending of the time series where filter taps are undefined, and spectral filters typically require an attenuating window function that effectively reduces the number of available independent samples. The regression band-stop filter is implemented using the Fourier frequency shift property and is termed the RFSF (Regression Frequency Shift Filter).

An application of the RFSF is in the processing SZ phase coding where a band-stop filter is required to eliminate the strong trip echo after which the weak trip signal’s velocity can be effectively estimated. The RFSF significantly reduces measurement error so that improved recovery of the weak trip velocity is possible. The RFSF is demonstrated on NEXRAD SZ phase coded data. The resulting recovered velocities are compared to the equivalent NEXRAD Level-2 velocity field. The comparison demonstrates that the more velocities are recovered in the region of overlaid echoes, and additionally, and the spatial standard deviation of the velocity field are reduced with the RFSF.