An improved hybrid spectrum width estimator

Spectrum width, a measurement of the variability of the radial wind velocities within a radar pulse volume, has been shown to be useful in the remote detection of in-cloud turbulence for aviation users. The standard pulse-pair estimator of spectrum width, which uses the zeroth and first lags of the autocorrelation function of the I&Q time-series, has been shown to have biases and large errors for low signal-to-noise ratios and/or small true spectrum widths. The other standard pulse-pair estimator, based on the first and second lags, does better in those regimes but has serious and unexpected saturation behavior. We present a proposed replacement of the current WSR-88D spectrum width estimator that integrates simpler estimators to produce a superior spectrum width estimate. This straightforward approach takes advantage of the relative strengths of each of the simpler estimators. The performance characteristics of this hybrid estimator are compared with those of the standard pulse-pair estimators, and potential impacts on the NEXRAD Turbulence Detection Algorithm are shown.