Abstract

Given an estimate of the spectrum of a time series sample drawn from a single pulse lidar return, there are several ways in which the signal frequency can be estimated. Texts on weather radar stress use of the first moment expression for estimation of the mean Doppler frequency shift, with or without threshold techniques to remove contributions from the spectrum of the noise.^1,2 For lidar, there is increasing interest in discrete spectral peak estimation (DSPE) techniques,³ where the frequency at which the spectral maximum occurs is accepted as the estimate. Indeed, the weighting algorithm proposed for use with radar wind profilers⁴ uses a DSPE as a ’first guess’ of the mean frequency, prior to formation of a weighted average of spectral elements in the neighborhood of this value that have power levels above a selected threshold. It evidently has more in common with the DSPE method than earlier moment algorithms.

© 1991 Optical Society of America