Abstract

The performances in terms of bias and standard deviation of the frequency estimators originally applied to Doppler Radar signals (FFT, Pulse-Pair and Poly-Pulse-Pair), are analysed, when working on Doppler Lidar signals characterized by low Signal to Noise Ratio (down to −10dB). Analytical equations were presented by Berger&Groginsky, and Zrnic, for the statistical error as a function of SNR and signal processor parameters. Those equations are rederived from the basic maths to make more explicit the various assumptions and limitations they imply. Those equations are then systematically compared to numerically derived performances (study made for the franco-german airbone Doppler lidar project WIND), based on simulated Doppler Lidar signals. Examples are presented of the estimator standard deviations when they are applied to signals with wide-band SNR ranging from −10dB to +10dB.

© 1991 Optical Society of America