Comparison of maximum likelihood estimators of mean wind velocity from radar/lidar returns

Wind patch returns from a radar/lidar may be accurately modeled by a Gaussian random process with a Gaussian power spectral density of unknown mean, standard deviation and power. The maximum likelihood estimator (MLE) of the unknown parameters is analyzed, its performance compared against other estimators of the mean wind velocity and the corresponding Cramer-Rao bound (CRB). Simulations show the MLE to have superior performance to the poly-pulse-pair techniques at all signal-to-noise ratios. We present also a simplification of the CRB that results in the computation of the variance of mean wind velocity uncoupled from the return and wind velocity dispersion.