Abstract

The frequency estimators that perform the best on Doppler lidar signals are somehow matched to the signal characteristics like spectral width or SNR. In the Levin estimator for instance, the fitted filter is tuned in height and width to be as close as possible to the expected signal power spectrum. In practical applications, the tunable estimator parameters are set to values corresponding to the atmospheric conditions that are most likely to be met (clear air stable conditions without significant shear and low turbulence level). Other conditions are however often met that unpredictably broaden or alter the returned spectrum. This may be particularly true for dedicated applications like wake vortex or wind shear detections. The performance of the estimation may then be highly degraded. The coupled estimation of both the signal mean frequency and width could then improve the radial velocity retrieval. Such a co-estimation is possible using an autoregressive moving average (ARMA) type of filter.

© 1995 Optical Society of America