Techniques of Wind Vector Estimation from Data Measured with a Scanning Coherent Doppler Lidar

Abstract

The results of a theoretical study of the feasibility of wind velocity vector estimation from data, measured with a scanning coherent Doppler lidar, are presented. The estimation techniques considered are (a) the direct sine wave fitting (DSWF) and the filtered sine wave fitting (FSWF), where at first the radial wind velocities are estimated and then the wind vector is estimated from the dependence of the radial velocity versus the azimuth angle of the scanning; and (b) the maximum of the function of accumulated spectra (MFAS) and the maximum likelihood for the wind vector estimation (WV ML), where the wind vector is estimated directly from data measured by a scanning lidar without intermediate estimation of the radial wind velocities. It has been shown that due to strong averaging of noise fluctuations in accumulated spectra, the WV ML and MFAS techniques allow one to estimate the wind vector with acceptable accuracy at an essentially lower signal-to-noise ratio (SNR) than the methods of the sine wave fitting, where noise can be the source of many spurious estimates of the radial wind velocity. The ability to find optimal criterion (in the case of MFAS) for acceptance or rejection of the wind vector estimate has been analyzed. The amount of measured data needed for spectral accumulation in order to realize optimal performance has been calculated.