Denoising Atmospheric Radar Signals Using Spectral-Based Subspace Method Applicable for PBS Wind Estimation

摘要

This paper mainly focuses on the advantages of subspace-based eigenvector (EV) spectral estimator to improve the power spectrum and the quality of calculations in spectrum parameter estimation. In general, the spectrum produced by most of subspace methods is sharply peaked at the frequency of complex sinusoids. Although subspace methods exhibit the advantage of spectral resolution, the retrieval of the actual spectrum width is not well observed in many cases, compared with standard Fourier estimates. Several simulation works are carried out to determine the unknown order of the signal correlation matrix, which significantly helps in obtaining the equivalent Fourier spectrum using EV along with numerous advantages of the subspace method for better estimation of spectrum parameters. Such advantages are useful in precisely obtaining the atmospheric moments (Doppler frequency, spectrum width, etc.) from the synthesized beams required for wind estimation by the postset beam steering technique. In addition, the systematic improvements done in EV are much useful for complete wind profiling up to $sim$20 km with a temporal resolution of $sim$26 s, which is reported for the first time.