Linear approximation of Rayleigh-Brillouin scattering spectra

摘要

Rayleigh-Brillouin scattering is the basis of many remote sensing techniques, including high spectral resolution lidar measurements of aerosols and wind. Rayleigh-Brillouin spectra can be accurately estimated using physics-based models like the so-called Tenti's S6 and Pan's S7 models. Unfortunately, these are computationally expensive and can be the bottleneck for real-time lidar processing and iterative parameter estimation problems. This short article describes a very efficient linear approximation of the Rayleigh-Brillouin spectra based on Principal Component Analysis (PCA). Using PCA, the outputs of the above models can be approximated with very high accuracy using a single matrix multiplication. The described method can be applied to the output of any detailed scattering model, thus it can be used for a wide range of problems, e.g., for scattering from different gases (Air, N-2, O-2, ...) and for different ranges of temperature and pressure. The precision of the approximation can be adapted to the requirements of the studied problem, and can easily exceed the actual accuracy of the reference models. (C) 2016 Optical Society of America