Bayesian approach for spatial super-resolution of heterodyne wind lidars

TL;DR

This paper introduces a Bayesian inverse problem approach to enhance the spatial resolution of wind speed measurements from heterodyne lidars, surpassing the traditional physical limits imposed by laser pulse duration.

Contribution

It presents a novel Bayesian inversion method that leverages spectrogram modeling and prior distributions to significantly improve spatial resolution in wind lidar measurements.

Findings

Achieved a resolution gain factor of 2 to 2.5 in simulated and experimental data.

Demonstrated effectiveness across different signal-to-noise ratios.

Surpassed the traditional resolution limit cτ in wind lidar measurements.

Abstract

Wind speed measurements using heterodyne lidars are limited in spatial resolution because of the current signal processing methods. This limit is equal to c $\tau$ ( c is the speed of light and $\tau$ is the laser pulse duration) corresponding to the length of the atmosphere contributing to the wind speed measurement at one distance. To go beyond this limit, we use an inverse problem approach based on a model of the spectrogram (concatenation of periodograms of each range) and prior distributions on our unknowns: backscattering amplitude and wind speed at each range. We apply our inversion method to simulated and experimental spectrograms, demonstrating a gain in resolution by a factor of 2 to 2.5 depending on the signal-to-noise ratio.