A straightforward signal-to-noise ratio estimator for elastic/Raman lidar signals

Abstract

In this paper we estimate the signal-to-noise ratio (SNR) at the opto-electronic receiver output of both elastic and Raman lidar channels by means of parametric estimation of the total noise variance affecting the lidar system. In the most general case, the total noise variance conveys contributions from photo-induced signal-shot, dark-shot and thermal noise components. While photo-inducted signal-shot variance is proportional to the received optical signal (lidar return signal plus background component), dark-shot and thermal noise variance components are not. This is the basis for parametric estimation, in which the equivalent noise variance in any receiving channel is characterized by means of a two-component vector modeling equivalent noise parameters. The algorithm is based on simultaneous low-pass and high-pass filtering of the observable lidar returns and on weighted constrained optimization of the proposed variance noise model when fitting an estimate of the observation noise. A noise simulator is used to compare different noisy lidar channels (i.e. with different pre-defined noise vectors or dominant noise regimes) with the two-component noise vectors estimate retrieved. Both shot-dominant and thermal-dominant noise regimes, as well as a hybrid case are studied. Finally, the algorithm is used to estimate the SNR from lidar returns from tropospheric elastic and Raman channels with satisfactory results.