The molecular channel of the space-based Doppler lidar ADM-Aeolus relies on a double Fabry-Perot (FP) interferometer. The difference in photon numbers transmitted by the two FPs divided by their sum-the so-called Rayleigh response-is a function of the central frequency of the spectrum of the laser light backscattered by the atmosphere, so that a proper inversion enables the measurement of Doppler shifts and line-of-sight wind velocities. In this paper, it is shown that the relationship between the Rayleigh response and the Doppler shift does not depend on the sole characteristics of the instrument, but also on the atmospheric pressure and temperature (through the Rayleigh-Brillouin effect), and the likely presence of a narrow-band radiation due to particle scattering. The impact of these on the precision of inverted Doppler shifts (or line-of-sight winds) is assessed showing that a correction is needed. As they are lacking the appropriate precision, climatology profiles of pressure, temperature or aerosols cannot be used as an input. It is proposed to use data predicted by a numerical weather prediction system instead. A possible correction scheme is proposed. Its implication on the quality of retrieved Rayleigh winds is discussed.

The first spectral moment estimation of a Doppler signal yk can be obtained using an FFT or an autocovariance [1,2] algorithm if the assumption on stationnarity and a slow variation are not met. The FFT algorithm has limited on resolution related to the number of sampled to be used (2N).

Unlike most space-based atmospheric sounding instruments, proposed lidar systems are characterized by a very small sampling cross-section (r ~ 10 m). Although lidars may be able to sample through partly cloudy scenes that would be considered "contaminated" for passive systems, there remains the issue of how representative and useful that information might be.