Bayesian statistical inference to remove periodic noise in the optical observations aboard a spacecraft

Appl Opt. 1988 Nov 1;27(21):4514-9. doi: 10.1364/AO.27.004514.

Abstract

Optical data taken aboard a spacecraft sometimes suffer from an unexpected modulation synchronized with the rotation or wobble of the spacecraft. This modulation may cause a serious error in estimating the volume emission rate from the observed column emission rate when an inversion method is used, since the differential operation in the inversion procedure is sensitive to noise. A new statistical technique that discriminates between random and periodic noise has been developed in this study. This technique, based on the concept of entropy maximization, gives the best-fit solution to the data. The criterion of fit used in this procedure is the logarithmic likelihood of a solution calculated in terms of the probability characteristics of the data. This method is very effective on data containing a periodic modulation due to the spin motion of a rocket as well as a random fluctuation. We use this technique in the analysis of 5577-A airglow data obtained aboard a rocket.