Alternative Retrieval Algorithm for the Ozone Mapping and Profiler Suite Limb Profiler.

摘要

Stray light significantly contaminates the signals from many instruments for remote sensing. The usual way of compensating for stray light is to use a model for the production of stray light by the instrument, to estimate what the measured signals would have been in the absence of stray light, and then to use the corrected signals to derive the sought-for geophysical or astrophysical quantities. Correcting the signals for stray light is an ill-posed inverse problem (very close to a deconvolution). In most remote sensing tasks, extracting the geophysical or astrophysical quantities from the signals is also an ill-posed inverse problem. Stacking these two ill-posed inverse problems often leads to unusably unstable results. Stable, accurate results can be obtained by bypassing the estimation of the corrected signals. The geophysical or astrophysical quantities are derived directly from the uncorrected signals, but the production of stray light by the instrument is included in the forward model, along with the external radiative transfer effects that are usually included in the forward mode. That eliminates the stacking of ill-posed inverse problems.