Use of a neuro-variational inversion to retrieve aerosol parameters and ocean constituents from ocean color imagery

The retrieval of ocean constituents (OC) from satellite ocean color measurements is very sensitive to the atmospheric correction. Improved atmospheric correction algorithms, which simultaneously estimate OC and aerosol optical properties (AP), have recently been developed. Advanced programming techniques such as Neural Network can help us in implementing improved atmospheric correction algorithms in operational satellite data processing. The method is based on the combination of MLP and classical variational inversion in the visible bands. This procedure enables to retrieve OC, specially Chl-a from visible wavelengths. It consists in inverting the direct model to retrieve OC, by minimizing the distance between the observed and MLP-calculated Top Of Atmosphere reflectance. To perform this inversion, we have modelled the oceanic and atmospheric radiative transfer equations with respect to the AP and OC by using MLP. The accuracy is better than 7%. We have focused on the influence of the information on AP for the retrieval of OC with pseudo-data for weakly absorbing aerosols. We show that Chl-a can be retrieved with an error of 19.7% if we have a perfect knowledge of three AP. Finally, results for East Coast of the United States are presented.