FAST COMPUTATIONAL SCHEME FOR INVERSE MODELING OF MULTISPECTRAL RADIANCES - APPLICATION FOR REMOTE-SENSING OF THE OCEAN

In a variety of remote sensing applications inverse modeling is used t o retrieve parameters of our environment from measurements of the radi ance spectrum. As the dimensionality of the data increases with the nu mber of spectral channels or as models become more complicated, invers e modeling becomes expensive in terms of computing time. We describe a technique to parameterize a given inverse model, which reduces the ne cessary computing time by orders of magnitude. An example that uses si mulated data of SeaWiFS, an advanced ocean color sensor to be launched in 1993, is given, whereby the concentrations of suspended matter, ph ytoplankton, and gelbstoff (yellow substance) in water are determined.