AbstractThe paper describes the case 2 coastal water algorithm, which has been developed for the ground processor of the Medium Resolution Imaging Spectrometer (MERIS). The spectrometer is operated on board the earth observation satellite
ENVISAT of the European Space Agency ESA. The algorithm derives the inherent optical properties (IOP) (1) absorption coefficient of phytoplankton pigment, (2) absorption coefficient of gelbstoff and total suspended matter after
bleaching the phytoplankton pigment fraction and (3) the scattering coefficient of total suspended matter (TSM). The IOPs of (1) and (3) are also converted into
the concentrations of chlorophyll a and TSM dry weight. The algorithm is based on a neural network (NN), which relates the bi-directional water leaving radiance reflectances with these IOPs. The network is trained with simulated reflectances.
The bio-optical model used for the simulations is based on a large data set collected mainly in European waters. One special feature is the combination of a forward NN and a backward NN, which allows testing if the measured spectrum is within the scope of the training set. The comparison with independent test data sets, in situ observations and with the case 1 water algorithm used for MERIS show a good agreement.