This report presents and discusses the application of Multi Layer Perceptron (MLP) neural networks to derive Chlorophyll-a concentration (Chl-a), absorption of the yellow substance at 412 nm (ays(412)) and concentration of the total suspended matter (TSM) from remote sensing reflectance RRS values. MLPs were developed on the basis of data collected within the framework of the Coastal Atmosphere and Sea Time Series (CoASTS) and Bio-Optical mapping of Marine Properties (BiOMaP) programs carried out by the Institute for Environment and Sustainability (IES) of the Joint Research Centre (Ispra, Italy). Investigated oceanographic regions include the Eastern Mediterranean Sea, the Ligurian Sea, the Northern Adriatic Sea, theWestern Black Sea, the English Channel and the Baltic Sea. The study verifies the applicability of MLPs to retrieve ocean color data products in each basin. For instance, the highest accuracy in retrieving Chl-a has been found in the Easter Mediterranean Sea and the Ligurian Sea (14 and 25 %, respectively). In the case of ays(412), the MLP is the most performing in the waters of the English Channel and the Baltic Sea (14 and 13%). Instead, the TSM retrieval is the most accurate in the Black Sea and at the Acqua Alta Oceanographic Tower (14 and 19%). To enhance mission specific ocean color resuls, MLP coefficients are also computed applying band-shift corrections to produce RRS spectra at wavelengths matching those of SeaWiFS, MODIS and MERIS. Resulting tables of MLP parameters are reported to permit independent applications of neural networks presented in this study.
|Publication status||Published - 1 Jan 2011|