Inferred IOP and particulate matter from MERIS and MODIS multi-spectral satellite images

Abstract

How to accurately determine and quantify particulate and dissolved matters at the water surface in coastal areas (case 2 waters) using passive ocean colour remote sensing is still a topical issue. The best actual models, like the OC5 algorithm from Gohin et al., 2002, are empirical, and usually restricted to a limited area; hence the need to gather numerous oceanographic data is strong. To improve our understanding of in-water radiation law, we tested a numerical bio-optical model with several components including bottom reflectance. The software called WASI (Water Colour Simulator) was developed by Peter Gege (Gege, 2004) and integrates forward and inverse modelling for eight common types of optical in-situ measurements in aquatic environments (in particular remote sensing reflectance, absorption or attenuation), using well-established analytical models. In forward mode, it simulates reflectance spectra using the IOP or the particulate and dissolved matters. In inverse mode, an input hyperspectral reflectance spectra measured with a spectro-radiometer is analyzed by fitting iteratively possible reflectance curves under constraints on some bio-optical parameters (simplex method) so as to provide the IOP and the particle concentrations. We used in-situ data acquired during the oceanographic mission OPTIC-CONGO (Gulf of Guinea) to evaluate the accuracy of the model results. The substance concentrations derived by the WASI model were compared with the in-situ measurements and the results are encouraging (Schmeltz et al., 2009). To apply this inversion to satellite data from MODIS (NASA) or ENVISAT/MERIS (ESA), we developed a multiple regression program adapted from Wernand (1997) to reconstruct hyperspectral reflectances from the multi-spectral satellite channels (10 bands for MODIS and 15 bands for MERIS). These reconstructed spectra are inverted in WASI to obtain the IOP and substances concentrations.