Blondeau-Patissier, D, Tilstone, GH, Martinez-Vicente, V, Moore, GF (2004). "Comparison of bio-physical marine products from SeaWiFS, MODIS and a bio-optical model with in situ measurements from Northern European waters". JOURNAL OF OPTICS A-PURE AND APPLIED OPTICS, 6(9), 875-889.
In this paper, we compare bio-physical marine products from SeaWiFS, MODIS and a novel bio-optical absorption model with in situ measurements of chlorophyll-a (Chla) and total suspended material (TSM) concentrations, normalized water-leaving radiances (nL(w)) and absorption coefficients of coloured dissolved organic matter (a(CDOM)), total particulate (a(total)) and phytoplankton (a(phy)) for 26 satellite match-ups in three Northern European seas. Cruises were undertaken in 2002 and 2003 in phytoplankton dominated open ocean waters of the Celtic Sea and optically complex waters of the Western English Channel (WEC) and North Sea. For all environments, Chla concentrations varied from 0.4 to 7.8 mg m(-3), TSM from 0.2 to 6.0 mg l(-1) and a(CDOM) at 440 nm from 0.02 to 0.30 m(-1). SeaWiFS OC4v4, with the Remote Sensing Data Analysis Service (RSDAS) atmospheric correction for turbid waters, showed the most accurate retrieval of in situ Chla (RMS = 0.24; n = 26), followed by MODIS chlor_a_3 (RMS = 0.40; n = 26). This suggested that improving the atmospheric correction over optically complex waters results in more accurate Chla concentrations compared to those obtained using more complicated Chla algorithms. We found that the SeaWiFS OC4v4 and the MODIS chlor_a_2 switching band ratio algorithms, which mainly use longer wavebands than 443 nm, were less affected by CDOM. They were both more accurate than chlor-MODIS in the higher CDOM waters of the North Sea. Compared to MODIS the absorption model was better at retrieving a(total) (RMS = 0.39; n = 78) and a(CDOM) (RMS 0.79; n = 12) in all study areas and TSM in the WEC (RMS = 0.04; n 10) but it underestimated Chla concentrations (RMS = 0.45; n = 26). The results are discussed in terms of atmospheric correction, sensor characteristics and the functioning and performance of Chla algorithms.