Improving low-quality satellite remote sensing reflectance at blue bands over coastal and inland waters

The satellite remote sensing reflectance (Rrs(lambda)) at two short blue bands (410 or 412 nm and 443 nm) are prone to large uncertainties in coastal and inland waters, prohibiting algorithms from generating reliable ocean color products associated with these bands. In this study, we developed an algorithm to estimate Rrs(41x) and Rrs (443) when the satellite Rrs(lambda) in blue bands suffer from large uncertainties. The algorithm first determines the Rrs(lambda) spectral shape from the satellite-measured Rrs(lambda) values at three wavelengths of 48x (486, 488, or 490), 55x (547, 551, or 555), and 67x (667, 670, or 671) nm. The algorithm then derives Rrs(41x) and Rrs(443) from the estimated Rrs(lambda) spectral shape with algebraic formulations. We assessed the algorithm performance with satellite (SeaWiFS, MODISA, and VIIRS-SNPP) and in situ Rrs(lambda) matchups from global waters. It is shown that the uncertainties of estimated Rrs(41x) and Rrs(443) are substantially smaller than the original satellite products when applicable. Besides, implementation of the algorithm contributes to a significant increase in the number of utilizable Rrs(41x) and Rrs(443) values. The algorithm is relatively stable and is best applicable to the satellite Rrs(lambda) spectra for which the Rrs(48x) and Rrs(55x) measurements are subject to small uncertainties. The demonstrations support the application of the blue-band estimation algorithm to a wide range of coastal waters.