Estimation of chromophoric dissolved organic matter using above-surface hyperspectral remote sensing

摘要

A method for the inversion of hyperspectral remote sensing was developed to determine the absorption coefficient for chromophoric dissolved organic matter (CDOM) in the Mississippi and Atchafalaya river plume regions and the northern Gulf of Mexico, where water types vary from Case 1 to turbid Case 2. Above-surface hyperspectral remote sensing data were measured by a ship-mounted spectroradiometer and then used to estimate CDOM. Simultaneously, water absorption and attenuation coefficients, CDOM and chlorophyll fluorescence, turbidities, and other related water properties were also measured at very high resolution (0.5-2 m) using in situ, underwater, and flow-through (shipboard, pumped) optical sensors. We separate ag, the absorption coefficient a of CDOM, from adg (a of CDOM and nonalgal particles) based on two absorptionbackscattering relationships. The first is between ad (a of nonalgal particles) and bbp (total particulate backscattering coefficient), and the second is between a_p (a of total particles) and bbp. These two relationships are referred as a_d-based and a_p-based methods, respectively. Consequently, based on Lee's quasi-analytical algorithm (QAA), we developed the so-called Extended Quasi-Analytical Algorithm (QAA-E) to decompose a_(dg), using both a_d-based and a_p-based methods. The absorption-backscattering relationships and the QAA-E were tested using synthetic and in situ data from the International Ocean-Colour Coordinating Group (IOCCG) as well as our own field data. The results indicate the a_d-based method is relatively better than the a_p-based method. The accuracy of CDOM estimation is significantly improved by separating a_g from a_(dg) (R~2 = 0.81 and 0.65 for synthetic and in situ data, respectively). The sensitivities of the newly introduced coefficients were also analyzed to ensure QAA-E is robust.