Validation of semi-analytical inversion models for inherent optical properties from ocean color in coastal Yellow Sea and East China Sea

Abstract

The performances of three semi-analytical retrieval models for water inherent optical properties were validated in the coastal Yellow Sea and East China Sea, including the Quasi-Analytical Algorithm (QAA), the Garver-Siegel-Maritorena model (GSM) and the Over Constrained Linear Matrix (LM). The model-retrieved parameters, namely absorption coefficients of phytoplankton (a ph), colored dissolved and detrital particulate matter (a dg), total absorption coefficients (a t), and backscattering coefficient of particles (b bp), were compared. The bio-optical datasets collected from a Yellow Sea and East China Sea cruise in April and September 2003 were used in the study. The QAA model performed the best in retrieval for all the coefficients, showing log-transformed root mean square errors of 0.306 for a ph, 0.268 for a dg, 0.144 for a t, and 0.273 for b bp at 443 nm. The LM model showed a slightly larger deviation than the QAA model with a similar error trend for absorption coefficients, but it returned the largest uncertainties for b bp, with log-transformed root mean square error up to 0.646. The GSM model, however, yielded the largest and fluctuating errors along with wavelength for absorption coefficient retrievals. Substituting the fitting parameters from measured data for the empirical spectral parameters, all three models returned better results. These improvements demonstrated that semi-analytical algorithms designed for ocean water need regional modifications before applying to coastal areas. The QAA algorithm may be the most suitable model for retrieval for the Yellow Sea and East China Sea, and future model refinements should concentrate on regional modeling of inherent optical properties.