Evaluation of Empirical and Semianalytical Spectral Reflectance Models for Surface Suspended Sediment Concentration in the Highly Variable Estuarine and Coastal Waters of East China

Development and validation of the surface suspended sediment concentration (SSC) models derived from the surface remote-sensing reflectance spectra [R_rs(λ)] are important in satellite monitoring of estuarine and coastal waters. Seven empirical and seven semianalytical spectral reflectance models for evaluation of the surface SSC were compared with one another and with laboratory tank (one dataset) and in situ measurements (two datasets) performed in different natural waters of East China. All models were presented in the form of ${R-{{{\rm rs}}}}$ spectral ratios, in which wavelengths were selected from the list of NASA's satellite sensor, MODIS unsaturated central wavelengths. A statistical analysis has been performed to find the best models and spectral ratios for remote-sensing monitoring purposes. Analysis has shown that empirical models are generally superior to the semianalytical models for solution existence, prediction accuracy, and correlation with the observed SSC values. However, all semianalytical models using the red to green spectral ratio have demonstrated approximately the same accuracy and correlation as empirical models, what provides an additional support for using more simple easily calculated empirical models. Additionally, relationships between SSC and inherent optical properties (IOPs) (absorption and backscattering coefficients) and between IOPs and R_rs(λ) provided by the semianalytical models have their own benefits for aquatic optics and remote sensing purposes.