Remote-sensing inversion model of surface water suspended sediment concentration based on in situ measured spectrum in Hangzhou Bay, China

Suspended sediment concentration (SSC) is an important parameter for monitoring coastal water quality. Suspended particles are also the main optically active substances for ocean color remote sensing. It is important to study the surface reflectance spectra features of coastal turbid water, as it can be the basis for establishing more accurate remote-sensing inversion models. In this study, Hangzhou Bay, China, was selected as the study area. Two in situ measurement and sampling stations in the estuary of the Qiantang River which flows into Hangzhou Bay were set up separately. Above-water spectrum observation method, which the NASA recommended, was adopted to measure the reflection spectrum of turbid waters. Surface water samples were simultaneously collected to obtain the corresponding SSC data. The results showed that the total suspended particle concentrations in the Hangzhou Bay were typically high, and the inorganic suspended particle concentrations were far greater than the phytoplankton concentrations, which averages 705 mg/L and 1.16 mg/m(3). The SSC at two sampling stations both showed significant temporal variability, particularly appearing short-period rapid fluctuations accompanying the tidal cycle. The measured surface water reflectance spectra all showed typical curve characteristics of high turbid water, and as the SSC increased, the corresponding reflectivity of surface water also increased. The increments at different wavelengths were variational, with two reflectance peaks appearing at 650-700-nm and near the 800-nm wavelength of spectral curves, respectively. The first derivative of spectral curves showed that the first reflectance peak location appeared to be a red shift phenomenon with the SSC increasing. The correlation coefficients between the SSC of surface water and the remote-sensing reflectance according to moderate resolution imaging spectra-radiometer (MODIS) channels' central wavelength were different significantly, which were larger at MODIS long-wavelength channels (> 650 nm) and smaller at MODIS short-wavelength channels (400-550 nm). The value of determination coefficient R-2 was 0.82 when the reflectance ratio of MODIS band 2 to band 1 was selected as the SSC sensitive bands combination and exponential regression analysis was employed. Therefore, the reflectance ratio of MODIS band 2 to band 1 can be adopted as the main band combination for establishing surface water SSC remote-sensing inversion model in the Hangzhou Bay.