Retrieval of spectral backscattering from spectral scattering based on spectral partitioning technique

Abstract

Direct measurement of backscattering using sensors has technological limitations such as saturation in highly turbid waters and the effect of absorption in strongly absorbing highly productive waters. In the present study, a quantitative method is developed to estimate the particulate spectral backscattering coefficient based on spectral partitioning of the scattering coefficient. The input slope of the non-algal backscattering spectra was determined from the scattering slope and the magnitude was quantified from the water turbidity. The contribution of phytoplankton to the backscattering was determined as a linear combination of the spectral dependency factors and the chlorophyll concentration. The proposed method of conversion, from scattering to backscattering, when compared with two existing methods (Haltrin and Petzold average ratio), resulted in lower relative errors (mean relative error: -0.214–0.037, mean net Bias: -0.259–0.019, root mean square error: 0.006–0.015, coefficient of determination: 0.91–0.98), and substantial improvement was also observed in the spectral shape and magnitude of the backscattering coefficients.