Retrieval of Particle Scattering Coefficients and Concentrations by Genetic Algorithms in Stratified Lake Water

We retrieved the mass-specific scattering coefficient b(sm)(lambda) = 0.60.(lambda/650)(-1.82) of the inhomogeneous and optically complex water column of eastern Lake Constance in May 2012. In-situ measured and modelled remote-sensing reflectance R-rs(lambda) were matched via a parameter search procedure using genetic algorithms. The optical modelling consisted of solving the azimuthally-averaged Radiative Transfer Equation, forced with in-situ suspended matter concentration (sm) data. b(sm)(lambda) was univocally determined at red wavelengths. In contrast, we encountered unresolved spectral ambiguity at blue wavelengths due to the absence of organic absorption in our dataset. Despite this, a surprisingly good sm retrieval regression is achieved (R-2 > 0.95 with respect to independent data) using our b(sm)(lambda). Acquisition of accurate inherent optical properties in future field campaigns is needed to verify the estimated b(sm)(lambda) and related assumptions.