Numerical modelling of transspectral processes in natural waters: implications for remote sensing
Based on a previously developed and thoroughly validated hydrooptical model, numerical simulations of the spectral composition of water leaving radiance are presented. These simulations take into account absorption, elastic scattering, water Raman (inelastic) scattering as well as the fluorescence of chlorophyll (chl) and dissolved organics (doc). The results obtained for forward modelling were also used for the inverse problem: retrieval of water quality parameters from water volume reflectance (R) spectra. The Levenberg-Marquardt multivariate optimization procedure was used for this purpose. Unlike water Raman scattering, the chl and doc fluorescence has an impact on R, so the retrieval results can change substantially for waters rich in chl or doc. Suspended minerals (sm) suppress both the chl and doc fluorescence influence on R. The retrieval results indicate that chl can be accurately assessed if the concentration of sm is not low and the doc concentration is < 2 mgCl-1. For waters devoid of doc, the concentration of chl can be accurately retrieved even if the sm concentration is very low. Retrieval errors prove to be strongly dependent on the fluorescence yield value of both chl and doc.