MODELLING THE REMOTE SENSING REFLECTANCE OF HIGHLY TURBID WATERS

Abstract

In ocean colour remote sensing, the subsurface remote-sensing reflectance rrs of optically deep waters can be linked to its absorption (a) and backscattering coefficients (bb) by various semi-empirical models. In particular, rrs is expressed as a function of a parameter u which is related to the absorption and backscattering coefficients. By running HydroLight and Monte Carlo simulations in waters with high scattering coefficient values, it was found that the existing, commonly used models underestimated rrs at high u. A polynomial of at least fourth order was required to sufficiently describe the relationship across all values of u. Monte Carlo simulations and a water tank experiment were used to show that the rate of increase in rrs is due to increased multiple scattering by suspended particles in highly turbid waters. A water tank experiment was also conducted with finely powdered calcium carbonate as the suspended sediment. The results from the experiment likewise validated the quartic relation. A general model relating rrs and u was derived for waters of any turbidity, excluding correction for trans-spectral effects. Application of this derived relationship will avoid significant errors in waters of high turbidity.