ALGORITHMS FOR REMOTE-SENSING OF HIGH-CONCENTRATION, INORGANIC SUSPENDED SEDIMENT

A numerical simulation of a two-flow reflectance equation was used to fit aircraft radiance data taken from a high turbidity macrotidal region. The coefficients of the model were fitted on a trial-and-error basis but with reference to known physical conditions. Both a concentration and a wavelength dependence of the optical coefficients were necessary in order to produce the final scattering to absorption ratios. The simulation study also indicated that the maximum upwelling wavelength could be a pertinent parameter in helping to identify water masses. The simulated remote sensing algorithms worked equally well with both the Landsat MSS data and with Nimbus-7 CZCS data, the latter taken from two regions, the Bay of Fundy and the Beaufort Sea. The comparative analysis between image ground truthing and model predictions encountered three major limitations: (I) uncertainty in the atmospheric conditions and hence correction routines, (2) the influence of highly absorbing dissolved material associated with river runoff, and (3) the exact spectral shape for broad wavebands. Within these limitations the functional form of the remote sensing algorithms were similar to those suggested by Holyer, but also incorporated ratios of wavebands. Finally, the numerical simulations were used to predict in situ sediment algorithms for current and future multispectral sensors onboard Landsat-4, and-5 (MSS and Thematic Mapper), SPOT and MOSS satellite systems. © 1990 Taylor and Francis Ltd.