Simultaneous Determination of Water Constituent Concentrations and Partial Least Squares

This paper deals with a physically based bio-optical model for the imultaneous determination of concentrations of water constituents chlorophyll-a (CHL), total suspended matter (TSM), and coloured, dissolved organic matter (CDOM). The model is based on the dependency of absorption and backscatter coefficients on the constituent concentrations. A fundamental remote sensing equation is rewritten to a regression model which is used to estimate the concentrations of CHL and TSM. A weighted regression analysis is performed and the weights are determined as the weights chosen in a partial least squares (PLS) or canonical covariance analysis of in situ measurements of CHL, TSM and CDOM with in situ spectra in the 400-730 nm region. Resampled weights are used with geometrically corrected and calibrated airborne optical casi data to produce maps of jointly estimated CHL and TSM contents. Also, considerations to give a better understanding of the PLS technique than offered by the NIPALS algorithm are given and canonical correlations analysis is briefly described.