Some insights of spectral optimization in ocean color inversion

Over the past few decades, various algorithms have been developed for the retrieval of water constituents from the measurement of ocean color radiometry, and one of those approaches is spectral optimization. This approach defines an error function (or cost function) between the observed spectral remote sensing reflectance and an estimated spectral remote sensing reflectance over the range of observed wavelengths, with the latter modeled using a few variables that represent the optically active properties (such as the absorption coefficient of phytoplankton and the backscattering coefficient of particles). The values of the variables when the error function reaches a minimum are the optimized properties. The applications of this approach implicitly assume that there is only one global minimum condition, and that any local minimum (if exist) can be avoided through the numerical optimization scheme. Here, with data from numerical simulations, we show the shape of the error surface as a mechanism to visualize the solution space for the model variables. Further, using two established models as examples, we demonstrate how the solution space changes under different model assumptions as well as the impacts on the quality of the retrieved water properties. © 2011 SPIE.