Application of Experimental Design Theory to Ocean Remote Sensing With a Hyperspectral Detector

摘要

A method based on a theory of experimental Optimal Design (OD) was developed in Russia to select the best combination of wavelengths so an optically remote spectral sensor may optimally estimate oceanic chlorophyll concentration. It gives the number of spectral bands, their center wavelength, and each band's spectral width to make the best estimate of chlorophyll concentration in a fixed observation time. A hyperspectral sensor is convenient because it eliminates the need for variable optical filters. An ideal sensor would have an infinite number of infinitesimally narrow spectral bands. Optimal designs are given for both ideal and real sensors. The designs were computed using ocean radiance spectra simulated by a Monte Carlo model for a range of chlorophyll concentrations. The computed OD was tested for robustness over a wide range of experimental conditions. The methods of color index and principal component analysis were also applied. The optimal design method gives more accurate chlorophyll estimates.