Abstract

The ocean color spectrum is defined as the ratio of the spectrum of light upwelled from the sea to the spectrum of light incident on the sea surface. Ocean color spectra, observed from an airplane flown over waters off Oregon, are analyzed. The original spectra are resolved into fifty-five wavelength bands, each 5 nm wide. The shapes of these spectra are parametrized by, and shown to be accurately recoverable from, their first four principal components Y(i)(i = 1, 2, 3, 4). These Y(i) are the scalar projections of each spectrum on the first four characteristic vectors of the sample covariance matrix. Empirically determined characteristic vectors like these, but derived from a much larger and more globally representative sample of color spectra, may someday provide an efficient, standard basis for parameterizing ocean color spectra. Regression equations are found with which phytoplankton pigment concentration and water transparency may be estimated as linear functions of the parameters Y(i). Pigment concentration estimates thus obtained are imprecise. The poor fit is at least partly due to the inappropriateness of the linear regression model and the neglect of other optically important substances typically present in sea water.

Full Text
Paper version not known

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call