Retrieval of the Fluorescence and Chlorophyll Concentration from Reflectance Spectra Using Polarization Discrimination and Improved Baseline Approach

In previous work* we have shown that it is possible to separate the elastic scattering from the chlorophyll fluorescence signal in water leaving radiance using a polarization discrimination technique which makes use of the fact that the elastically scattered component is always at least partially polarized, while the fluorescence signal is unpolarized. The technique's validity was confirmed by excellent agreement between the spectral shape of fluorescence extracted in this manner and laser induced fluorescence of the same species. The approach works well even if different concentrations of clay are added to the algae samples and for rough water conditions if temporal (or spatial) averaging is included. The technique has also been successfully tested during in-situ field measurements along eastern Long Island. While generally robust, the efficacy of the fluorescence retrieval depends on the degree of polarization of the water leaving radiance and the near linearity of the relationship between the two orthogonal polarization components of elastic scattered light over a limited spectral range of interest in the vicinity of the fluorescence spectra. To evaluate the scope and limitations of the technique, we compare polarization effects and the efficacy of the fluorescence retrieval by simulating the total and polarized components of water leaving radiances originating from elastic scattering and fluorescence in case 1 and case 2 waters using radiative transfer. This is achieved by superimposing the contribution of known fluorescence spectra onto calculated elastic reflectance for case 1 and 2 waters. Then by using our polarization discrimination procedure we invert the total resulting spectral data back to obtain fluorescence spectra, and compare it for accuracy with the added fluorescence spectra. It was found that a comprehensive radiative transfer program gives excellent results for both case 1 and 2 waters, permitting accurate predictions of degrees of polarization and extraction of fluorescence. The simulations showed that the range of efficacy of the technique covered both case 1 and 2 waters generally, but was limited in its applicability where bottom albedo represents a significant component of unpolarized water leaving radiance. Since the above analyses showed that the traditional height over baseline fluorescence extraction method strongly overestimates fluorescence value for case 2 waters, we examined and evaluated an approach which does not require polarization discrimination, but does make use of the known Chl fluorescence spectra. We believe it to be a significant improvement for use with case 2 waters. Basically the new approach matches inversion algorithms outside of, but in the immediate vicinity of the Ch fluorescence band, with the total reflected spectra and extrapolates these to the spectral region under the fluorescence. It is then shown that peak of the difference between these and the total observed spectrum (including the fluorescence) should match the fluorescence peak. Operating with these constraints, it is possible to use automated curve fitting techniques to determine both Chl concentration and fluorescence height with much better accuracies than are attainable with the traditional height over baseline method. This approach is also expected to be useful for optimizing determination and use of satellite observation bands. * S. Ahmed et al. Separation of Fluorescence and Elastic Scattering from Algae in Seawater Using Polarization Discrimination. Opt. Comm., 235, 23-30, 2004.