665 
OCEAN COLOR REMOTE SENSING 
USING POLARIZATION PROPERTIES OF REFLECTED SUNLIGHT 
R. FROUIN 1 , E. POULIQUEN 1 , F.-M. BREON 2 
U Scripps Institution of Oceanography, 
La Jolla, California 92093-0221, USA 
2 : Laboratoire de Modélisation du Climat et de l’Environnement 
CEA/DSM, 91191 Gif sur Yvette, France 
ABSTRACT: 
The effects of the atmosphere and surface on sunlight backscatterred to space by the ocean may be 
substantially reduced by using the unpolarized component of reflectance instead of total 
reflectance. At 450 nm, a wavelength of interest in ocean color remote sensing, and for typical 
conditions, 45% of the unpolarized reflectance may originate from the water body instead of 20% 
of the total reflectance, which represents a gain of a factor 2.2 in useful signal for water 
composition retrieval. The best viewing geometries are adjacent to the glitter region; they 
correspond to scattering angles around 100°, but they may change slightly depending on the 
polarization characteristics of the aerosols. As aerosol optical thickness increases, the atmosphere 
becomes less efficient at polarizing sunlight, and the enhancement of the water body contribution 
to unpolarized reflectance is reduced. Since the perturbing effects are smaller on unpolarized 
reflectance, at least for some viewing geometries, they may be more easily corrected, leading to a 
more accurate water-leaving signal and, therefore, more accurate estimates of phytoplankton 
pigment concentration. 
KEY WORDS: Polarization, Ocean Color, Atmospheric Effects, Aerosols 
1- INTRODUCTION 
A major problem in ocean color remote sensing from space is the correction of top-of-atmosphere 
reflectance, the signal measured, for atmospheric and surface interference. At the wavelengths of 
interest, located in the blue-green region of the solar spectrum, as much as 90% of the photons 
reflected and/or backscattered to space may not have interacted with the water body and, thus, do 
not contain information on water composition. Since the perturbing effects are large, extraction of 
the useful signal requires accurate computation of those effects, which must be known to within a 
few percent to yield reasonably accurate estimates of phytoplankton pigment concentration. 
Uncertainties in radiometric calibration, aerosol characteristics, and sunglint and foam effects 
make the task arduous and, in some cases, impossible. 
To reduce the influence of the atmosphere and surface on the top-of-atmosphere signal, one 
may exploit the polarization properties of reflected sunlight. Because molecules, aerosols, and 
hydrosols polarize sunlight differentially, there may be viewing geometries for which the 
contribution of the water body to the polarized or unpolarized components of total reflectance may 
be enhanced. Exploiting those properties would obviously require measurements of total 
reflectance and polarization rate. These measurements will be made by (or may be envisioned on) 
future sensors, such as the Polarization and Directionality of the Earth Reflectance (POLDER) 
imager (Deschamps et al., 1994). If the enhancement is substantial, correction of the perturbing 
effects becomes easier since accuracy constraints may not be as severe, leading to a more accurate 
water-leaving signal (the signal of interest). 
Since the signal backscattered by the water body is polarized weakly after transmission accross 
the air-sea interface and through the atmosphere (see section 3), we focus our study on unpolarized 
reflectance. Using a Monte Carlo code we simulate the relative contribution of the water body to 
top-of-atmosphere unpolarized reflectance for varied conditions. We compare this contribution 
with that of the water body to total reflectance, identifying the viewing geometries that minimize 
atmospheric and surface effects. We quantify, for those geometries, the increase in the ratio of 
useful to total signal when using unpolarized reflectance instead of total reflectance.