The classic atmospheric correction algorithm, routinely applied to second-generation ocean-color sensors such as
SeaWiFS, MODIS, and MERIS, consists of (i) estimating the aerosol reflectance in the red and near infrared (NIR) where
the ocean is considered black (i.e., totally absorbing), and (ii) extrapolating the estimated aerosol reflectance to shorter
wavelengths. The marine reflectance is then retrieved by subtraction. Variants and improvements have been made over the
years to deal with non-null reflectance in the red and near infrared, a general situation in estuaries and the coastal zone, but
the solutions proposed so far still suffer some limitations, due to uncertainties in marine reflectance modeling in the near
infrared or difficulty to extrapolate the aerosol signal to the blue when using observations in the shortwave infrared (SWIR),
a spectral range far from the ocean-color wavelengths. To estimate the marine signal (i.e., the product of marine reflectance
and atmospheric transmittance) in the near infrared, the proposed approach is to decompose the aerosol reflectance in the
near infrared to shortwave infrared into principal components. Since aerosol scattering is smooth spectrally, a few
components are generally sufficient to represent the perturbing signal, i.e., the aerosol reflectance in the near infrared can
be determined from measurements in the shortwave infrared where the ocean is black. This gives access to the marine
signal in the near infrared, which can then be used in the classic atmospheric correction algorithm. The methodology is
evaluated theoretically from simulations of the top-of-atmosphere reflectance for a wide range of geophysical conditions
and angular geometries and applied to actual MODIS imagery acquired over the Gulf of Mexico. The number of discarded
pixels is reduced by over 80% using the PC modeling to determine the marine signal in the near infrared prior to applying
the classic atmospheric correction algorithm.
|