14:30–14:45 | ABSTRACT 1165
SPECTRALLY SELF-CONSISTENT GASEOUS ABSORPTION CORRECTION FOR PACE OCI
Accurate correction of gaseous absorption is critical for hyperspectral ocean color remote sensing, where subtle spectral features carry information on phytoplankton composition, fluorescence, and optical properties. Conventional approaches rely on ancillary ozone, water vapor, and nitrogen dioxide products combined with radiative transfer calculations to estimate band-specific transmittance. However, effective gaseous absorption depends not only on absorber amount but also on scattering-enhanced photon path length, surface-atmosphere coupling, and viewing geometry, introducing scene-dependent variability that is difficult to constrain externally. We present an ancillary-independent approach that exploits hyperspectral self-consistency to recover gas-corrected top-of-atmosphere reflectance directly from observed spectra. Across the contiguous 340-890 nm spectral range of the PACE Ocean Color Instrument, the gas-free reflectance can be accurately reconstructed as a linear combination of the full hyperspectral reflectance vector. This result follows from the low intrinsic dimensionality of ocean-atmosphere reflectance, which lies on a smooth radiative manifold controlled by a limited number of physical degrees of freedom. Because both gas-affected and gas-free spectra depend on the same latent state variables, a global linear operator provides an accurate mapping between them. The operator is derived from a large ensemble of radiative transfer simulations spanning realistic variations in geometry, aerosols, gases, and surface conditions, and is applied directly in instrument space without ancillary inputs. Validation experiments demonstrate high reconstruction accuracy and improved spectral consistency relative to traditional methods, even in the presence of measurement noise. This approach offers a robust pathway toward spectrally cleaner hyperspectral retrievals for next-generation ocean color missions.
Robert Frouin, Scripps Institution of Oceanography, [email protected]
Jing Tan, Scripps Institution of Oceanography, [email protected]
