Modeling polarimetric effects on aquatic remote sensing retrievals for Landsat 10

Polarization effects in the coupled atmosphere–surface interaction can introduce systematic radiometric errors in satellite observations, particularly for aquatic remote sensing where the water-leaving signal is small relative to the total top-of-atmosphere (TOA) radiance. The upcoming Landsat 10 mission will deploy the Landsat Instrument Suite (LandIS), which provides expanded spectral coverage in the visible and near-infrared (VNIR) and improved spatial resolution compared to previous Landsat missions, enabling enhanced retrieval of in-water constituents such as chlorophyll-a (CHL), total suspended sediments (TSS), and colored-dissolved organic matter (CDOM). In this study, we evaluate the impact of sensor polarization sensitivity on aquatic constituent retrievals for the LandIS VNIR bands using an end-to-end forward modeling and uncertainty quantification framework. Water-leaving radiance is simulated using the Hydrolight radiative transfer model across a broad range of Case-2 inland and coastal water conditions defined by varying CHL, TSS, and CDOM concentrations. Polarization effects are incorporated through a Fresnel Mueller matrix formalism at the air–water interface and within subsurface scattering. Atmospheric polarization is modeled using the polarized radiative transfer model MODTRAN-P. The resulting polarized TOA radiance is spectrally sampled to the LandIS band configuration and perturbed with instrument noise and polarization sensitivity consistent with mission requirements. Under nominal LandIS SNR and polarization-factor requirements, retrieval RMSE values at solar zenith angle (SZA)=40 deg are 1.15 μg/L for CHL, 1.34 mg/L for TSS, and 0.19 1/m for CDOM. Under high–solar zenith angle conditions of 76 deg, retrieval errors increase to 2.56 μg/L for CHL, 2.65 mg/L for TSS, and 0.71 1/m for CDOM, demonstrating the increased influence of polarization sensitivity under oblique illumination geometries. The expanded LandIS VNIR configuration also improves retrieval performance relative to the heritage Landsat-8/9 OLI band set, reducing RMSE from 3.67 to 1.15 μg/L for CHL, 3.12 to 1.34 mg/L for TSS, and 0.56 to 0.19 1/m for CDOM. The modeling framework developed in this study provides a physics-based approach for quantifying the impact of polarization on aquatic remote sensing retrievals and for linking instrument polarization performance to science product requirements for future satellite missions such as Landsat 10.