Abstract
Ocean color remote sensing is vital for monitoring marine ecosystems, but accurate retrieval of ocean color parameters remains challenging in regions affected by strongly absorbing aerosols such as dust, biomass burning, and industrial pollution. To address this issue, we leveraged the Ocean Color–XGBoost with Radiative Transfer simulation (OC-XGBRT) atmospheric correction (AC) algorithm to improve the accuracy of remote sensing reflectance ( R r s (λ)) in blue bands (412, 443, and 488 nm) from the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor under the absorbing aerosol conditions. We integrated the OC-XGBRT AC algorithm with traditional bio-optical retrieval models (Ocean Color Index (OCI), KD2M, and Quasi-Analytical Algorithm (QAA)) to estimate chlorophyll-a concentration (CHL), diffuse attenuation coefficient for downwelling irradiance at 490 nm ( K d ( 490 ) ), and absorption coefficient of detritus and gelbstoff at 443 nm ( a d g ( 443 ) ), and further compared the retrieval results with those derived from three widely used AC algorithms (NASA NIR, OC-SMART, and POLYMER). Results of validation against in situ measurements from the SeaWiFS Bio-optical Archive and Storage (SeaBASS), Aerosol Robotic Network-Ocean Color (AERONET-OC), and GLObal Reflectance community dataset for Imaging and optical sensing of Aquatic environments (GLORIA) demonstrated that the OC-XGBRT outperformed three comparative AC algorithms (NASA NIR, OC-SMART, and POLYMER) under the absorbing aerosol conditions. Specifically, the proposed algorithm achieved markedly higher coefficient of determination (R 2 ) values (0.57 for CHL and 0.45 for a d g ( 443 ) ) and lower Mean Absolute Percentage Deviation (MAPD) values (58.4% for CHL and 63.7% for a d g ( 443 ) ), representing a clear improvement over other algorithms under absorbing aerosol conditions. Applications in five typical oceanic regions frequently influenced by absorbing aerosols, including the west coast of North Africa, the Persian Gulf, the west coast of North America, the Black Sea, and the Chinese Bohai-Yellow Sea, showed that the proposed approach substantially improved the spatial coverage and temporal continuity of ocean color products, especially for a d g ( 443 ) , which is most sensitive to the interference of absorbing aerosol. Overall, OC-XGBRT algorithm provides an effective solution for improving ocean color products in complex atmospheric conditions and supports more reliable monitoring of coastal ecosystems.