Enhancing ocean color retrieval in sunglint regions: the efficacy of IPNOA over conventional atmospheric correction methods

Sunglint significantly impacts the extraction of ocean color information, particularly for sensors lacking tilt capabilities. Traditional atmospheric correction algorithms often fail to retrieve effective data in high-sunglint regions. The polynomial-based POLYMER method, applied to MERIS data, effectively addresses sunglint, although its accuracy decreases by about 15% in such conditions. To enhance data reliability in sunglint regions, we propose the Improved polynomial nonlinear optimization approach (IPNOA), a revision of the POLYMER atmospheric correction. IPNOA employs the QAA-RGR (quasi-analytical algorithm-red-green-bands-ratio) to refine the bio-optical ocean reflectance model. Additionally, due to the nonlinear optimization algorithm's sensitivity to initial values, this study uses global 8-day average oceanic optical properties at 4 km resolution as the initial setting. The performance of IPNOA was initially evaluated using a synthetic dataset, with retrieved remote sensing reflectance (Rrs) closely matching the simulated Rrs across all wavelengths. The mean absolute percentage error (MAPE) remained below 1% for non-sunglint, moderate sunglint, and high sunglint conditions. Further analysis of in situ data revealed that IPNOA performs better, exceptionally at 412 nm, with a MAPE of 5.27% in sunglint regions. When processed by POLYMER, the dataset exhibited a MAPE of 68.47%. Finally, an analysis of global data from MODIS, VIIRS, and HY1C/D on July 15, 2022, showed good agreement among the three on a global scale. Above all, these results indicate that the IPNOA algorithm has strong potential for retrieving valid products in moderate, even high sunglint regions, offering practical benefits for expanding the spatial coverage of ocean color satellite data.