Assessing Landsat-8 atmospheric correction schemes in low to moderate turbidity waters from a global perspective.

Atmospheric correction is one of the major challenges in ocean color remote sensing, thus threatening comprehensive evaluation of water quality within aquatic environments. In this study, five state-of-the-art atmospheric correction (AC) processors (i.e. Acolite, C2RCC, iCOR, L2gen, and Polymer) were applied to Operational Land Imager (OLI) Landsat-8 scenes and evaluated against in situ measurements across various types of waters worldwide. A total of 262 matchups between in situ measured and satellite-derived remote sensing reflectance (R (rs)) at 20 sites were obtained between August 2013 and August 2021. Classification of optical water types (OWTs) was carried out using in situ measurements with matched satellite observations. OWT-specific analysis demonstrated that L2gen produced the most accurate R (rs) with R (2) >= 0.74 and root mean squared error (RMSE) <= 0.0018 sr(-1) for the four visible bands of OLI, followed by Polymer, C2RCC, iCOR, and Acolite. In terms of R (rs) spectral similarity, C2RCC yielded the lowest spectral angle (SA) of 8.55 degrees, followed by L2gen (SA = 9.20 degrees). The advantage and disadvantage of each AC scheme were discussed. Recommendations to improve the accuracy for atmospheric correction were made, such as polarization observations and concurrent aerosol and ocean color measurements.