Semi-analytical algorithm for estimating particulate absorption coefficient in inland water based on OLCI images

The absorption coefficient of particulate matter (a_p) is a crucial inherent optical property of water, significantly influencing the underwater light environment. In this study, a semi-analytical model incorporating three spectral bands from Sentinel-3 OLCI imagery was proposed for retrieving a_p(674) in turbid inland waters. It eliminates irrelevant factors in the bio-optical model by inputting remote sensing reflectance at 674, 709, and 779 nm, and uses R_rs(674)/[R_rs(560)+R_rs(709)] as the optical indicator for estimating the coefficient 1-1/k, where k represents the ratio between a_p(674) and a_p(709). The estimation accuracy of the algorithm was extensively assessed by using datasets from 13 lakes in China and six lakes in Belgium. The results demonstrated that the proposed algorithm has excellent estimation performance in turbid water. Compared to the QAA-750E algorithm, the median absolute percentage error (MdAPD) decreased from 50.17% to 17.68%, the unbiased mean absolute percentage error (UMAPD) dropped from 44.09% to 21.82%, and the root mean square error (RMSD) improved from 0.61 m^-1 to 0.24 m^-1. The algorithm was successfully applied to map a_p(674) of lakes in the Eastern Plains from 2016 to 2023 using OLCI images, and the spatial and temporal trends in a_p(674) were briefly analyzed. This developed algorithm demonstrates strong potential for long-term a_p(674) monitoring in inland waters worldwide.