Characterizing the 3-D Structure of Particle Beam Attenuation Coefficient in the Northern South China Sea

The particle beam attenuation coefficient (<italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">c_p</i>, m<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">-1</sup>) is a core parameter in bio-optical research, and its three-dimensional variability has attracted increasing attention in the South China Sea (SCS). However, satellite remote sensing algorithms for retrieving local <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">c_p</i> at 660 nm (<italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">c_p</i>660) profiles remain underdeveloped. In this study, based on field observations, we identified three representative <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">c_p</i>660 vertical structures in this area: uniform, Gauss-like, and exponential decay types. We found the ratio of water depth to mixed layer depth can serve as a reliable indicator for distinguishing these profile types. Based on the above finding, we further developed satellite-based inversion algorithms to derive three-dimensional <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">c_p</i>660 distributions in the upper SCS, achieving accuracies within 35%. These include a Bayesian optimization algorithm for the Gauss-like type, an empirical constant algorithm for the exponential decay type, and a machine learning algorithm applicable to both. Unlike the Bayesian and empirical constant algorithms, which primarily rely on sea surface information, the machine learning approach—integrating temperature and salinity profiles—could offer improved accuracy and robustness. The algorithms we proposed in this research would provide valuable tools for advancing the understanding of vertical distributions of phytoplankton biomass, functional groups, and primary productivity in the SCS.