satellite derived 3D cp660 in the upper 125m water layer of the northern South China Sea

Using field observation data, we developed and evaluated satellite-based inversion approaches for estimating the three-dimensional distribution of cp660 in the upper layer of the SCS. Three methods were proposed: an empirical constant approach for cp660 profiles following an exponential decay structure; a Bayesian optimization algorithm for Gauss-like profiles; and a machine learning model applicable to both profile types. Evaluation results demonstrated that all three methods effectively captured the spatiotemporal variability of the cp660 vertical structure in this region. Using the above algorithms, we generated monthly three-dimensional distribution of cp660 for the northern SCS spanning 2003–2023.