Suspended sediment concentration (SSC) is an important indicator of sediment dynamics, material transport, and ecological change in nearshore seas. Satellite observations have revealed a significant decline in surface SSC in the Bohai Sea over the past two decades. However, previous studies have mainly relied on single-factor correlations or qualitative analyses, making it difficult to distinguish the relative contributions of wind, tidal currents, and riverine sediment discharge under coupled conditions. Traditional hydrodynamic–sediment models are computationally demanding and sensitive to boundary conditions and parameterization, whereas remote-sensing-based studies commonly use correlation analysis or linear regression, which cannot adequately capture nonlinear interactions among multiple drivers. To address these limitations, this study developed an interpretable data-driven framework integrating a convolutional neural network (CNN) with SHAP. The CNN was used to extract nonlinear relationships from multi-source environmental data, while SHAP provided model interpretation and quantitative attribution. MODIS-derived SSC was used as the target variable, with observation-corrected ERA5 winds, TPXO tidal currents, and Yellow River sediment discharge as inputs. The model achieved a test-set correlation coefficient of 0.86 and successfully reproduced SSC temporal variability, including responses to spring–neap tidal transitions, wind changes, and extreme sediment discharge events. SHAP analysis identified tidal currents as the dominant control on Bohai Sea SSC, contributing 55% of the mean absolute attribution and ranking first across 98% of the effective sea area. Wind and Yellow River sediment discharge contributed 31% and 14%, respectively. Relative to the spatiotemporal mean SSC, their mean absolute impacts were 31%, 17%, and 8%, respectively. Clear spatial heterogeneity was also observed. The positive tidal-current contribution peaked near the southwestern coast of Bohai Bay at approximately 49%, whereas the negative contribution reached about 50% from the central and northern Bohai Sea to Liaodong Bay. Wind showed its highest positive contribution in northern Liaodong Bay and the Bohai Strait at about 38%, while its negative contribution reached 45% in the central-western and southern nearshore regions. The positive contribution of Yellow River sediment discharge peaked north of Laizhou Bay at about 37%, whereas its negative contribution reached approximately 35% in the eastern Bohai Strait. This framework enables accurate SSC simulation while quantifying the relative importance and spatial differences of multiple drivers, providing a practical approach for identifying SSC controls and supporting sediment-process research and ecological management in the Bohai Sea and Yellow River estuary.

暂无