Abstract
Located in the northern South China Sea (SCS), the Guangdong Sea areas exhibit a highly complex hydrodynamic structure driven by the combined effects of tides, monsoons, and offshore current systems, serving as a core region for China’s marine economy and offshore engineering. Although the Regional Ocean Modeling System (ROMS) is widely applied in current simulations, its accuracy is often constrained by the inadequate adaptability of its parameterization schemes to the regional environment. Furthermore, systematic parameter optimization tailored to this specific domain remains scarce. To address these limitations, this study conducts an observation-driven parameter optimization for surface current simulations in the western Guangdong Sea areas, aiming to enhance the reliability of hydrodynamic simulations and forecasting. A three-dimensional ROMS hydrodynamic model was employed to systematically design 18 physical parameterization experiments. The model’s performance was rigorously evaluated against 26 h continuous in situ current measurements from four observation stations, utilizing statistical metrics including the correlation coefficient (R), root mean square error (RMSE), Taylor diagrams, and the MMS standardized evaluation. The results indicate that the Mellor–Yamada vertical mixing scheme yields the optimal regional adaptability. For horizontal diffusion, the biharmonic scheme outperforms the Laplacian approach. Regarding bottom friction, the logarithmic formulation demonstrates superior accuracy compared to the quadratic and linear schemes, with the latter proven unsuitable for this region. A comprehensive evaluation identifies the ‘MY–Biharmonic–Logarithmic’ combination as the optimal parameterization configuration for the western Guangdong Sea areas. This study establishes an adaptable ROMS parameterization framework for the western Guangdong Sea areas and elucidates the influence mechanisms of key physical parameters on simulation outcomes. These findings not only provide high-precision hydrodynamic support for short-term pollutant dispersion forecasting, and disaster mitigation in this region but also offer valuable methodological references for numerical modeling in the broader SCS and analogous complex coastal environments.