Evaluating Solar Radiation Attenuation Schemes in ROMS: Improving Upper Ocean Thermal Simulation by Incorporating Vertically Resolved Optical Profiles From BGC‐Argo Data

Abstract

Accurate simulation of solar radiation attenuation is critical for modeling the upper ocean's thermal and dynamical state. Traditional schemes, such as PS77 and M02, rely on surface‐only chlorophyll‐based proxies and assume vertical homogeneity, often failing to capture subsurface optical structures. This study evaluates a vertically resolved irradiance attenuation scheme (KVIS) constructed through a machine‐learning‐based data fusion framework that integrates BGC‐Argo vertical profiles with satellite remote sensing observations. Using this framework, three‐dimensional KVIS fields are generated and assessed via 1D mechanism analysis and 3D ROMS simulations across three oceanographic distinct regions. Results demonstrate that KVIS consistently outperforms both PS77 and M02 across all study areas. Relative to PS77, KVIS reduces temperature root‐mean‐square error (RMSE) by 8.6%–19.2% depending on the region, and by 4.3%–19.1% relative to M02. The physical mechanism underlying this improvement is that KVIS, by resolving the true vertical profile of light attenuation, allows more solar energy to penetrate below the immediate surface layer. This deeper heat distribution reduces artificial near‐surface stratification and prevents excessive surface heat trapping, leading to a more realistic mixed layer depth and improved subsurface thermal structure. These findings establish KVIS as a more reliable and physically consistent scheme for upper‐ocean temperature and circulation modeling.