Abstract
In oceanographic research, reconstructing the three-dimensional (3D) distribution of temperature and salinity is essential for understanding global climate dynamics, predicting marine environmental changes, and evaluating their impacts on ecosystems. While previous studies have largely concentrated on the effects of various modeling approaches on reconstructing oceanic variables, limited attention has been paid to the role of surface waves in reconstruction. This study, based on sea surface data, employs a deep learning-based neural network model, U-Net, to reconstruct 3D temperature and salinity across the North Pacific and Equatorial Pacific within the upper 200 m. The input of wave information includes the significant wave height (SWH), Langmuir number (La), and Langmuir enhancement factor (ε); the latter two indicate the strength of Langmuir turbulence, which promotes vertical mixing in the ocean surface layer and thereby affects profiles of temperature and salinity. The results indicate that incorporating wave information, particularly the La and ε, significantly enhances the model’s ability to reconstruct ocean temperature and salinity. This highlights the critical role of surface waves in enhancing the reconstruction of 3D ocean temperature and salinity.