Short-term sea surface temperature (SST) forecasts are crucial for operational oceanology. This study introduces a specialized Transformer model (U-Transformer) to forecast global short-term SST variability and compares its performance with Convolutional Long Short-Term Memory (ConvLSTM) and Residual Neural Network (ResNet) models. The U-Transformer model forecast consistently outperformed the ConvLSTM and ResNet models, especially in regions with active mesoscale eddies. Globally, the U-Transformer model achieved SST root mean square errors (RMSEs) ranging from 0.2 °C at a 1-day lead time to 0.54 °C at a 10-day lead time during 2020–2022, with anomaly correlation coefficients (ACCs) decreasing from 0.97 to 0.79, respectively. However, in regions characterized by active mesoscale eddies, RMSEs from the U-Transformer model exceeded the global averages by at least 40%, with values in the Gulf Stream region reaching more than twice the global average. Additionally, ACC values in active mesoscale eddy regions declined more sharply with forecast lead time compared to the global averages, decreasing from approximately 0.96 at a 1-day lead time to 0.73 at a 10-day lead time. Specifically, the ACC value dropped to 0.89 in the Gulf Stream region at a 3-day lead time, while maintaining 0.92 globally. These findings underscore the importance of advanced approaches to enhance SST forecast accuracy in challenging active mesoscale eddy regions.

Marine heatwaves (MHWs) are becoming increasingly frequent and intense around China, impacting marine ecosystems and coastal communities. Accurate forecasting of MHWs is crucial for their management and mitigation. In this study, we assess the forecasting ability of the global eddy-resolving ocean forecast system LICOM Forecast System (LFS) for the MHW events in October 2021 around China. Our results show that the 1-day lead forecast by the LFS accounts for up to 79% of the observed MHWs, with the highest skill during the initial and decay periods. The forecasted duration and intensity of the MHW event are consistent with observations but with some deviations in specific regions of the Yellow and South China seas. A detailed analysis of the heat budget reveals that the forecasted shortwave radiation flux is a key factor in the accuracy of the forecasted MHW duration and intensity. The oceanic dynamic term also greatly contributes to the accuracy in the southern Yellow Sea. In addition, the increasing bias of the forecasted duration and intensity with lead time are mainly caused by the underestimated shortwave radiation. Our findings suggest that improving the accuracy of oceanic dynamic processes and surface radiation fluxes in the LFS could be a promising direction to enhance the forecasting ability of marine extreme events such as MHWs.

This study investigates the wave-induced vertical mixing mechanism and systematically compares the application of two non-breaking wave parameterization schemes (Bv and Pw) in oceanic numerical simulations of the South China Sea, according to two key physical variables: sea surface temperature (SST) and the vertical mixing coefficient. The goal is to explore the effects of different parameterization methods on the upper-ocean temperature distribution in the South China Sea. The results indicate that although both schemes enhance vertical mixing in the upper ocean, they do so through different mechanisms. The Bv scheme directly increases the vertical mixing coefficient, demonstrating significantly stronger mixing intensity, while the Pw scheme impacts mixing indirectly by modulating turbulent kinetic energy generation, resulting in comparatively weaker mixing. SST simulation results show that the Bv scheme is more effective in reducing SST in both winter and summer, with broader spatial improvements. Further analysis of the mixing coefficient confirms that, compared to the Pw scheme, the Bv scheme not only strengthens surface mixing but also penetrates deeper into the water column.