Improved Skills for Predicting Sub-Seasonal to Seasonal Global Sea Surface Temperature.

Accurate prediction of global sea surface temperature at sub-seasonal to seasonal (S2S) timescales is critical for drought and flood forecasting, as well as for improving disaster preparedness in human society. Operational physics-based numerical models are run to predict S2S sea surface temperature and corresponding climate indices, such as El Niño-Southern Oscillation. However, these models are hampered by computational inefficiencies, rapidly dissipating ocean-atmosphere initial conditions, and significant uncertainty and biases. Here, we introduce a novel spatiotemporal deep learning neural network to model the nonlinear and complex coupled atmosphere-ocean weather systems. This model, named as CAS-Canglong, incorporates climatic and temporal features and employs a self-attention mechanism to enhance the prediction of global S2S sea surface temperature pattern. Compared to the physics-based models, it shows higher computational efficiency and more predictive capability, improving one to three months sea surface temperature predictive skill by 55.6 ± 14.1% to 63.4 ± 15.5% in six ocean regions with dominant influence on S2S variability over land. This achievement underscores the huge potential of deep learning for largely improving forecasting skills at the S2S timescales over land.