Principled Operator Learning in Ocean Dynamics: The Role of Temporal Structure
This addresses the problem of unstable and physically inconsistent ocean predictions for weather and climate forecasting applications, representing an incremental improvement over existing neural operator methods.
The paper tackles the challenge of long-term prediction stability and physical fidelity in neural operators for ocean forecasting by incorporating temporal Fourier modes into the Fourier Neural Operator (FNO). The resulting FNOtD model substantially improves stability and consistency with physical dynamics, achieving competitive predictive skill with a state-of-the-art numerical ocean model at significantly lower computational cost.
Neural operators are becoming the default tools to learn solutions to governing partial differential equations (PDEs) in weather and ocean forecasting applications. Despite early promising achievements, significant challenges remain, including long-term prediction stability and adherence to physical laws, particularly for high-frequency processes. In this paper, we take a step toward addressing these challenges in high-resolution ocean prediction by incorporating temporal Fourier modes, demonstrating how this modification enhances physical fidelity. This study compares the standard Fourier Neural Operator (FNO) with its variant, FNOtD, which has been modified to internalize the dispersion relation while learning the solution operator for ocean PDEs. The results demonstrate that entangling space and time in the training of integral kernels enables the model to capture multiscale wave propagation and effectively learn ocean dynamics. FNOtD substantially improves long-term prediction stability and consistency with underlying physical dynamics in challenging high-frequency settings compared to the standard FNO. It also provides competitive predictive skill relative to a state-of-the-art numerical ocean model, while requiring significantly lower computational cost.