Search papers, labs, and topics across Lattice.
The authors develop Neptune, an end-to-end deep learning framework designed to emulate global ocean and sea-ice dynamics for subseasonal-to-seasonal (S2S) forecasting up to 60 days at resolutions down to 0.25掳. Addressing the extreme computational cost and software complexity of traditional numerical Ocean General Circulation Models, the architecture couples Convolutional Neural Networks with Spherical Fourier Neural Operators (SFNOs) to resolve both local boundary features and multiscale spherical interactions. Conditioned on daily atmospheric forcing, Neptune maintains physical stability over long rollouts while accurately forecasting 3D ocean state variables, sea ice concentration, and key coupled climate modes like ENSO and the Indian Ocean Dipole.
Neural operators can now stably emulate full 3D global ocean and sea-ice dynamics out to 60 days at 0.25掳 resolution, breaking the compute bottleneck of numerical ocean circulation models.
Subseasonal-to-seasonal (S2S) forecasting is societally critical, supporting decision-making in sectors ranging from water and agricultural management to disaster risk reduction, energy planning, and insurance. Achieving reliable predictions at these timescales requires representing the ocean and its dynamics, but traditional physics-based Ocean General Circulation Models (OGCMs), are computationally expensive and difficult to develop and improve because of the code complexity. In this work, we propose Neptune, an end-to-end data-driven framework for global ocean and sea-ice components emulation tailored for S2S timescales, up to 60 days. Neptune combines Convolutional Neural Networks (CNNs) and Spherical Fourier Neural Operators (SFNOs) to effectively capture local features and global cross-scale interactions, thereby obtaining a coherent representation of the ocean state. Forced by prescribed daily atmospheric fields, Neptune emulates ocean state variables, from temperature and salinity, to zonal and meridional currents, from sea surface height to sea ice thickness and concentration, with daily outputs at the ocean surface and through the water column. Specifically, we propose two variants of Neptune, Neptune-1 and Neptune-025, capable of emulating the ocean state at 1掳 and 0.25掳 resolution, respectively. Evaluated against a suite of metrics, including statistics (RMSE, CRPS and ACC), physical coherency (Ocean Heat Content, Eddy Kinetic Energy and Ice Brier Score) and climate indices (ENSO and Z20 metric, IOD), Neptune successfully reproduces the spatio-temporal evolution of the oceanic fields up to 60 days, and is stable over long timescales. Neptune provides compelling evidence that end-to-end data-driven ocean emulators can become a powerful component of next-generation S2S forecasting systems, emulating ocean state at high spatio-temporal resolution.
