OceanMoE improves long-term ocean variable forecasting with adaptive models
OceanMoE: Structured Conditional Sparse Computation for Long-Horizon Multivariate Ocean Forecasting
Machine Learning
Summary
Forecasting multiple ocean features like temperature and currents together is tricky because they change differently across locations and variables. The paper presents OceanMoE, a model that balances shared ocean information with specialized predictions by selecting the right small groups of experts for each place and variable. This approach helps make better long-term ocean predictions by adapting to local and variable differences while still learning from common ocean patterns. Tests showed OceanMoE reduced forecast errors compared to traditional methods and changed how experts were chosen based on what was being predicted and where.
What this means in practice
- •For meteorological agencies: Improve long-range ocean forecasts by using OceanMoE’s adaptive modeling to better capture variable and local differences.
- •For climate modelers: Incorporate OceanMoE’s structured expert routing to enhance multivariate ocean predictions within large-scale Earth system simulations.
Authors
Yishun Zhu, Jian Wang
Abstract
Multivariate ocean forecasting must exploit shared evolution in a coupled ocean system while adapting to the heterogeneous statistical and dynamical characteristics of different prediction variables and locations. Fully shared models may lack the flexibility to handle this heterogeneity, whereas fully independent models discard the common ocean context shared across variables. The key question is how to retain shared context in a unified model while allowing computation to specialize according to the prediction target and local state. We propose OceanMoE, a structured conditional sparse Mixture-of-Experts framework that combines sharing and specialization for multivariate ocean forecasting. OceanMoE fuses cross-variable information to construct target-specific local representations and uses them to perform content-conditioned sparse routing at each spatial location, with the number of active experts adapted to router confidence. In the decoder, routing is augmented with a learned geographic bias parameterized by spherical-harmonic spatial bases, while shared residual and seasonal pathways provide common cross-variable and month-dependent context. Experiments on long-horizon autoregressive ORAS5 forecasting show that OceanMoE lowers aggregate forecasting error in both evaluated settings and maintains lower geometric-mean normalized RMSE than the corresponding baselines over most later rollout months. Routing analyses further show that expert allocation varies with prediction targets and spatial locations. These results support structured conditional computation as a modeling strategy for balancing shared ocean context with adaptive specialization.