HybridOM: Hybrid Physics-Based and Data-Driven Global Ocean Modeling with Efficient Regional Downscaling

Global ocean modeling is vital for climate science but struggles to balance computational efficiency with accuracy. Traditional numerical solvers are accurate but computationally expensive, while pure deep learning approaches, though fast, often lack physical consistency and long-term stability. To address this, we introduce HybridOM, a framework integrating a lightweight, differentiable numerical solver as a skeleton to enforce physical laws, with a neural network as the flesh to correct subgrid-scale dynamics. To enable efficient high-resolution modeling, we further introduce a physics-informed regional downscaling mechanism based on flux gating. This design achieves the inference efficiency of AI-based methods while preserving the accuracy and robustness of physical models. Extensive experiments on the GLORYS12V1 and OceanBench dataset validate HybridOM's performance in two distinct regimes: long-term subseasonal-to-seasonal simulation and short-term operational forecasting coupled with the FuXi-2.0 weather model. Results demonstrate that HybridOM achieves state-of-the-art accuracy while maintaining physical consistency, offering a robust solution for next-generation ocean digital twins.

Accurate ocean forecasts are important for understanding climate, marine ecosystems, and extreme events, but today’s ocean models face a difficult trade-off: physics-based simulators are reliable but expensive to run, while AI models can be fast but may drift away from physically realistic ocean behavior. We developed HybridOM, a new ocean modeling system that combines both ideas. It keeps a simplified physics engine to preserve large-scale ocean motion, while using machine learning to correct the parts that are too complex or costly to simulate directly. The same framework can simulate the global ocean, forecast future ocean states, and refine coarse global predictions into higher-resolution regional details. Our experiments show that HybridOM improves accuracy while better preserving important physical properties of the ocean, such as currents, heat content, and particle transport. It is also much faster than traditional numerical ocean solvers and can be adapted to higher-resolution settings more efficiently. This work points toward ocean forecasting systems that are both scientifically trustworthy and practical to run at large scale.