NeuralOGCM: Differentiable Ocean Modeling with Learnable Physics

2025-12-12

AI Summary

High-fidelity scientific simulation has long faced a fundamental trade-off between computational efficiency and physical fidelity. To address this, we propose the first differentiable ocean circulation model, centered on a differentiable dynamical solver wherein key physical parameters—such as diffusion coefficients—are treated as learnable variables. We further integrate neural networks to correct subgrid-scale processes and discretization errors. Our approach achieves end-to-end learnability of the core physical equations—the first such integration in ocean modeling of implicit differentiable ODE solvers (via Torchdiffeq) with physics-informed neural networks (PINNs). The model rigorously enforces physical consistency and energy conservation, ensuring long-term stability over multi-year integrations. Compared to conventional numerical methods, it achieves several-fold speedup; relative to purely data-driven baselines, it significantly improves predictive accuracy while preserving interpretability and physical plausibility.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Machine Learning: Deep Neural Architectures and Foundation Models

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

High-precision scientific simulation faces a long-standing trade-off between computational efficiency and physical fidelity. To address this challenge, we propose NeuralOGCM, an ocean modeling framework that fuses differentiable programming with deep learning. At the core of NeuralOGCM is a fully differentiable dynamical solver, which leverages physics knowledge as its core inductive bias. The learnable physics integration captures large-scale, deterministic physical evolution, and transforms key physical parameters (e.g., diffusion coefficients) into learnable parameters, enabling the model to autonomously optimize its physical core via end-to-end training. Concurrently, a deep neural network learns to correct for subgrid-scale processes and discretization errors not captured by the physics model. Both components work in synergy, with their outputs integrated by a unified ODE solver. Experiments demonstrate that NeuralOGCM maintains long-term stability and physical consistency, significantly outperforming traditional numerical models in speed and pure AI baselines in accuracy. Our work paves a new path for building fast, stable, and physically-plausible models for scientific computing.

Problem

• Balancing computational efficiency with physical fidelity in ocean modeling
• Integrating differentiable programming and deep learning for improved simulations
• Optimizing physical parameters and correcting subgrid-scale errors autonomously

Innovation

• Differentiable ocean model with learnable physics parameters
• Deep neural network corrects subgrid-scale errors
• Unified ODE solver integrates physics and AI components