Simultaneous emulation and downscaling with physically-consistent deep learning-based regional ocean emulators

2025-01-09

AI Summary

To address the challenge of high-fidelity, long-term ocean simulation in complex regions such as the Gulf of Mexico, this paper proposes the first physics-consistent deep learning framework for joint regional ocean simulation and downscaling. Methodologically, it integrates physics-constrained generative modeling, autoregressive spatiotemporal modeling, and multi-scale feature alignment, employing a PDE-informed loss function to suppress hallucination and long-term drift. The framework enables stable, decade-scale autoregressive integration of sea surface variables at 8 km resolution while simultaneously downscaling to 4 km and performing bias correction. Experiments demonstrate substantial improvements in both short-term forecasting and long-term statistical fidelity (e.g., mean and variability), complete elimination of unphysical drift, and unprecedented synergy among high spatial resolution, multi-decadal stability, and dynamical interpretability—establishing a new paradigm for regional ocean prediction.

Technology Category

• Machine Learning: Deep Generative Models & Autoencoders
• Computer Vision: Low Level & Physics-based Vision
• Search and Optimization: Sampling/Simulation-based Search

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies

Abstract

Building on top of the success in AI-based atmospheric emulation, we propose an AI-based ocean emulation and downscaling framework focusing on the high-resolution regional ocean over Gulf of Mexico. Regional ocean emulation presents unique challenges owing to the complex bathymetry and lateral boundary conditions as well as from fundamental biases in deep learning-based frameworks, such as instability and hallucinations. In this paper, we develop a deep learning-based framework to autoregressively integrate ocean-surface variables over the Gulf of Mexico at $8$ Km spatial resolution without unphysical drifts over decadal time scales and simulataneously downscale and bias-correct it to $4$ Km resolution using a physics-constrained generative model. The framework shows both short-term skills as well as accurate long-term statistics in terms of mean and variability.

Problem

• Ocean Simulation
• Complex Marine Areas
• AI Simulation Stability

Innovation

• Advanced Deep Learning
• High-Resolution Ocean Modeling
• Gulf of Mexico Simulation