Simultaneous Emulation and Downscaling With Physically Consistent Deep Learning‐Based Regional Ocean Emulators

• Published in: Journal of geophysical research. Machine learning and computation Vol. 2; no. 3
• Main Authors: Lupin‐Jimenez, Leonard, Darman, Moein, Hazarika, Subhashis, Wu, Tianning, Gray, Michael, He, Ruyoing, Wong, Anthony, Chattopadhyay, Ashesh
• Format: Journal Article
• Language: English
• Published: 01.09.2025
• DOI: 10.1029/2025JH000851
• ISSN: 2993-5210

Building upon recent advancements in AI‐driven atmospheric emulation, we present a novel framework for AI‐based ocean emulation, downscaling, and bias correction, with a specific focus on high‐resolution modeling of the regional ocean in the Gulf of Mexico. Emulating regional ocean dynamics poses distinct challenges due to intricate bathymetry, complex lateral boundary conditions, and inherent limitations of deep learning models, including instability and the potential for hallucinations. In this study, we introduce a deep learning framework that autoregressively integrates ocean surface variables at 8 km spatial resolution over the Gulf of Mexico, maintaining physical consistency over decadal time scales. Simultaneously, the framework downscales and bias‐corrects the outputs to 4 km resolution using a physics‐informed generative model. Our approach demonstrates short‐term predictive skill comparable to high‐resolution physics‐based simulations, while also accurately capturing long‐term statistical properties, including temporal mean and variability. Recent advances in artificial intelligence (AI) have shown that deep learning models can effectively reproduce complex atmospheric behavior. In this study, we apply similar techniques to simulate ocean dynamics in the Gulf of Mexico, a region that is especially difficult to model due to its complex seafloor topography (bathymetry) and the influence of surrounding land and open ocean boundaries. We developed an AI‐based method that predicts how ocean surface conditions change over time at high spatial resolution (8 km) while also correcting for biases and enhancing resolution to 4 km using physics‐informed techniques. Unlike many AI models that can become unstable or produce unrealistic outputs over time, our approach remains physically consistent even over decades. It achieves short‐term accuracy comparable to traditional high‐resolution simulations and also captures long‐term patterns such as average conditions and variability, offering a more efficient way to study regional ocean systems. An AI‐based physically consistent long‐term regional emulator has been developed for the Gulf of Mexico region A deterministic and stochastic downscaling model has been developed to super‐resolve and bias‐correct the low‐resolution predictions to high resolution The performance of the framework has been compared with AI‐based and high‐resolution physics‐based baselines on short‐ and long‐term metrics