Simultaneous emulation and downscaling with physically-consistent deep learning-based regional ocean emulators
This work addresses the problem of high-resolution ocean modeling for climate and environmental applications, representing an incremental advance in AI-based emulation methods.
The paper tackled the challenge of regional ocean emulation and downscaling for the Gulf of Mexico, achieving stable autoregressive integration at 8 km resolution and downscaling to 4 km without unphysical drifts over decadal timescales.
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.