Regional Ocean Forecasting with Hierarchical Graph Neural Networks

TL;DR

This paper introduces SeaCast, a graph neural network model for high-resolution regional ocean forecasting that is faster and more efficient than traditional numerical methods, validated on Mediterranean Sea data.

Contribution

The paper presents a novel graph neural network architecture, SeaCast, tailored for regional ocean forecasting, integrating external forcing data and handling complex ocean geometries.

Findings

SeaCast achieves high accuracy in Mediterranean Sea forecasts.

The model outperforms traditional numerical solvers in speed and efficiency.

Validation shows strong agreement with operational models.

Abstract

Accurate ocean forecasting systems are vital for understanding marine dynamics, which play a crucial role in environmental management and climate adaptation strategies. Traditional numerical solvers, while effective, are computationally expensive and time-consuming. Recent advancements in machine learning have revolutionized weather forecasting, offering fast and energy-efficient alternatives. Building on these advancements, we introduce SeaCast, a neural network designed for high-resolution, medium-range ocean forecasting. SeaCast employs a graph-based framework to effectively handle the complex geometry of ocean grids and integrates external forcing data tailored to the regional ocean context. Our approach is validated through experiments at a high spatial resolution using the operational numerical model of the Mediterranean Sea provided by the Copernicus Marine Service, along with both numerical and data-driven atmospheric forcings.