Regional Ocean Forecasting with Hierarchical Graph Neural Networks
This work addresses the need for efficient ocean forecasting for environmental management and climate adaptation, though it is incremental as it builds on existing machine learning advancements in weather forecasting.
The paper tackles ocean forecasting by introducing SeaCast, a neural network for high-resolution, medium-range predictions, validated on Mediterranean Sea data with numerical and data-driven forcings, achieving results comparable to traditional models but with reduced computational cost.
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.