Transferable and Principled Neural Operators for 3D Ocean Forecasting in Regional Seas: A Baltic Sea Case

Objective

Accurate and timely ocean forecasting is essential for enhancing climate resilience, ensuring maritime safety, and supporting sustainable marine resource management. Operational ocean forecasting systems largely rely on physics-based numerical models; however, their computational demands, reliance on spatial and temporal discretization, and dependence on parameterized subgrid-scale processes limit their scalability, adaptability, and capacity to produce large, rapid ensemble forecasts. As climate change increases the frequency and intensity of extreme ocean events, the need for fast and accurate forecasting becomes even more critical to support early warning systems and informed decision-making. To address these challenges, this project will develop a transferable and principled neural operator model for 3D ocean forecasting. Neural operators are designed to learn the solution operators of partial differential equations (PDEs), enabling mappings between infinite-dimensional function spaces. Once trained, they enable rapid, high-resolution ensemble forecasting while reducing dependence on costly computational infrastructure. This project brings together expertise in machine learning and operational ocean forecasting, enabling a mutually beneficial exchange of knowledge between the researcher and the host institution. At the University of Copenhagen, the fellow will leverage extensive datasets, computational infrastructure, and expert guidance to develop this data-driven model, rigorously evaluate its predictive skill, and benchmark its performance against existing operational ocean forecasting systems. Dissemination activities will target scientific communities where immediate impact is anticipated, including those focused on climate adaptation, natural hazard mitigation, and sustainable ocean management. Together with the host’s capabilities and track record, this ambitious project is well-positioned for success, supporting the fellow’s career development.