A groundbreaking artificial intelligence system is transforming ocean forecasting in the Mediterranean, generating detailed 15-day predictions in just seconds while significantly reducing computational costs.
Developed by researchers at the Centro Euro-Mediterraneo sui Cambiamenti Climatici (CMCC) in collaboration with the University of Helsinki, the new high-resolution forecasting system, known as SeaCast, leverages AI to deliver faster and more energy-efficient forecasts than traditional physics-based models. The research describing the system has been published in Scientific Reports.
A New Benchmark in Regional Ocean Forecasting
SeaCast stands out from existing global AI forecasting models by integrating both oceanic and atmospheric variables, capturing the complex regional dynamics that define the Mediterranean basin. While most AI ocean models operate at relatively low resolutions and rely primarily on ocean data, SeaCast runs at approximately 4-kilometer resolution (1/24°), matching the resolution of CMCC’s operational Mediterranean forecasting system, MedFS.
The system produces forecasts down to a depth of 200 meters and was trained using CMCC’s Mediterranean reanalysis datasets, which are publicly available through the Copernicus Marine Service.
A key technical innovation is its graph-based neural network, designed to handle intricate coastlines and lateral boundary conditions long-standing challenges in regional ocean modeling.
Faster, More Efficient, More Accurate
SeaCast not only extends the forecasting window from 10 to 15 days but also consistently outperforms the current Copernicus operational model over the standard 10-day horizon.
The gains in efficiency are striking. While the conventional numerical forecasting system requires roughly 70 minutes on 89 central processing units (CPUs) to produce a 10-day forecast, SeaCast can generate a 15-day forecast in about 20 seconds using a single graphics processing unit (GPU).
Researchers say this dramatic reduction in processing time enables rapid “what-if” scenario testing and probabilistic ensemble forecasting tools that are critical for assessing uncertainty and supporting evidence-based decision-making.
Practical Benefits Across the Mediterranean
Accurate ocean forecasts are essential for sectors including shipping, aquaculture, environmental monitoring and coastal risk management. By providing rapid, high-resolution predictions, SeaCast could enhance early warning capabilities and improve planning and risk mitigation strategies across Mediterranean coastal communities.
“This achievement demonstrates how bringing together oceanography, atmospheric science, and AI expertise produces tangible results, and can unlock a new generation of regional ocean forecasts,” said Emanuela Clementi, CMCC researcher and co-author of the study. She added that combining physical knowledge with advanced AI methods significantly improves forecast accuracy while dramatically lowering computational costs.
Daniel Holmberg, first author of the study and a researcher at the University of Helsinki, highlighted the collaborative nature of the project, noting that close cooperation between teams in Bologna and Lecce played a key role in the system’s development.
Why Atmosphere Ocean Coupling Matters
One of SeaCast’s most significant innovations is its integration of atmospheric forcing data during model training and forecasting. Results show that incorporating atmospheric information substantially improves surface-level forecast accuracy, where ocean conditions are strongly influenced by winds, temperature and pressure systems.
Sensitivity experiments further revealed which atmospheric variables contribute most to enhanced marine predictions. The study also found that extending training periods to include up to 35 years of historical data significantly improves model performance.
CMCC researchers are now working to integrate SeaCast into operational forecasting chains alongside traditional physics-based systems. The team says the model establishes a new benchmark for regional, high-resolution AI ocean forecasting and could serve as a blueprint for similar systems worldwide.
As climate variability intensifies and coastal risks grow, faster and smarter forecasting tools such as SeaCast may become indispensable for safeguarding Mediterranean communities and supporting sustainable marine management.