AI for Weather and Climate: How Machine Learning Boosts Forecasting

Oceanographic predictions: How HPC can help train reliable AI models    
Paraskevi Vourlioti       

Dr. Paraskevi Vourlioti is the Chief Technical Officer at Neuralio A.I., responsible for leading projects in climate-driven land evaluation and renewable energy solutions. She holds a PhD in Meteorology from the University of Cologne, specializing in data assimilation and short-term weather forecasts for energy applications. With a strong background in AI, HPC, and operational weather forecasting, she has contributed to various EU projects, ESA tenders and authored numerous publications in meteorology and renewable energy. For this project, she carried out research in collaboration with AlongRoute Data I.K.E., which specialises in oceanography and marine weather forecasting.

HPC is transforming oceanographic forecasting by enabling the development of reliable AI models. This talk highlights how the Vienna Scientific Cluster (VSC5) helped us train advanced GNN-GRU models to predict Significant Wave Height (SWH) with highly accurate six-hour forecasts. Discover how HPC is accelerating ocean predictions and improving maritime decision-making. The work was carried out following a strategic partnership between two deep-tech startups, namely AlongRoute Data I.K.E. and Neuralio A.I..

AIFS: a data-driven probabilistic weather forecasting system   
Mariana Clare   
      
Mariana Clare is a researcher at the European Centre for Medium Range Weather Forecasts (ECMWF), where she works on building a machine learning model for weather forecasting. She is particularly interested in how to capture the model uncertainty in these data-driven approaches. She recently received a PhD from Imperial College London, focussing on developing advanced numerical and statistical techniques to quantify uncertainty in coastal ocean models. By training she is a mathematician, having done her undergraduate degree in Mathematics at the University of Oxford.   

AI is driving progress in many fields and weather forecasting is no exception. The past few years have seen several data-driven weather forecasting models that are comparable in skill to leading traditional models. Over the past year, ECMWF has been developing its own data-driven forecasting system, the AIFS. This talk will give an overview of AIFS and discuss its performance for a few significant weather events. It will also showcase ongoing research efforts at ECMWF, including data-driven probabilistic ensemble forecasting, and learning to predict directly from observations.