Physics-driven approaches in concert with AI-based methods to predict extreme space weather events
AIxtreme-I is focused on space weather problems. The integration of physical knowledge and machine learning techniques will improve the accuracy with which numerical models simulate the propagation processes in space weather contexts, will allow accurate and computational efficient parameter identification processes which will lead to improved models and prediction in solar and heliophysics.
Mechanistic approaches in concert with AI-based methods to predict extreme weather events
AIxtreme-II will be focused on weather problems. Thanks to the combined exploitation of deterministic weather prediction models and efficient data-driven AI-based algorithms, weather forecasts with unprecedented accuracy in relation to key meteorological observables such as wind, temperature and cumulated precipitation, will become available on the Mediterranean Sea.
Figure: A deterministic NWP forecast (from our WRF model running operatively at DICCA-UNIGE) for the 10-𝑚 wind speed at the Italian SYNOP station of Lampedusa. The raw forecast has been converted to a probabilistic forecast using the QRF. The 50% and 90% prediction intervals are shown as overlain grey ribbons, while the dashed black line is the median. Red bullets are the materialized observations.