Home > TERATEC FORUM > Workshops

TERATEC 2025 Forum
The European meeting for Experts in High Power Digital
Simulation . HPC/HPDA . Artificial Intelligence . Quantum Computing

Wednesday May 21
Workshop 01 - 4:15 pm to 6:15 pm

HPC and AI serving the Climate: Challenges and Opportunities
Chaired by Hatem Ltaief, Principal Research Scientist, KAUST and Florent Ventimiglia, Head of Computing, Software Engineering and Data Valorization Department, CNES

Using AI approaches for climate science
By David Salas, Head of the Large-Scale Meteorology and Climate Group, CNRM

One of the main thrusts of climate and climate change research involves running numerous numerical simulations with climate models, typically thousands of years in the future, in order to estimate uncertainties. This has prompted the climate research community (as well as the weather forecasting community) to evaluate model emulation approaches using AI, with convincing initial results.

However, these approaches can have limitations, such as a lack of consistency in the set of simulated variables, even if it is possible to introduce physical constraints. This observation has motivated hybrid modeling approaches, which involve replacing certain parts of a climate model with AI algorithms. These include the computationally-intensive parts, or those representing the effect on climate of phenomena not explicitly resolved by the model (parameterizations).

These approaches are producing promising results, but at the cost of painstaking work to ensure the stability of hybrid models, which must be capable of simulating climate over thousands of years. In addition, AI methods can also be mobilized for equation discovery approaches to guide model development, or to deepen analyses of simulation or observation data on climate, its variability and extremes.

Biography: David Salas y Mélia obtained a PhD in ocean and atmospheric physics from the University of Toulouse in 2000. Until 2024, his research focused on understanding and estimating past and future climate trends through the use of models. In particular, he developed a model of Antarctic and Arctic sea ice integrated into climate models and Météo-France's global weather forecasting model. From 2016 to 2024, he was director of the department at the Centre National de Recherches Météorologiques (CNRM, Toulouse) in charge of Météo-France's climate and air quality research. He is now a researcher on the CNRM team in charge of developing a weather forecasting model using Artificial Intelligence.

Register now and get your badge here >>>


 

© Ter@tec - All rights reserved - Lawful mention