Topic: Machine Learning
Thursday, 23 April 2026 from 7-9h UTC (*NOTE - NEW DATE*)
Organizers and Conveners: Michael Goodliff (RIKEN) and James Taylor (RIKEN)
Advances in machine learning are rapidly transforming the way we model, predict, and understand complex geophysical systems. At the same time, data assimilation remains a cornerstone of numerical prediction, providing a rigorous framework for combining observations with dynamical models. This event will explore the growing synergy between machine learning and data assimilation, highlighting new methods that enhance forecast skill, improve uncertainty quantification, and enable efficient use of increasingly diverse observational data. Through invited talks and discussions, we aim to foster cross-disciplinary exchange and identify emerging directions for next-generation prediction systems.
Detailed program with abstracts
Yopad link with asked questions
Presentations:
- AI-DOP: A machine-learned path from Earth System observations to a global weather forecast
Mihai Alexe, Eulalie Boucher, Peter Lean, Ewan Pinnington, Simon Lang, Patrick Laloyaux, Tomas Kral, Tobias Necker, Tony McNally
- Efficient ensemble forecasting with generative flow models and data assimilation
Tobias Finn, Marc Bocquet, Freddy Bouchet, Etienne Mémin
- Forward and inverse modelling of parametric dynamical system in a physics-aware latent space
Qiyao Zhou, Xujia Zhu, Pierre Joli, Yu Cong, Sibo Cheng
- Unifying background-error covariance modelling for midlatitude and tropical atmospheric data assimilation
Bostjan Melinc, Uros Perkan, Lukas Kugler, Ziga Zaplotnik
- High-Resolution Ensemble Uncertainty Quantification Using Diffusion-Based Machine Learning
Swapan Mallick, Kasper Tølløse, Per Dahlgren, Jelena Bojarova, Xiaohua Yang, Eivind Støylen, Harald Schyberg
Time Zones:
07 – 09 UTC
Europe: 08 – 10 am BST (London) | 09 – 11 am CEST (Berlin)
Asia/Australia: 03 – 05 pm CST (Shanghai) | 04 – 06 pm JST (Tokyo) | 05 – 07 pm AEDT (Sydney)
Americas: 00 – 02 am PDT (San Fran.) | 01 – 03 am MDT (Denver) | 03 – 05 am EDT (New York)
How to attend to ISDA-Online events?
Simply subscribe to the ISDA-Online mailing list
