AlgoEarth & CESOC Workshop:
Advancing AI Climate Models
Prof. Dr. Martin Schultz (Research Center Jülich and University of Cologne)
Dr. Florentine Weber (Research Center Jülich and University of Cologne)
4th -5th March 2026, Cologne
Venue: University of Cologne, Pohligstraße 3, lecture hall 4.001
Registration for in-person participation is closed now. If you are interested in attending the workshop remotely, please contact: florentine.weber
uni-koeln.de
Machine learning is increasingly used in weather and climate science. While AI-based weather forecasts can now rival or outperform traditional Numerical Weather Prediction at short time scales, extending these methods to climate - especially under changing conditions - remains a major challenge. Observational data are limited, and the climate system involves complex interactions and feedbacks that are not easily captured. This workshop asks: how can we push AI beyond forecasting and emulation to help understand and model the climate of tomorrow?
The workshop within the framework of Algorithmic Earth System Science and CESOC at the University of Cologne builds on the successful conference series on Machine Learning for Earth System Sciences (MLESS) but focuses specifically on discussing ways to tackle the fundamental problem of AI climate modeling. Even though remarkable progress has been made in AI-based weather forecasting and climate model emulators, fully operational AI climate simulations remain extremely difficult - largely due to sparse observational records and complex interactions across atmosphere, ocean, land, cryosphere, and biosphere. To achieve breakthroughs in this field, it is necessary to re-think how current deep learning networks are built and how architecture choices may limit the applicability to climate feedbacks.
This workshop aims to bring together world-leading experts to explore the potential and limitations of AI climate modelling and raise awareness for this fascinating yet highly challenging field. We would be thrilled to have you onboard.
Our goal is to connect expertise across algorithm design, machine learning, and Earth system science, fostering interdisciplinary thinking around next-generation AI methods designed for emerging global observation platforms and their use in weather and climate prediction.
Invited Speakers
We are excited to welcome international experts from:
NVIDIA, ECMWF, CSIC (Spain), JSC/FZJ, DKRZ, KIT, Hereon, CMCC (Italia), Chalmers University of Technology
Topics will include:
- Non-autoregressive climate emulation
- The DestinE Climate Emulator
- ICON and cloud microphysics
- Generative ML for climate and weather
- Customised AI Technology
- Data-driven global ocean models
- And more…!
Panel Discussions
- Pure versus hybrid models: what are the most promising approaches? What are the limitations of current approaches? Do you see potential for completely new modelling concepts/architectures?
- How can we ensure the consistency and robustness of AI climate models? How can we turn climate emulators into full-fledged climate models? How can we do sciemnce with AI climate models and make new discoveries?
If you see yourself as a pioneer shaping the next generation of climate modelling - someone who thinks beyond boundaries and wants to move the field forward together with others - then we warmly invite you to register.
Registration