Machine Learning Across ESM: Subgrid-Scale Parameterizations, Emulation, and Hybrid Modeling

[Simon Driscoll]

Dear colleagues,

We are very excited to announce our session “Developments in Machine Learning Across Earth System Modeling: Subgrid-Scale Parameterizations, Emulation, and Hybrid Modeling” to be held at AGU26 in San Francisco this year (7-11 December). 

We are delighted to have already confirmed our invited speakers, Prof Greg Hakim (University of Washington), and Dr Kara D. Lamb (Columbia University), and welcome contributions from all across this research space.

We hope our session is of interest to you and colleagues - please share widely with all potentially interested colleagues and we look forward to seeing everyone in San Francisco! 🙂 If you have any questions please feel free to reach out to us.

Abstract Submission

You can submit an abstract here: https://agu.confex.com/agu/agu26/prelim.cgi/Session/280793 Abstract submissions are due by Wednesday, 5 August at 23:59 EDT/ 03:59 UTC.

Conveners

Simon Driscoll (University of Cambridge), Sara Shamekh (New York University), Ching-Yao Lai (Stanford University), Karan Jakhar (Pravāh).

Further Details

Machine learning is reshaping the representation of complex physical processes in Earth system models, offering new avenues for parameterization, emulation, and hybrid modeling. This session focuses on the use of machine learning to emulate computationally expensive or unresolved processes, accelerate physical simulations, enable data-driven discoveries, and improve representation across domains such as convection, turbulence, radiation, hydrology, sea ice, and other Earth system components.

Topics include (but are not limited to):

- Subgrid-scale parameterizations via machine learning
- Emulators of physical processes, model components, or whole weather and climate models (including end-to-end learning and foundation models)
- Data-driven discoveries
- Hybrid ML-physics modeling frameworks
- Physics-informed neural networks, neural operators, and differentiable programming
- Reinforcement learning and other approaches for ensuring physical consistency, stability, and optimizing model behavior
- Calibration and parameter optimization using ML
- Verification and explainability (XAI) of data-driven models (including AI forecasting) 
- Coupling of ML models with physical models
- Cross-domain applications (atmosphere, ocean, cryosphere, land).

Best,

Simon Driscoll, Sara Shamekh, Ching-Yao Lai, Karan Jakhar

Dr. Simon Driscoll

Assistant Research Professor

Department of Applied Mathematics and Theoretical Physics (ICCS)

University of Cambridge

Office: H1.20 (note new office)

Telephone: +44(0)1223 764064

(Also Queens' College, Silver St, Cambridge CB3 9ET)





Queens’ College | University of Cambridge bridgetothefuture.co.uk   All data is held securely in line with data protection regulations and our full Privacy Policy is available on our website. Charity no. 1137495 | Registered with the Fundraising Regulator