High-resolution modeling and machine learning

Many of the subgrid-scale physics in weather and climate models are key sources of error for the representation of clouds, precipitation and circulating systems in the atmosphere. While for some of the subgrid-scale physics can in principale be resolved, but only over small domains and short time periods, for others the underlying equations are unknown. 

As data-driven weather forecasting has proven to be good at different lead times, we are also exploring how machine learning can improve weather forecasts from hours to weeks and months.

Finally, the HiRAD-Gen project will develop innovative generative machine-learning models to downscale coarse-resolution atmospheric data to high-resolution outputs. HiRAD-Gen (High-Resolution Atmospheric Downscaling Using Generative Machine Learning Models) funded by external page PASC will start on 1. Jan. 2025

Projects

1. external page Swiss AI, under the auspices of the ETH AI Centre, aims to develop an open foundation climate system model with fine-tuning capabilities down to a few kilometres over the greater Alpine region. external page Fanny Lehmann will join the initiative as a postdoc and external page Levi Lingsch as a PhD student, both in collaboration with SAM.
   
2. In a joint project with the external page Swiss Data Science Center (SDSC), we explore ways to build physics-regulated machine learning subgrid-scale parameterizations that leverage existing accurate but computational expensive parameterizations. Our target model is ICON, which will be at the heart of the knew exascale platform (EXCLAIM) envisioned by the Center for Climate Systems Modelling (C2SM) and its partners via an OpenETH grant. In this project ICON external page has been coupled to an ML parameterization developed in Python.
3. In a collaboration with the ETH AI Center, we explore ways how explainability method of trained neural networks can help to explain the physics underlying regional climate patterns. Interested in an AI Ph.D. or postdoc fellowship? Please get in contact with us (before applying).
   

Publications

• The preprint of the first project related publication "Revisiting Machine Learning Approaches for Short- and Longwave Radiation Inference in Weather and Climate Models, Part II: Online Performance" is now under review and available for download from external page ESS Open Archive.
  • This is the first time that external page ICON has not only been coupled to an ML parameterization developed in Python, but both run in tandem on GPUs.           
    
• The preprint of the first project related publication "Revisiting Machine Learning Approaches for Short- and Longwave Radiation Inference in Weather and Climate Models, Part I: Offline Performance" is now under review and available for download from external page ESS Open Archive.

Conferences and other acitivies

• The team presented their work at the external page 2022 EMS Annual Meeting, Bonn, Germany (04/09/2022 – 09/09/2022). [external page SessionOSA1.9]
  
• Guillaume Bertoli presented his work on building a physics-constrained radiation scheme at the  external page Centre Henri Lebesgue workshop in external page Moulin Mer, Logonna-Daoulas, Finistère, France
  
• In HS2020 and again HS2021, a team of student participating in the Data Science Lab explore whether ML method are able to predict cyclone tracks. 
  
• Guillaume Bertoli started to explore how ML algorithms can emulate the radiation code in ICON. Guillaume graduated with a Ph.D. in external page numerical analysis at the departement of mathematics at the University of Geneva and previously worked on strang splitting and time stepping methods [external page Superconvergence of the Strang splitting when using the Crank-Nicolson scheme][external page Strang splitting method for semilinear parabolic problems].
• Kenza Amara started here Ph.D. at the AI center on explainbility.