Climate Modeling
Climate Modeling for the future of the planet
Refining Climate Predictions
The technology behind climate models was first created 50 years ago. Much has changed since then, and there is now an opportunity to make use of the latest advances in supercomputing, modern programming languages, and machine learning to improve climate models and enable more certain projections of local trends of average and extreme temperature and precipitation change in our rapidly warming climate. We're building modern machine learning (ML) into current climate models to improve their performance in key areas and ultimately to refine climate change predictions. Our ML is trained on ultra-realistic ‘digital twin’ simulations of the Earth’s atmosphere that exploit the world’s fastest supercomputers.


Better Climate Models using Finer Grids
In the same way photos have become clearer because screens now pack in more pixels, fine grid ‘global storm resolving models’ (GSRMs) based on grids with less than 5 km (3 miles) horizontal spacing and 50 or more vertical grid levels spanning the depth of the atmosphere) can now provide a detailed and actionable ‘digital twin’ of our world, enabling realistic simulation of airflow around mountain peaks and within thunderstorm systems that generate much of the world’s most intense rainfall.
GSRMs are currently too costly to run for more than a year or two, so they are not yet practical for climate modeling. An earlier AI2 Climate Modeling project team pioneered the use of a domain-specific language (DSL) for making a GSRM run more efficiently on modern high-performance computers using diverse hardware accelerators such as GPUs, while still presenting developers with elegant, understandable, maintainable code. This project was transitioned to NOAA and NASA partners at the end of 2022.
Smarter, More Accurate Simulations with Machine Learning
Present-day GSRM simulations can be leveraged to increase the accuracy of currently affordable global climate models by training machine learning (ML) to replace their less accurate components—something that has never been done in operational climate models before. We partner with a leading climate modeling center, NOAA’s Geophysical Fluid Dynamics Laboratory (GFDL) using their X-SHiELD GSRM, a modified version of the U.S. global weather forecasting model, to design new GSRM simulations and create ML training datasets across multiple climates. Our group is a world leader in this area.


Create Open-Source and Collaborative Solutions
We're developing open-source software so the broader climate modeling community can easily adopt our advances. Our partnership with GFDL ensures our work builds on their valuable experience and has the quickest impact. This collaboration brings our team’s innovation together with GFDL’s climate modeling experience and computing resources to achieve quicker impacts that can set an example for other climate modeling centers.
Recent Updates
Pace: Python/GT4Py version of GFDL's FV3 dycore - published in GMD
May 17, 2023Pace is an Python-based implementation of the nonhydrostatic FV3 dynamical core and GFDL cloud microphysics scheme written using the GT4Py domain…
Kwa et al. 1 year nudge-to-3 km corrective ML results published in JAMES
May 15, 2023Published version of preprint highlighted here last fall. We use GFDL's new year-long X-SHiELD simulation to train corrective ML for coarse 200 km…
Bretherton commentary for GRL on reservoir computing for AGCM emulation
April 30, 2023Geophysical Research Letters invited Chris Bretherton to write a Commentary about a recent paper by Arcomano et al. 2023 applying reservoir computing…
Watt-Meyer et al. 2021 recognized by GRL as most-downloaded
March 30, 2023Geophysical Research Letters informed us this paper was among their most downloaded papers of 2021. As of this posting it has 36 Google Scholar…
GFDL ML article features Spencer Clark and AI2 climate modeling research
March 2, 2023NOAA/GFDL, our collaborating climate modeling center, just featured a short article on innovative GFDL-related efforts to use ML in climate modeling…
Recent Papers
Improving the reliability of ML-corrected climate models with novelty detection.
Clayton Sanford, Anna Kwa, Oliver Watt-Meyer, Spencer K. Clark, Noah D. Brenowitz, Jeremy McGibbon, and Christopher S. BrethertonESSOAr • 2023 The use of machine learning (ML) for the online correction of coarse-resolution atmospheric models has proven effective in reducing biases in near-surface temperature and precipitation rate. However, this often introduces biases in the upper atmosphere and…Pace v0.2: a Python-based performance-portable atmospheric model
Johann Dahm, Eddie Davis, Florian Deconinck, Oliver Elbert, Rhea George, Jeremy McGibbon, Tobias Wicky,Elynn Wu,Christopher Kung, Tal Ben-Nun, Lucas Harris , Linus Groner, Oliver FuhrerGeoscientific Model Development • 2023 Progress in leveraging current and emerging high-performance computing infrastructures using traditional weather and climate models has been slow. This has become known more broadly as the software productivity gap. With the end of Moore's law driving forward…Machine-Learned Climate Model Corrections From a Global Storm-Resolving Model: Performance Across the Annual Cycle
Anna Kwa, Spencer K. Clark, Brian Henn, Noah D. Brenowitz, Jeremy McGibbon, Oliver Watt-Meyer, W. Andre Perkins, Lucas Harris, Christopher S. BrethertonJournal of Advances in Modeling Earth Systems • 2023 One approach to improving the accuracy of a coarse-grid global climate model is to add machine-learned (ML) state-dependent corrections to the prognosed model tendencies, such that the climate model evolves more like a reference fine-grid global storm…Old dog, new trick: Reservoir computing advances machine learning for climate modeling
Christopher S. BrethertonESSOAr • 2023 Physics-informed machine learning (ML) applied to geophysical simulation is developing explosively. Recently, graph neural net and vision transformer architectures have shown 1-7 day global weather forecast skill superior to any conventional model with…Improving stratocumulus cloud amounts in a 200-m resolution multi-scale modeling framework through tuning of its interior physics
Liran Peng, Michael Pritchard, Peter N. Blossey, Walter M. Hannah, Christopher S. Bretherton, Christopher R. Terai, and Andrea M. JenneyESSOAr (submitted to the American Geophysical Union journal JAMES) • 2023 High-Resolution Multi-scale Modeling Frameworks (HR) -- global climate models that embed separate, convection-resolving models with high enough resolution to resolve boundary layer eddies -- have exciting potential for investigating low cloud feedback…