Machine Learning for Earth System Modeling ICML 2024

Who am I? 🌊 Climate Scientist Ocean transport of Heat, Carbon Oxygen Impact of small scale processes on global climate variability. 🤓 Open Science Nerd Maintainer: Pangeo CMIP6 Cloud Data xMIP/xGCM ⚙ Integration Engineer Manager for Data and Computation - NSF-LEAP Lead of Open Research - m2lines M²LInES jbusecke juliusbusecke.com @JuliusBusecke @CodeAndCurrents@hachyderm.io @codeandcurrents.bsky.social

We are starting to feel the consequences of the climate crisis! Climate-related risks to health, livelihoods, food security, water supply, human security, and economic growth are projected to increase with global warming of 1.5°C and increase further with 2°C. - IPCC Report - https://www.wsj.com https://www.twitter.com https://edition.cnn.com/

We generally know what to do...

We generally know what to do... But we have not done a great job at it so far! Its still going up! 😩

The large majority of modelling studies could not construct pathways characterized by lack of international cooperation, inequality and poverty that were able to limit global warming to 1.5°C. (high con f idence) - IPCC Report - We need everyone on board!

The large majority of modelling studies could not construct pathways characterized by lack of international cooperation, inequality and poverty that were able to limit global warming to 1.5°C. (high con f idence) - IPCC Report - We need everyone on board!

What do we have? Observations Numerical Simulations Theory IPCC Chapter 9

What do we have? Observations Numerical Simulations Theory IPCC Chapter 9

What do we have? Observations Numerical Simulations Theory IPCC Chapter 9

Challenges: Scale and Model Bias Only feasible for short time reduced complexity Current climate projections https://vimeo.com/259423826

Challenges: Scale and Model Bias Only feasible for short time reduced complexity Current climate projections https://vimeo.com/259423826

Challenges: Scale and Model Bias Slide: Pierre Gentine

Challenges: Scale and Model Bias Slide: Pierre Gentine

Promising results but much more to be done! Slides: Pierre Gentine https://leap.columbia.edu/ https://m2lines.github.io/

What do we have? Tons of Data! https://earthdata.nasa.gov/eosdis/cloud-evolution SWOT NISAR Estimated Size of NASA observational data holdings Coupled Model Intercomparison Project (CMIP) CMIP6: ~20 PB CMIP7 (starting soon): 100+ PB https://wcrp-cmip.org/cmip7/ https://wcrp-cmip.org/ Interested in working with CMIP6 data in the cloud? Check out the Pangeo Cloud Data! https://github.com/leap-stc/ cmip6-leap-feedstock

Climate 🤝 ML: ClimSim A large multi-scale dataset for hybrid physics-ML climate emulation Winner will be announced soon!

Food for thought How can we improve existing climate forecasts and more e ff iciently integrate observations? How can we make climate simulation data more useful to stakeholders? The climate crisis is a global issue. How can we ensure that all global communities can contribute and use climate data based on their needs?

Thanks 🙏 Speakers Organizers Program Committee Members You

Schedule