Enter Note Done
Go to previous page in this tab
Session
  • Presentation | IN43A: AI Foundation Models for Earth, Space, and Planetary Sciences III Oral
  • Oral
  • Share Icon
    Bookmark Icon
  • [ONLINE] IN43A-07: Quantitative Evaluation of Foundation Models for Weather Forecasting
    • Schedule
    Notes
  • Online
    Online
    Set Timezone

Generic 'disconnected' Message
Author(s):
Katherine Breen, NASA Goddard Space Flight Center (First Author, Presenting Author)
Jordan Caraballo-Vega, NASA Goddard Space Flight Center
Jian Li, NASA Goddard Space Flight Center
Glenn Tamkin, NASA Center for Climate Simulation
Arlindo daSilva, NASA Goddard Space Flight Center
Donifan Barahona, NASA GSFC
Kara Lamb, Columbia University
Nathan Arnold, Colorado State University
Michael Hendrickson, NASA Goddard Institute for Space Studies
Min-Jeong Kim, NOAA Science Center
Andrea Molod, NASA Goddard Space Flight Center
Melanie Frost, NASA Data Science Group
William Putman, NASA GSFC
Mark Carroll, NASA Goddard Space Flight Center


This study explores how large, pre-trained artificial intelligence (AI) models, called foundation models (FMs), can help improve weather forecasts and the representation of key meteorological variables.


To make sure these models are useful for science, we focused on addressing the following questions:




  • Can we reproduce published findings and verify that FM output is physically realistic?




  • Can we make a workflow that is accessible and useful for a variety of domain experts?




  • How do FM forecasts compare with our operational weather forecasting system?




This work helps set a clear standard for using AI in weather forecasting. Over time, the goal is to make AI weather models more useful, trustworthy, and available to Earth science researchers.




Scientific Discipline
Suggested Itineraries
Neighborhood
Type
Where to Watch
Main Session
Discussion