NSF NCAR researcher awarded large resource grant in national AI initiative

by Shira Feldman

On Monday, May 6, David John "DJ" Gagne and his collaborators received a large resource allocation award for the purpose of conducting their research project to develop, train, and evaluate global and regional artificial intelligence numerical weather prediction (AI NWP) models in an open, scaleable, science-focused framework. 

 

Gagne received this grant as part of a National Science Foundation (NSF)-led program called the National Artificial Intelligence Research Resource (NAIRR) Pilot. The NAIRR Pilot originates from President Joe Biden's Executive Order on the Safe, Secure and Trustworthy Development and Use of AI. The NAIRR Pilot’s objective is to provide AI researchers and students with access to critical AI resources and data, and describes itself as “a vision for a shared national research infrastructure for responsible discovery and innovation in AI.”

 

David John Gagne headshot

David John Gagne and colleagues recently received a large resource award from the NSF-led NAIRR Pilot to conduct their AI research project, CREDIT.

Gagne, who is among the first recipients of NAIRR Pilot support, is a Machine Learning Scientist at the NSF National Center for Atmospheric Research (NCAR) in Boulder, Colorado. Gagne is a staff member in the Computational and Information Systems Lab (CISL) and heads the Machine Integration and Learning for Earth Systems (MILES) group.

 

The NAIRR Pilot resource award provides Gagne and his collaborators with 500,000 node-hours—equal to three million GPU-hours—on the Summit supercomputer at the Department of Energy's (DOE) Oak Ridge National Laboratory (ORNL).

 

The project in question is called the Community Runnable Earth Digital Intelligence Twin Training and Reforecasts (CREDIT) project. The CREDIT project will investigate how choices in the data processing, architecture, loss functions, and evaluation lead to more accurate, stable, physically consistent, and valuable AI NWP models, and produce large sets of reforecasts for further analysis and evaluation by the broader community.

 

This video shows a forecast stemming from the CREDIT project.

 

Gagne has co-led the development of the project's code, along with his two colleagues and fellow NCAR Machine Learning Scientists: CISL's John Schreck, and Will Chapman of NCAR's Climate and Global Dynamics Lab (CGD).

 

"Today marks a pivotal moment in the advancement of AI research as we announce the first round of NAIRR pilot projects,” said NSF Director Sethuraman Panchanathan. “The NAIRR pilot, fueled by the need to advance responsible AI research and broaden access to cutting-edge resources needed for AI research, symbolizes a firm stride towards democratizing access to vital AI tools across the talented communities in all corners of our country." 

 

The CREDIT project is one of 35 projects that will be supported with computational time as part of the first round of NAIRR Pilot awards.