National Discovery Cloud for Climate workshop overview

Trey Stafford

2024-03-26

Workshop overview

• San Diego, CA March 6-7, 2024
• Focus on defining NSF vision for accessible, cloud-based climate science
• Similarities to National Artificial Intelligence Research Resource Pilot (NAIRR)
• ~30 projects represented
• Mostly PIs in attendance
• Trey represented the QGreenland-Net project

Workshop focused on defining a vision for the NSF’s activities supporting the creation of a more integrated cloud-based system for climate science.

The NDC-C is a 2-year pilot that aims to lay the foundations for an NDC that will cater to a wider science community.

“The National Science Foundation (NSF) initiated the development of a National Discovery Cloud for Climate (NDC-C) to “federate advanced compute, data, software, and networking resources, democratizing access to a cyberinfrastructure ecosystem that is increasingly necessary to further climate-related science and engineering”. The NDC-C builds on many existing efforts supported by the CISE and GEO directorates across the entire science vertical from hardware to foundational infrastructure to software services to core climate research to community activities.” - https://www.nationaldiscoverycloud.org/home

CISE == Directorate for Computer and Information Science and Engineering GEO == Geosciences Directorate

There were ~30 projects represented, ranging from infrastructure work (e.g., new HPC resources) to research projects focused on answering earth-science questions using cloud-computing resources (e.g., permafrost landscape analysis in the Arctic). Discussions ranged from “grandiose” visions of a cloud-based computational environment that users interact with through AI natural-language interfaces to what we can do as a disparate group of individually-funded projects in the next 6-18 months.

Workshop activities

• Guided discussions
  • Collective impact
  • Equity and education
• Breakout groups
• Networking/collaboration time

Workshop outcomes

• Reccomendations for the NSF NDC-C via a workshop report
• Slides describing each project
• Establishment of actual and potential collaborations

This is an image of the QGreenland-Net slide created at the workshop.

Collaborations explored

• EAGER: Scalable Climate Modeling using Message-Passing Recurrent Neural Networks
• Elements: A workflow for efficient and reproducible permafrost geomorphology analysis
• Pelican: Advancing the Open Science Data Federation Platform
• National Data Platform Pilot: Services for Equitable Open Access to Data

• “EAGER: Scalable Climate Modeling using Message-Passing Recurrent Neural Networks” project are working on AI models that could be leveraged by QGreenland-Net to suggest/automate the creation of data transformation workflows.
• “Elements: A workflow for efficient and reproducible permafrost geomorphology analysis” project is interested in using the tools we develop to find and extract data of contextual importance to study sites in Greenland
• “Pelican: Advancing the Open Science Data Federation Platform” project and I spoke about potentially adding QGreenland data to the Open Science Data Federation Platform (OSDF), which provides cloud-friendly access to science data via a global network of data caches.
• “National Data Platform Pilot: Services for Equitable Open Access to Data” project spoke with me about the possibility of connecting the ADC/DataOne to the National Data Platform, which provides AI tools to support data discoverability.

Resources

• NDC-C NSF website
• NDC-C workshop website
• NDC-C workshop notes
• More photos from Trey’s trip