The National Oceanic and Atmospheric Administration (NOAA) announced on July 27 that it will move its operational weather prediction supercomputing system to commercial cloud infrastructure, marking a major shift in how the agency develops and runs the models that underpin U.S. weather forecasting.
According to a NOAA press release, Google Cloud will serve as the primary provider of high-performance computing infrastructure for NOAA’s Weather and Climate Operational Supercomputing System (WCOSS).
NOAA said the migration will support current and future improvements to its weather forecast models, including the Global Forecast System and Global Ensemble Forecast System. The agency plans to transition operational workloads through early 2027 and complete the cloud migration by December 2027.
“Cloud-based high-performance computing will accelerate the transition of research into operations by eliminating traditional bottlenecks of on-premise systems,” NOAA Administrator Neil Jacobs said. “Not only will this new approach provide access to the latest chips, it will advance weather forecasting by aligning operational and research compute and provide capabilities for elastic on-demand model development.”
The WCOSS provides the computing power behind NOAA’s operational weather and climate models.
Moving the system from agency-owned hardware to commercial cloud infrastructure is intended to provide greater flexibility, allow more frequent model updates, improve system resilience, and enable NOAA to rapidly scale computing capacity during periods of peak demand such as hurricane season.
NOAA said the transition will occur in two phases. The first phase establishes the agency’s cloud-based supercomputing capability, while the second phase migrates individual weather prediction models into the new operational environment.
The agency said the modernization effort is expected to provide earlier and more accurate forecasts and warnings for extreme weather events by giving researchers and operational forecasters access to more adaptable computing resources.
“We are charting a course to global leadership in cloud-based numerical weather prediction,” said Ken Graham, director of NOAA’s National Weather Service. “This is a significant achievement for NOAA; one that will ensure our critical modeling operations can seamlessly adapt to emerging science, and strengthen our mission to protect life and property.”
In a blog post, Google said the modernization effort positions NOAA among the first operational numerical weather prediction centers worldwide to move forecasting operations to commercial cloud infrastructure.
NOAA will rely on Google’s H4D virtual machines powered by fifth-generation AMD EPYC processors to support its high-performance computing workloads, according to the blog post.
Beyond infrastructure, NOAA said the partnership will expand its use of artificial intelligence for weather forecasting. The agency is already using Google DeepMind as the framework for its Artificial Intelligence Global Forecast System (AIGFS) model suite.
According to NOAA, the next generation of machine learning forecasting tools developed with Google will reduce the time required to generate a global weather forecast from hours to seconds.
The cloud transition continues a focus on technology that has been building since NOAA released a strategy to coordinate efforts to use AI across the agency in 2020.
Earlier this year, NOAA announced contracts to develop cloud-based technology for forecasters across its 122 local Weather Forecast Offices and other National Weather Service centers.
NOAA said its weather forecasting systems and models are developed and managed by the National Weather Service’s Modeling and Development organization in College Park, Md.