NOAA, Microsoft Partner To Put Climate Models In The Clouds

The future of meteorological simulation and forecasting may well be bound for the clouds — well, cloud datacenters anyway. 

The US National Oceanic and Atmospheric Administration (NOAA) on Wednesday announced a strategic partnership with Microsoft Azure to explore the use of cloud computing to advance climate research.

This isn't, however, a cash grab in which Microsoft can avail itself of government funding. The year-long project is being conducted under a collaborative research and development agreement (CRADA) that allows NOAA to contribute staff, facilities, and intellectual property to the project, but not funding.

Microsoft, for its part, can contribute funding, though in this case, it's more likely the cloud giant will provide access to compute resources and expertise. The Register reached out to NOAA and Microsoft for comment; we'll let you know if we hear anything back.

NOAA says the collaboration will explore the use of Microsoft's cloud computing resources to advance five key areas of research and development. These include:

• Assisting NOAA's Earth Prediction Innovation Center (EPIC) pilot Earth system models on Microsoft Azure.
• Utilizing machine learning to improve NOAA's climate and forecast models.
• Integrating Microsoft's data collection, processing, and storage capabilities to simplify the dissemination of the agency's fisheries survey.
• Developing a searchable catalog of oceanic observations.
• Designing weather modeling and forecasting systems that can incorporate data from both internal and external resources.

The partnership follows an earlier collaboration between NOAA, Microsoft, and satellite services provider Xplore this summer. The project — also conducted under a CRADA — demonstrated the use of Azure's Orbital platform to connect and downlink data from NOAA's polar satellites, extending their effective lifespan.

The year-long project is also exploring the use of Xplore's Major Tom mission control software to transmit commands to the NOAA-18 satellite.

• US weather forecasters triple supercomputing oomph with latest machines
• Processing data... in space: AWS powers Earth observation satellite payload
• Is your datacenter safe from the next X-class solar flare?
• NASA scraps budget-busting GeoCarb greenhouse gas observatory project

NOAA's supercomputing muscle

It remains to be seen whether NOAA's collaboration with Microsoft will spur greater investment in cloud resources. In either case, the agency's reliance on supercomputers isn't going anywhere anytime soon.

For one, it's no secret that cloud resources are far more expensive than deploying hardware on-prem. Second, the NOAA just flipped the switch on two shiny new supercomputers this summer that it says are 3x more powerful than their predecessors, enabling longer, more detailed forecasts.

Located in Phoenix, Arizona, and Manassas, Virginia, respectively, the Cactus and Dogwood supercomputers were developed by General Dynamics in collaboration with HPE at a cost of more than $150 million. According to NOAA, the redundant systems support larger, higher resolution models than previously possible.

This "will allow us to better capture the physical processes that are going on in the Earth's systems, like formation of clouds, the formation of precipitation, and so forth," Brian Gross, director of the Environmental Modeling Center for the National Weather Service, told The Register in a June interview.

The systems are expected to cost as much as $505 million over their lifespan.

Microsoft's UK climate collab

With that said, there is precedent for migrating meteorological and climate models to the cloud. Early last year, the UK's Meteorological Office (MET) contracted Microsoft to host its own climate models on Azure's supercomputing-as-a-service platform.

Under the deal, for which the UK government would contribute £1.2 billion, Microsoft deployed HPE's Cray EX supercomputers alongside its own compute and data services in four quadrants across the south of the UK. Meanwhile, Azure's data archive systems provided the storage for more than four exabytes of data used by MET's models. And in the spirit of the project's climate focus, the systems would be powered by renewable energy.

Having narrowly avoided a lawsuit by French supercomputer vendor Atos late this spring, that project is clear to move forward. ®