IBM Using Supercomputing To Improve Local Weather Prediction At Global Scale

IBM’s “The Weather Company” announced a partnership with the University Corporation for Atmospheric Research (UCAR) and the National Center for Atmospheric Research (NCAR) to improve weather prediction globally at local levels. The collaboration will match IBM’s OpenPOWER-based supercomputer technology and expertise with advanced meteorological science from The Weather Company and NCAR’s community weather model.

Prediction Of Small-Scale Weather Events

The group will produce rapidly-updating storm-scale models that can predict weather events in smaller local areas. This is a significant progress from today’s models that can predict weather events such as hurricanes or snowstorms accurately enough only for larger regions.

The new model promises to improve weather and climate forecasting through better analysis of small-scale phenomena such as thunderstorms that can impact the weather. The model will be able to cover the entire planet with such small-scale analysis, including areas of the globe that had been previously underserved by existing regional models for weather prediction.

Powered By IBM Supercomputers

One of the main use cases for supercomputers is weather and climate simulation and forecasting. IBM’s The Weather Company will partner with UCAR will design a supercomputing system using IBM’s next-generation Power9 processors, which are scheduled to be delivered to the first customers at the end of this year.

“IBM is one of only a few organizations in the world that has the capability to develop a model to run at this global, granular scale,” said Mary Glackin, head of weather science and operations for The Weather Company, an IBM Business, to Tom’s Hardware in an email.“As advocates for science, we embrace strong public-private partnerships that understand the value science brings to society, such as our continued efforts with UCAR to advance atmospheric and computational sciences,” he added.

A Model Optimized For Cloud Scale

Although regional models for thunderstorm prediction have existed for about a decade, they don’t scale well to the entire planet. The Weather Company will adopt NCAR’s Model for Prediction Across Scales (MPAS) community model to run more efficiently on these next-generation supercomputers. This should make a global weather prediction model that can also predict small-scale events possible. Not only that, but it should also make possible accurate long-range forecasts days or even months in advance.

"This is a major public-private partnership that will advance weather prediction and generate significant benefits for businesses making critical decisions based on weather forecasts," said UCAR President Antonio J. Busalacchi. "We are gratified that taxpayer investments in the development of weather models are now helping U.S. industries compete in the global marketplace," he noted.