Nvidia Earth-2 platform gets StormCast new AI model for more nuanced bad weather warnings

Nvidia's Earth-2 platform that enables simulation and visualization of weather at a global scale apparently was promising enough to get a few companies in the business of weather emulation/prediction interested enough to ink a few cooperation agreements with Team Green several months ago. While detailed reports on how much exactly those companies now rely on Nvidia's hardware and software are hard to come by, we do have confirmation that Huang-led juggernaut is still interested in maintaining that specific software suite and giving it new powers indicating that things are going well.

On the 19th of August, the company announced the release of StormCast as an addition to Earth-2 that can make predictions of the future based on past events. The Lawrence Berkeley National Laboratory as well as the University of Washington get a mention as research/development partners; Nvidia's media release claims that thanks to the improvements, Earth-2 now delivers much better local (within a radius of around 30 km) bad weather predictions, striking an unbelievable balance between resolution, precision, time and cost of making predictions:

... up to 10% more accurate than the U.S. National Oceanic and Atmospheric Administration (NOAA)’s state-of-the-art 3-kilometer operational CAM

If applied correctly, the new AI model can help make early-warning systems for severe events that much more timely and precise, potentially saving lives and money. The media release names Taiwan's National Science and Technology Center for Disaster Reduction as one of the first institutions to make use of this new technology.