Now loading...
Precise weather predictions play a vital role in safeguarding lives and ecosystems while informing critical choices across sectors like farming, power generation, public safety, and beyond.
Experts in meteorology, government bodies, climate technology developers, and businesses are actively deploying, refining, and expanding upon these cutting-edge AI systems on their own AI hardware to drive forward scientific advancements.
Within the scope of weather prediction, the AI platform Brightband, which participates in NVIDIA’s Inception program through its Sustainable Futures effort, has begun operational use of the Earth-2 Medium Range model to deliver daily worldwide forecasts. Julian Green, Brightband’s co-founder and chief executive, highlighted the thrill surrounding the rise of AI-driven weather technologies. He noted that models like Earth-2 Medium Range are accelerating progress, and its open-source nature fosters faster innovation by enabling comparisons and enhancements within the broader weather community.
Israel’s national weather authority is integrating Earth-2 CorrDiff into its daily operations and intends to adopt Earth-2 Nowcasting for producing detailed forecasts as often as eight times a day. This setup allows officials to better address severe weather events while slashing computing expenses. Amir Givati, the service’s director, reported a 90 percent drop in processing time for 2.5-kilometer resolution forecasts compared to traditional methods on CPU systems. He added that during a recent heavy rain event, their AI-enhanced model using CorrDiff outperformed all other tools in forecasting six-hour rainfall totals.
The Weather Company is testing Earth-2 Nowcasting for pinpointing intense local storms, while the National Weather Service explores these tools to streamline its processes.
For energy prediction and grid management, TotalEnergies is assessing Earth-2 Nowcasting to sharpen short-term hazard detection and strategic planning. Emmanuel Le Borgne, the company’s product manager for climate and weather forecasting, described Earth-2 as a significant leap in deploying sophisticated weather analytics broadly. He emphasized that tools like Earth-2 Nowcasting are transformative for energy operations, where rapid insights into nearby conditions can make all the difference in mere minutes.
Italian energy giant Eni is rigorously evaluating Earth-2 models, such as FourCastNet and CorrDiff, to refine forecasts through downscaling techniques, generating probability-based, fine-grained projections for weather patterns and natural gas needs up to weeks in advance.
China’s GCL, a major producer of solar components and a worldwide energy provider, has implemented Earth-2 models in its solar power forecasting system. Relative to conventional numerical methods, Earth-2 delivers superior accuracy at reduced expense, markedly boosting the reliability of GCL’s solar energy output estimates.
The Southwest Power Pool, partnering with Hitachi, employs Earth-2 Nowcasting and FourCastNet3 to refine wind predictions for the current day and the day ahead. This initiative aligns with the pool’s goals of bolstering grid stability and supporting smarter decisions throughout its region.
In evaluating economic consequences, S&P Global Energy leverages Earth-2 CorrDiff to convert broad climate information into targeted regional analyses for hazard evaluation. Meanwhile, the international insurer AXA is applying FourCastNet to simulate thousands of potential hurricane paths as part of its research into model testing, technique refinement, and comparisons with current approaches.
