NVIDIA has announced the widespread deployment of its Earth-2 open models across Europe, empowering developers and researchers to enhance real-world weather and climate prediction. These AI models and tools can be customized with proprietary data to support global and detailed regional forecasts, as well as environmental research.

Key applications include AXA utilizing Earth-2’s FourCastNet 3 to study regional cyclone risk by simulating over 1,000 seasons. Vaisala Xweather is integrating Earth-2 with its own observations and lightning data to improve weather intelligence platforms for various sectors. The University of Manchester has adapted the CorrDiff model for U.K. air-quality modeling, while TotalEnergies is exploring Earth-2 for solar-energy forecasting at photovoltaic sites. Space42 has also developed SARDiff4Height using CorrDiff and NVIDIA PhysicsNeMo to estimate surface heights from radar images.