Digital Event Horizon
NVIDIA has announced a new family of open-source weather and climate models that promise to revolutionize the field of weather forecasting. The new models, part of the NVIDIA Earth-2 family, include cutting-edge AI models for nowcasting, medium-range forecasts, and global data assimilation. With these powerful tools, developers can build accurate and reliable forecasts in minutes, rather than hours or days.
NVIDIA announces groundbreaking open-source models for weather forecasting with unprecedented accuracy and speed. The NVIDIA Earth-2 family comprises a suite of accelerated tools and models for building sovereign weather and climate predictions. StormScope, powered by generative AI, enables country-scale forecasts into kilometer-resolution predictions in just minutes. Earth-2 Medium Range predicts medium-range forecasts (up to 15 days) with high accuracy across multiple variables. HealDA produces initial conditions for weather prediction globally, generating snapshots of the current atmosphere in seconds on GPUs.
NVIDIA has made a groundbreaking announcement that is set to transform the field of weather forecasting forever. The company's latest open-source models, part of the NVIDIA Earth-2 family, promise to bring unprecedented accuracy and speed to the prediction of severe weather events, medium-range forecasts, and global data assimilation. These cutting-edge AI models are designed to enable developers to build sovereign weather and climate predictions that they fully own and control.
The NVIDIA Earth-2 family comprises a suite of accelerated tools and models that enables developers to bring together disparate weather and climate AI capabilities. This comprehensive ecosystem is designed to be easy-to-use, thanks to an extensive range of open-source software, including Earth2Studio for creating inference pipelines and Physics Nemo for training models. With this powerful toolkit at their disposal, developers can now create custom weather and climate simulations that are tailored to their specific needs.
At the heart of the NVIDIA Earth-2 family is a new generation of AI models that are specifically designed to tackle some of the most complex challenges in weather forecasting. The first model to be released is Earth-2 Nowcasting, powered by a revolutionary new architecture called StormScope. This cutting-edge technology uses generative AI to make country-scale forecasts into kilometer-resolution predictions of local storms and hazardous weather in just minutes.
What sets StormScope apart from traditional physics-based weather-prediction models is its ability to directly predict satellite and radar data using AI. This allows it to generate the first predictions that outperform traditional models on short-term precipitation forecasting, by simulating storm dynamics directly. The model is trained directly on globally available geostationary satellite observations (GOES) over the contiguous US (CONUS), but this approach could be applied to train versions of the model over other regions with similar satellite coverage.
Another significant release from NVIDIA Earth-2 is Earth-2 Medium Range, powered by a new model architecture called Atlas. This model enables high-accuracy weather prediction for medium-range forecasts - or forecasts of up to 15 days in advance - across 70+ weather variables including temperature, pressure, wind and humidity. The Atlas model uses a latent diffusion transformer architecture to predict incremental changes in the atmosphere so as to preserve critical atmospheric structures and reduce forecasting errors.
On standard benchmarks, Earth-2 Medium Range has outperformed leading open models such as GenCast on the most common forecasting variables measured by the industry. This is a significant achievement, given that weather forecasting is one of the most complex and challenging tasks in AI research.
The final model to be released from NVIDIA Earth-2 is Earth-2 Global Data Assimilation, powered by a new model architecture called HealDA. This model produces initial conditions for weather prediction - snapshots of the current atmosphere, including temperature, wind speed, humidity and air pressure, at thousands of locations around the globe. What's particularly exciting about HealDA is that it can generate initial conditions in seconds on GPUs instead of hours on supercomputers.
When coupled with Earth-2 Medium Range, this results in the most skillful forecasting predictions produced by an open, entirely AI pipeline. The potential implications for weather forecasting are profound, enabling developers to build sovereign weather and climate predictions that they fully own and control.
The NVIDIA Earth-2 family of models is set to revolutionize the field of weather forecasting, making it possible for developers to build accurate and reliable forecasts in minutes, rather than hours or days. With its cutting-edge AI models and powerful toolkit, NVIDIA Earth-2 promises to transform the way we approach weather forecasting forever.
Related Information:
https://www.digitaleventhorizon.com/articles/NVIDIA-Earth-2-Revolutionizes-Weather-Forecasting-with-Cutting-Edge-AI-Models-deh.shtml
Published: Mon Jan 26 09:04:00 2026 by llama3.2 3B Q4_K_M
