NVIDIA is worked up to announce three latest open-source models as a part of the NVIDIA Earth-2 family, making it easier than ever to construct weather forecasting capabilities across the weather stack, including tasks resembling data assimilation, forecasting, nowcasting, downscaling and more. As well as, developers can quickly start constructing weather and climate simulations by utilizing NVIDIA open source software: Earth2Studio for creating inference pipelines and Physics Nemo for training models.
NVIDIA Earth-2 comprises a set of accelerated tools and models which enables developers to bring together typically disparate weather and climate AI capabilities. Because Earth-2 is totally open, developers can customize and fine-tune their simulations to their specific needs, using their very own data and their very own infrastructure to construct sovereign weather and climate predictions they fully own and control. Earth-2:
- Is a collection of leading open weather and climate models
- Is straightforward-to-use because of an ecosystem of open source software
- Allows you to create your personal sovereign capabilities
Earth-2 Nowcasting: Kilometer-Scale Severe Weather Prediction
Out now on Hugging Face: Earth-2 Nowcasting, powered by a brand new model architecture called StormScope, using generative AI to make country-scale forecasts into kilometer‑resolution, zero- to six-hour predictions of local storms and unsafe weather in only minutes. Earth-2 Nowcasting can generate the primary predictions that outperform traditional, physics-based weather-prediction models on short-term precipitation forecasting by simulating storm dynamics directly. It harnesses AI to directly predict satellite and radar data.
This version is trained directly on globally available geostationary satellite observations (GOES) over the contiguous US (CONUS). Nevertheless, this method might be applied to coach versions of the model over other regions with similar satellite coverage.
Research Paper: Learning Accurate Storm-Scale Evolution from Observations
Earth-2 Medium Range: Highly accurate 15-Day Global Forecasts
Out now on Hugging Face: Earth-2 Medium Range, powered by a brand new model architecture called Atlas, enabling high-accuracy weather prediction for medium-range forecasts — or forecasts of as much as 15 days upfront — across 70+ weather variables including temperature, pressure, wind and humidity. It uses a latent diffusion transformer architecture to predict incremental changes within the atmosphere in order to preserve critical atmospheric structures and reduce forecasting errors. On standard benchmarks, it outperforms leading open models resembling GenCast on essentially the most common forecasting variables measured by the industry.
Research Paper: Demystifying Data-Driven Probabilistic Medium-Range
Weather Forecasting
Earth-2 Global Data Assimilation: An End-to-End AI Pipeline
Coming soon to Hugging Face: Earth-2 Global Data Assimilation, powered by a brand new model architecture called HealDA, which produces initial conditions for weather prediction — snapshots of the present atmosphere, including the temperature, wind speed, humidity and air pressure, at hundreds of locations across the globe. Earth-2 Global Data Assimilation can generate initial conditions in seconds on GPUs as a substitute of hours on supercomputers. When coupled with Earth-2 Medium Range, this leads to essentially the most skillful forecasting predictions produced by an open, entirely AI pipeline.
These models join established open NVIDIA weather and climate models resembling FourcastNet3, CorrDiff, cBottle, DLESym and more.
NVIDIA Earth2Studio is an open-source Python ecosystem for quickly creating powerful AI weather and climate simulations. It provides all of the obligatory inference tools to start with the brand new model checkpoints on Hugging Face. It’s as easy as:
Hugging Face Package for Earth-2 Nowcasting
Research Paper: Learning Accurate Storm-Scale Evolution from Observations
Hugging Face Package for Earth-2 Medium-Range
Research Paper: Demystifying Data-Driven Probabilistic Medium-Range
Weather Forecasting
