IBM and NASA have developed a new AI foundation model for a wide range of climate and weather applications, with contributions from the Department of Energy’s Oak Ridge National Laboratory. The new open-source model, named Prithvi WxC, is built to tackle both short-term weather forecasting and long-term climate projections with flexibility and scalability. In keeping with NASA’s open science policies, the new AI model is available for download on Hugging Face.

Due to its innovative design and training methodology, the weather and climate foundation model can address a much wider range of applications compared to current weather AI models. This is detailed in a recently published arXiv paper titled “Prithvi WxC: Foundation Model for Weather and Climate.”

The development of Prithvi WxC was triggered by the realization that AI emulators can rival traditional numerical weather prediction models running on HPC systems. However, most existing weather and climate models have primarily focused on single-use cases, particularly mid-range forecasting. Prithvi WxC addresses this gap by providing a flexible solution that is adaptable to various weather and climate applications.

Prithvi WxC features 2.3 billion parameters and is pre-trained on 40 years of weather and climate data from NASA’s Modern-Era Retrospective Analysis for Research and Applications, Version 2 (MERRA-2).

“The NASA foundation model will help us produce a tool that people can use: weather, seasonal, and climate projections to help inform decisions on how to prepare, respond, and mitigate,” Karen St. Germain, director of the Earth Science Division of NASA’s Science Mission Directorate, said in a statement included in the release.

Prithvi-WxC is an extension of the ongoing collaboration between IBM and NASA, enhancing the suite of geospatial foundation models within the “Prithvi” family. Last year, the two industry leaders introduced the Prithvi geospatial AI foundation model.

They have also worked together on initiatives at NASA’s Jet Propulsion Laboratory (JPL), where IBM Watson’s cognitive computing capabilities have been employed to analyze and extract insights from vast amounts of scientific data.

Additionally, in 2018, IBM partnered with NASA and Hewlett Packard Enterprise to develop and test the “Spaceborne Computer,” designed to evaluate high-performance computing in the harsh conditions of space.

The Prithvi WxC has been launched in two distinct versions: one focused on climate and weather data downscaling, and the other on gravity wave parameterization.

The downscaling model works by transforming low-resolution variables into high-resolution outputs, offering improvements of up to 12 times the original resolution, thereby enhancing forecasting and climate projections.

The gravity wave model aims to help scientists more accurately estimate gravity wave generation, which affects climate and weather patterns but is not adequately represented in traditional models.

“This space has seen the emergence of large AI models that focus on a fixed dataset and single use case — primarily forecasting. We have designed our weather and climate foundation model to go beyond such limitations so that it can be tuned to a variety of inputs and uses,” said Juan Bernabe-Moreno, Director of IBM Research Europe (UK and Ireland) and IBM’s Accelerated Discovery Lead for Climate and Sustainability.

Bernabe-Moreno noted that the new AI model is capable of operating at both local and global scales. It can play a crucial role in understanding meteorological phenomena like hurricanes and atmospheric rivers. Furthermore, it has the potential to enhance climate change mitigation efforts by improving the resolution of climate models and offering more precise predictions of climate risks.

Karen St. Germain, director of NASA’s Earth Science Division, highlighted the significance of tools like this model in delivering actionable science. She stated, “The rapid changes we’re witnessing on our home planet demand this strategy to meet the urgency of the moment.”

IBM and NASA will continue to collaborate on a series of innovative projects aimed at extracting new insights from space and earth observation data. This partnership seeks to leverage advanced technologies, including AI and ML, to enhance the analysis of vast amounts of data collected from various sources, such as satellites and ground-based sensors.