IBM and NASA have announced the release of Surya, the world’s first open-source AI foundation model trained on high-resolution solar observation data. Now available on Hugging Face, Surya represents a breakthrough in space weather forecasting—offering powerful new tools to anticipate solar storms that can disrupt satellites, power grids, GPS, and telecommunications.
Why It Matters
The Sun, 93 million miles away, plays a critical role in sustaining life—but it also poses risks to our technology-dependent civilization. Solar flares and coronal mass ejections can:
- Damage satellites and spacecraft
- Disrupt aviation navigation
- Knock out power grids
- Threaten astronaut safety
- Even impact global food production through GPS disruptions
A single severe solar storm could cost the global economy up to $2.4 trillion over five years (Lloyd’s report). Recent solar events have already disrupted GPS, forced flight diversions, and damaged satellites—demonstrating the urgent need for accurate forecasting.
What Makes Surya Different
Traditional space weather prediction has been limited by partial, low-resolution solar data. Surya overcomes these barriers by training on nine years of high-resolution solar imagery from NASA’s Solar Dynamics Observatory.
Key highlights:
- 16% boost in solar flare classification accuracy compared to older methods
- First AI model capable of visual solar flare prediction, showing where a flare may occur up to two hours in advance
- Trained on datasets 10x larger than typical AI workloads, requiring custom multi-architecture AI solutions
- Supports critical forecasting tasks: flare prediction, solar wind speed estimation, EUV spectra prediction, and active region tracking
“Think of this as a weather forecast for space,” said Juan Bernabe-Moreno, Director of IBM Research Europe. “Surya helps us anticipate solar storms with unprecedented precision—an essential step in protecting modern technology.”
A Tool for Global Science
By releasing Surya and its curated heliophysics dataset openly on Hugging Face, IBM and NASA are democratizing access for the global research community. Scientists, governments, and industries worldwide can now build specialized applications—whether for aviation, energy infrastructure, or space exploration.
Kevin Murphy, NASA’s Chief Science Data Officer, emphasized:
“This model empowers broader understanding of the Sun’s behavior at unmatched speed and scale, helping protect the critical systems we all depend on here on Earth.”
IBM + NASA’s AI Collaboration
Surya joins the Prithvi family of foundation models, which includes:
- Prithvi-Weather: Open-source model for climate and weather forecasting
- Prithvi-Geospatial: Tools for mapping and environmental monitoring
This collaboration exemplifies IBM’s commitment to using AI not just for business transformation, but as a driver of scientific discovery across Earth and space.
Looking Ahead
As humanity ventures deeper into space and grows more reliant on satellite networks, space weather prediction will be as vital as Earth weather forecasting. Surya’s release is more than a scientific achievement—it’s a critical safeguard for our technological civilization.
