IBM and NASA Create a Digital Twin of the Sun to Predict Solar Disruptions

• Surya is the first foundation model trained to simulate the Sun’s behavior using satellite imagery and magnetic data.
• The open-source model and its benchmark suite, SuryaBench, enable earlier warnings for solar flares and other space weather events.

IBM and NASA have released Surya, a digital twin of the Sun built from nine years of imagery and magnetic data collected by NASA’s Solar Dynamic Observatory (SDO). The open-source model, available on Hugging Face, GitHub, and via IBM’s TerraTorch library, simulates the Sun’s behavior to help forecast solar storms that can disrupt satellites, power grids, and communications on Earth.

“We’ve been on this journey of pushing the limits of technology with NASA since 2023, delivering pioneering foundational AI models to gain an unprecedented understanding of our planet Earth,” said Juan Bernabé-Moreno, the IBM director in charge of the scientific collaboration with NASA, in a news release. “With Surya we have created the first foundation model to look the Sun in the eye and forecast its moods.”

Surya was trained using high-resolution images and solar magnetic data, processed through an AI model built to capture both small details and large-scale patterns. According to IBM Research, the model predicts solar flares up to two hours in advance, double the current warning time, and achieved a 16% gain in classification accuracy in testing.

The release also includes SuryaBench, an open-source collection of datasets and evaluation tools built to support solar forecasting and research. It simplifies work on tasks like flare detection, solar wind analysis, and magnetic structure tracking, all of whihc are key for understanding how solar activity affects life and infrastructure on Earth.

🌀 Tom’s Take:

IBM and NASA’s earlier Prithvi models turned satellite data into a digital twin of Earth to support weather and climate prediction. Now Surya does the same for the Sun. These models help us plan for what’s next and open up entirely new ways to see our world and universe through data.

Source: IBM Research