- Scientists hope Surya extracts insights from the Sun’s complex magnetic processes
- Researchers processed nine years of imagery from the Solar Dynamics Observatory
- Surya achieved a reported 16% improvement in flare classification accuracy
IBM and NASA have introduced Surya, the first open source foundation model for solar physics.
IBM says the AI model, whose name comes from the Sanskrit word for the Sun, is trained to forecast solar activity such as flares and storms that can disrupt satellites, navigation systems, and power grids.
It has been made available through Hugging Face, GitHub, and IBM’s TerraTorch library, alongside a dataset collection called SuryaBench.
From Earth data to solar forecasts
The project comes as reliance on space-based technology expands, from aviation and communication to future deep space missions.
Predicting solar weather remains a difficult task, given that these events originate millions of miles away on a body whose physics are still only partly understood.
“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.
“With Surya we have created the first foundation model to look the Sun in the eye and forecast its moods.”
This collaboration follows earlier work by IBM and NASA on AI-driven models for Earth and weather prediction, which led to the development of the Prithvi model that analyzed satellite data to aid studies of climate and atmospheric systems.
With Surya, they are attempting something similar for the Sun, turning years of high-resolution solar imagery from NASA’s Solar Dynamics Observatory into a kind of digital twin.
Scientists hope the model will allow forecasts that go beyond whether a flare will happen.
Early reports suggest Surya can generate high-resolution visual predictions of flares up to two hours before they occur, doubling the lead time of traditional methods.
That would mean additional preparation time for astronauts and operators of critical infrastructure on Earth.
To build Surya, researchers processed nine years of imagery from the Solar Dynamics Observatory, which captures the Sun every 12 seconds at multiple wavelengths.
They employed a long-short vision transformer with spectral gating to manage the immense data load.
The model was trained not only to analyze current conditions but to infer what future observations would look like, testing its accuracy against real data.
“We want to give Earth the longest lead time possible,” said Andrés Muñoz-Jaramillo, a solar physicist at the Southwest Research Institute and a lead scientist on the project.
“Our hope is that the model has learned all the critical processes behind our star’s evolution through time so that we can extract actionable insights.”
Like other large language models and AI tools, Surya raises questions about whether its outputs should be treated as discovery or as augmentation of human expertise.
However, its backers emphasize automation and efficiency, pointing to a claimed 16% improvement in flare classification accuracy.
Still, forecasting remains far from certain, as the Sun’s activity involves many processes that remain poorly understood.
While Surya is described as a step toward better anticipation of solar threats, researchers are careful not to present it as a final answer.
Instead, they frame it as a bridge that may help scientists work with massive data more effectively.
As with any AI writer or LLM, its predictions are limited by the data it has been trained on and the assumptions built into its design.
