NASA and IBM have announced an open source artificial intelligence model that may revolutionize the way scientists forecast the sun’s weather. And it is a model called Surya which would be trained using more than a decade of solar data to provide earlier signals about flares and eruptions that could damage satellites, astronauts or even power grids
In the previous year, the strongest geomagnetic storm of the last 20 years crashed to Earth. The auroras that showed the skies gave the effect of imagery having been painted as they could be easily spotted way south in the United States than would otherwise be the case. Such storms however are a cause of concern to energy providers. An example is that in 1989, a nine-hour blackout in Quebec was caused by plasma ejections of the Sun. In this case, improved preparedness did not accord North America any severe service outages.The case made space weather prediction crucial the open source artificial intelligence model by NASA and IBM will overcome this dilemma
Surya: An Artificial Intelligence Telescope of the Sun
The company IBM referred to Surya as an AI telescope to the Sun, aimed at predicting the presence of flares and rendering visualizations of eminent solar activity. The model, trained on nine years of high resolution imagery collected by the Solar Dynamics Observatory (SDO) in NASA was constructed using 366 million parameters. Juan Bernabe-Moreno who is the director of IBM Research Europe, UK and Ireland stipulated that the science team at NASA was at the core in developing its capabilities.
Initial evaluation demonstrated that Surya was able to determine more frequently than all earlier systems, whether a flare would take place within 24 hours of the moment by 16 percent. It is also able to generate predictions of the appearance of the Sun which has been able to predict imagery two hours ahead. Bernabe-Moreno observed that two hours might be a short time but it can be a turning point to the energy providers dealing with responsive grids.
Uncertainty impacts on funding endangers NASA science programs
The giving of Surya is timely when the scientific undertakings in NASA are under pressure of funding. The science budget of the agency has been proposed to be cut by about half by the administration of President Trump. Solar Dynamics Observatory was also reduced in funding as announced by The Planetary Society and is set to have its annual budget reduced by 4 million to a budget of only 8 million.
The proposed reductions would slash almost 4,000 jobs at NASA out of about twenty percent of its employees. Congressional lawmakers have responded but the current budget to operate in 2026 remains in the unknown as many missions are up in the air. Bernabe-Moreno stressed that Surya could not exist without NASA. In the first months, they have been telling us what the model should do, the science team under NASA have been telling us. Even as the money dries up, the AI platform has become an enduring resource to the overall scientific community.
The Prediction of Solar Storms: A Challenge.
Solar storms come often when the Sun emits huge quantities of energy and particles. Their occurrence, namely solar flares and coronal mass ejections, can cause disruption in radio connections, can destroy satellites and expose astronauts to radiation. In severe scenarios they would lead to disruption of the power grids around the world. Despite knowing the conditions that are associated with flares, scientists have yet to predict with certainty when the flare will occur and what magnitude it will be. It is only these little destabilizations that we know occur, but we do not when, she said.
Surya is taught on more than 250 terabytes of solar images, gathered at various wavelengths by SDO. With this swathe of data, scientists hope the AI will be able to pick up minute pre-flare signals that humans are incapable of noticing. In the prediction of flares, the model not only predicted the time but also the location and intensity of the flares, two hours before their occurrence, effectively doubling the warning time, compared to the existing methods.
What the Sun does not see
Tracking space weather is what Surya does literally, but according to researchers, its circuits could find their way into far more applications. Bernabe-Moreno wants to integrate the model in older IBM and NASA weather and climate models to explore solar and terrestrial weather interaction, including the possible drivers of solar activity in lightning. Surya is only a foundation model and not specific to one task, so the scientists think that it could be modified to study other planetary objects and astrophysical phenomena.
Just like big language models that have been conditioned on text, Surya can be used as a general purpose tool to discover new trends in solar physics and other fields. The most promising application of it, at present, is the safeguarding of essential infrastructure. As countries have become more dependent on satellite communications, digital data, and electricity grids, even a handful of hours notice may spell the difference between mild inconveniences and significant shutdowns. With scientists affirming that the Earth is overdue regarding a solar superstorm, the need for such predictive tools has never been more important.