NASA, IBM’s ‘Hot’ New AI Model Unlocks Secrets of Sun

Publisher’s note: This article was updated on August 20, 2025 to correct the number of years of training data used and the accuracy of the model. The original article indicates that the model was formed over 14 years of data from the solar dynamic observatory and exceeded the existing benchmarks of 15%; The model was in fact formed over 9 years of data and exceeded the existing benchmarks of 16%.

NASA reveals heat in solar science with the launch of the fundamental model of Suryiophysics, an artificial intelligence model (AI) formed over 9 years of observations from the NASA solar dynamic observatory.

Developed by NASA in partnership with IBM and others, Surya uses progress in AI to analyze large quantities of solar data, helping scientists better understand solar eruptions and predict the space weather that threatens satellites, electrical networks and communication systems. The model can be used to provide precocious warnings to satellite operators and helps scientists predict how the ultraviolet exit from the sun affects the upper atmosphere of the earth.

Preliminary results show that Surya makes progress in forecasting solar push, a long -standing challenge in heliophysics. Surya, with its ability to generate visual predictions of two -hour solar rockets in the future, marks a major step towards the use of AI for the meteorological prediction of the operational space. These initial results exceed the existing benchmarks of 16%. By providing open access to the model on Huggingface and the GitHub code, NASA encourages the community of sciences and applications to test and explore this AI model for innovative solutions that exploit the unique value of continuous, stable and long -term data sets of the solar dynamic observatory.

The success of the model is based directly on the long -term database of the Solar Dynamics observatory. Launched in 2010, the NASA solar dynamic observatory provided an uninterrupted and high resolution of the sun for almost 15 years by capturing images every 12 seconds in several wavelengths, as well as precise magnetic field measurements. This stable and well -calibrated data set, covering an entire solar cycle, is particularly suitable for the formation of AI models like Surya, allowing them to detect subtle models in solar behavior that shorter data sets would lack.

Surya’s strength lies in its foundation model architecture, which directly draws raw solar data. Unlike traditional AI systems that require extensive labeling, Surya can adapt quickly to new tasks and applications. Applications include monitoring of active regions, forecasting the escape activity, forecasting solar wind speed and the integration of data from other observatories, in particular the solar solar mission and the Heliospherical Observatory of NASA and the NASA solar probe.

“We are progressing data focused on data by integrating NASA’s deep scientific expertise into advanced AI models,” said Kevin Murphy, director of scientific data at NASA headquarters in Washington. “By developing a foundation model formed on NASA heliophysics data, we facilitate the analysis of the complexities of the sun’s behavior with unprecedented speed and precision. This model allows a broader understanding of the impact of solar activity which has an impact on the systems and critical technologies on which we rely here on Earth. ”

Solar storms present significant risks for our company dependent on technology. Powerful solar events energize the ionosphere of the earth, causing substantial GPS errors or a complete signal loss for satellite communications. They also present risks for electrical networks, as the geomagnetically induced currents of coronal mass ejections can overload transformers and trigger widespread breakdowns.

In commercial aviation, solar eruptions can disrupt radio and navigation systems while exposing high altitude flights to the increase in radiation. The stakes are even higher for human space flights. Astronauts to the Moon or Mars may need to depend on precise predictions of the shelter of intense radiation during events of solar particles.

The influence of the sun extends to the growing number of satellites with low terrestrial orbit, including those which provide a global internet at high speed. As the solar activity intensifies, it heats the upper atmosphere of the earth, the increase in the drag which slows down the satellites, draws them from the orbit and causes premature reinstatement. Satellite operators often find it difficult to predict where and when solar eruptions can affect these satellites.

“Our company is based on technologies that are very sensitive to space weather,” said Joseph Westlake, director of the Heliophysics division at NASA headquarters. “Just as we use meteorology to predict earth weather, space weather forecast predict the conditions and events of the spatial environment that can affect the land and our technologies. The application of AI to the data of our heliophysical missions is a vital step in the increase in our space meteorological defense to protect our modern world. ”

While Surya is designed to study the sun, its architecture and methodology are adaptable through scientific fields. From planetary science to earth observation, the project throws the fundamental infrastructure for similar IA efforts in various fields.

Surya is part of a broader Pusée de la NASA to develop scientific tools with free access and fueled by AI. Model and training data sets are available for free online for researchers, educators and students around the world, which reduces obstacles to participating and triggering new discoveries.

The training of Surya was supported in part by the National Pilot of Research Resources on Artificial Intelligence (NAIRR), an initiative led by the National Science Foundation (NSF) which offers researchers access to advanced IT, data sets and AI tools. The NAIRR pilot brings together federal resources and industry, such as NVIDIA’s calculation power, to extend access to the infrastructure necessary for the search for advanced AI.

“This project shows how the NAIRR driver unites federal and industry resources to accelerate scientific breakthroughs,” said Katie Antypas, director of the advanced Cyberinfrastructure Office of the NSF. “With the support of Nvidia and NSF, we do not only allow today’s research, we lay the foundations for a national AI network to conduct the discoveries of tomorrow.”

Surya is part of a larger effort defended and supported by the NASA office of the main scientific data services and the heliophysical division, the NSF and universities in partnership to advance NASA scientific missions thanks to innovative data science and AI models. The Architecture of AI of Surya was developed jointly by the team of implementation concepts and advanced concepts (Impact) under the Office of Data Science and IT for Marshall Space Flight Center in NASA in Huntsville, in Alabama; IBM; and a collaborative science team.

The scientific team, assembled by the NASA headquarters, was made up of experts from the Southwest Research Institute in San Antonio, Texas; The University of Alabama in Huntsville in Huntsville, Alabama; The University of Colorado Boulder in Boulder, Colorado; Georgia State University in Atlanta, Georgia; Princeton University in Princeton, New Jersey; The Heliophysical Division of the NASA SMD; Goddard Space Flight Center from NASA in Greenbelt, Maryland; Pasadena NASA jet laboratory, California; And the Seti Institute in Mountain View, California.

For a driving behind the scenes in the architecture, industry and academic collaborations of Surya, the challenges behind the development of the model, read the blog post on the NASA scientific data portal:

https://Science.data.nasa.gov/features-events/inside-surya-solar-ai-model

For more information on NASA’s strategy to develop foundation models for science, visit:

https://Science.nasa.gov/artificial-intelligence-science