The Sun, a constant yet unpredictable presence, continues to challenge scientists with its complex behavior. To address this, IBM and NASA have unveiled Surya, a foundation model specifically designed to decode solar phenomena. Announced on August 20, Surya combines advanced artificial intelligence with years of heliophysics data to improve predictions of solar flares and storms. These bursts of electromagnetic energy pose risks to astronauts, satellites, communication systems, and power grids on Earth, making accurate forecasting vital for both science and society.
Surya’s training draws on nine years of observations from NASA’s Solar Dynamics Observatory (SDO). This satellite has monitored the Sun since 2010, capturing high-resolution images every 12 seconds and collecting data across multiple wavelengths. These measurements reveal temperature variations in the solar atmosphere and track magnetic field dynamics, both crucial for understanding how energy flows through the star. Yet, the massive scale and complexity of this information have historically overwhelmed researchers. Surya addresses this by transforming disparate datasets into a unified framework, enabling more consistent and efficient analysis.
At the heart of the innovation is the creation of a solar digital twin—a dynamic, virtual representation of the Sun that updates continuously as new data arrives. This model allows scientists to experiment with simulations, manipulate solar conditions, and study outcomes in ways that physical observation alone cannot provide. To achieve this, Surya integrates a long-range vision transformer capable of analyzing very high-resolution imagery, detecting relationships across distant regions of the Sun’s surface and atmosphere. Enhancing this system is a spectral gating mechanism that reduces memory demands and filters out noise, ensuring higher-quality results.
Unlike traditional algorithms that rely heavily on labeled datasets, Surya learns directly from raw data. This flexibility makes it adaptable for a range of predictive tasks, such as assessing solar wind patterns or identifying regions prone to flare activity. During testing, Surya incorporated observations from additional spacecraft, including the Parker Solar Probe and the Solar and Heliospheric Observatory (SOHO), further expanding its versatility. Results demonstrated that the system improved solar flare classification accuracy by 16 percent and provided a two-hour warning before flare onset—twice the lead time offered by conventional models. This breakthrough highlights the practical value of the solar digital twin in safeguarding critical technology.
NASA emphasizes that while the model was developed for heliophysics, its architecture is broad enough to benefit other domains, including planetary science and Earth observation. Kevin Murphy, NASA’s director of data science, noted that Surya’s design enables researchers to analyze solar activity with unprecedented speed, delivering insights that directly impact Earth’s technological systems. By extending predictive capacity, Surya reduces the risks associated with geomagnetic storms, which could otherwise cripple telecommunications, disrupt GPS, and destabilize electrical grids worldwide.
Lead scientist Andrés Muñoz-Jaramillo underscores the importance of maximizing lead time for potential solar disasters. He explains that the project’s goal is to capture essential processes in solar evolution and translate them into actionable knowledge. In this sense, Surya not only advances the science of heliophysics but also strengthens global resilience. By merging artificial intelligence with continuous data streams, the solar digital twin becomes a critical tool in predicting and mitigating the Sun’s impacts on Earth. As the model evolves, it may also set a precedent for how AI and digital twin technologies can revolutionize the study of other planetary bodies.
Ultimately, Surya exemplifies how cutting-edge computation can transform raw space data into predictive power. With its adaptive learning, integration of multiple instruments, and innovative design, the solar digital twin offers humanity a longer warning window against solar threats—an achievement that could prove indispensable in an increasingly technology-dependent world.