NYU Abu Dhabi Scientists Unveil AI System to Predict Solar Winds Days in Advance

In 2025, scientists at New York University Abu Dhabi (NYUAD) unveiled a groundbreaking artificial intelligence system capable of predicting harmful solar winds up to four days in advance. Unlike conventional AI models that rely on text data, this system analyzes high-resolution images of the Sun, offering unprecedented accuracy in forecasting space weather events. The findings, published in The Astrophysical Journal Supplement Series, mark a 45% improvement over existing forecasting models and a 20% improvement compared to earlier AI-based attempts.

Why Solar Wind Forecasting Matters

Solar winds, streams of charged particles released from the Sun’s outer atmosphere, can disrupt satellites, GPS navigation, aviation communication, and even terrestrial power grids. In severe cases, solar storms have caused blackouts and satellite malfunctions, costing billions of dollars. A longer warning window provides critical time for space agencies, power operators, and telecommunications networks to protect vulnerable infrastructure.

A Multimodal AI Model: Images + Data

The NYUAD team developed a multimodal encoder-decoder neural network that combines two types of input:

• Ultraviolet images of the Sun captured by NASA’s Solar Dynamics Observatory (SDO)
• Historical solar wind speed measurements recorded at the L1 Lagrange point, the key monitoring location between Earth and the Sun

By learning patterns in solar activity from both images and data, the AI model identifies subtle precursors to solar wind acceleration. This dual approach significantly outperforms physics-based models that often struggle with the Sun’s complex magnetic dynamics.

Practical Applications: Safeguarding Earth and Space Assets

The new AI forecasting model could transform global space weather monitoring efforts:

• Satellite Protection: Operators can activate safe modes or adjust orbits to minimize damage.
• Navigation Systems: GPS reliability improves for aviation, shipping, and defense applications.
• Power Grids: Utilities can prepare for potential geomagnetic disturbances, reducing risks of large-scale outages.
• Space Exploration: Missions to the Moon and Mars will benefit from advanced warnings of high-radiation events.

Researchers emphasize that this technology represents a major step toward operational space weather forecasting, especially as solar activity intensifies during the ongoing solar maximum cycle.

Challenges and Future Research

While highly promising, the system faces challenges. Forecasting extreme solar events like coronal mass ejections (CMEs) remains complex, and the model’s performance across different solar cycles requires further validation. Additionally, making AI models more interpretable ,  so scientists understand why certain predictions are made, is a priority for building trust in real-world applications.

Nevertheless, the NYUAD breakthrough demonstrates how AI can extend humanity’s predictive capabilities beyond Earth, offering a critical defense layer for both space-based assets and ground infrastructure.