Satellites at Risk? New AI Predicts Space Weather With Breakthrough Accuracy

By New York University December 2, 2025

Collected at: https://scitechdaily.com/satellites-at-risk-new-ai-predicts-space-weather-with-breakthrough-accuracy/

Researchers created an AI system that can predict solar wind conditions up to four days before they reach Earth, offering far greater accuracy than current methods.

By training the model on ultraviolet images of the Sun from NASA’s Solar Dynamics Observatory, the team uncovered patterns that help forecast potentially dangerous space weather.

AI Advances Solar Wind Prediction

Scientists at NYU Abu Dhabi (NYUAD) have created a new artificial intelligence (AI) system capable of predicting solar wind speeds up to four days before they reach Earth. The research, published in The Astrophysical Journal Supplement Series, reports a level of accuracy that clearly surpasses existing forecasting techniques.

Why Solar Wind Forecasting Matters

Solar wind consists of a constant flow of charged particles streaming outward from the Sun. When this flow intensifies, it can trigger space weather events that disturb Earth’s atmosphere, push satellites off their planned orbits, damage their electronics, and disrupt power systems on the ground. A dramatic example occurred in 2022, when a powerful solar wind surge caused SpaceX to lose 40 Starlink satellites, underscoring the importance of early warnings.

AI Trained on NASA Solar Images

The NYUAD research team, led by Postdoctoral Associate Dattaraj Dhuri and Co-Principal Investigator at the Center for Space Science (CASS) Shravan Hanasoge, developed the AI by pairing high-resolution ultraviolet (UV) images from NASA’s Solar Dynamics Observatory with historical solar wind measurements.

Unlike widely used AI language models that process text, this system studies detailed images of the Sun to detect visual patterns linked to changes in the solar wind. Their approach increases forecasting accuracy by 45 percent compared to operational models currently in use, and by 20 percent when measured against earlier AI-driven methods.

Improving Protection for Technology on Earth and in Orbit

“This is a major step forward in protecting the satellites, navigation systems, and power infrastructure that modern life depends on,” said Dhuri, lead author of the study. “By combining advanced AI with solar observations, we can give early warnings that help safeguard critical technology on Earth and in space.”

Strengthening Preparedness for Space Weather

This progress highlights the growing role of AI in solving one of space science’s most persistent challenges: anticipating the behavior of the solar wind. More dependable predictions allow scientists and engineers to better prepare for upcoming space weather events, improving the resilience of essential systems and infrastructure worldwide.

Reference: “A Multimodal Encoder–Decoder Neural Network for Forecasting Solar Wind Speed at L1” by Dattaraj B. Dhuri, Shravan M. Hanasoge, Harsh Joon, Gopika SM, Dipankar Das and Bharat Kaul, 8 September 2025, The Astrophysical Journal Supplement Series.
DOI: 10.3847/1538-4365/adf436