Abu Dhabi team train AI system to help save Earth satellites and power grids from solar storms

Model uses Nasa data to forecast powerful solar winds days in advance

Solar winds are streams of charged particles released by the Sun that can intensify into storms. Photo: NYUAD

Abu Dhabi scientists have developed a powerful artificial intelligence model that can predict the speed of solar winds up to four days ahead.

The model developed at the New York University Abu Dhabi (NYUAD) will help Earth prepare for solar storms that threaten satellites, navigation systems and power grids.

Their study, published this month in The Astrophysical Journal Supplement Series on September 8, explains how the AI system was trained to read high-resolution ultraviolet images of the Sun so it can identify patterns of solar winds.

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

The Northern Lights visible through clouds over Saltburn-by-the-Sea in North Yorkshire, England. PA

Why is it important to forecast solar winds?

Solar winds are continuous streams of charged particles released by the Sun that can sometimes intensify into powerful storms. And when these bursts collide with the Earth’s magnetic field, they create remarkable displays such as the Northern Lights.

The university’s model uses images from Nasa’s Solar Dynamics Observatory along with solar wind records. It then looks for small changes on the Sun’s surface that hint when the solar wind might pick up.

Findings in the study showed that there was a 45 per cent improvement in forecast accuracy by the AI model compared to existing models. There was also a 20 per cent improvement over earlier AI-based approaches.

Researchers want to forecast these events because of the potential damage they can cause to infrastructure and satellites.

Elon Musk’s SpaceX lost 40 of its Starlink satellites in 2022 when intense solar wind created a drag in the atmosphere, causing the devices to be pulled out of orbit.

In 1989, a geomagnetic storm caused by solar wind triggered a blackout in Quebec, Canada, plunging millions into darkness for 12 hours.

The most powerful event on record, the Carrington Event of 1859, sparked fires at telegraph stations and disrupted global communications. It is named after Richard Carrington, an English astronomer who, with colleagues, assessed the phenomenon from stations across London and the south of England.

Nasa has said that if a storm of that magnitude struck today, its effects could lead to crippled economies as power is disrupted and communication systems across continents collapse.

Dagger at sharp end of today's forecasts

The NYUAD’s AI model adds momentum to a global trend of using AI to forecast conditions in space.

Nasa released a model in 2023 called the Deep Learning Geomagnetic Perturbation, or Dagger, which can issue warnings of solar storms 30 minutes before they strike Earth. Dagger uses AI and real-time satellite data to predict geomagnetic disturbances worldwide, producing updates every minute.

Nasa’s model offer’s critical short-term alerts, while the NYUAD system extends the forecast window to four days. That gives satellite operators, telecom companies and energy companies more time to implement safety measures.