Here Comes the Sun: USU Computer Scientists Develop Models to Predict Extreme Solar Phenomena

During this past fall, the sun emitted several strong solar flares — sudden, powerful bursts of radiation from its surface. Each of these, on Sept. 14, Oct. 31 and Nov. 6, were classified as intense flares capable of affecting earthly activities.

Previous solar storms occurred in 1859, 1972, 1989, 2000, 2003, 2006 and 2022, pelting the Earth in varying degrees and reminding earthlings of the sun’s intense power and potential for disruption. Communications systems, ranging from telegraph networks in 1859 to Starlink satellites in 2022, were among critical equipment bearing the brunt of the sun’s wrath in each episode.

“Solar activity can cause disruptions, ranging from minor inconveniences to dangerous outages on Earth,” says Utah State University computer scientist Soukaina Filali Boubrahimi.

Her USU colleague Shah Muhammad Hamdi says such activity also poses hazards to humans.

“Solar activity creates a radiation exposure hazard for aircraft travelers as well as astronauts aboard the International Space Station,” he says.

In 1989, they note, a geomagnetic storm from the sun caused a nine-hour outage of Canada’s Hydro-Québec electricity transmission system.

“We depend on satellites for so many of our daily activities, including using mobile phones, surfing the Web and TV viewing, as well as our energy, defense and space systems,” says Filali Boubrahimi, assistant professor in USU’s Department of Computer Science.

“We need the ability to better predict solar activity and prepare for these potentially harmful conditions,” says Hamdi, also an assistant professor in the Department of Computer Science.

To that end, Filali Boubrahimi and Hamdi are leading efforts to develop novel machine learning models to better predict solar flares and thereby enable Earth’s inhabitants and space travelers to better prepare for the sun’s stormy behavior. In 2023, Filali Boubrahimi was awarded a National Science Foundation Faculty Early Career Development CAREER award to pursue this challenge. A sole principal investigator, Filali Boubrahimi is the recipient of a five-year, $691,972 CAREER grant to support her project, End-to-End Active Region-based Heliospheric Forecasting System using Multi-spacecraft Data and Machine Learning.

In 2023, Hamdi was awarded a National Science Foundation SHINE (Solar, Heliospheric, and INterplanetary Environment) award to pursue similar challenges. A sole principal investigator, Hamdi is the recipient of a three-year, $437,703 SHINE grant to support his project, Understanding the Relationships of Photospheric Vector Magnetic Field Parameters in Solar Flare Occurrences using Graph-based Machine Learning Models.

While NSF-funded solar observatories, including Big Bear Solar Observatory, and scientific agencies such as NASA and NOAA have honed observations of the sun’s activity cycles, significant barriers to accurate prediction of solar flares remain. To address these challenges, Filali Boubrahimi and Hamdi, and their graduate students Omar Bahri, MohammedReza EskandariNasab, Pouya Hosseinzadeh, Peiyu Li, and Onur Vural are developing a comprehensive forecasting system that leverages data from multiple spacecraft and employs machine learning techniques to predict solar flares.

With the CAREER grant, the team of Filali Boubrahimi is developing a high-spatial resolution active regions vector magnetogram dataset, spanning two solar cycles and based on real data collected from NASA’s space-borne Solar Dynamics Observatory (SDO) and Solar and Heliospheric Observatory (SOHO), as well as Japan’s Hinode satellite.

With the SHINE grant, Hamdi’s team investigates the connections between photospheric vector magnetic field parameters and the occurrence of solar flares. By utilizing graph-based machine learning models, the research seeks to uncover patterns and relationships that could improve the understanding and prediction of solar flare events.

“Our goal is to create a comprehensive solar flare catalog combing images and data, and to leverage it to develop predictive, data-driven models ready to be deployed in an operational environment,” Filali Boubrahimi says. “We will share our findings with the scientific community in important annual meetings, including the NSF’s Solar, Heliospheric, and INterplanetary Environment workshop,” adds Hamdi.

Filali Boubrahimi’s CAREER grant bolsters funding from a three-year NSF grant she received in 2022. Hamdi’s SHINE grant bolsters funding from a two-year NSF CRII grant he received in 2022. In addition to providing research opportunities for graduate and undergraduate students in their labs, they plan to continue offering learning activities for participants in USU’s Native American Summer Mentorship Program.

They say mining the solar activity datasets is timely, because NASA and NOAA’s Solar Cycle Prediction Panel declared the beginning of Solar Cycle 25 in December 2019, starting with a solar minimum — or time of lessened solar activity.

“The panel foresees this current cycle will ramp up toward a solar maximum in July 2025,” Filali Boubrahimi says.

“By that time, we hope to have our models ready for high solar activity,” Hamdi adds.