With a $593,864 grant, the three-year project, led by New Jersey Institute of Technology (NJIT)’s Yan Xu at NJIT's Institute for Space Weather Sciences (ISWS) and Jason Wang at the university’s Ying Wu College of Computing, will develop AI-powered space weather forecasting capabilities that could offer solar researchers new insights into the complex magnetic processes in regions of the Sun's atmosphere that trigger such eruptions.
According to the researchers, the new SolarDM AI-powered forecasting system could boost early-warning detection of these eruptive events on Earth by days, while offering vital insights to the space weather science community as activity on the Sun ramps up over the course of the current 11-year solar cycle, which began in 2019.
“Major solar eruptions are powered by magnetic processes taking place in the solar corona, where we’ve lacked critical data due to poor observation conditions and insufficient instruments,” Xu, the project’s principal investigator, said in a statement. “Observations of the atmospheric layer underneath are crucial to study 3D magnetic fields. SolarDM’s data insights potentially give us a way to map the magnetic landscape of this region, allowing us to better predict these powerful eruptions."
Solar physicists have long studied the structure and evolution of magnetic fields in the corona (the Sun’s upper atmosphere). According to NJIT, the breaking and reconnecting of these field lines are known to power explosive events capable of disrupting technologies on Earth, such as satellite operations.
However, challenges persist in observing the magnetic field conditions in the second layer of the Sun’s atmosphere, the chromosphere, a rarely visible region positioned above the lowest layer of the star’s atmosphere, the photosphere.
To address this, the NJIT team is leveraging advanced artificial intelligence to generate synthetic vector magnetograms - computer-generated images of magnetic field dynamics in the photosphere and chromosphere - providing data that could shed light on the precursors to solar eruptions.
The SolarDM AI system will be trained using simulations of the Sun's magnetic field and observational data from NSF's Synoptic Optical Long-term Investigations of the Sun (SOLIS) — one of the world’s most advanced solar telescopes for long-term monitoring of the Sun, currently stationed at NJIT’s Big Bear Solar Observatory. In addition, data from NASA’s missions will be used to augment the training set.
“Due to the differences between the instruments on board the ground-based and space-borne observatories, it is extremely challenging to obtain high-quality alignments of the data needed for training and testing the AI system,” said Wang. “The forecast horizon of state-of-the-art solar eruption forecasting systems is 24 hours. If successful, with SolarDM’s generated vector magnetograms, it is expected that the new AI system can extend the forecast horizon from 24 hours to three days.”
Xu and Wang said the AI modelling system will use the data to predict when and where eruptions are likely to occur across millions of miles of the solar atmosphere, and explain why it has arrived at those conclusions.
“Insights into why the AI model is making its forecasts could significantly enhance our understanding of the underlying physics that are behind these powerful events,” said Xu.
Promoted content: Does social media work for engineers – and how can you make it work for you?
So in addition to doing their own job, engineers are expected to do the marketing department´s work for them as well? Sorry, wait a minute, I know the...