UNH researchers categorize aurora image database using artificial intelligence
by Clarence Oxford
Los Angeles CA (SPX) Jan 10, 2025
The aurora borealis, or northern lights, is famend for its mesmerizing show of sunshine within the night time sky. Nevertheless, this pure phenomenon-driven by photo voltaic exercise and carried by photo voltaic winds-can disrupt essential communications and safety infrastructure on Earth. Researchers on the College of New Hampshire (UNH) have leveraged artificial intelligence to categorize and label an unlimited database of aurora photographs. This revolutionary effort goals to reinforce scientific understanding and enhance forecasting of geomagnetic storms.
The examine, lately featured within the Journal of Geophysical Analysis, utilized superior artificial intelligence and machine studying instruments to course of over 706 million auroral photographs from NASA’s Time Historical past of Occasions and Macroscale Interactions throughout Substorms (THEMIS) mission. THEMIS, comprising twin spacecraft, captures photographs of the night time sky each three seconds from 23 stations throughout North America, offering an unparalleled view of auroral exercise.
“The huge information set is a invaluable useful resource that may assist researchers perceive how the photo voltaic wind interacts with the Earth’s magnetosphere, the protecting bubble that shields us from charged particles streaming from the solar,” stated Jeremiah Johnson, affiliate professor of utilized engineering and sciences and the examine’s lead writer. “However till now, its big measurement restricted how successfully we will use that information.”
To deal with this problem, the researchers developed an revolutionary algorithm to effectively kind and label THEMIS all-sky photographs (ASI) collected between 2008 and 2022. The algorithm categorized photographs into six distinct teams: arc, diffuse, discrete, cloudy, moon, and clear/no aurora. This systematic annotation makes it simpler for scientists to filter, kind, and analyze the info.
“The labeled database might reveal additional perception into auroral dynamics however at a really primary stage, we aimed to arrange the THEMIS all-sky image database in order that the huge quantity of historic information it incorporates can be utilized extra successfully by researchers and supply a big sufficient pattern for future research,” stated Johnson.
The hassle not solely organizes a beforehand unwieldy dataset but in addition gives researchers with instruments to unlock new insights into how auroral phenomena evolve over time and reply to photo voltaic occasions.
Analysis Report:Automatic Detection and Classification of Aurora in THEMIS All-Sky Images
Associated Hyperlinks
University of New Hampshire
Solar Science News at SpaceDaily