While AI poses a significant threat to our livelihoods, some people are actually trying to use it for good, like the scientists using AI to help predict when the northern lights will occur.
Anyone who’s ever tried to catch a glimpse of the northern lights understands the frustration. Sometimes you’ll get lucky and see the shimmering display. Other times, all the conditions are just right, and yet the auroras are a no-show.
A team of researchers from the University of New Hampshire has developed an AI tool capable of sorting over 706 million images taken by NASA’s THEMIS (Time History of Events and Macroscale Interactions during Substorms) satellite between 2008 and 2022.
The aurora borealis occurs when charged particles from the sun, usually released by coronal mass ejections or solar flares, interact with the Earth’s atmosphere, creating a bright, shimmering, colorful display in the sky.
Predicting when the northern lights will be visible has always been a little tricky. We know they become more frequent as solar activity peaks during the sun’s 11-year solar cycle, which is why the auroras have been so frequent in the past year or so. The ability to measure varying conditions to determine, say, “the auroras will be shimmering overhead on Tuesday at 9:15 PM,” has always been a little out of reach. It’s not quite as reliable as trying to predict rainfall.
The new dataset will allow researchers down the line to conduct their own studies to help us better understand and better predict when the northern lights will occur, and where. AI’s ability to wrangle together millions of images in an instant and categorize and cross-reference them based on a wide variety of criteria will help researchers in the years to come to understand this beautiful phenomenon better than we could’ve ever imagined.
The post AI Could Help Us Predict the Northern Lights appeared first on VICE.
