As on earth, there are also storms in space. While storms on earth are tangible, space storms are less evident as they consist of charged particles. The effects of these storms can be pleasant – geomagnetic storms are responsible for the northern lights – but can also pose serious problems like blackouts, failing satellites, and health threats for astronauts and people on high-altitude flights. For reference, in February 2022, 38 commercial satellites crashed back to earth due to a solar storm. Assuming total cost to get a satellite deployed of USD 500.000, this solar storm cost USD 19.000.000 in addition to the connection issues satellite users experienced.

Like the weather on earth, researchers built models to predict these solar events state-of-the-art models are based on physics models and require an entire supercomputer to run. Researchers from Japan now have created a machine learning (ML) emulator based on those physics models that can run a space weather forecast much quicker (a million times quicker to be exact) and, theoretically, on a simple laptop.

Article: Machine Learning‐Based Emulator for the Physics‐Based Simulation of Auroral Current System – Kataoka – 2024 – Space Weather – Wiley Online Library