AI quake prediction may no longer be pipedream

Nov 02, 2017, 7:14 AM EDT
(Source: Martin Luff/flickr)
(Source: Martin Luff/flickr)

The elusive science of predicting the timing and magnitude of earthquakes could soon be within our reach as researchers experiment with artificial intelligence to crack the code of this holy grail of seismology.

Scientists at the Los Alamos National Laboratory and the University of Cambridge are employing machine learning to analyze the subtle “creaking and grinding” of the fault zones between two quakes, to figure out when the underground forces would trigger another rupture, notes Seeker.

Previous efforts to foretell earthquakes have focused on observing foreshocks, disturbances in electromagnetic fields and unusual animal behavior, but to no fruitful results, writes Scientific American. Recent advances in machine-learning algorithms and the ability of supercomputers to process huge amounts of data hand more power to ongoing experiments.