Our current method of listening for signs of earthquakes has one major problem: cities. They’re loud, with many human-caused vibrations in the ground drowning out the sound of weaker seismic activity. But a deep-learning algorithm can be taught to overcome this problem. A research team from Stanford has developed an algorithm that can filter out city noise in order to detect seismic activity that was previously going unnoticed.
Known as UrbanDenoiser, the algorithm was trained on 80,000 samples of urban seismic noise from Long Beach and 33,751 samples of earthquake noise from San Jacinto, learning to differentiate between the two and ignore the former. In a comparison to data from a 2014 earthquake in La Habra, Calif., UrbanDenoiser detected four times more earthquake sounds than the seismometers at the time.
“Showing that [the algorithm] works in a noisy urban environment is very useful, because noise in urban environments can be a nightmare to deal with, and very challenging,” said Paula Koelemeijer, a seismologist at Royal Holloway University of London.
