When applied to the Long Beach area data sets, the algorithms detected significantly more earthquakes and made it easier to figure out how and where they started. And when applied to data from a In the 2014 La Habra earthquake, also in California, the team observed four times more seismic detections in the “murmured” data compared to the officially recorded number.
It’s not the only work applying AI to earthquake hunting. Penn State researchers have trained in-depth algorithms to accurately predict how changes in measurements might indicate upcoming earthquakes— a task that has puzzled experts for centuries. And members of the Stanford team have previously trained models for phase-picking, or measuring the arrival times of seismic waves within an earthquake signal, which can be used to estimate the earthquake’s location.
In-depth algorithms are particularly useful for earthquake monitoring because they can offload human seismologists, said Paula Koelemeijer, a seismologist at the Royal Holloway University of London who was not involved in the study.
In the past, seismologists would look at graphs produced by sensors that record the movement of the ground during an earthquake, and they would identify patterns by sight. Deep learning could make that process faster and more accurate by helping to cut through large amounts of data, says Koelemeijer.
“Show that” [the algorithm] working in a noisy urban environment is very beneficial, because noise in urban environments can be a nightmare to deal with, and very challenging,” she says.