Earthquake prediction with machine learning shows promise
Using a 2D tabletop simulator (image, as seen through a polarized lens) that models the build-up and release of stress along an artificial fault, the researchers demonstrated that by listening to the acoustic signal emitted by a laboratory fault, machine learning can predict the time remaining before it fails with “remarkable” accuracy.
“At any given instant, the noise coming from the lab fault zone provides quantitative information on when the fault will slip,” says Paul Johnson, a Los Alamos National Laboratory fellow and lead investigator on the research. “The novelty of our work is the use of machine learning to discover and understand new physics of failure, through examination of the recorded auditory signal from the experimental setup. I think the future of earthquake physics will rely heavily on machine learning to process massive amounts of raw seismic data.”
In addition to earthquake forecasting, say the researchers, their work could be applicable to nondestructive testing of industrial materials, avalanches, and other events.
One surprising result of their work, the researchers found, was that machine learning identified a signal from the “fault zone” that had previously been thought to be low-amplitude noise as able to provide forecasting information throughout the laboratory quake cycle.
“These signals resemble Earth tremors that occur in association with slow earthquakes on tectonic faults in the lower crust,” says Johnson. “There is reason to expect such signals from Earth faults in the seismogenic zone for slowly slipping faults.”
The researchers analyzed data from a laboratory fault system that contains fault gouge – ground-up material created by the stone blocks sliding past one another. An accelerometer was used to record the acoustic emission emanating from the shearing layers.
“As the material approaches failure, it begins to show the characteristics of a critical stress regime, including many small shear failures that emit impulsive acoustic emissions,” says Johnson. “This unstable state concludes with an actual lab quake, in which the shearing block rapidly displaces, the friction and shear stress decrease precipitously, and the gouge layers simultaneously compact.”
The ability to predict failure for an artificial fault in the lab, say the researchers, gives hope that prediction of actual earthquakes is possible. In fact, they have begun working with real data and are researching ways to isolate the signal that identifies when an earthquake is about to occur.
For more, see “Machine learning predicts laboratory earthquakes.”
Related articles:
IoT friendly seismic sensor helps prevent secondary damages
Magnetic field tech looks to ‘hover’ buildings in earthquakes
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
