Researchers at Los Alamos National Lab find that AI could help predict the timing of a quake.

Scientists have long held the view that earthquakes are impossible to predict, being random events. But the potential for damage these natural events have, including the loss of life, have led researchers to explore whether that could be changed using technology such as AI and machine learning.

Scientists at Los Alamos National Lab in New Mexico, led by Dr Paul Johnson, a staff scientist at the lab’s geophysics group, conducted experiments that attempted to predict the timing of a quake, rather than its magnitude, location or time. The test was surprisingly successful.

Simulated quakes

The scientists used an earthquake simulation machine at Penn State to create an artificial tremor. They then fed copious amounts of data from this simulated quake into a machine learning model.

When they ran the model to see its efficacy they were surprised. “It worked almost immediately,” Johnson said. “And we were just stunned at how quickly it worked and how easily it worked on a problem that we had struggled with for so long.”

Real world, slow-slip quakes

The scientists then applied their model to a kind of real-world tremor known as a “slow slip earthquake.”

These imperceptible earthquakes happen when one tectonic plate impacts another over extended periods of time, even as long as a month.

In February 2011, two such slow slip earthquakes started creeping along the Japan Trench and a month later, set off a magnitude 9.0 earthquake that shook Japan for nearly six minutes. It triggered a tsunami and a nuclear disaster that together took over 20,000 lives.

The Los Alamos scientists then tested their findings in the Pacific Northwest on these silent quakes.

They found that the AI-driven computers could successfully predict the next slow-slip right after the previous one.

But forecasting killer quakes is still a long way off

However, it is still not possible for scientists to predict the tragic real-life mega quakes that strike the world with tragic loss of life.

“For seismogenic earthquakes, the ones we really care about as citizens, we’re not there yet,” Johnson said.

Related Story: Fugaku, The World’s Fastest SuperComputer Predicts Tsunami Flooding