Earthquakes are one of the most destructive natural disasters. The 2011 Tōhoku earthquake and tsunami cause Japan a damage of three hundred and sixty Billion USD. Earthquakes damage doesn’t end when the ground stop shaking, actually, it is the start of another disaster called aftershock.
Aftershocks are a smaller earthquake. They are the lower in magnitude but cause severe damage than the primary earthquakes. Aftershocks start to form as the crust around the displaced fault plane start to adjusts, they mostly came in the first few hours and days that follow the previous earthquake.
Though scientists have got success in predicting the size and timing of these aftershocks but finding it’s exact location is still an impossible task. Now that could change, as researchers from Google and Harvard University have come up with a pretty interesting solution.
Researchers from Harvard University and Google’s AI division have created an Artificial Intelligence system that comes up with the best way to predict where future aftershocks will occur. In a paper published in the journal Nature this week, researchers show how deep learning algorithms can analyze a database of earthquakes worldwide to predict where aftershocks might happen.
“Aftershock forecasting, in particular, is a challenge that’s well-suited to machine learning because there are so many physical phenomena that could influence aftershock behavior and machine learning is extremely good at teasing out those relationships,” Phoebe DeVries, a post-doctoral fellow at Harvard University and co-author on the study, said in a statement.
The researchers trained this AI on more than 131,000 mainshock–aftershock pairs, then selected unrelated 30,000 pairs for a test. In their test, they found that the artificial intelligence system was more reliable at predicting the site of aftershocks with much more accuracy. On a scale of accuracy running from 0 to 1 — in which 1 is a perfectly accurate model and 0.5 is a faulty score, this new AI system gives the accuracy rate of 0.849.
In the paper, researchers reveal that the one reason behind its enormous accuracy is its use of two complex metrics, that had not previously been thought to be correlated with aftershocks, called maximum shear stress change and the von-Mises yield criterion. Meade tells ScienceDaily, this factor is often used in fields like metallurgy, “but has never been popular in earthquake science.” Now, with the findings of this new model, geologists can investigate its relevance.
The working technique of this AI is pretty similar to the facial recognition system, the difference is just here it analyzes the past data of earthquakes instead of pixel arrangements patterns.
Meade came up with the idea of using artificially intelligent neural networks on aftershocks a few years ago. His next step is to try to predict the magnitude of earthquakes with the help of AI.
“I think there’s a quiet revolution in thinking about earthquake prediction. It’s not an idea that’s totally out there anymore. And while this result is interesting, I think this is part of a revolution in general about rebuilding all of science in the artificial intelligence era” Meade said in a statement.
While the system has shown impressive results, it still has some limitation. This AI model only focuses on aftershocks caused by permanent changes to the ground and ignores the follow-up quakes. The speed of this model is also too slow to work in real-time which make it useless in real-world condition as aftershocks mostly come in the first few hours and days that follow an earthquake.
As Phoebe DeVries, a Harvard postdoc who helped lead the research, told BBC, “We’re quite far away from having this be useful in any operational sense at all. We view this as a very motivating first step.”
You can find this research paper from here.
More in AI :