In scientists vs. earthquakes, can AI tip the scales to the scientists?

I don’t know about you, but earthquakes are a pretty terrifying phenomenon to me. And if they aren’t to you, just have a gander at the Pulitzer Prize-winning piece about the ‘really big one’ that will decimate the Pacific Northwest someday and tell me you’re not at least a little bit terrified. But one of the oft overlooked — or simply uniformed — terrors of earthquakes is that there’s rarely just the earthquake; there’s almost always an aftershock (or shocks) that can be nearly as large as the initial quake, and sometimes far more deadly.

So what does all this have to do with you? It’s another real-world example of how A.I. is changing the hard sciences for the better, with potentially thousands of lives that could be positively impacted (or outright saved) by it.

The problem with aftershocks

The scene is Christchurch, New Zealand; a magnitude-7.1 earthquake struck near the city in September of 2010, but luckily didn’t kill anyone. But more than 5 months later, a magnitude-6.3 aftershock hit closer to the city centre and ended up causing 185 deaths.

Scientists have gotten relatively advanced at predicting the scale of aftershocks, but according to the journal Nature, it’s incredibly difficult to forecast where the quakes will happen: “Until now, most scientists used a technique that calculates how an earthquake changes the stress in nearby rocks and then predicts how likely that change would result in an aftershock in a particular location. This stress-failure method can explain aftershock patterns successfully for many large earthquakes, but it doesn’t always work.”

A.I. to the rescue?

The good news for seismologists struggling with this particular problem? Earthquakes produce a lot of data… and we’ve been watching and recording for a long period of time now. And the more data you can feed a sophisticated neural network, the better it can discern patterns and calculate results. And that’s just what the scientists did. They looked at more than 130,000 mainshock and aftershock events, including some of the most powerful in our recorded history (like the devastating magnitude-9.1 event that hit Japan in March 2011).

Here’s how Nature describes the method and results:

“The researchers used these data to train a neural network that modelled a grid of cells, 5 kilometres to a side, surrounding each main shock. They told the network that an earthquake had occurred, and fed it data on how the stress changed at the centre of each grid cell. Then the scientists asked it to provide the probability that each grid cell would generate one or more aftershocks. The network treated each cell as its own little isolated problem to solve, rather than calculating how stress rippled sequentially through the rocks.

“When the researchers tested their system on 30,000 mainshock-aftershock events, the neural-network forecast predicted aftershock locations more accurately than did the usual stress-failure method. Perhaps more importantly, DeVries says, the neural network also hinted at some of the physical changes that might have been happening in the ground after the main shock. It pointed to certain parameters as potentially important — ones that describe stress changes in materials such as metals, but that researchers don’t often use to study earthquakes.

So not only can the A.I. better predict the location of aftershocks, it also ascertained a novel and creative way of looking at seismic effects of earthquakes that researchers hadn’t really employed previously.

This is a perfect example at the types of big, important, impactful problems A.I. is well suited to solve (or at least help solve). There are so many real-world applications for smart implementation of A.I., some of which could save thousands of lives if done correctly.

That’s why we’re so excited to work (and write) in this field; if we get it right, it really could change the world for better and for always.