Mobile phones to predict earthquakes? Researchers have an idea

One of the hardest natural disasters to prepare for, and predict, are earthquakes. Technology exists that can help predict when earthquakes might occur, but they are very expensive to build out and maintain.

However, a group of scientists published a report that believe the technology contained in our smartphones is enough to develop an ad hoc, crowdsourced, earthquake early warning system.

At first glance, the idea makes perfect sense, as our smartphones already have just about everything needed to provide data needed for a central system to analyze: GPS, accelerometer, and data connectivity. On top of that, there are far more smartphones in use than there are industrial-grade seismic sensors.

The basis of the technology is already in use, but the sheer quantity of “consumer” data presents challenges that amount to being able to filter out the noise of less precise instruments, and identify false alarms. However, the sheer number of available devices may make that task a little simpler than it sounds.

The way early warning systems work is through picking up tremors that occur before the big shake-up. From those readings, projections can be made to predict where an earthquake will happen.

Researchers looked at data from GPS connected devices in Japan from the 2011 Tohoku earthquake that created a massive tsunami. Filtering through the noise to data-sets that were as “crude” as consumer devices, they found accurate earthquake readings within about one minute of the first tremors of the quake. That is pretty impressive given a sample size of about 460 phones.

Creating a similar model for the San Francisco area, the scientists determined that a warning could have been issued within 5 seconds of the first tremors to hit during the 1989 World Series.

A pilot project is going to be performed in Chile. The goal is to build an early warning network out of commercially available smartphones. These devices will be used solely for the purpose of earthquake detection, a first step toward seeing how feasible it would be to work with data from devices enduring the rigors of consumer ownership.

sources: Science Analysis via The Verge