Early warning on earthquakes can help save lives, but many countries can't afford them. That's why scientists are turning to another location sensor already widespread in many countries: the smartphone. A single smartphone makes for a crappy earthquake sensor — but get enough of them reporting, and it won't matter.
A new study, published today in Science Advances, says that the right network of cell phones might be able to substitute for modern seismograph arrays, providing a crucial early warning in the event of a quake. The study looks at historical earthquake data and modern smartphone hardware (based on the Nexus 5) and comes away with a map of how a smartphone-based earthquake detector might work. As it turns out, a phone's GPS is more powerful than you might think.
A modern phone has almost everything you could want in an earthquake sensor
Early warning systems are designed to pick up the first tremors of an earthquake, projecting where the incoming quake is centered and how strong it's likely to be. When they work, the systems are able to give citizens and first responders crucial time to prepare for the quake. There are already seismograph-based systems in place in California, Mexico, and Japan, but poorer countries often don't have the means to implement and maintain them. This new method wouldn’t be as good as most scientific earthquake sensors, but those can cost tens of thousands of dollars each, making a smartphone-based sensor a lot cheaper. For countries that can’t afford a seismograph-based system (which includes much of the Southern Hemisphere), it could make a crucial difference in catching quakes early.
A modern phone has almost everything you could want in an earthquake sensor: specifically, a GPS-powered location sensor, an accelerometer, and multiple data connections. There are also a lot of them, even in poor countries, so a distributed system could count on getting data points from multiple angles. Of course, a phone moves around a lot more than a traditional earthquake sensor, but the hope is that what phones lose in quality, they can make up for in quantity. Given that there will likely be millions of phones in the vicinity of a quake, researchers hope there will be a few thousand that will be in a position to serve as earthquake sensors. Even with less accurate GPS sensors, those phones could still be very good at picking up geological shifts. "People are used to thinking of GPS as location," Minson says, "but it turns out that the change in position of your phone from [GPS] is very, very accurate."
What phones lose in quality, they can make up for in quantity
So the researchers set out to see if it was possible to create the cell phone warning system. They looked at two earthquakes: one real and one hypothetical. "In an ideal world, we'd test it by looking at cellphone observations of earthquakes," says USGS researcher Sara Minson. "Unfortunately, those data don't exist."
Instead, the team looked at early warning data from the 2011 Tohoku earthquake in Japan, which was detected by 462 GPS-based earthquake sensors in the surrounding area. Of course, those sensors were working with more sophisticated instruments than you'd find in a smartphone. So the researchers removed data until these sensors’ readings were as crude as those from smartphones’ GPS and accelerometers. The result was a noisier data set, but one that could still detect an earthquake just 77 seconds after the first tremors, using the right algorithms.
That simulation still assumes those 462 phones are behaving like good earthquake detectors, lying flat on a surface in clear view of a GPS satellite. So the researchers simulated a Magnitude 7 earthquake on the Hayward Fault near San Francisco, similar to the quake that hit the region during the World Series in 1989. Researchers used population maps to project how many phones would be in range, and they compared that with how much data would be needed to sense the quake. If just under 4,700 phones — that is, 0.2 percent of the units in the Bay Area — were actively reporting data, the earthquake would be reported only five seconds after the initial tremor.
"We really want to prove this can be done with smartphones."
That's not necessarily good news, as researchers are quick to admit, since it’s still unclear how many phones would really participate. Even if an earthquake-sensing app became widely adopted, it's not obvious what percentage of phones would be actively reporting at a given time. Most phones only check GPS intermittently, usually when there’s a clear view of the sky, which helps conserve the battery while keeping the phone’s location roughly up to date. To work as an earthquake sensor, a phone would have to check GPS more frequently, which would drain the battery faster and could scare users off the app entirely. As a result, the percentage of actively reporting phones might easily be less than 0.2 percent of the total population.
A bigger problem comes from how phones typically manage that data. When an Android or iOS app calls for the phone's location, the app gets a smoothed version of the GPS, which automatically combines GPS data with other inputs for a less jumpy estimate of where you are. USGS wants the raw data from the phone's GPS chip, but because of the way Android and iOS manage location data, there’s no clear way for an application to grab the raw data before it’s smoothed out. It might be possible with a jailbroken iPhone or a custom Android fork, but USGS is still muddling through the programming challenges involved.
In the meantime, the easiest solution is simply bundling a new GPS chip alongside the phone. The team's next move is a pilot project in Chile, which will build an early warning system out of commercially available phones bundled with an external GPS chip. The country doesn’t have an early warning system, so a working system based on cell phones could serve a crucial purpose, even if it doesn’t work quite as well as the more expensive seismograph-based systems. Each phone will be entirely dedicated to earthquake sensing, so it won't face all of the challenges of an earthquake-sensing app, but the researchers hope it will help to solve a number of the intermediate challenges. "We really want to prove this can be done with smartphones," Minson says.