March 7, 2012 at 6:58 PM ET
One year after Japan's earthquake warning system was put to its sternest real-world test, U.S. researchers have built a system that could provide the same type of advance alerts for quake-prone California — the only problem is that they can't afford to get it ready for prime time.
"I've got a system that works in my office," said Thomas Heaton, director of Caltech's Earthquake Engineering Research Laboratory. "It works for maybe 100 of us who are prototyping the system. It's been a grassroots effort where a number of scientists have cobbled it together as a demonstration project. But to turn it into a system where literally 50 million Americans would have everything linked into it? It's not ready for that."
The California network, known as Earthquake Early Warning or ShakeAlert, has been in development since long before the magnitude-9.0 quake and tsunami that swept over Japan last March 11. It operates much like the Japanese network does: Readings from about 400 seismic monitoring stations around California are processed on a real-time basis, and when a quake is detected, computer software figures out how long it will take seismic waves to reach your location.
The system takes advantage of the fact that two types of seismic waves emanate from the epicenter: The first waves to arrive are primary waves, or P waves, which are followed by slower secondary waves, or S waves. The S waves, which travel through Earth's crust at a speed of about 2 miles per second, produce more up-and-down motion and tend to be more damaging. The P waves serve as precursors, enabling experts to estimate the intensity and arrival time for the S waves that will follow.
If the projected intensity is above the level you're worried about, your computer will start sounding an alarm and clicking through a countdown, as seen in the video above.
"Right now it's working as well as you could hope for a kludged-together demonstration project from a bunch of professors," Heaton told me. He can adjust the controls downward to be alerted about minor quakes heading toward Caltech in Pasadena, or turn them up so high he can work undisturbed in his office.
"You can go days without anything, and then a day comes when there's a cluster," he said.
The Japanese system, which was developed at an estimated cost of $500 million, turned in a stellar performance during last year's quake. As the video below demonstrates, Tokyo residents had as much as 30 seconds' warning before the shaking began.
Thirty seconds may not sound like much warning, but it's enough time to shut off gas mains and issue a warning to take cover. In Japan, the warnings are flashed via radio and TV, as well as through computer links and mobile phones. Automated broadcast alerts can be set to turn on a car's emergency flashers and warn drivers to slow down and pull over. The same principle is applied to safeguarding Japan's extensive rail system: Thanks to automated warnings, two dozen trains that were operating in the earthquake zone on March 11 were brought to a halt within seconds, with no reports of serious injuries or damage.
Bugs in the system
During last year's catastrophe, the biggest problem had to do with the fact that the closer residents were to the quake's epicenter, the less warning they received. Another issue was that the complexity of the initial seismic shock and the aftershocks caused the system to become overloaded, leading to a temporary shutdown.
Heaton and his colleagues are encountering similar bugs in the California system. "They're always being engineered to be better systems and less buggy, but we'll never eliminate all the bugs," he said. Right now, the team is working on an Android app version of ShakeAlert. Even the app would be unsuitable for mass distribution, however.
"The technology exists to deploy it, but strategically, I don't see how we could ever support it," Heaton said.
Going public with ShakeAlert would require a more concerted effort, backed by the expertise and funds that are typically associated with federal government agencies such as the U.S. Geological Survey. So far, the USGS has spent about $2 million on ShakeAlert, and the Gordon and Betty Moore Foundation is backing the research with $6 million in contributions to Caltech, the University of California at Berkeley and the University of Washington over the next three years. Other supporters include Google.org and Deutsche Telekom's Silicon Valley Innovation Center.
The California Integrated Seismic Network estimates that a statewide quake warning system would cost about $80 million over five years, while the cost of a similar system for the Pacific Northwest has been estimated at $70 million. But it might take additional funding to get the system as fully linked in with society as Japan's system is now.
"Ultimately, when it does run, you don't want university professors running it," Heaton said, with a tone of amiable self-deprecation. "We're the least reliable people to run something like that."
Realistically, will ShakeAlert ever be ready for prime time? Heaton thinks it might take more than a catastrophic earthquake on the other side of the world to get Americans motivated about earthquake alerts at home.
"My experience at this point in my life is that it's hard to get people to focus on things like this unless something bad happens," he said. "It's been really peaceful and quiet in the western U.S. for quite some time now. ... We're very concentrated on our own issues. We were shocked by what happened [in Japan], but not enough to actually do something."
If it's hard to put in a system based on well-tested geophysics that provides a warning just seconds in advance of the Big One, it's a lot harder to extend the lead time to hours, or days. But people keep trying.
"One prediction that we have learned to make following earthquakes, and this one is a very strong prediction, is that several people will claim to have predicted the earthquake," Heaton joked.
Some researchers are trying to determine whether a statistical analysis of earthquake clustering can lead to better assessments of the chances that a big earthquake will follow smaller tremors. This month's issue of Physics World looks into the prospects for short-term probabilistic forecasting, as well as the controversy surrounding the researchers who didn't predict the deadly 2009 L'Aquila earthquake in Italy (and are now facing manslaughter charges).
Heaton is doubtful that statistics could ever predict the onset of future quakes with the kind of reliability people expect. He noted that 50 percent of all earthquakes have foreshocks, and one quake out of 20 turns out to be a foreshock for a larger quake. "We can say, yeah, earthquakes come in clumps, but to get more particular and specific — personally, I don't think it's very helpful," he said. "What are people going to do with that information, anyway?"
It's possible that some as-yet-unknown mechanism might provide advance indications that a big quake is coming. "There are interesting observations that seem to be reliable about phenomena that are totally mysterious to us," Heaton acknowledged. "Many of them concern electrical phenomena."
Heaton even keeps an open mind about claims that animal behavior can be analyzed to predict future earthquakes.
"I think we know some things that animals are unlikely to do — that is, pick up vibrations from the earth," he told me. "There may be other things out there that are happening that we don't understand very well. So I'm not going to say 'never' to something like that. But the more we think about the problem, the more we recognize that once an earthquake starts, at some point, trying to predict how big it will get before it stops seems to be a particularly difficult dynamics problem."
More about the Japan quake anniversary:
Alan Boyle is msnbc.com's science editor. Connect with the Cosmic Log community by "liking" the log's Facebook page, following @b0yle on Twitter or adding Cosmic Log's Google+ page to your circle. You can also check out "The Case for Pluto," my book about the controversial dwarf planet and the search for other worlds.