In earthquakes, alerts may turn machines into action heroes

The UC Berkeley Seismological Laboratory issued an alert about the recent earthquake in California’s Napa Valley 10 seconds before it struck. That may not seem like much time — unless you’re a child of the 1950s and 1960s who was trained in school to duck and cover the second you saw a large bright nuclear flash.

Earthquake early warning systems can deliver alerts of impending seismic activity a few seconds to as long as four minutes before the tremors begin. The systems don’t predict earthquakes, but a quake’s energy waves move slowly enough to create an opportunity for a warning. The length of warning depends on the distance from the earthquake’s center.

Even if it sends an alert just a few seconds before an event, an earthquake warning system can help save lives and prevent property damage. But the U.S. has yet to fund an earthquake early warning  system. That’s not the case in Japan; that nation has a warning system that issued alerts that triggered the shutdown of the transit system when a 9.0-magnitude earthquake struck offshore in the Pacific Ocean on March 11, 2011. No trains derailed.

The cost of building and operating an alert system for the West Coast of the United States has been estimated at approximately $120 million for the first five years.

But investing in a fully built alert system that’s integrated into schools, offices and other types of buildings could give rise to a new industry, said William Leith, a senior science adviser at the U.S. Geological Survey. Leith offered testimony on the subject of early warning systems to a subcommittee of the U.S. House of Representatives last June.

“Consultants will advise users on how to use alerts to take protective actions,” he said. “Mass notification companies will customize alerts for their clients. Automated control producers will make and install equipment to take actions and sound alarms at user facilities. Entrepreneurs will undoubtedly develop creative new applications specific to various industry sectors.”

Joshua Bashioum, the founder of Early Warning Labs in Santa Monica, Calif., is in the vanguard of this industry. His year-old, privately funded company is building hardware systems that can interface with building operational systems and IT networks. What Early Warning Labs intends to do is take earthquake alert data, calculate the intensity at client locations and project the risk of damage at those locations. It will then push out machine-to-machine commands.

The possible action scenarios are almost endless. An automated command could turn on a data center’s emergency generators and begin disaster recovery procedures. It could alert surgeons in an operating room of an impending quake, get schoolchildren to take cover, automatically open garage doors at fire houses to prevent jamming, turn off gas pipelines, shut down high-tech manufacturing assembly lines and brace equipment, stop elevators and set off audible alarms. That’s not to mention the potential for triggering messages on TV and radio and even activating citywide sirens.

Bashioum said that until the government makes earthquake early warning systems fully functional, private-sector companies can’t use the data to issue alerts. All they do now are test installations.

“Our efforts right now are focused on identifying what needs to be done,” said Bashioum, who is among those speaking next week at the International Conference on Earthquake Early Warning at the University of California, Berkeley. Saving lives is the main goal, he said.

Large companies may be interested as well.

David Jonker, the senior director of Big Data Initiatives at SAP, said in an earthquake early warning, where every second counts, the challenge is to issue a warning as fast as possible. There won’t be time to read disks, and he believes in-memory systems will be the preferred approach for processing warning data.

SAP is already in the warning business; it provides technology for NY-Alert, New York’s all-hazard alert and notification system.

Jonker said SMS, as well as user responses, are too slow. “You are very much talking about a machine-to-machine play,” he said.

What is clear is that building an early warning system will take time, and that includes the time needed to train the public in how to respond to warnings.

“We’re going to be focused on getting the science right and the warning generated correctly, and then we’re going to depend on our public sector and corporate partners to figure out how we are going to push it out,” said Bill Steele, director of outreach and information services at the Pacific Northwest Seismic Network.