April 22, 2011

Earthquake early warning in action

Posted by Austin Elliott

No country would have been better prepared for such a massive earthquake than Japan. Straddling the boundaries between four converging tectonic plates, Japan is one of the most earthquake prone nations in the world, and is probably the most earthquake savvy. They have world-class networks of monitoring instruments including seismometers, tide gauges, and GPS locating stations; they routinely practice for massive quakes, which they are often visited by; and they have something Californians at this point still only dream of: an earthquake early warning system.

Screenshot of Japan’s earthquake early warning system in action during the Tohoku quake. The upper panel shows a projected shaking intensity level at the computer’s location, the middle panel shows a countdown until shaking begins, and the lower panel shows the current calculation of the earthquake’s size.

The idea behind it is simple: the electromagnetic waves needed to transmit a warning travel effectively instantaneously, while seismic waves travel at “only” a few kilometers per second, delaying their arrival to distant locations. It’s exactly like lightning and thunder–the flash is nearly instantaneous for everyone, but the immediate cracking thunder up close differs from the delayed rolling rumble heard from far away, which is why we count the time between the flash and the bang to judge how far away a strike was (~five seconds per mile, in case anyone needed a reminder). The analogy isn’t entirely complete. Earthquakes aren’t the product of some abrupt electromagnetic signal like thunder is produced by lightning. In effect what early warning systems do is take the loud, close thunderclap and warn people tens of miles away that they’ll hear a rumble soon.

For people close to the epicenter of an earthquake, the time between detection of the quake and the onset of shaking where they are is minimal, so early warning is of little use. People farther away, however, have potentially tens of seconds’ warning that the seismic waves are on their way. Unfortunately for the utility of the system, the intensity of shaking dies off with distance, so it’s the people closest to the epicenter who need the most warning. Too bad… for most quakes. The system is at its most ideal for a massive quake like the 9.0 last month, which took nearly three minutes to produce and ended up rupturing hundreds of kilometers along the coast, effectively stretching the epicenter–or at least the source of seismic waves–so that cities that weren’t near the epicenter still ended up being close to a part of the fault that ruptured. In that case, the 30 second warning for the onset of shaking helpfully prepared people for the intensifying shaking after 3 minutes of rupture propagation.

In an earlier post I highlighted an amateur video showing the warning system in action. Thanks to the similarly excited Dr. Matthew d’Alessio at Cal State Northridge, I can guide you to an excerpt from one of Japan’s TV networks as the alert for the 9.0 quake preempts coverage of parliamentary proceedings:


The video gives an eerie perspective of the event unfolding: the quake is detected and reported long before shaking reaches Parliament, and the staff at the news studio have only just begun to realize the gravity of the event as the shaking intensifies. Only once the shaking has died down does attention shift to the dire tsunami warning.

The mere seconds of warning afforded by earthquake detection apparently pale in comparison to the minutes we have for tornadoes, days we have for hurricanes, and weeks we have for flood events, but they do allow you to drop crucial tasks and focus on the calamity at hand before it gets the best of you, a function automated systems are even better at!

The technology behind this is impressive, and by no means beyond the grasp of the U.S., but the development of a useful system requires a major budget, and the U.S. is busy… prioritizing. But that’s a subject for many a different blog.

In Japan, the software that runs these systems is developed by scientists and private companies, and has a variety of manifestations.

I’ll leave you to marvel at the system in action in a few different cases. As more information is processed from the ever-increasing number of seismometers detecting the earthquake, the epicenter and magnitude are revised, altering the estimates of arrival time and shaking level at the site, in a format unfortunately reminiscent of a painful scene in a certain NBC mini-series…


Here’s a little wall-mounted device that tells you all you need to know–don’t mind that this guy fakes the quake (maybe a test?):


Here’s a family taking appropriate action in an earlier, much milder quake:


The revision of magnitude calculations is common–in most cases it happens well after the earthquake, as data comes in from seismic stations around the world that help constrain just how much energy was released. It’s especially common after the so-called great earthquakes, which produce such large waves for so long that they drown out their own signals on seismograms, making them especially difficult to analyze. In any earthquake, however, realtime assessment of the magnitude is a dicey game. That’s not to say it produces inaccurate results from the data–there just isn’t enough data when only a few seismometers have been shaken yet! As more sensors detect the earthquake, a more robust magnitude and location can be determined. It’s more or less the same reason you have to watch replays from multiple angles to truly disagree with sports referees.