Published online 14 May 2008 | Nature | doi:10.1038/news.2008.824


Japan's earthquake warning system fails again

But officials insist system will work for bigger quakes.

tokyo satelliteTokyo, seen here from a satellite, may not get enough warning about medium-sized quakes.NASA

A magnitude-6.7 earthquake that struck on 8 May less than 200 kilometres northeast of Tokyo caused only minor damage, but it took a heavy toll on the public’s confidence in Japan’s earthquake warning system.

Although the quake rattled the greater Tokyo area, and about half a dozen people were reported to have been injured by falling picture frames, vases and speakers, no major damage was reported.

Nevertheless, Japan’s early warning system should have kicked in. It didn’t. Officials at the Japan Meteorological Agency (JMA) say that the system, meant to give at least 10 seconds' warning of an earthquake, missed the incident because it was a borderline case.

The system is based on Japan’s 'seismic intensity' ratings. Whereas magnitude measures the energy released at the seismic focus of an earthquake, seismic intensity rates the destructive strength of the tremor as felt at the surface.

Below the threshold

Last Thursday’s 40 kilometre-deep quake shook most of the area with a seismic intensity of level 4 or lower. Warnings are only sent out with an intensity of at least 5 (which has roughly 30 times the intensity of level-4 shaking).

A tremor in the lower-5 range can put cracks in walls and knock over bookcases. And crucially, some of northern Chiba prefecture, including the city of Mito, did however register shaking that reached level 5.

Estimates of the seismic intensity, based on an analysis of the magnitude and the depth of the earthquake as well as the geographical features of the earth between the focus and the surface, are possible within seconds of the quake occurring. The early warning system works by picking up primary waves, or 'P-waves', released by the earthquake. These are non-destructive waves that can arrive tens of seconds ahead of destructive secondary waves.

“We’ve tried to be perfectly clear about the limits of the system.”

Akihiko Wakayama
Japan Meteorological Agency

JMA’s seismic detectors relay P-wave data to computers that automatically calculate the magnitude and depth, and estimate the impact at different points on land. Even after the first calculation, data from Japan’s many seismometers continue to be relayed to the computers, producing a series of recalculations.

On 8 May, the first nine sets of data â€" the first spanning a time window of 9.3 seconds immediately after the earthquake; and the ninth spanning 58.3 seconds after the earthquake â€" gave varied estimates of magnitude (ranging from a 6.0 to a 6.9) and depth (ranging from 10 kilometres to 70 kilometres).

It wasn’t until this ninth analysis, almost a minute after the quake and 40 seconds after the ground in Mito had already shaken, that a lower-5 estimate was produced.

Later analysis of data from dozens of seismometers gave the most accurate reading: a magnitude-6.7 quake at a depth of 40 kilometres, which produced a lower-5 shaking in Mito and one other town.

Repeated failure

It wasn’t the first time that the system has missed the mark since it went into action last October. A warning of lower-5 shaking on 28 April arrived at the southern island of Miyakojima 6 seconds after its residents had already felt the tremor. And JMA's computers underestimated the seismic intensity of a quake on 26 January in Ishikawa prefecture on the Japan Sea coast (see 'Early warning system underestimates quake').


Akihiko Wakayama, deputy director of the JMA's earthquake and tsunami observations division, says the agency is receiving complaints from government officials and the public. “People are saying the system is useless, but we’ve tried to be perfectly clear about the limits of the system,†* he says.

Wakayama says the system will work much better for larger earthquakes, giving people time to stop trains and leave elevators to avoid mishap. “Smaller earthquakes have less clear data, but large earthquakes give larger data sets. The larger the earthquake, the faster we can get a reading,†* he says. Fortunately for the Japanese people, but unfortunately for those wanting to show the merits of the system, moderate earthquakes greatly outnumber large devastating ones. 

Commenting is now closed.