Earthquake forecasts are bogus

by Robert J. Geller

Contributing Writer

At least once a year, Japan’s government issues alarming but bogus earthquake forecasts (e.g.,”Hokkaido’s risk of M9 quake at 40 percent,” The Japan Times, Dec. 20) which are routinely and unquestioningly reported by the media. This must stop.

Present science is, alas, unable to reliably say that one place in Japan is more or less dangerous than another, when earthquakes will strike or how big they will be when they do. There are effective ways to mitigate, although not eliminate, earthquake hazards, but that’s a subject for another column.

Relying on the now discredited belief that large earthquakes occurred cyclically, in the mid-1970s the Japanese government adopted earthquake countermeasures based on the idea that a megaquake was “overdue” off the Pacific coast of the Tokai district (from Shizuoka to Nagoya). Through repetition, the meme that “the ‘Tokai earthquake’ is imminent” became an almost unquestioned belief in Japan.

In about 2000 the government switched from claiming that the Tokai earthquake was imminent to the claim that the “Nankai Trough Great Earthquake” (from Shizuoka to Kochi, including the Tokai, Tonankai and Nankai regions) was imminent. On the basis of similar arguments about earthquake cycles, the government has also been taking countermeasures against a supposedly imminent M7 earthquake directly under the Tokyo area since the early 1980s.

A forecasting model is just a mathematical game unless it has been shown, going forward in time, to agree with data recorded after the forecasts were made. How have government forecasts done over the past 40 years? As for great offshore earthquakes, the M9 Tohoku earthquake, which caused about 19,000 deaths and tremendous economic damage, hit an area labeled low-risk, while nothing has happened in the allegedly high-risk Tokai-Tonakai-Nankai region. This can be scored as a failure. As for large shallow earthquakes directly under Japan, none has hit Tokyo, while the “unexpected” 1995 Kobe and 2016 Kumamoto earthquakes, among others, hit areas not labeled as high-risk. Thus we can confidently say that the government’s forecasts are, based on the test of experience, wrong.

The government’s forecasts are too vague to be useful anyway. The latest forecast says there is a 7-40 percent chance of a megaquake off the Pacific coast of Hokkaido in the next 30 years, but these percentages are meaningless. If a megaquake hits the government can claim success, but if it doesn’t it can say that was the other 60-93 percent. In any case, the government won’t be called to account for 30 years, by which time almost everyone will have forgotten the forecast. And since the government issues new forecasts every year, the old ones disappear into the memory hole.

In summary, the government’s 30-year forecasts are based on an incorrect model of periodically repeating quakes, have been repeatedly wrong in practice and are too vague to be useful. The public can ignore them and the media should stop reporting them. This doesn’t mean that Hokkaido, Tokai, or Nankai aren’t at risk, just that we can’t say they’re at more risk than other regions.

But wait, there’s more. In 1978, the Large scale Earthquake Countermeasures Act was enacted by the Diet. The law is based on the premise that “precursors” can be detected with enough accuracy to issue an alarm that the Tokai earthquake will strike within 72 hours. The Meteorological Agency is the lead agency for this prediction system, which is still operational.

On paper, the prediction system, which has never been activated, works as follows. If anomalous data are recorded a committee of six seismologists will advise the Meteorological Agency director whether a prediction is warranted. If so, the agency director will advise the prime minister, who, with the approval of the Cabinet, will declare a legally binding state of emergency that places Shizuoka and five other prefectures under lockdown. Expressways and rail lines will be closed, schools and factories shut down and so on. Since there never has been an operational test, no one knows how people would cope when food stocks and gasoline supplies run out.

As a rough estimate of the cost of a state of emergency, let’s say 0.1 percent of Japan’s GDP, or ¥500 billion, per day. With such a high cost, one might assume that the scientific basis for the prediction system had been fully vetted when the 1978 law was passed. Not so. No reliable and generally accepted evidence for earthquake “precursors” existed in 1978, and none exists now. The government has presented numerical simulations of precursory “pre-slip,” but this is just wishful thinking, because the phenomenon has never been reliably observed anywhere in the world.

The seismologists who established the Tokai earthquake prediction system in 1978 in the absence of sufficient scientific underpinning were highly irresponsible, as were their students and their students’ students, who maintained the facade for 40 years. This September a government committee finally acknowledged that the prediction of the Tokai earthquake three days before its expected occurrence was impossible and the prediction system has de facto been abandoned. However, no one has yet been held responsible for deceiving the public for the past 40 years.

The seismologists who pretended for the past 40 years that they could predict the Tokai earthquake 72 hours in advance are basically the same people who have been issuing bogus 30-year earthquake forecasts for the past 15 years. It’s hard to see any reason for viewing their pronouncements as authoritative. In fact their credibility has been repeatedly questioned.

For example, professor Hitoshi Takeuchi of the University of Tokyo said almost the same things in a newspaper interview (Asahi, Aug. 27, 1979) that I’ve said here. And I’ve said similar things many times in both English (e.g., Nature, 1991, 2011, 2017) and Japanese (e.g., in an invited article to mark the 20th anniversary of the government’s Headquarters for Promotion of Earthquake Research).

Still, in the worst tradition of press club journalism, the media continue to fill their columns uncritically with almost verbatim reports of bogus government earthquake forecasts. This must cease.

Robert J. Geller is a professor emeritus (seismology) at the University of Tokyo, where he taught from August 1984 until March 2017.