Commentary / Japan

Scrap scientifically unsound quake forecasts

by Robert J. Geller

Contributing Writer

What, if anything, were you thinking about earthquakes on the morning of March 11, 2011?

Most likely you weren’t thinking about earthquakes at all. But if you were, it was probably that, as Japan’s government regularly said, and the media regularly reported, Japan was mainly at risk due to two “expected” earthquakes: the magnitude 9 “Nankai Trough Great Earthquake” (off the Pacific coast from Shizuoka all the way to Kochi) and a magnitude 7 earthquake under Tokyo. Each of these “expected” earthquakes supposedly had roughly a 70 percent probability of striking in the next 30 years.

Later that day, at 2:46 p.m. to be precise, the magnitude 9.0 Great East Japan Earthquake struck Japan’s Pacific Coast from Iwate to Ibaraki prefectures, causing over 18,000 deaths and the tsunami that lead to the Fukushima nuclear disaster. As this devastating earthquake struck a completely different part of Japan than the “expected Great Nanakai Trough Earthquake,” the seismologists chosen as advisers by Japan’s government were reduced to repeatedly stating to TV interviewers that it was “completely unexpected” (“sotei gai“).

The 2011 Tohoku earthquake was not the only “unexpected earthquake” in recent years. The April 14-16, 2016, Kumamoto earthquake sequence (the largest event was magnitude 7.3) caused 272 deaths, and the Sept. 6, 2018 earthquake in the Iburi area of Hokkaido caused 42 deaths. Meanwhile, no damaging earthquakes have occurred in the officially “expected” Tokyo and Nankai regions. What are we to make of this? The obvious conclusion to be drawn by anyone with a modicum of common sense — admittedly an uncommon attribute these days — is that the government earthquake forecasts, are, in a word, wrong.

This is the key point, so let’s go through it again. The government isn’t just making up its earthquake forecasts out of thin air. It has a mathematical model (based on fallacious ideas — see below) that it uses to generate the quake forecasts. But a model is only a model — unproven, in other words — until it’s been validated by showing it agrees with observed data. This is the fundamental principle of science. Furthermore, the data to which a model’s forecasts are compared must be data recorded after the forecasts were issued. With me so far?

OK, good. Now for the reveal. As discussed above, the earthquakes “expected” by the government (Nankai, Tokyo) didn’t occur, while earthquakes “unexpected” by the government (Tohoku, Kumamoto, Hokkaido) did occur. What this is telling us is not only that the particular forecasts made by the government are wrong, but, more importantly, that the model used to generate the forecasts is wrong. Thus, going forward, all present and future forecasts based on the government’s model must be presumed to be wrong, and should not be used as the basis for public policy.

There’s nothing shameful about having made incorrect earthquake forecasts using a model that seemed plausible but ultimately turned out to be wrong. That’s how science works. We churn through hypotheses at a fantastic rate, testing them and discarding the ones that don’t work. In the end the success rate of proposed hypotheses in any field of science isn’t too much better than the success rate of sperm competing to fertilize an egg.

Three things are shameful, however.

First, Japan’s government was irresponsible in issuing its forecasts to the public before its model was tested; it should have waited for validation, just as is the case with new pharmaceuticals.

Second, now that we know the forecasting model is wrong, Japan’s government should withdraw it from use.

Third, it should be obvious to the media that the government’s earthquake forecasts are wrong, but they continue to uncritically report them.

The reasons for the first two are obvious. Japan’s government appoints its own advisers, who are chosen because they can be counted on to rubber stamp whatever the government wants to do. Having once adopted a policy, however misguided, the government’s resistance to changing it, which is an implicit admission of error, is overwhelming.

As for the third, most reporters who cover government earthquake forecasts aren’t trained in science. So I can’t blame any who uncritically reported government earthquake forecasts as fact before the Tohoku earthquake. But now, eight years later, and after not only Tohoku but also Kumamoto and Hokkaido, the media, starting with NHK, still report government forecasts as unquestioned fact, even though the empirical evidence makes it clear the forecasts are wrong. This is a regrettable symptom of the sheeplike mentality induced by the press club system, whereby no journalist is willing to report something different from anyone else.

If people and local governments in Nankai and Tokyo take earthquake countermeasures that’s not so bad. Even though the government’s forecasts are scientific nonsense, every part of Japan, including Nankai and Tokyo, is at risk from earthquakes. The problem is the complacency induced in the rest of Japan, which leads to needlessly high casualties and economic losses when an earthquake does occur.

The example of Kumamoto is notorious. Until 2016, the home page of the prefectural government was enticing businesses to build their new factories in Kumamoto Prefecture, which it touted as a “safe area” with a low incidence of big earthquakes.

There is a sound scientific reason for why the government’s earthquake forecasts fail. They are based on the idea that earthquakes repeat more or less in cycles. This idea was popular 100 years ago when seismology was just getting started and was still prevalent 50 years ago, but is now known to be incorrect. It stands to reason, as is in fact the case, that a model based on incorrect premises will yield incorrect forecasts.

Meanwhile, back in the real world, the government should scrap its scientifically unsound earthquake forecasts, but if it doesn’t the media shouldn’t hesitate to point out the flaws. And Japan’s earthquake countermeasures should be based on the fact that large damaging earthquakes can strike anywhere at any time, rather than only focusing on particular areas.

Robert J. Geller is a professor emeritus (seismology) at the University of Tokyo.

GET THE BEST OF THE JAPAN TIMES
IN FIVE EASY PIECES WITH TAKE 5