This is the first in a series of investigative reports from the Chunichi Shimbun about how figures for the probability of a Nankai Trough earthquake were manipulated for a government expert panel report released in February 2018.
It was a few days before Feb. 9, 2018, when the Earthquake Research Committee, the government’s expert panel, announced it had revised its projections about the chance of a major earthquake in the Nankai Trough, off Japan’s Pacific coast, in the next 30 years from about 70 percent to between 70 and 80 percent.
I obtained the information on the revised figure from a source and began coming up with story angles such as whether disaster prevention measures were sufficient and how much damage a huge quake in the Tokai region could cause.
First, to get an expert comment on the revision, I called Takeshi Sagiya, a seismologist and a professor at Nagoya University, expecting to hear him sound the alarm on disaster preparedness. But the comment he made was unexpected.
“Personally, I think it is extremely misleading,” he said.
“If the government presented a figure of 80 percent, then people would think the earthquake coming next would be a Nankai Trough quake and disaster prevention measures would be concentrated on that particular quake,” Sagiya said. “It’s alright if the actual level of danger is the same as the predicted figure. But that’s not the case. This is a complete misunderstanding. Figures only fuel a sense of crisis. I think this is problematic.”
A misunderstanding? Problematic? I was perplexed by the comments. Then Sagiya said something even more shocking.
“The probability for the Nankai Trough quake came out high because the Nankai Trough quake is getting special treatment,” he said. “The figures are inflated because of special treatment and ulterior motives.”
These unanticipated words — inflated, special treatment and ulterior motives — puzzled me even more and prompted me to investigate further.
“A different method to calculate the prediction was used just for the Nankai Trough quake. We shouldn’t call that science. Seismologists consider it untrustworthy,” Sagiya said. “If they used the same method as for other regions, the figure would drop to about 20 percent. Many seismologists claim the same method should be used (for the Nankai Trough earthquake), but people in charge of disaster prevention claim the figure shouldn’t be lowered at this point.”
Earthquake predictions are first presented by seismologists at a subcommittee on ocean-trench earthquakes to be approved by the Earthquake Research Committee, a part of the government’s Headquarters for Earthquake Research Promotion headed by the minister of education, culture, sports, science and technology.
During subcommittee meetings in 2012, seismologists called for the need to use consistent calculation methods to ensure the credibility of predictions. The 70 percent figure for the Nankai Trough quake was set to be significantly lowered. But the proposal allegedly faced strong opposition.
It was hard for me to understand the mechanisms of the two different methods used to calculate the probabilities of earthquakes just by listening to Sagiya’s explanations over the phone.
But it was clear why only the estimate for the Nankai Trough quake was exceptionally high — a different calculation method seismologists believed wasn’t credible was applied. On top of that, there was a hidden agenda explaining why some people were opposed to using a consistent method nationwide.
“Do you know how the probability of the Nankai Trough quake is calculated? It’s based on upheavals of (a stratum at) Murotsu Port, northwest of Cape Muroto in Kochi Prefecture, which have been recorded three times between 1707 and 1945,” Sagiya said. “But how can you get (the probability rate) only by how much the land rose in the past? The method is full of questionable factors. I wonder how many researchers actually think the figure is reliable. But it never changed.”
Before calling Sagiya, I called the head office of the Chunichi Shimbun newspaper and an editor in charge said it would be better not to play up the news because readers would overreact to the figure. After talking with Sagiya, I thought the editor made the right call.
A short story was published on the day of the announcement, saying that the government’s earthquake committee “slightly revised upward from last year the probability of a magnitude 8 to 9 earthquake hitting the Nankai Trough within 30 years to between 70 and 80 percent.”
I was feeling uneasy, thinking there must be something more to the story, remembering Sagiya’s comments just before he hung up the phone: “I’m sure the discussions made in the subcommittee meetings are all recorded in the minutes.”
I made an information disclosure request to the education ministry and, after about a month, obtained the minutes, which revealed numerous surprising comments that had never been disclosed.
Surprisingly, the minutes showed that experts on the panel were strongly opposed to the methods used to calculate the probability of the Nankai Trough earthquake.
A subcommittee on ocean-trench earthquakes has prominent seismologists listed as members, and they are tasked with assessing the latest probability figures of quakes nationwide.
Minutes of the subcommittee’s meetings held between 2012 and 2013 — obtained by the Chunichi Shimbun — showed that although the specific names of those who made the remarks were blacked out, almost all of the members expressed doubt over the calculation method for the chance of the Nankai Trough quake occurring within the next 30 years.
Up until 2012, the probability for such a quake was said to be around 60 to 70 percent.
One of the members said: “If we calculate the probability now using the method we’ve been using all along, we will come up with 70 or 80 percent for the next 30 years. But we should mention somewhere that the figure can be 20 percent if we use a different method.”
It is utterly surprising to see that there had been discussions as far back as eight years ago on the credibility of the figures. So why did they release the figures?
Another member questioned the method that results in high probabilities, saying: “If you allow me to make a scientific argument, I undoubtedly think it is not valid to maintain (the previous method). There are opinions that the time-predictable model (used in the method) is not valid at least in this subcommittee, so I’m not convinced about disclosing the figure.”
The “time-predictable model” apparently refers to a special method used only for calculating the probability of the Nankai Trough quake — something Sagiya described as a method that allowed figures to be inflated and given special treatment.
The Earthquake Research Committee announced in May 2013 that the chance of a quake of magnitude 8 or more happening in the Nankai Trough within 30 years was between 60 and 70 percent. But no explanations appear to have been given on the calculation method, let alone the time-predictable model.
The press release handed out at the time of the 2013 announcement does not mention the 20 percent figure at all in its summary referred to as the “main text.” Despite all the heated debate on the issue, it is buried in the latter part of the bulky document, as if to deliberately make it inconspicuous.
The method that was described in the minutes as the basis for the 20 percent figure is applied to calculate probabilities for all the other earthquakes predicted in Japan. It is meant to determine the probability by calculating the average frequency of occurrence from past data on quakes.
Quakes in the Nankai Trough, however, have varied greatly in frequency in the past, striking the region once in a few decades at times and once every few centuries at other times. The period of “within 30 years” used to calculate a probability seems long from the viewpoint of human lives but is short in terms of seismic cycles. Therefore, it is relatively natural for earthquake probabilities to come out low.
Moreover, by reading the minutes, it became clear that seismologists were skeptical about calculating probabilities in the first place.
The minutes included numerous comments by members showing negative reactions to presenting probabilities.
“It has various problems,” “It doesn’t mean much,” “I have no idea why we are presenting this,” “We shouldn’t present this,” “We can’t do this, no matter what,” and so on.
Impact makes a difference
One member’s comment said it all.
“The probability figure has a big impact. So it’s totally understandable that people want to rely on it, like they rely on the probability of rain, which is familiar and easy to understand. But it is difficult to make good use of earthquake probabilities, since earthquakes don’t happen that often.”
Even rainfall forecasts didn’t prove right with high probability until real-time data on changes in the atmosphere became available. Why did this time-predictable model end up predicting a probability as high as 80 percent for the Nankai Trough earthquake? Why did the yearly probability figure continue to rise like a rainfall forecast for the coming hours?
The answers were all in the minutes.
Keiichi Ozawa is a Chunichi Shimbun reporter who covered nuclear power and earthquakes after an earthquake-triggered tsunami caused meltdowns at the Fukushima No. 1 nuclear power plant in March 2011. Ozawa received the science journalist award, presented by the Japanese Association of Science & Technology Journalists in June, for this series published between October and December 2019.