A seismologist at the University of Tokyo is blasting Japanese officialdom for building seismic hazard maps based on the idea that some earthquakes could be predicted, and called on the government to admit that they can't.

Robert Geller, a professor of seismology, wrote a column in Nature that charges the Japanese government has been using a flawed model for predicting where the worst earthquakes are most likely to occur.

The Japanese government's hazard maps show the southeastern coastal regions (centered approximately on Tokyo) as most likely to undergo violent shaking. Japanese seismologists divided the area off the coasts of Honshu and Shikoku into three regions: Tokai, Tonankai and Nankai.

Geller says that the focus on the Tokai earthquake blinded Japanese scientists to the very real dangers of earthquakes elsewhere. The country's worst earthquakes over the last 30 years have all been far removed from Tokai, Geller notes. In fact no large earthquakes have occurred in the Tokai region at all since 1975. But the Japanese government still classifies it as the most hazardous.

Part of the problem, Geller says, is the seismic gap theory, which first became popular in the 1970s. It says that seismically active areas that haven't been through an earthquake for longer periods are more likely to experience one. Another is the theory that that a small uplift can predict a larger shock. That theory arose after a single report of a precursor to an earthquake in 1944.

In the 1970s, there was also a lot of promising work on earthquake prediction, and Japanese authorities became convinced they could give early warning of a Tokai quake. The government eventually codified into law that the Japanese Meteorological Agency must operate an alert system based on the assumption that such warnings are possible.

Geller notes that the promise of earthquake prediction turned out to be just that -- promise. None of the precursors to quakes that were proposed have proven of predictive value.

The theory that small slips would predict the big quakes hasn't been borne out by data from GPS satellites. Basing even a large-scale program of observational research on the 1944 data would be uncalled for. It beggars belief, then, that the Japanese government operates a legally binding earthquake-prediction system on this basis, Geller wrote.

Yet the JMA is required to continuously watch for such precursors (mostly based on how fast the Earth's crust is moving) and then decide whether to declare a state of emergency.

Geller argues that had the Japanese scientists simply looked into historical records they would have found that huge tsunamis have hit the northwestern coast several times over the past few centuries. While there was no way to predict the magnitude 9 quake that hit the country March 11, it was possible to see that big tsunamis were a good possibility, which could have been factored into the design of the Fukushima Daiichi nuclear plant.