On March 19th, 2012, I gave a presentation at the Pacific Basin Nuclear Power Conference in Pusan, Korea.

In that presentation I used a slide from the ANS Summer meeting in 2011. This slide indicated that a study by TEPCO in 2006 showed that the recurrence frequency of a tsunami greater than 5.7m was higher than 1/10,000 years.

The slide depicts a recurrence frequency of about 1/400 years.

I was absolutely wrong to use this slide. The information I was given that the slide was about Fukushima Daiichi was incorrect. I did not verify the source of the slide.
The prediction of 1/400 in the slide was based on the location of Yamada in Iwate-ken on the Sanriku coast. Please understand that the coastline in Sanriku is completely different than the coast on the Sendai Plain, where Fukushima Daiichi is located. The coast on Sanriku has a completely different topography and the tsunami effects historically are amplified by the hundreds of small inlets, coves, and bays which proliferate Sanriku and focus the tsunami. For example, even if one believes that the height of the tsunami was 14m, as reported by TEPCO, the height of the tsunami on the Sanriku coast exceeded 26m.

One must be very careful to make probabilistic inferences only from data which comes from comparable sources. The Sanriku Coast is another world in comparison to the Sendai Plain.

I was not careful.