I recall reading decades ago that there have been past tsunamis from Japanese quakes that killed tens of thousands. I think these were centuries ago (I'm not checking now). Is this 2011 tsunami really unprecedented for Japan? Coreholes in some of the flat inland areas that might be subsiding should preserve a record of past tsunamis? There should be a reasonable historic record also.

The slip distribution that Brian posted has maximum slips of less than 20 m. This part of the subduction zone is actually bigger than what just broke. So, this is not even the worst case earthquake that should have been designed for (although the local effects probably could not have been much worse).

I had a job interview a decade ago at the Nuclear regulatory commission. I did not get the job, which is a good thing, because it looked boring, but I had zero funding then. I was surprised that someone there said that many mid-continent nuclear power plants are designed for 0.15 g. I'm sure that this is an oversimplification on my part. This is in areas where large earthquakes are not known historically and are not expected. But, we don't understand intraplate (within plates) earthquake occurrence very well. I suppose the odds of an earthquake large enough and close enough to one of these plants to cause a major problem is very low for, say, some 1000 year period. But, how much does it cost to overdesign? At Caltech, buildings are overdesigned way beyond the building codes we were told at a meeting there (or, I suppose that might only apply to newer buildings?).

Chris