This is a rather naive approach to calculating risk for putting nuclear reactors in earthquake zones. I should of course have looked up an official approach, and become an earthquake scientist and statistician. I assume such reports will be forthcoming for each site of concern. However, since the public is deeply involved without such expertise, I will enter my considerations with the same aplomb.
One would first think that to build a nuclear reactor in an earthquake zone, one would need to know that it would be strong enough so that there is less than a one in ten chance of it having an earthquake stronger than it is built for, or a larger Tsunami than it is designed for over its lifetime. That means over its lifetime of 60 years, you have to build for a 60 x 10 = 600 year quake, i.e. the strongest quake expected in 600 years.
However, as we have seen with the Fukushima catastrophe, even if one plant in the world is destroyed and contaminates a fair sized area, it is an extreme setback for the entire nuclear industry, with many countries abandoning nuclear power. This is of course not the criteria used in case of an airplane crash, or even on the sinking of the Titanic on cruise ships, or the destruction of two skyscrapers. It was the criteria used on the Zeppelins, however. In any case, we realistically adopt this criteria. Say then that there are 30 reactors in earthquake zones around the Pacific Rim (California has four, and Japan has about 25 nearest the major faults). (I will do a better world-wide survey in the future.) If we only want a one in ten chance of another earthquake caused destruction anywhere in an earthquake zone, we have to multiply the 600 year safety factor by 30, and get that we must protect each site against the most severe earthquake expected in 18,000 years (which I will call 20,000 years). I don’t think the earthquake geologists can in general make this estimate for each of the sites. The conclusion could be that the nuclear industry could continue, but it is not wise to build reactors in either earthquake or tsunami zones.
The above is really a maximum estimate regarding each reactor as a separate seismic entity. In fact, many reactors are present at each site, so one could just consider the joint vulnerability of the number of sites in earthquake zones. Thus California has only two sites, and Japan has six on the more vulnerable eastern coast, giving a total of eight. We then multiply the 600 year safety for one site by eight to get 4800 or about 5,000 years. So with the one-in-ten limit for a failure on the Pacific Rim in the 60 years of an active reactor, each site must be protected against the worst quake or tsunami expected in 5,000 years. This may be possible, and should be seen as a target.
The Fukushima quake was the fifth worst of recorded earthquakes since 1900. It was a 9.0, and totally unexpected for the nearby fault, and it was only half the distance from the Japanese coast to the major subduction fault.
Of course while nuclear opponents and proponents are looking at the Japanese quake from the viewpoint of the long term fight over nuclear power, the real tragedy is the 16,000 lives lost and 4,000 missing, and the communities destroyed by the quake and tsunami. The communities were protected by tsunami walls, tsunami drills, and warnings. However, the quake was of such a rare magnitude as to overwhelm them. In our local press, the worry was over whether you lived within the 50 mile limit of the San Onofre nuclear reactors or not. No attention was paid to the fact that Southern California coastal communities have NO tsunami walls, no real siren systems or evacuation signs, and little tsunami education. There are also 50,000 older buildings in Los Angeles that are not up to current earthquake standards. The real lesson of the Japanese quake should be that quakes and tsunamis can occur, even if rarely, that are far above the expectations of quake analyses of the last several hundred years.