Imperfect science: Uncertainty, diversity, and experts

Seismic safety issues related to nuclear reactors in the eastern United States pose special challenges to the Earth and engineering sciences, given the severe consequences that can attend even very infrequent earthquakes. To deal with low-probability, potentially damaging ground motions, two major probabilistic seismic hazard analyses were conducted in the 1980s for nuclear reactors in the eastern United States, that part of the country east of the Rocky Mountains. The first study was performed by the Lawrence Livermore National Laboratory (LLNL) [Bernreuter et al., 1989] and was supported by the U.S. Nuclear Regulatory Commission (USNRC). The second was commissioned by the Seismicity Owners Group of the Electric Power Research Institute (EPRI), [1989]. These studies generally agreed in terms of median hazard estimates, but mean hazard estimates at individual sites varied considerably, in several cases by 2 orders of magnitude or more.