Okay, we all know how an earthquake knocks us down. We are standing there, admiring the birdies, when there is a big lurch to the left, and we fall down. We feel that we have experienced a lateral acceleration, which can be measured as a percentage of the actual downward acceleration (force) that is gravity. We can probably take 30% g lateral and keep standing. Any more and we fall.
The early seismologists noted this and designed instruments to measure it. They were mostly pendulums and masses on springs. These were incredibly crude, and needed a lot of motion to get them to work. That chance happened with the famous El Centro earthquake, right in the middle of a huge soil basin.
This was the single biggest thing to happen to earthquake engineering. Everybody+dog used this ground motion for design. Problem was they applied it everywhere to any area and foundation. Applying that to some situations would be the equivalent of crushing all the surrounding rock to powder. Waves only propagate if the stress disturbance stays within the linear zone. It's like watching those stupid huge wake boats churning up huge breaking waves and thinking they will smash your dock. They don't, they turn into something that will propagate.
PGA only works for the very low frequencies of soil basins. Stupidity piled onto stupidity when they took this stuff east for nuclear plants. As well, they considered seismic motion as infinite sinusoids which was more horrible. This combined to make rock sites horrible, and soft soil sites good.
If the frequency starts to go above 1 Hz, then PGA has no more link with standard physics. At high frequencies, instruments record a high PGA which makes solid rock sites terrible. In actuality they have a low PGV and are good what ails you. I remember the high PGA for the east had all the dams sliding into the ocean, in their dreamworld of phoney analysis.
The use of PGA will be deadly in the next earthquake for buildings that have been 'designed'. PGA has no amplification on soft soil. PGV is physically linked to the induced shear strain and has meaning related to damage on all soil types and in all areas of the world. It links smoothly to the Modified Mercalli intensity scale. You need a PGV of about 50 cm/s for structural damage. 10-20 cm/s will knock down the ceiling tiles and bonk you on the head.
The old company designed nuclear plants using El Centro on solid rock. Everything failed testing and analysis, when we looked back on it. But somehow everything passed. (We could never figure out how) They are using the same methodology for the retrofit. I hope they know the secret techniques of the Ancients. :)
*PGA stands for peak ground acceleration. It represents the maximum swing of a pendulum.
*PGV is peak ground velocity. It is derived from the integration of the time series of acceleration. Thanks to the chip revolution and the necessity to have accelerometers in your phone, it is becoming quite easy to measure these things.