A question I attacked was how soon did the triggering begin.
Here is a seismogram recorded at the Long Valley Caldera, a volcanic
center near Mammoth Lakes, California, which was one of the areas that
had the largest number of triggered earthquakes.
In this seismogram the earliest triggered events we can see are several
minutes after the shaking from the mainshock arrives. But there could
be earlier events that we can't see because they are covered up by the
shaking of the mainshock. To search for such events I tried filtering
the seismograms.
The filtering shown below works because at large distances (a few hundred km or more)
from the earthquake the shaking has little high frequency content left.
This is why people far from an earthquake experience "rolling motion" instead
of the sharp motions felt by those near the fault. However, the small
earthquakes that occurred near the seismometers, but far from Landers, do
have high frequency energy.
By removing the low frequencies we can enhance the small events
near the seismometer.
Note that no triggered events are seen in the seismogram recorded at Parkfield. There were also no increase of activity at Parkfield so this record demonstrates that the filtering process did not create the triggered events by mistake.
The earliest triggered events occurred within a minute of when the strong shaking reached each area. This information can be used by those developing theoretical models of what caused these triggered events. If their model doesn't create triggered earthquakes that soon then it needs to be changed.