I-MAG STS Corporation
From several readers: Do you think it more likely a tsunami would strike the
eastern or western coasts of England?
Generally, if all one has is forty to one hundred years of measurements (so
latitude, longitude and depth of the epicenter as well as Richter and date/time),
this is not enough information to predict major earthquakes, let alone get close on
time or location. Part of the problem is that a century is longer than most
seismologists live, but a blink in geological time. Most seismological data older
than forty years is fairly suspect in terms of accuracy. However, statistics has
methods for dealing with this challenge. We look at not only the history of where
and when events were, but also for areas where they were not. A prevailing theory
is that tension builds between plates as they move so that, absent an explanation
like being under a river delta, there should be a increasing chance of a major event
the longer a fault segment has been increasing tension. Haiti 2010 is often given as
an example of this. There are problems with determining if smaller earthquakes
are relieving tension or are foreshocks leading to a larger event. Then there are
tuples - two or more earthquakes of comparable strength occur very close in space
and time. Our models compensate for tuples, but have no explanation why they
The hundreds of earthquakes between the eastern coast of Honshu and the
western edge of the Pacific Plate have no known precedent. Nor is it understood
why there was and is so much activity there as opposed to at plate edges.
Part of the answer if pure probabilities - the square area of the generating region
versus the length of the coastline (with apologies to the late Benoit Mandelbrot).