How the “chaotic” process of plate tectonics works, and why scientists are getting better at predicting major shakeups.

Question: What causes earthquakes?

Arthur Lerner-Lam: Well earthquakes are a weird kind of phenomenon in the following sense.  When we talk about the theory of plate tectonics, the working hypothesis is that it’s a smooth motion of the plates, that the plates move over geologic time with predictable rates.  It’s as if they’re kind of a slow movement of traffic if you will.  Everybody is sort moving at a rate that stays constant in time, but obviously earthquakes happen.  Earthquakes predominantly occur where the plates interact with one another along what we call plate boundaries obviously.  Sometimes those plate boundaries are sideways like we have in the San Andreas Fault or in the earthquake that happened in Haiti.  Sometimes those boundaries are compressional like what happened in Chile or what might happen off the coast of Seattle and sometimes those boundaries are what we call extensional.  The best example of this would be the basin and range in the western United States around the longitude of Salt Lake City going all the way west to the Sierra.  So those three types of boundaries sort of describe how the plates interact, but earthquakes of course are not a continuous process by any means.  They’re really what we call a stick-slip process, so while the hypothesis of plate tectonics or the working model is that the plates move at a relatively constant rate, at the boundaries they stick and the slip.  The boundaries don’t move at a relatively constant rate.  Instead the boundaries stick, and when they stick the movement of the plates stresses that boundary and eventually that stress overcomes the strength of that boundary and it breaks and that breakage is an earthquake.  That’s sort of our… what we call a stick-slip model, or we have other names for it, but that’s basically what happens. 

The problem that we face as earthquake seismologists is that sticking and slipping is almost a chaotic process, not quite, but almost and so it’s very difficult to predict on a day-to-day basis.  Over geologic time of course it’s very predictable.  We can get an idea of where the active plate boundaries are.  Those plate boundaries of course are major faults in the earth and we know where the earthquakes will occur, but because we can’t predict the stick and slip cycle very well, certainly over human timescales, we can’t really get what we call a prediction, but there is a deeper understanding of it that allows us to forecast instead of predict and by forecast I mean well first of all we know where they occur because we know where the plate boundaries are, but we can get a sense of what the probability of an earthquake might be in two ways.  One, by looking at the past history and if we have enough earthquakes, we can sort of get a statistical model of when earthquakes might occur, and that gives us some probability, and the other is to actually do a bit of forward modeling and actually take the plate motions, take a model of the friction along the plate boundary, take a model of how the stress builds up and sort of predict when and where or I should say forecast when the next earthquake might occur, and it’s gotten to the point where we can probably do that over the span of a decade or two, perhaps 30 years, to the extent that we can sort of provide an actuarial table. 

If you wanted to go out and buy insurance in California, as a matter of fact, there is such an actuarial table to help you determine what your rates are or I should help the insurance companies determine what they’re going to charge you.  Of course we didn’t have that for Haiti.  We barely had that for Chile, but the point is we’re getting better at it, so you know one of the exciting parts of earthquake seismology is that the better we get at forecasting earthquakes the closer we might actually come to a prediction that is useful on human timescales.