Under the Hood of the Earthquake Machine

Watson Lecture Preview


What makes an earthquake go off? Why are earthquakes so difficult to forecast? Professor of Mechanical Engineering and Geophysics Nadia Lapusta gives us a close-up look at the moving parts, as it were, at 8:00 p.m. on Wednesday, February 13, 2013, in Caltech's Beckman Auditorium. Admission is free.


Q: What do you do?

A: I study friction as it relates to earthquakes. At a depth of five miles, which is the average depth at which large earthquakes in Southern California occur, the compression on the two sides of the fault is roughly equivalent to a pressure of 1,500 atmospheres. So you can imagine that friction plays an important role. I make computational models that combine our theories about friction with laboratory studies of how materials behave. We try to reproduce what seismologists, geodesists, and geologists see actual earthquakes doing, in order to infer the physical laws that govern them.

Our planet's surface is made up of a bunch of plates that are always moving, and an earthquake happens when the locked boundaries of the plates rapidly catch up with the slow motion of the plates themselves. You get a sudden shearing—a sideways motion that generates the destructive waves that we perceive as shaking.

A number of factors affect this process. If you rub your palms together, you generate heat. An earthquake is a very intensive rubbing of palms, if you will, and so a lot of heat is produced—enough to weaken the rocks and perhaps even melt them.

However, there are pore fluids permeating the rocks—we often get our drinking water from underground aquifers, for example. As these fluids heat up, they expand, which modifies the shearing process. They produce expanding cushions of steam, essentially, which reduce the friction.

The waves generated by the shearing motion put an additional load on the fault ahead of the shear zone, so they actually affect how the shearing progresses. The shear tip sprouts at about three kilometers per second, or 6,700 miles per hour. So an earthquake is a highly dynamic, nonlinear system.

To make things even more interesting, a fault doesn't just sit still for hundreds of years, waiting for the next big earthquake. It's more like a living thing—there are slow slippages between earthquakes that constantly redistribute the forces in the system, and the exact point where an earthquake initiates depends a lot on these slow motions. So we simulate thousands of years of fault history that includes a few occasional, very fast events that last for a few seconds. These calculations are very time-consuming and memory-intense. The Geological and Planetary Sciences Division's supercomputer has several thousand processors, and we routinely use 200 to 400 of them, sometimes for weeks at a time. We would happily use the entire machine, but of course people would yell at us.


Q: How did you get into this line of work?

A: I've loved both mathematics and physics since I was a child. I was born in Ukraine, where my mom was a professor of applied mathematics and my dad was a civil engineer. They used to give me math and physics problems from a very early age. I did my undergraduate studies in applied mathematics in Kiev, and I was thinking of going into materials science. I came to the U.S. for graduate school, and my advisor at Harvard was working on materials failure and on earthquakes, which I found very interesting because it combined math and physics with a problem relevant to society.

My PhD was on frictional sliding and some initial models of earthquakes. Caltech is actually the perfect place to continue that, because it has world-class expertise in all relevant disciplines. I have wonderful colleagues, and the really fun part is working with them. I enjoy interacting with the experimentalists and talking to the people who make field observations or do radar measurements from satellites. They have different perspectives, different terminologies, and different views of the problem, so it's fun to try to explain to them what you mean, and to try to understand what they mean. And the most fun, of course, is when you come to an understanding that leads to new science in the end.


Q: Speaking of societal relevance, what does your work mean for us here in L.A.?

A: Large earthquakes, fortunately, are relatively rare, so we don't have detailed observations of very many of them. Our models, however, allow us to explore scenarios for potentially very damaging earthquakes that we haven't experienced. For example, faults have locked segments and creeping segments. The San Andreas fault has a creeping segment between Los Angeles and San Francisco, and the assumption has been that this segment will confine a large earthquake to either the southern or the northern part of the fault. Only one large urban area would be affected. However, our models show that a through-going rupture may be possible. If that happens, both Los Angeles and San Francisco are affected, and you have a much bigger problem on your hands.


Named for the late Caltech professor Earnest C. Watson, who founded the series in 1922, the Watson Lectures present Caltech and JPL researchers describing their work to the public. Many past Watson Lectures are available online at Caltech's iTunes U site.

Written by Douglas Smith