Earthquakes and Structure

One of the basic questions seismologists attempt to answer is why do earthquakes start in one place and end in another. With Donna Eberhart-Phillips, I have been attempting to answer this question by imaging the rocks underground by using seismic waves. Earthquakes in California occur up to 20 km benearth the earth's surface so we can't see all of the important structures on the surface. However, different rocks transmit seismic waves at different speeds. By using recordings of many earthquakes at many stations we can attempt to map these variations in seismic wave velocities underground, much as CAT scans are used by doctors to create three dimensional images of the inside of a patient's body (in fact, seismologists have been doing this for slightly longer than doctors).

Here is an example of what we have found for the 1989 Loma Prieta earthquake south of San Francisco. To examine the structure underground we will look at "cross-sections" which are slices through the earth.

Key to Figures On each cross section we show the speed at which compressional (or P) seismic waves travel as a color. These speeds range from 3 to 7 kilometers per second. The slower velocities are shown in red and the faster velocities are in blue. On top of the velocity model we show the locations of aftershocks in green and background seismicity (earthquakes which occured before the mainshock) in yellow. Contours of constant velocity are also shown in black to emphasize the shape of the velocity structure.
Along Fault Section Here we are looking at a slice of the earth along the length of the fault which ruptured during the magnitude 7.1 Loma Prieta mainshock. The outer white line shows the area where the fault slipped 1 meter and The slip model is from David Wald's 1991 paper in BSSA. the inner white lines show the areas where it slipped 2 meters. Note that these areas fall within a region of generally higher velocity as shown by the bluer colors and the contour that is bent upwards.
Across Fault Section 1 This image shows a slice across the fault which can be seen by looking at the aftershocks (in green). If you just look at the velocity model it is hard to find the fault and there is relatively high velocity material on both sides. This slice was taken in an area where the mainshock had its maximum slip.
Across Fault Section 2 Here is another slice across the fault, but from an area southeast of where the mainshock took place. Now the San Andreas fault can be found by looking at the yellow background earthquakes. From looking at the velocity model you can see that there is a sharp velocity contrast across the fault with slow material on one side and faster material on the other side.
These basic patterns that Donna and I have found seem to explain how geologic structures at depth controls earthquakes. The first is that the large mainshocks occur in areas where the fault is not well developed and high velocities occur on both sides of the fault. The second is that where the fault is very clear and has very low velocities on one side that the fault tends to slip slowly and constantly and produces only small earthquakes. In some places, the low velocities may be due to high pressure fluids being trapped in the rocks. If these fluids leak into the fault they may weaken the fault and prevent it from storing a lot of energy to release in large earthquakes.

These same patterns have been found not only for the Loma Prieta earthquake, but for the 1984 M6.2 Morgan Hill earthquake, the M6 Parkfield earthquakes, and the 1983 M6.7 Coalinga earthquake.