Earthquakes that have a source dimension in time and space smaller than the period and wavelength of the waves of interest can be viewed as a point source to a good approximation. However, for a few events the effect of the finite source on the waveforms is significant even at long periods. In those cases we can use synthetic waveforms computed for different source models to extract information about the kinematics of the rupture.

When earthquakes propagate unilaterally along long faults they radiate waves that look very different depending on the direction we look at them from. Waves propagating in the rupture direction produce a very narrow but high pulse, but those propagating in the anti rupture direction are low amplitude and spread out.

Application to the 2001, Mw 7.9 Kunlun, China, earthquake

Vala Hjorleifsdottir looked at the Mw 7.9, Nov 14, 2001, Kunlun, China event, which propagated eastward over a 400 km long section of the Kunlun fault. By quantifying the azimuthal phase-shift between observed data and synthetics calculated for different rupture speeds, we can put bounds on what values of rupture speeds are consistent with the data.

9/29/06rect.gif" width="450" height="280">

A comparison between data and synthetics for Rayleigh-waves computed for source models with two distinct rupture speeds, 2.5 km/s and 4.5 km/s respectively. Notice that the synthetics arrive later than the data in the rupture direction (90 degrees) for the slower rupture speed but later in the anti-rupture direction (270 degrees) and vice versa for the faster rupture speed. We estimated the rupture speed to be 3.4-3.8 km/s. This value is very close to the shear wave velocity of the material the fault broke, indicating that not much energy is going into the fracture and friction on the fault. Currently we are working on methods to quantify the variation of rupture speed in time and space.

Application to the 2002 Mw 7.9 Denali fault, Alaska, earthquake

Dr. Chen Ji constructed the slip history of the 2002 Mw 7.9 Denali fault earthquake, which ruptured a fault plane 320 km long. He then used the SEM code to evaluate his result. After taking several cutting-edge results, such as 3D mantle tomography model S20RTS [Ritsema et al., 1999], CRUST2.0 crustal model [Laske et al., http://mahi.ucsd.edu/Gabi/rem.html, 2001], ETOPO5, and finite fault slip model of the earthquake [Ji et al., 2004] into account, we were able to fit the broadband teleseismic body waves (f<0.2 Hz), and long period surface waves (T>40 sec) fairly well.

Fault geometry of the November 3, 2002, Denali earthquake. Below: The five red boxes labeled A--E that denotes the surface projections of the five fault segments that were involved in the rupture. Bottom: Slip distribution along the five fault segments A--E shown.

To the right of each station are shown the data in black and the 3D SEM synthetic seismograms in red. Both data and synthetics are bandpass filtered between periods of 5 and 150 seconds. The epicenter is denoted by the black star. Top: P waves on the vertical component. Bottom: S waves on the transverse component. Note the strong amplification in the direction of rupture, as stations in N-America record much larger amplitudes than those in other parts of the world.

(to top)

HOME | GLOBAL | REGIONAL | NORMAL MODES | GLACIAL REBOUND | BEOWULF | PEOPLE | REPRINTS | DOWNLOADS