RUPTURE OF THE 2004 SUMATRA-ANDEMAN EARTHQUAKE

The devastating earthquake of Dec. 26, 2004 broke a large portion of the Sumatra-Andeman trench. However, it proved difficult to rapidly determine the area and timing of rupture, largely because the later parts of the (very long) direct body-wave arrival overlapped with later arrivals that usually don't need to be accounted for. We developed a new approach to image regions which radiated high-frequency energy without a priori assumptions on the fault geometry or timing. We used the Hi-Net array of Japan, which includes ~700 high quality borehole seismometers, as an antenna to map the progression of slip by monitoring the direction of high-frequency radiation. A back-projection method was applied, in which seismograms are shifted in time so that energy from a specific location near the source will stack coherently. All locations in the vicinity of the earthquake are considered, and the progression of rupture is mapped out by the space-time occurrence of coherent radiation.

Our analysis finds a steady, near-constant propagation of rupture starting near the hypocentral region and progressing northward for ~1300 km along the trench for about 500 seconds. Thus, the Sumatra earthquake seems to have had the character of a large-scale slip pulse. The average velocity of rupture propagation is about 2.8 km/sec, similar to that for a few other great subduction zone earthquakes. We compared the aftershock zones and rupture durations of four great earthquakes of the recorded era; the rupture of the 2004 event, in both duration and extent is the longest ever recorded. The shape or aspect ratio of the aftershock zones of great earthquakes vary significantly. A comparison of the space-time history of moment release (i.e., map of inferred slip distribution) with that obtained from analysis of longer-period seismograms indicates that short-period waves radiated proportionally from same parts of the fault as longer-period waves. In particular, although slip appears to be greater in the southern half of the rupture zone, the character of the rupture process (e.g. rupture velocity) appears not to change. Evidence for slow slip in the north remains elusive and equivocal.