Rupture during 8 minutes propagating to the NNW for the 2004 Sumatra-Andaman Islands Earthquake - is this earthquake larger than the 1960 Chile earthquake?

Anthony Lomax

Anthony Lomax Scientific Software, Mouans-Sartoux, France, anthony@alomax.net, www.alomax.net

updated 16 Feb 2005

See also: Rapid estimation of faulting extent for large earthquakes by locating the end of rupture: application to the 2004, Mw=9.0 South Asia mega-thrust

This figure shows short period, 1-second Gaussian-filtered seismograms from the 26 December, 2004, Mw=9, Sumatra-Andaman Islands earthquake for two pairs of stations.  Each pair includes stations at about the same great-circle distance (GCD) from the hypocenter at opposing azimuths along an arc through the likely rupture zone as defined by early aftershocks (NEIC, 2005) and by plate boundaries.   Two stations are to the NNW of the hypocenter (OBN in Russia at 70° GCD, 328° azimuth, and ULHL in Kyrgyzstan at 43° GCD, 338° azimuth) and two stations are to the SSE of the hypocenter (NWAO in Australia at 41° GCD, 152° azimuth, and CASY in Antarctica at 70° GCD, 173° azimuth).  P and S indicate the predicted arrival times using model ak135 (Kennett et al., 1995) for the first P and S waves from the hypocenter.

The two closer stations ULHL and NWAO exhibit some short period signal after the first S arrival that could be S energy.  However, the farther stations OBN and CASY show little signal after the first S arrival indicating that the short period signal at all stations is primarily P energy radiated directly from the source.  The amplitude, duration and timing relations of these signals thus give us important information about the source process.

The first P signal arrives at about the same time at the two closer stations (ULHL and NWAO) and at the two farther stations (OBN and CASY), consistent with each pair being at the same distance from the hypocenter.  The smaller duration of the stronger signal at ULHL relative to that at NWAO and at OBN relative to that at CASY is consistent with rupture propagation from the hypocenter towards the NNW, i.e. CASY and NWAO are further than ULHL and OBN respectively from a point of rupture termination to the NNW of the hypocenter.

The duration of the short period signal on these and other seismograms implies rupture duration of around 8 minutes (also inferred by Jeffrey Park in an earlier contribution to the IRIS Special Report for this event).  Taking the point of rupture termination near the northern limit of the early aftershocks (NEIC, 2005) implies a rupture length of around 1200 km (see also: Rapid estimation of faulting extent for large earthquakes).  These results imply an average rupture velocity of about 1200 km / 480 sec or about 2.5 km/sec.

A short period seismogram from the great 1960 Chile earthquake (see http://neic.usgs.gov/neis/eq_depot/world/1960_05_22_seismogram.html) shows a duration of stronger signal of about 4 minutes.  If this duration corresponds to the rupture duration for the 1960 event, this result implies that the 2004 Sumatra-Andaman earthquake may be larger than the 1960 Chile earthquake!

Interactive viewing of these seismograms with SeisGram2K (requires Java plugin):

References

Kennett, B.L.N. Engdahl, E.R. & Buland R., (1995).  Constraints on seismic velocities in the Earth from travel times, Geophys. J. Int., 122, 108-124

NEIC, (2005).  Magnitude 9.0 OFF THE WEST COAST OF NORTHERN SUMATRA, http://neic.usgs.gov/neis/bulletin/neic_slav.html

Credits

Alberto Michelini first noted the relevance of the 1960 short period seismogram to this analysis.  2004 data provided by the IRIS Global Seismographic Network, and distributed through the IRIS Data Management System. 

Anthony Lomax, Anthony Lomax Scientific Software, Mouans-Sartoux, France, anthony@alomax.net, www.alomax.net