Back to the deformation and Stress Change Modeling home page

Back to the team's online recent papers pages


The Size and Duration of the Sumatra-Andaman Earthquake from Far-Field Static Offsets

Science, Vol 308, Issue 5729, 1769-1772, DOI: 10.1126/science.1113746, 2005.
[Printable article (1.7 Mb)]


P. Banerjee,
Wadia Institute of Himalayan Geology, Dehra Dun, 248001, India.


Fred F. Pollitz
U.S. Geological Survey, Menlo Park, California, USA


R. Bürgmann3
Department of Earth and Planetary Science and Berkeley Seismological Laboratory,
University of California, Berkeley, CA 94720, USA.

Non-technical summary. The 26 December 2004 Mw=9.2 Sumatra-Andaman earthquake is the largest earthquake ever recorded in the era of modern space geodesy. It produced static offsets of at least several millimeters at distances up to 4500 km away from the source region. We utilize data from 41 continuously operating GPS stations at distances > 500 km from the source region (Figure 1A) and 5 campaign sites from the near-source region (Figure 2) to constrain a segmented coseismic slip model. Modeling of this dataset reveals that: (1) total slip exceeded 5 meters along the entire ~1300 km length of the aftershock zone, (2) source size as measured by the dominant moment tensor components is 25% - 35% greater for the total static offset than for the longest measurable seismic period of 1 hour (0S2 period) (Figure 3), and (3) post-1 hour rupture of the northern ~650 km of the rupture (Andaman segment) is correlated with M>5 aftershock activity of the Andaman Islands portion of the megathrust.


Figure 1. (A) The GPS data are from stations belonging to the IGS, the BAKUSURTANAL Indonesian network, the Sumatran GPS Array operated by the Tectonic Observatory at Caltech and the Indonesian Institute of Science (LIPI), and the Indian "National Program on GPS" Network of the Department of Science & Technology made available by the India GPS Data Centre. Purple dots indicate aftershocks. (B) Black arrows are observed coseismic offsets at 41 continuous GPS sites which recorded the 2004 Sumatra earthquake. Ellipses indicate 95% confidence limits on the horizontal velocity vector. Red arrows indicate the predicted coseismic offsets on Model M3 of Banerjee et al. (2005) at observed sites plus selected other sites where continuous and campaign GPS will potentially be available.


Figure 2. Multi-plane fault geometry of the earthquake sequence. Superimposed are the hypocenters of M>4 earthquakes occurring from 26 Dec 2004 to 5 Jan 2005 (NEIC catalog); the subset of CMT aftershock focal mechanisms with reverse slip (plunge of tension axis > 45º); the 0, 50, and 100 km isodepth contours from Gudmundsson et al. (1998).


Figure 3. Moment tensor component Mrr derived from the CMT solution (periods < 300 sec), the seismic slip inversion of Ammon et al. (2005) (periods < 2000 sec), and free oscillations (Stein and Okal, 2005) compared with that of Models M1 and M3 derived from the far-field static offset. An average rake of 110º and average fault dip of 13.5º was used to convert scalar seismic moment of 6.0 x 1022 Nm estimated in Ammon et al. (2005) to the corresponding Mrr. Black curve is the rate of M>5 earthquakes on the Andaman segment (i.e., north of 8ºN) as a function of time after the mainshock, smoothed using a Gaussian scaling time of 0.25 hours. Shallow aftershocks in the back-arc region (8ºN to 10ºN and east of ~93.5ºE) are excluded.