We imaged the source of the April 11 2012 M8.6 earthquake off-shore Sumatra by back-projection of the seismic waveforms recorded by large arrays at teleseismic distance. This source imaging technique allows to track the location and migration of the multiple sources of high-frequency radiation that compose an earthquake rupture front. Our particular approach is based on a high-resolution array-processing technique, MUSIC (Multiple Signal Classification -- Schmidt et al,1982; Goldstein and Archuleta,1991; Meng et al, 2011, 2012a). This technique achieves higher resolution than conventional beamforming. In addition, we developed a "reference window" strategy to remove the systematic "swimming" artifact (Meng et al, 2012b).
Data selection and processing
We considered the seismic data recorded by various seismic arrays at epicentral distances between 30 and 90 degrees.
The large aperture and dense spacing of the European and Japanese Hi-net network (shown below)
provided good spatial resolution for this event.
The P waveforms were filtered between 0.5 and 1 Hz,
selected by the signal-to-noise ratio and mutual coherency of their initial 10 seconds,
and aligned by multi-channel cross-correlation.
Selected stations from the European and Hi-net (Japan) networks.
We applied the MUSIC back-projection technique on 10 seconds long sliding windows.
The movies below show the raw results: the warm colors indicate the position of the high frequency radiation on the fault, as seen by the Europe and Japan networks, respectively.
Other figures below show more detail and an interpretation of the spatio-temporal evolution of this extremely complex earthquake rupture.
Back-projection source imaging based on seismic data from Europe (left) and Japan (right) networks.
The colors indicate the amplitude of the MUSIC pseudo-spectrum, back-projected to the source region, normalized by the beamforming power,
in logarithmic scale with units of dB.
The white star is the mainshock epicenter and the green circles are the epicenters of the first day of aftershocks from the NEIC catalog.
Time relative to hypocentral arrival time is shown on top.
The trench and coastline are shown by white curves.
Top: Spatio-temporal distribution of high-frequency radiation as seen from Japan (left) and Europe (right).
Colored dots indicate the position of peak high-frequency radiation identified from each frame of the movies above.
Squares correspond to significant secondary peaks with MUSIC pseudo-spectrum at least 50% as large as the main peak in the same frame.
The size of the symbols scales with beamforming amplitude.
Their color indicates their timing relative to hypocentral time (color scale on right).
Black dots are the epicenters of the first day of aftershocks from the NEIC catalog.
Gray dashed lines indicate possible fault planes consistent with global CMT solutions.
A possible interpretation of the rupture history.
Stage 1: the earthquake started as a bilateral rupture, about 150 km long,
on a fault striking NNW (here coined "fault A").
This strike is consistent with one of the planes of the CMT solution (show on the left).
This stage generated the strongest radiation.
Stage 2: the rupture then branched into a different, almost orthogonal fault ("fault B") consistent with the conjugate plane of the CMT solution.
This stage was bilateral and broke about 300 km.
Rupture on the NNE segment of fault B reached near the trench.
Stage 3: the rupture finally branched into a third orthogonal fault ("fault C"), unilaterally for about 100 km heading WNW, until it reached near the NinetyEast ridge.
Stage 4: a weak, late rupture possibly involved a different fault strand parallel to fault C.
The location of the M8.2 aftershock, two hours later (not shown here), is consistent with rupture
along the SSE segment of fault C.
Spatio-temporal details of the rupture:
timing and position of the high frequency radiators relative to the hypocenter.
The position reported in this figure is not along a single axis, but along
one or the other axis defined as X and Y in the sketch (inset)
and indicated by labels on the left in the main plot.
The results from both arrays are plotted together.
Circles are results from Europe and squares from Japan. Filled and open symbols
indicate principal and secondary radiators, respectively.
The two arrays are globally consistent and in some stages they provide complementary information on the rupture process.
Colors indicate position orthogonal the current axis
(stability of the color indicates that rupture is mainly along the chosen axis).
The slopes in this figure indicate an overall rupture speed of order 2 km/s,
although there is variability at smaller scales.
The NE segment of the rupture on fault B was delayed by about 15 s.
Source amplitude as function of time
Beamforming source amplitude as a function of time seen by the European array (green) and Japanese Hi-net (blue). The yellow and red dots indicate the amplitude of the secondary sources.
L. Meng, J.-P. Ampuero, Y. Luo, W. Wu and S. Ni
Apparent frequency-dependent source properties induced by artifacts in teleseismic back-projection,Submitted to Earth, Planets and Space, special issue "The 2011 Tohoku Earthquake"
L. Meng, A. Inbal and J.-P. Ampuero (2011). A window into the complexity of the dynamic rupture of the 2011 Mw 9 Tohoku-Oki earthquake Geophys. Res. Let., 38, L00G07, doi:10.1029/2011GL048118
Meng, L., J.-P. Ampuero, A. Sladen, and H. Rendon (2012), High-resolution backprojection at regional distance: Application to the Haiti M7.0 earthquake and comparisons with finite source studies, J. Geophys. Res., doi:10.1029/2011JB008702, in press.
Schmidt, R. "Multiple emitter location and signal parameter estimation," Antennas and Propagation, IEEE Transactions , vol.34, no.3, pp. 276- 280, Mar 1986
Goldstein, P. and R. J. Archuleta (1991). Deterministic Frequency-Wavenumber Methods and Direct Measurement of Rupture Propagation During Earthquakes Using a Dense Array: Theory and Methods, J. Geophys. Res., 96, 6173-6185.