Mw 8.3 Illapel, Chile

2015 September 16 22:54:33 UTC

Rupture imaging from the backprojection of P-waves (preliminary results)

Last update: 19/09/2015

Data and processing

We use P-wave velocity records at North-American stations, filtered between 0.5 and 1.0 Hz.
Signals are realigned according to theoretical arrival time from the USGS epicenter, using multichannel cross-correlation – see "Trace alignment".
We use the back projection method described in Satriano et al. (2012).

Stations
Station distribution - Cylindrical Projection
Stations
Station distribution - Stereographic Projection


USArray stations used for the back projection analysis.
(click on the images for a larger version)


Array response function and beam power

Back projection resolution is estimated by the array response function (ARF), computed at the lowest frequency of the band (0.5 Hz). The ARF allows to correctly evaluate the beam power images obtained from the back projection analysis.
Here we use to approaches for back projection imaging:
— linear trace stack weighted by trace semblance;
— pure trace semblance.
These two approaches have different resolution and enhance different features.

Linear stack, weighted by semblance



Array response function @0.5Hz Integrated energy
Array response function @0.5Hz - Linear stack, semblance weight
Maximum beam power over time Integrated energy
Maximum beam power over time - Linear stack, semblance weight
Array response function (left) and maximum beam power over time (right) for semblance-weighted linear stack. The star is the epicenter. Dark gray dots are aftershocks, located by SSN.
(click on the images for a larger version)

Pure semblance



Array response function @0.5Hz Integrated energy
Array response function @0.5Hz - Semblance
Maximum beam power over time Integrated energy
Maximum beam power over time - Semblance
Array response function (left) and maximum beam power over time (right) for pure semblance stack. The star is the epicenter. Light and dark gray dots are foreshocks and aftershocks, respectively, located by SSN.
(click on the images for a larger version)



Sources of High-Frequency emission

We extract over time local maxima of back projection beam power. Peak time is corrected for the directivity effect. These back projection peaks are locations of high-frequency (~0.5 Hz) coherent energy emitters.

For semblance-weighted linear stack, the circle amplitude is mostly related to the actual signal energy (with some weighting based on coherency). For pure semblance stack, the circle amplitude is proportional to waveform coherency.

Back projection peaks -
Linear stack, semblance weight 		Back projection peaks - Semblance
BP peaks
Back projection peaks - Linear stack, semblance weight
BP peaks
Back projection peaks - Semblance
Time coded back projection peaks for semblance-weighted linear stack (left) and semblance stack (right). Circle amplitude is proportional to beam power. The star is the epicenter. Light and dark gray dots are foreshocks and aftershocks, respectively, located by SSN. Isolines correspond to the 3, 5 and 7 m slip contours from from the USGS preliminary finite fault solution (V1).
(click on the images for a larger version)


Projected peaks along-dip and along-strike show similar trends for both back projection methods.
Smblance stack better enhances the rupture start, and shows coherent sources moving mostly northwards, with apparent velocity of ~1 km/s.

Back projection peaks, projected -
Linear stack, semblance weight 		Back projection peaks, projected - Semblance
BP peaks
Back projection peaks, projected - Linear stack, semblance weight
BP peaks
Back projection peaks, projected - Semblance
Time coded back projection peaks for semblance-weighted linear stack (left) and semblance stack (right), projected along-dip (top) and along-strike (bottom). Circle amplitude is proportional to beam power. Lines indicate reference rupture velocities.
(click on the images for a larger version)



Energy time function

We look at coherent energy release as a function of time by integrating back projection beam power on space at every time step.
The function associated to semblance stack is highly distorted, since it enhances waveform coherency at origin time.

Energy time function - Linear stack, semblance weight Energy time function - Semblance
Energy time function
Energy time function - Linear stack, semblance weight
Energy time function
Energy time function - Semblance
Space-integrated beam power as a function of time for semblance-weighted linear stack (left) and semblance stack (right).
(click on the images for a larger version)



Trace alignment

Trace alignment - Linear stack, semblance weight Trace alignment - Semblance
Trace alignment
Trace alignment - Linear stack, semblance weight
Trace alignment
Trace alignment - Semblance
Trace alignment for semblance-weighted linear stack (left) and semblance stack (right). Top panel is trace alignment according to theoretical arrival time from origin. Following panels show trace alignment corresponding to each of the detected back projection peaks. Blue curve is semblance function. Vertical blue line is the arrival time corresponding to back projection peak.
(click on the images for a larger version)



References

C. Satriano, E. Kiraly, P. Bernard, J.-P. Vilotte (2012). The 2012 Mw 8.6 Sumatra earthquake: evidence of westward sequential seismic ruptures associated to the reactivation of a N-S ocean fabric, Geophys. Res. Lett., 39(15), L15302, doi 10.1029/2012GL052387.