Repeating Earthquakes Olivier Lengliné - IPGS Strasbourg Cargese school

Please interrupt Questions / remarks

1 – Review of Repeating earthquake observations & interpretations 2 – Two examples of application

Observations - Waveforms Nadeau & Johnson, 1998

Parkfield, California – Mw6.0 USGS Bakun et al., 2005 De Bilt, The Netherlands

Uchida et al., 2012 Time (s) Off Kamaishi, Japan – M4.9

Chen et al., 2008 Chihshang fault, Taiwan

Time (s) 9 events 13 events 19 events Soultz-Sous-Forêts geothermal reservoir, France BRGM

San-Andreas Fault Schaff & Beroza, 1998 Rubinstein et al., 2012

u(t) = Source * Path * Station

Station is the same Change in medium property, [e.g Poupinet et al., 1984] Change in source properties, [e.g. Lengliné & Got, 2011]

Poupinet et al., 1984 Lengliné and Got, 2011 Directivity Velocity variations

u(t) = Source * Path * Station Station the same Change in medium property, [e.g Poupinet et al., 1984] Change in source properties, [e.g. Lengliné & Got, 2011] ! Homogeneous medium  waveform similarity

Observations - Locations Waldhauser et al., 2004

Murray & Langbein, 2006 Parkfield

Off Kamaishi Okada et al., 2002 Relative moment released normalized by each maximum value Moment release distribution

Earthquake relative relocation  Uncertainties P-wave picks  Uncertainties of the velocity model

Earthquake relative relocation  Uncertainties P-wave picks  Uncertainties of the velocity model  More precise data: time delays estimated from cross-correlation  Ray geometry – rotation  Do not correct absolute position

Earthquake relative relocation  Uncertainties P-wave picks  Uncertainties of the velocity model  More precise data: time delays estimated from cross-correlation  Ray geometry – rotation  Do not correct absolute position From cross-correlation  centroid location Got et al., 1994 Waldhauser & Ellsworth, 2000

Earthquake relative relocation  Uncertainties P-wave picks  Uncertainties of the velocity model  More precise data: time delays estimated from cross-correlation  Ray geometry – rotation  Do not correct absolute position From cross-correlation  centroid location Got et al., 1994 Waldhauser & Ellsworth, 2000 See Tutorial this afternoon for Methods

Lengliné & Marsan, 2008 Size = Assumed stress drop + circular crack + moment – magnitude relation

Bourouis & Bernard, 2007 Chen et al., 2008 Soultz-sous-Forêts Taiwan Radius estimated from corner frequency

Murray & Langbein, 2006 Rau et al., 2007 Clusters of co-located, similar waveforms earthquakes, appears at the transition between fully locked and fully creeping areas

Waldhauser & Schaff, 2008 Example from Northern-California Parkfield Is it related to fault slip velocity ?

Rubin et al., 1999 San Andreas Fault Streaks of microearthquakes – along slip direction Rheological / frictional / geological / geometrical transition ?

Observations - Timing

YearNumber μ Δt = 24.5 yr σ Δt = 9.5 yr COV = 0.37 Time (years) Earthquake number Parkfield

Repeaters off Kamaishi Repeating interval = /- 0.5 yrs Time (years)

Waldhauser et al., 2004 Distance along strike (km) Year San-Andreas fault at Parkfield

Waldhauser et al., 2004 Distance along strike (km) Year Periodic repeating ruptures

Rubinstein et al., 2012 Quasi-periodic behavior of the slip activity

Aseismic slip No interacting asperity The simplest model A locked seismic patch embedded in a fully creeping zone

Slip on the creeping part Slip on the seismic asperity Time Slip

 Aseismic slip on the fault = seismic slip Time Slip d seis

 Aseismic slip on the fault = seismic slip  Elastic solution for a circular crack

 Aseismic slip on the fault = seismic slip  Elastic solution for a circular crack

 Aseismic slip on the fault = seismic slip  Elastic solution for a circular crack  Constant stress drop

Chen et al., 2007

1st Hypothesis The constant stress drop hypothesis is not correct Empirical fit to the data then suggests in order to have T r ~ M 0 1/6 Implies that the stress-drop is higher for small events. Stress levels reach 2 GPa for the smallest events (more than 10 times laboratory strength) This result is at odds with estimates based on seismic spectra Relation not consistent with established scaling relations for large earthquakes.

Imanishi & Ellsworth, 2006

Chen & Lapusta, 2009

But not the estimated plate velocity – streaks close to locked section  reduced velocity ?

Slip on the creeping part Slip on the seismic asperity Time Slip Seismic slip

Uchida, 2014 Off Kamaishi repeating sequence following Tohoku, 2011, Mw9 earthquake

Lengliné & Marsan 2008 Schaff & Beroza, 1998

Following Parkfield, 2004, Mw6 event

Response of a velocity strengthening area to a stress-step Marone, 1991 The Omori like decay of RES is well rendered by the slip evolution of the creeping area following a stress step

Nadeau & McEvilly, 1999

Bourouis & Bernard, 2007

Bouchon et al., 2011

Kato & Nakagawa, 2014 Kato et al., 2012

Repeating earthquake are local (sparse) creep-meter at depth Difficult to quantify if the seismic slip reflects the surrounding aseismic loading

Time after 01/01/1984 (years) Number of earthquakes Complications to the idealized picture Repeating sequence of small micro-earthquakes at Parkfield

Time after 01/01/1984 (years) Number of earthquakes Complications to the idealized picture Repeating sequence of small micro-earthquakes at Parkfield

Interactions from nearby small events Chen et al., 2013 More isolated events = more periodic

Vidale et al., 1994 How can strength of the interface build up so quickly between 2 events ? Healing of the interface

What is an asperity ? (geometrical/frictional/geological …) What is the lifetime of an asperity ? In which case do we observe periodicity ? (density of asperity) Are repeating LFE earthquakes obeying a similar mechanism ? Questions

2 examples of use of repeating earthquake sequences -Earthquake detection and time activity (with P. Ampuero) -Variation of source properties (with L. Lamourette, L. Vivin, N. Cuenot, J. Schmittbuhl)

Parkfield

Landweber deconvolution Example for one pair at one station

Landweber deconvolution All pairs at all stations

Sparse deconvolution 54 new detected events in the first 20s following a repeating earthquakes

Stack aftershock sequence Typical rupture duration

Wang et al., 2014

Omori’s law extended almost up to the rupture duration Implies a very low c-value and thus a very large stress changes in the R&S Dieterich framework Seismicity rate Time (t/t a ) No flatenning of the earthquake rate at early times Is this particular to the repeating earthquakes ?

Station surface sites 150 Hz sampling frequency months long circulation test 411 earthquakes recorded Largest magnitude event M2.3

4 groups of similar events Relocation suggest a similar location Each group have at least one event larger than 1.4 4/6 of the largest events of the circulation are included in these groups

SVD analysis (Rubinstein & Ellsworth, 2010 ) Up to a factor x 300 of moment ratio

SVD analysis (Rubinstein & Ellsworth, 2010 ) Up to a factor x 300 of moment ratio

For the largest event of each group

Corner frequency of the largest event of each group f c ~ [10-20] Hz

Wiener filter (equivalent to spectral ratio) Same rupture area

The difference of seismic moment reflects a difference of seismic slip/ stress drop Increase of pore pressure lowers the normal stress on the fault plane 2 effects: Shear failure promoted (reach the Coulomb enveloppe) Stabilizes the slip Several instances of aseismic movements have been suggested in the Soultz reservoir We are observing a transition from unstable to stable slip on the interface Bourouis & Bernard, 2007

Thank you