Under the macroscope : the dynamics of very large earthquakes

Transcription

1 Under the macroscope : the dynamics of very large earthquakes Jean Pablo Paul Ampuero Caltech Seismolab / Université Côte d Azur, IRD, Géoa

2 Enabled by global earthquake source product Lingling Ye (Caltech), Mar=n Vallée (IPGP) and Gavin Hayes 5US

3 namic rupture models of e 2016 Mw 7.8 Kaikoura, NZ rthquake seismic back- projec=on sen Meng, UCLA) Thomas Ulrich and Alice Gab SeisSol Team (LMU, Munich

4 Trade- offs in earthquake source studies CI.PAS. SH φ=320.5, =65.7 CN.LLLB. SH φ=331.2, =78.9 IU.CCM.00 SH φ=343.5, =56.5 IU.DWPF.00 SH φ=350.3, =44.8 CN.RES. SH φ=354.5, =91.7 IU.SSPA.00 SH φ=356.2, = φ=292.1, φ=122.3, =85.1 =90.8 IU.CASY.00 SH CI.BAR. φ=181.7, =97.7 P φ=320.2, IU.SBA.00 SH =63.7 φ=190.7, =80.1 CN.LLLB. IU.SNZO.00 SH P φ=224.5, =94.8 φ=331.2, =78.9 IU.POHA.00 SH φ=291.2, =88.1 IU.RSSD.00 P IU.KIP.00 SH φ=336.7, φ=292.1, =90.8 =66.2 IU.CCM.00 P φ=343.5, =56.5 IU.DWPF.00 P φ=350.3, = IU.SSPA.00 P φ=356.2, = CN.SCHQ. SH 2698 φ= 4.1, = CN.DRLN. SH φ=11.4, = II.BORG.00 SH 2892 φ=19.7, =90.0 II.ESK.00 SH φ=32.5, = II.CMLA SH 120 φ=38.7, Time =70.2 (s) G.ECH. SH φ=41.4, =96.0 IU.PAB.00 SH φ=46.4, =85.1 IU.LSZ.00 SH φ=108.1, =96.7 II.SUR.00 SH φ=122.3, =85.1 IU.CASY.00 SH φ=181.7, =97.7 IU.SBA.00 SH φ=190.7, =80.1 CN.FRB. P IU.SNZO.00 SH φ= 2.3, =79.8 φ=224.5, CN.DRLN. =94.8 P φ=11.4, =66.8 IU.POHA.00 II.BORG.00 SH P φ=19.7, =90.0 φ=291.2, =88.1 G.ECH. P φ=41.4, =96.0 IU.KIP.00 SH IU.PAB.00 P φ=292.1, φ=46.4, =90.8 =85.1 II.SACV.00 P φ=60.3, =58.3 GT.DBIC. P φ=77.1, =71.8 II.ASCN.00 P φ=89.4, =58.4 IU.TSUM.00 P φ=108.8, =85.9 II.SUR.00 P φ=122.3, =85.1 II.HOPE.00 P φ=151.5, =47.6 IU.PMSA.00 P φ=174.6, =48.9 IU.SPA.00 P φ=180.0, =73.8 IU.CASY.00 P φ=181.7, =97.7 IU.SBA.00 P φ=190.7, =80.1 G.DRV.00 P φ=192.7, =93.6 IU.SNZO.00 P φ=224.5, =94.8 σ E = 3.05 MPa Time (s) Dep φ = 310.0, δ = 18.0, λ = Depth Distance a V r = 2.40 km/s, Var. = Shear Stress (MPa) H 0 = 29.6 km, H c = 18.1 km Distance along Strike (km) CN.FRB. P 491 CI.PAS. SH φ= 2.3, =79.8 φ=320.5, =65.7 CN.DRLN. P 600 CN.LLLB. SH φ=11.4, =66.8 φ=331.2, =78.9 II.BORG.00 P 373 IU.CCM.00 SH φ=19.7, Coseismic =90.0 Slip(m) φ=343.5, =56.5 G.ECH. P 274 IU.DWPF.00 SH φ=41.4, =96.0 φ=350.3, =44.8 IU.PAB.00 P 578 CN.RES. SH φ=46.4, =85.1 φ=354.5, =91.7 II.SACV.00 P 715 IU.SSPA.00 SH φ=60.3, =58.3 φ=356.2, =56.7 GT.DBIC. P 636 φ=77.1, =71.8 II.ASCN.00 P 835 φ=89.4, =58.4 IU.TSUM.00 P 481 φ=108.8, =85.9 II.SUR.00 P 527 φ=122.3, =85.1 II.HOPE.00 P 717 φ=151.5, =47.6 IU.PMSA.00 P 384 φ=174.6, =48.9 IU.SPA.00 P 292 φ=180.0, =73.8 IU.CASY.00 P 113 φ=181.7, =97.7 IU.SBA.00 P 331 φ=190.7, =80.1 G.DRV.00 P 166 φ=192.7, =93.6 IU.SNZO.00 P 130 φ=224.5, =94.8 IU.PTCN.00 P 543 φ=250.6, =53.2 G.PPT. P 373 φ=256.3, =72.2 IU.KIP.00 P 121 φ=292.1, =90.8 CI.BAR. P 184 φ=320.2, = CN.LLLB. P 257 φ=331.2, =78.9 IU.RSSD.00 Shear P Stress (MPa) 20 φ=336.7, =66.2 E r = 5.16e+16 IU.CCM.00 P E r /M = 1.10e 05 φ=343.5, = fc = 5.2 mhz IU.DWPF.00 P n 0 = φ=350.3, =44.8 n 1 = 1.74, r 1 = 0.17 IU.SSPA.00 P 469 φ=356.2, = n 2 = 1.58, σ = 0.11 r 2 = 0.23 MPa p 2778 High fidelity e along Strike (km) 4.9 Less uncertainty Moment rate Depth (km) 17.2 σ 0.15 = 5.12 MPa Time σ E = 3.05 MPa Mo_rate (x10 20 Nm/s) Depth (km) _Peru Moment (N m) M 0 = 4.67 x Nm 29.6 M w = e along Strike (km) T c = 67.7 s, T d = s 2.s 2.s 9 High defini=on More detail Frequency (Hz) CN.SCHQ. SH 0 Coseismic Slip(m) φ= 4.1, 120 = CN.DRLN. SH Time (sec) φ=11.4, =66.8 II.BORG.00 SH φ=19.7, =90.0 II.ESK.00 SH φ=32.5, =92.6 II.CMLA.00 SH m)

5 lobal source studies et al (JGR 2016) 116 M7+ shallow subduction zone thrust earthquakes inite source inversions with teleseismic data, Hz Robust source time functions (STF, moment rate) Uniform method and careful manual analysis

6 obal source studies t al (JGR 2016) 16 M7+ shallow subduction zone thrust earthquakes nite source inversions with teleseismic data, Hz obust source time functions (STF, moment rate) niform method and careful manual analysis

7 uesjons to address under the macroscope What are the common features of earthquakes? Do small and large earthquakes start equal? Are earthquakes self- similar at all magnitudes? How are earthquakes different from each other? Is there such a thing as a freak event? What do those similari=es and differences tell us about earthquake dynamics?

8 What general patterns do the STF follow?. Bin STFs by magnitude, 20 nearest neighbours. In each bin, at each point in time, compute median STF

9 Median STFs have linear onset, same for all magnitudes Mw>7.

10 Normalize each STF by its duration Scale them such that they integrate to 1 Compute median of normalized STF On average, all STFs can be scaled to a very simple, quasi-triangular shape

11

12

13

14

15 uctuajons around e median STF Fit a function to STFs: Mul$plica$ve noise STF residu magnitud normalize fi\ng fun siduals

16 STF fluctua=ons are mul=plica=ve and Gaussian Empirical cumula=ve distribu=on of STF residuals

17 STF fluctua=ons are mul=plica=ve, Gaussian and Brownian

18 Implications for moment / duration scaling. Linear growth suggests M 0 ~ T 2 scaling. In contrast to the widely reported M 0 ~ T 3 scaling à scaling break?

19

20 Summary of observed STF characterisjcs, Mw>7 All STFs can be scaled to a common, quasi- triangular shape Onsets are linear and the same for all Fluctua=ons are mul=plica=ve, Gaussian and Brownian

21 Why is linear moment rate growth surprising? Self-similar model for small earthquakes: Circular rupture with constant stress drop and constant rupture speed M 0 t 2!! ery large earthquakes: elongated rupture nce seismogenic width is saturated: oment grows slower than quadratic ut linear trend is observed after 5-10 s, efore rupture saturates the seismogenic width Seismogenic width Slip rate (m/s)

22 Implications for Rupture Growth Scaling. Observed STF growth is linear. If rupturing area grows as. and average slip grows as. Seismic moment. Moment rate exponent. Since we observe linear growth!"#!!!(!)!!!(!)!!!! (!)!(!)!(!)!!!!! =! +! 1!!"# ~ 1! +! ~!. Self-similar pulse or crack!!! = = 3 à How can we lower the moment rate growth?. Lower alpha, lower beta, or combination of both?. Pulse-like rupture with areas of systematic slip deficits?

23 Slip velocity: Seismic Aseismic Locked Average stress Peak slip velocity Extracted from Junle Jiang and Nadia Lapusta s dynamic earthquake cycle simula=ons.

24 Nuclea$on Intermediate- size event unzipping part of the lower edge of the coupled zone (Junle Jiang, Calte Propaga$on Pre- stress Arrest Final stress

25 April , Mw 7.8 Gorkha, al earthquake upture confined in depth igh- frequency deeper than low- equency slip, concentrated along e deep edge of the locked zone ish contours: slip (frequencies < 0.1 Hz) red circles: High- frequency radia=on (1 Hz) ac et al (Nature Geoscience, 2015)

26 CONCLUSIONS Today we have enough data to uncover general patterns of earthquake rupture Focusing on temporal evolution facilitates testing of conceptual rupture models A few things are certain. Large earthquakes are small earthquakes that did not stop. Individual earthquakes have large variability, but on average they follow a remarkably simple pattern. Observed pattern systematically deviates from standard models after few seconds. Pattern makes rupture evolution weakly predictable More questions than answers. Physical origin of the pattern?. What dynamical models can explain the linear STF growth?. What causes break of self-similarity at ~1s? F o o d thought! f o r

27 hat s next? Analysis of strike- slip ruptures Source studies with uncertainty quan=fica=on Develop methods for systema=c analysis across the magnitude range of scaling transi=on from M6 to M8+ à break of self- similarity, scaling of rupture aspect ra=o Develop dynamic rupture models consistent with these observa=ons

28 thank you Pulsars Pulses?