Foreshock Activity Related to Enhanced Aftershock Production D. - PowerPoint PPT Presentation

Full Introduction of Article Foreshock Activity Related to Enhanced Aftershock Production D. Marsan, A. Helmstetter, M. Bouchon, and P. Dublanchet (2014) Speaker: Ta-Wei CHANG (Ide Lab) Jun. 10 th , 2017

Introduction • Causal effect between fore- and main-shock? • Pre-event aseismic deformation Rapid loading of asperities • • Transient uncoupling extend from pre- to post-seismic phase? • Add to normal after-slip, affect aftershocks • Aftershocks bears information of initial fault uncoupling?

Data • Worldwide catalog by Advanced National Seismic System (ANSS) • 1/1/1980 ~ 9/19/2013 • 𝑛 ≥ 4.0 • Mainshocks: • 𝑛 ≥ 6.5 after 1/1/1981 • To avoid contamination by aftershocks of previous mainshocks: • Not preceded by 𝑛 ≥ 6.0 within a year • Within 𝑒𝑗𝑡𝑢𝑏𝑜𝑑𝑓 = 10×𝑀 𝑛 = 0.05×10 4.56 (𝑙𝑛) • 𝑛 : magnitude of first shock; 𝑀 𝑛 : rupture radius • 612 in total

Methods • Probability p that precursory seismicity acceleration is by chance: • Algorithm by Bouchon et al., [ 2013 ] • Before, within 50 (km) of mainshock • For each mainshock: 𝑂 H ≥ 𝑂 N 𝑜 N 1 N 2 𝑜 = 0 𝑜 + 1; T = 𝑈 2 −𝑈 2 L 𝑂 H < 𝑂 N -T L 0 • Max n from various T: 1, 5, 10 days; 1, 2, 3, 6, 12 months • Compare with Monte Carlo simulation: • Get n for 1000 Poisson time series in data sampling frequency # (< =>?@A BCDE> G < DACE ) • 𝑞 = H444

Relationship Between Fore- and Aftershock Activities Time series of event count • S𝐵: 𝑞 ≤ 0.1; 110 𝑛𝑏𝑗𝑜𝑡ℎ𝑝𝑑𝑙𝑡; 𝑡𝑗𝑕𝑜𝑗𝑔𝑗𝑑𝑏𝑜𝑢 𝑏𝑑𝑑𝑓𝑚𝑓𝑠𝑏𝑢𝑗𝑝𝑜 𝐶: 𝑞 > 0.1; 502 𝑛𝑏𝑗𝑜𝑡ℎ𝑝𝑑𝑙𝑡; 𝑝𝑢ℎ𝑓𝑠𝑡 Earthquake count 40.7 events 40% increase 29.1 events Dashed line: corrected for difference in magnitude of mainshocks (explained later)

Relationship Between Fore- and Aftershock Activities Regional Dependence • S𝐵: 𝑞 ≤ 0.1; 110 𝑛𝑏𝑗𝑜𝑡ℎ𝑝𝑑𝑙𝑡; 𝑡𝑗𝑕𝑜𝑗𝑔𝑗𝑑𝑏𝑜𝑢 𝑏𝑑𝑑𝑓𝑚𝑓𝑠𝑏𝑢𝑗𝑝𝑜 𝐶: 𝑞 > 0.1; 502 𝑛𝑏𝑗𝑜𝑡ℎ𝑝𝑑𝑙𝑡; 𝑝𝑢ℎ𝑓𝑠𝑡 Location of mainshocks à geographically independent

Relationship Between Fore- and Aftershock Activities Regional Difference • High correlation: precursory acceleration and number of aftershocks • Valid on regional scale 9: 𝑞𝑝𝑡𝑗𝑢𝑗𝑤𝑓 𝑑𝑝𝑠𝑠𝑓𝑚𝑏𝑢𝑗𝑝𝑜 3: 𝑏𝑛𝑐𝑗𝑕𝑣𝑝𝑣𝑡 • Of 14 regions: _ 2: 𝑝𝑞𝑞𝑝𝑡𝑗𝑢𝑓 • Exception 1: Izu-Bonin (one strong aftershock sequence) • Exception 2: Northern America (oceanic transform faults + continental sources) Oceanic transform faults: intensive foreshocks; little aftershocks • Gofar Transform Fault: pre-seismic acceleration creep halted by • mainshock Fault-specific: high porosity & thermal gradient? •

Relationship Between Fore- and Aftershock Activities Regional Differences— Positively Correlated

Relationship Between Fore- and Aftershock Activities Regional Differences— Other Cases Opposite Pink: large aftershock sequence removed (Became ambiguous) Ambiguous

Relationship Between Fore- and Aftershock Activities Possible Mechanisms Mainshocks of A are larger than B? 1. More productive • Regions of A: higher long-term seismicity? 2. More aftershocks • Preshocks causing acceleration also trigger their own aftershocks? 3. Aftershock: mainshock- and preshock-contributed • Aftershocks of A are larger than that of B? 4. Triggering more of their own aftershock • Magnitude completeness difference? 5.

Relationship Between Fore- and Aftershock Activities Mechanism 1 • Mainshocks of A are larger than B? Rejected! Mean magnitude: S𝑛 e : 6.88 𝑛 g : 6.90 , actually opposite • Productivity law: 𝑂 ∝ 𝑓 i6 • N: number of aftershocks; m: magnitude of mainshock • # Aftershocks— magnitude of mainshock • Cutoff: 50 (km); 𝛽 = 1.82; 𝑁 lem = 7.4 o p o q L = 1.03 •

Relationship Between Fore- and Aftershock Activities Mechanism 2 • Regions of A: higher long-term seismicity? Rejected! • No clear clustering • Independent of tectonic region • Preseismic rate: S𝐵: 3.04 𝐶: 3.27 • Only difference: acceleration!

Relationship Between Fore- and Aftershock Activities Mechanism 3 • Preshocks causing acceleration also trigger their own aftershocks? • Rate of preshocks triggering aftershocks: , where K m = S𝑙𝑓 H.}N6 ; 𝑛 ≤ 7.4 ro rm = s(6) L mtu • 𝑙𝑓 H.}N×~.• ; 𝑛 > 7.4 • Average over 612 events: 𝐿 = 3.6; 𝑙 = 8.20×10 •‚ • For each mainshock, within 1 year after event: • For preshock at: −365 < 𝑢 < < 0 (𝑁𝑏𝑗𝑜𝑡ℎ𝑝𝑑𝑙) „ = „ : 𝑡𝑗𝑛𝑗𝑚𝑏𝑠) H ‰‚5tu•m ? HH4 ∑ 𝐿(𝑛 < ) log • 𝑂 (𝑂 ‹ �< ƒ u•m ? „ = 1.20 „ − 𝑂 ‹ 𝑂 • ƒ „ ) „ − 𝑂 ‹ ƒ − 𝑂 ‹ = 11.6 (9.55 𝑢𝑗𝑛𝑓𝑡 𝑏𝑡 𝑛𝑣𝑑ℎ 𝑏𝑡 𝑂 • Actual 𝑂 ƒ • Needs k 9.55 times larger in 20 days prior to mainshock to explain this

Relationship Between Fore- and Aftershock Activities Mechanism 4 • Aftershocks of A are larger than that of B? • Productivity of A’s and B’s aftershocks: 𝐿 ƒ = 1.54𝐿 ‹ • Difference in magnitude • Almost equal b-values: S𝑐 ƒ : 1.21 ± 0.06 𝑐 ‹ : 1.18 ± 0.03 • Foreshock of Tohoku-Oki counts it as aftershock • After removal: 𝐿 ƒ = 1.08𝐿 ‹ • Doesn’t suppress correlation: gain in production of A is 34% than 40% • This sequence needs careful analyzation for being the end of 2-month long swarm

Relationship Between Fore- and Aftershock Activities Mechanism 5 • Magnitude completeness difference? Rejected! • Same cutoff magnitude for A and B • Correlation also exists when changing from 𝑛 • = 4.0 𝑢𝑝 𝑛 • = 5.0

�� Comparison With Clustering Models • Foreshock— aftershock relationship: can’t be explained by clustering models • Null hypothesis: ETAS Assume acceleration: earthquake clustering properties • Doesn’t predict correlation between acceleration and aftershock! • • ETAS: triggering rate density: • 𝜇 𝑦, 𝑧, 𝑢 = 𝜈 + ∑ 𝜇 ’ (𝑦, 𝑧, 𝑢) Earthquake 𝑗 before time 𝑢 ; 𝜇 ’ : rate density of triggering by earthquake 𝑗 • (–—˜) (”•H)• “ 𝜇 ’ 𝑦, 𝑧, 𝑢 = 𝐵 ’ 𝑓 i6 “ (𝑢 − 𝑢 ’ + 0.08) •H × • (–œ˜) L › ) N™(š › t• “ › Triggering: ≤ 10 𝑧𝑓𝑏𝑠𝑡 • Region: 2000×2000 (𝑙𝑛 N ) ; periodic boundary condition • Background seismicity: 3 earthquakes/ day • Random magnitude from ANSS; 𝑛 ≥ 4 •

Comparison With Clustering Models • To reproduce number of aftershock/ mainshock: • 𝛽 = 2.3; 𝑀 4 = 8 𝑛 ; 𝛿 = 3.5 • 𝛽: largest within 𝛽 < 𝑐 𝑚𝑜10 ; larger: enhance acceleration • Dispersion of data points: 𝐵 ’ = 3×10 •~ 1.8𝑣 + 0.1 , 0 < 𝑣 < 1 • 1000 synthetic data, each 50000 years long, S𝐵: 95.9 𝐶: 588.6 • Applied correction of difference in magnitude of mainshocks (as mechanism 1) • Aftershocks by foreshock (as mechanism 3) „ − 𝑂 ‹ „ • To see ∆= 𝑂 ƒ − 𝑂 ‹ − 𝑂 = 10.4 ƒ Probability: 0.13% (Gaussian fit) (5 out of 1000) • • Fluctuation of ETAS: unlikely to explain • Tectonically wide-spread • Inward migration of foreshocks: observed; weaker than simulation

Comparison With Clustering Models Data ETAS-modeled

Spreading of Aftershock Zone • Aseismic process required! • Ensemble-averaged clustering law and observation error not enough • A is 20% more wide-spread in space than B • Foreshocks: 1 year • Aftershocks: 20 days • Epicentral distance normalized: 𝑀 = 0.005×10 4.56 • Rotated so y=0 best perpendicular to 1-year aftershock distribution • Circle: unit circle • Ellipse: 1 s confidence ellipse • Upper-right: smaller: B

Spreading of Aftershock Zone • A is more wide-spread in space than B • ETAS: 5% only (20- days) • 1-day: 25% more • 20-day: 20% more • 1-year: 11% more (may be contaminated by background) • Scaling of aftershock productivity— magnitude (or rupture length): • 𝑓 H.}N×6 = 𝑀(𝑛) ›×˜.¤› E>¥(˜¦) L = 𝑀(𝑛) H.5} • 23% more in L to explain 40% more aftershocks • Similarity between 20% and 23% • SSpatial spreading Triggering of aftershocks of A : same ETAS-failed process! • Transient diffusing aseismic deformation-/ creeping- triggering fore- and aftershocks

Discussion & Conclusion • Earthquake process: complex! • Creeping without earthquake/ earthquake without creeping • This study: • More aftershock following foreshock Aseismic deformation transient • Compound rupture • • Observable in averaged sense only • Compound rupture with slow initial phase recorded; but: • Fewer aftershocks for oceanic earthquakes • Observed here is in averaged sense; not individually • Tectonic implication too regionally specific • Better constrain the a priori probability of large ruptures by slow deformation transients?