1. -Liu and Rice (2005), Aseismic slip transients emerge spontaneously in three-dimensional rate and state modeling of subduction earthquake sequences, JGR -Liu and Rice (2007), Spontaneous and triggered aseismic deformation transients in a subduction fault model, JGR Potential role of fluids in modulating stable and unstable slip Nadaya Cubas GE 277/169b

2. Common features: -slip velocity 10 to 100* V plate (10 -9 to 10 -7 m/s) -transients migrate along trench at speed from km/day to km/year -limited to a depth range near and below end of the seismogenic zone. -quasiperiodicity of aseismic transients observed in some shallow subduction zones from few months up to several (10) years -Associated with non-volcanic tremors -Evidence for high Pore Fluid Pressure Slow slip events

3. Liu and Rice 2005: Slow slip events can arise spontaneously rate and state formulation, with T (depth) dependence (a-b) parameter for certain (low) effective normal stress with the simplest representation of metamorphic fluid release (typically 50 MPa or 100 MPa) in the seismogenic zone and downdip region. Slow slip events

4. Model Dietrich-Ruina 3D thrust fault with rate- and state- dependent friction law

5. Transition weakening/strengthening (2) P lower than lithostatic – constant with time,  n eff from 50 to 150MPa (100MPa) (3) Lf = 120mm Parameters: (1) Temperature (depth) dependence of material parameter (a-b), for granite gouge under hydrothermal conditions

6. Non uniform slip mode, Complex slip, with strongly locked zones Results: in the seismogenic zone with small perturbations of (a-b) parameters

7. V plate downdip locked Transient, V transient = 3km/year 10*V plate shallow aseismic slip during interseismic Results: near and down dip the transition zone Aseismic transients

8. For type II, new transients can be triggered by ongoing transients, due to stress transfer triggered events have features, such as nucleation, propagation of the front of the faster slipping zone, and relocking, analogous to a seismic event, except that the former have much lower slip velocity (10 to 102 times Vpl) and are limited to downdip distances around 50 to 60 km in the simulations.

9. Common characteristics compared to GPS observations: 1.Transients have typical aseismic slip rates averaging around 10 to 10 2 times Vpl. 2.Transients migrate along strike at a speed of several to tens of km/yr in simulations. 1.Transients are limited to depths around the downdip end of seismogenic zone defined in the model. They generally cannot rupture into the updip portion because the fault there is very firmly locked. Nor can slip be enhanced in the velocity strengthening region that is further downdip. CCL: Can arise spontaneously near neutral stability Transition of frictional properties from velocity weakening to velocity strengthening, combined with great earthquake sequences that leave zones of heterogenous slip and stress alteration along strike, seem to be the cause of the aseismic slip event occurrence. They may reflect episodic relaxation, during the interseismic period, of stress concentrated around the transition zone.

10. Common characteristics compared to GPS observations: CCL: Can arise spontaneously near neutral stability Transition of frictional properties from velocity weakening to velocity strengthening, combined with great earthquake sequences that leave zones of heterogenous slip and stress alteration along strike, seem to be the cause of the aseismic slip event occurrence. They may reflect episodic relaxation, during the interseismic period, of stress concentrated around the transition zone. However: Long transient: tens of years rather than the several weeks span observed in some natural events - Large characteristic slip distance L, with smaller L : speed increase - The speed with which transient slipping zones propagate along the strike increases with the reduction of  ; e.g., that speed increases by a factor of 3 when the assumed  is reduced by a factor of 2, from 200 MPa to 100.

11. Liu and Rice 2007: rate and state formulation, with T (depth) dependence (a-b) parameter 2D to test smaller L Lithostatic pore pressure Reproduce oscillatory behavior, ranging from 3 months beneath the Bungo Channel, southwest Japan 11-19 months along the Cascadia 2–3 years near Gisborne, along the Hikurangi subduction zone in New Zealand 4 years in Guerrero, Mexico 6 years beneath the Bungo Channel 10 years or more, Alaska, Sunda Megathrust between the Mentawai Slow slip events

12. Evidences for HPP: 1.petrological analysis of phase equilibrium: dehydration conditions would be met for shallow dipping subduction zones around 350C and above, which means near and downdip of the frictional stability transition Fluids released from such dehydration reactions can significantly increase the pore pressure

13. Here, add one slide about HPP shallow NE japan – thermal pressurization presented before.

14. Evidences for HPP: 2. in northern Cascadia, nonvolcanic tremors, clearly occurring during the time period of the aseismic deformation transients, have hypocentral locations as inferred by Kao et al. [2005] which mostly correspond to the positive ‘‘unclamping’’ effective normal stress changes of less than 0.01 MPa on any hypothesized vertical fissures

15. Model 2D thrust fault with rate- and state- dependent friction law a)Realistic model: -effective normal stress is high (finite) in the seismogenic zone, -much lower from stability transition and further downdip. b) Simplified model: -RW patch completely locked (  eff infinitely high) -W extending up dip of the stability transition -Whole down-dip region at a much lower, uniform  eff, due to dehydration.

16. (2) P near lithostatic (3) Lf = 20  m (1) Temperature (depth) dependence of material parameter (a-b), for granite gouge under hydrothermal conditions Parameters:

17. h* critical cell size [Rice, 1993; Lapusta et al., 2000] Stable slip patch size for steady sliding with rate- weakening friction Results: with locked seismogenic zone Oscillation depends on W/h* W: (a-b) < 0 but very high pore pressure:  eff very low Shear modulus Poisson’s ratio

18. With L =  m,  eff = 2-3 Mpa: 14 months period However, oscillatory slip rate amplitude extremely low case b and c: lower than Vplate Results: with locked seismogenic zone Oscillation depends on W/h*

19. Rate weakening zone as being at a finitely higher  eff so that slip is allowed to penetrate naturally into it by a small amount show that the amplitude of the oscillatory slips increases substantially. Average period: 2 years Slip rate: 2 to 4 * Vplate Accumulated slip: 1 to 2 cm Results: with unlocked seismogenic zone Higher amplitude

20. Can arise spontaneously near neutral stability Or transient sequences can be triggered from the stress field alteration due to nearby earthquakes, or the pore pressure variations due to episodes of metamorphic fluid release along the subducting slab, or along-strike variations in seismic slips.

21. aseismic transients might act as a spatial- temporal connection between distant seismicity clusters in subduction zones Transient aseismic slip NF TF NF TF + Repetition in 2006 Transient sequences can be triggered by slab seismicity

22. Transients response will dependent on: Perturbation Introduction Time (Ttp /Tr0 ) : Accelerate or delay next thrust EQ Relative Location Along the Thrust Fault: Slightly varies interseismic time Magnitude of the Stress Perturbation: Pore pressure rise or drop can trigger transient >> And have direct consequences on te timing of future thrust EQ

23. Ccl Our studies can produce transients in abundance, of a wide range of recurrence intervals, including transients of relatively short periods (1 year) in cases for which there is a zone of near-lithostatic fluid pressurization around and downdip of the frictional stability transition, a situation independently supported by tremor triggering and petrology. However: However, with the simple model of a completely locked seismogenic zone updip of width W, the amount of deformation at the earth’s surface associated with the spontaneous short-term transients is at least an order of magnitude smaller than observed in Cascadia. Further, even with a relatively more realistic model of an unlocked seismogenic zone as in section 3.2, most of the aseismic moment in our modeling is contributed over a zone extending updip and moderately downdip from the stability transition, whereas a much broader downdip zone is thought to be activated during the transient events in the northern Cascadia and Guerrero subduction zones [Dragert et al., 2001; Hirn and Laigle, 2004; Larson et al., 2004 Have we made the downdip rate-strengthening zone too stable? Segall et al. (2010), Dilatant strengthening as a mechanism for slow slip event, JGR

24. The triggered transients, and the timing of future thrust earthquakes are mainly affected by three factors associated with the stress perturbation, namely, (1) the time within the earthquake cycle when the perturbation is introduced, : The time in the earthquake cycle when the stress perturbation is introduced can affect the number of triggered sequential transients and correspondingly advance or delay the next thrust earthquake. The same number of transients are triggered in case in Figure 15b, but they have faster slip rates (and shorter durations) than those in Figure 15a. In particular, the last transient reaches a maximum slip rate over 107 m/s, which is 102Vpl. Because of the strain released during the fast slipping event, EQ1 is delayed by 72.7 years compared to the unperturbed case, resulting in Tr1/Tr0 = 1.15 (delayed).

25. Migration updip explained by cumulative coulomb stress on the thrust fault after each transient the total slip during transient 1 results in a positive cumulative Coulomb stress Dtcoul (closer to failure) updip of the slipped region, where later transient 2 occurs, and so on. After transient (3), a large part of the seismic nucleation zone is under positive Dtcoul, leading to the rupture in the next thrust earthquake.

26. The triggered transients, and the timing of future thrust earthquakes are mainly affected by three factors associated with the stress perturbation, namely, (1) the time within the earthquake cycle when the perturbation is introduced, : The time in the earthquake cycle when the stress perturbation is introduced can affect the number of triggered sequential transients and correspondingly advance or delay the next thrust earthquake. (2) the relative location along the subduction fault, and (3) the magnitude of the stress perturbation. The triggered transients, and the (2) the relative location along the subduction fault,