5:15 PM - 7:15 PM
[SCG45-P47] Dependence of slow slip event sequences on initial conditions in numerical simulations
Keywords:numerical simulation, segmentation of slow slip events (SSEs), rate- and state- dependent friction law
The slip on the fault has been simulated based on laboratory-derived rate- and state- friction laws (RSF laws; Dieterich,1979; Ruina,1983). Heterogeneous fault slip is typically modeled by introducing heterogeneous frictional properties or the fault geometry. In fact, earthquakes modeled with velocity weakening (VW) property require some heterogeneity to arrest the rupture propagation in the model. In contrast, slow slip events (SSEs), transient aseismic slip that are also modeled with VW property, have been reported to occur in segmented regions along an elongated fault in strike direction, even in the absence of heterogeneity in VW region (Ohata, 2023). In this study, we report the dependence of SSEs on initial conditions also differs from that of earthquakes.
In modeling fault slip evolution following the RFS law, it is necessary to impose artificial initial conditions, including the distributions of slip velocity and state variable. For modeling earthquake cycles (ECS; Earthquake Cycle Simulation), it has been implicitly assumed that the influence of these initial conditions disappears after a certain amount of time evolution. Indeed, several studies of ECS have shown that there appear irregular patterns at an early stage of simulations affected by initial conditions, but the patterns converge to a regular cyclic one dependent on the frictional properties after several cycles. In these simulations, rapid and large stress changes on the whole fault occur due to seismic rapid slips, which seems to cancel out the effects of initial conditions. As a result, initial conditions have generally been considered negligible in ECS. Conversely, in case of SSEs on a uniform, elongated fault reported by Ohata (2023), we found that the initial conditions can continue to affect the slip pattern persistently.
We employ a long planar fault in the strike direction that subducts at Vpl = 6 cm/yr, embedded in an elastic full space. We set an elongated VW region surrounded by the velocity-strengthening (VS) region. In our previous research (Nishikiori et al., 2024SSJ), we found that the slip pattern in the VW region is significantly affected by the effective normal stress σ in the VS region. For the fault strength defined by |a – b|σ (Luo and Ampuero, 2018) with the frictional parameters a and b, high VS/VW strength ratio (more than 10) leads to chaotic patterns of SSEs, while low VS/VW strength ratio (around 1.0) produces periodic and persistent segmented SSEs. The reason for the difference in slip patterns is not well understood, but one possible cause is the difference in the degree of penetration of slip into the VS region. We had also partially recognized uniform (symmetric) and anti-symmetric initial conditions produce different segmented cycles on a fault with the same frictional properties. Based on this finding, we have proceeded with further analyses using various initial conditions.
While we fix the frictional properties of VW region, we examine the initial slip velocity distribution V(x) of a sinusoidal shape in the strike direction (x), V(x) = 0.9Vpl sin(2πx/λ) + Vpl, with various wavelength λ. Initial value of state variable is set to dc/V(x) (steady state) with the frictional parameter dc. When VS/VW strength ratio is around 10, we cannot tell the differences in the slip patterns for the different λ. It means that the influence of the initial condition disappears as time evolves, similar to the case of earthquakes. On the other hand, when the VS/VW strength ratio is 1.0, persistent segmentation of the SSE occurs, and the number of segments differs with λ. In this case, unlike earthquakes, mild stress changes caused by the slow slips are insufficient to cancel out the initial conditions. This highlights another key difference in the characteristics of homogeneous segmented SSEs and the earthquakes.
In modeling fault slip evolution following the RFS law, it is necessary to impose artificial initial conditions, including the distributions of slip velocity and state variable. For modeling earthquake cycles (ECS; Earthquake Cycle Simulation), it has been implicitly assumed that the influence of these initial conditions disappears after a certain amount of time evolution. Indeed, several studies of ECS have shown that there appear irregular patterns at an early stage of simulations affected by initial conditions, but the patterns converge to a regular cyclic one dependent on the frictional properties after several cycles. In these simulations, rapid and large stress changes on the whole fault occur due to seismic rapid slips, which seems to cancel out the effects of initial conditions. As a result, initial conditions have generally been considered negligible in ECS. Conversely, in case of SSEs on a uniform, elongated fault reported by Ohata (2023), we found that the initial conditions can continue to affect the slip pattern persistently.
We employ a long planar fault in the strike direction that subducts at Vpl = 6 cm/yr, embedded in an elastic full space. We set an elongated VW region surrounded by the velocity-strengthening (VS) region. In our previous research (Nishikiori et al., 2024SSJ), we found that the slip pattern in the VW region is significantly affected by the effective normal stress σ in the VS region. For the fault strength defined by |a – b|σ (Luo and Ampuero, 2018) with the frictional parameters a and b, high VS/VW strength ratio (more than 10) leads to chaotic patterns of SSEs, while low VS/VW strength ratio (around 1.0) produces periodic and persistent segmented SSEs. The reason for the difference in slip patterns is not well understood, but one possible cause is the difference in the degree of penetration of slip into the VS region. We had also partially recognized uniform (symmetric) and anti-symmetric initial conditions produce different segmented cycles on a fault with the same frictional properties. Based on this finding, we have proceeded with further analyses using various initial conditions.
While we fix the frictional properties of VW region, we examine the initial slip velocity distribution V(x) of a sinusoidal shape in the strike direction (x), V(x) = 0.9Vpl sin(2πx/λ) + Vpl, with various wavelength λ. Initial value of state variable is set to dc/V(x) (steady state) with the frictional parameter dc. When VS/VW strength ratio is around 10, we cannot tell the differences in the slip patterns for the different λ. It means that the influence of the initial condition disappears as time evolves, similar to the case of earthquakes. On the other hand, when the VS/VW strength ratio is 1.0, persistent segmentation of the SSE occurs, and the number of segments differs with λ. In this case, unlike earthquakes, mild stress changes caused by the slow slips are insufficient to cancel out the initial conditions. This highlights another key difference in the characteristics of homogeneous segmented SSEs and the earthquakes.