Aftershock triggering by postseismic stresses: A study based on Coulomb rate‐and‐state models

Abstract

The spatiotemporal clustering of earthquakes is a feature of medium‐ and short‐term seismicity, indicating that earthquakes interact. However, controversy exists about the physical mechanism behind aftershock triggering: static stress transfer and reloading by postseismic processes have been proposed as explanations. In this work, we use a Coulomb rate‐and‐state model to study the role of coseismic and postseismic stress changes on aftershocks and focus on two processes: creep on the main shock fault plane (afterslip) and secondary aftershock triggering by previous aftershocks. We model the seismic response to Coulomb stress changes using the Dieterich constitutive law and focus on two events: the Parkfield, Mw = 6.0, and the Tohoku, Mw = 9.0, earthquakes. We find that modeling secondary triggering systematically improves the maximum log likelihood fit of the sequences. The effect of afterslip is more subtle and difficult to assess for near‐fault events, where model errors are largest. More robust conclusions can be drawn for off‐fault aftershocks: following the Tohoku earthquake, afterslip promotes shallow crustal seismicity in the Fukushima region. Simple geometrical considerations indicate that afterslip‐induced stress changes may have been significant on trench parallel crustal fault systems following several of the largest recorded subduction earthquakes. Moreover, the time dependence of afterslip strongly enhances its triggering potential: seismicity triggered by an instantaneous stress change decays more quickly than seismicity triggered by gradual loading, and as a result we find afterslip to be particularly important between few weeks and few months after the main shock.

Delayed Seismicity Rate Changes Controlled by Static Stress Transfer

“…While such coupled Coulomb rate‐state (CRS) models introduce a number of additional unknown parameters, they have the advantage of explaining not only the sign of the coseismic stress change but also the magnitude of the rate step and the time‐dependent changes in the seismicity rate and can be applied to a mixed population of nucleation sources that respond to both ±ΔCFS. Previous studies have employed CRS modeling to investigate the spatiotemporal distribution of aftershock activity (e.g., Cattania et al, , ; Toda et al, ; Toda & Stein, ), including delayed stress shadows in populations of on‐ and off‐fault aftershocks (Helmstetter & Shaw, ; Marsan, ).…”

Modeling of Kashmir Aftershock Decay Based on Static Coulomb Stress Changes and Laboratory-Derived Rate-and-State Dependent Friction Law

“…Finally, it is important to note that the spatiotemporal distribution of aftershocks can be influenced by time dependent post-seismic processes such as induced fluid flow and afterslip (Cattania et al, 2015), which has been ignored in our study.…”

Connecting crustal seismicity and earthquake‐driven stress evolution in Southern California

“…This is quantified using a modified version of equation (B14) of Dieterich [] for the evolution of the state variable γ : where A is a fault constitutive parameter and S = τ − ( μ − α ) σ eff , where α is a positive dimensionless parameter [ Linker and Dieterich , ]. Changes in S can be interpreted as changes in CFF [ Hainzl et al , ; Cattania et al , , and references therein] by assuming that pore pressure changes are related to normal stress changes via a proportionality constant (Skempton's coefficient B ), d p = B d σ and defining an effective coefficient of friction so that .…”

“…In particular, up to 40% of aftershocks occurring close to a M ≥7 main shock globally are thought to be attributable to dynamic stressing [ Parsons , ]; van der Elst and Brodsky [] estimate that 15% to 60% of near‐field aftershocks of M 3–5.5 main shocks are attributable to dynamic stressing. Early aftershocks are also likely influenced by afterslip and secondary triggering [ Cattania et al , ]. The Coulomb rate‐and‐state model based on Dieterich [] presented in section 2 assumes that aftershock nucleation sites do not interact with each other; hence, secondary triggering is not included in our model.…”

“…Here we construct models of crustal stress evolution in Southern California in order to test the viability of static and viscoelastic stress transfer from several source earthquakes over a period of decades and compute the spatiotemporal distribution of triggered seismicity with a rate‐and‐state friction approach [ Dieterich , ]. Using a catalog of Southern California seismicity from 1981 to 2014 (E. Hauksson catalog available online), edited to 2288 M ≥3.5 events (Figure ), we apply the rate‐and‐state friction‐based log likelihood (LL) function [e.g., Zhuang et al , ; Cattania et al , ] to evaluate the ability of stress evolution models to explain the observed seismicity. One candidate model accounts for static stress steps at the times of the four M ≥6.7 events: the M 7.3 1992 Landers, M 6.7 1994 Northridge, M 7.1 1999 Hector Mine, and M 7.2 El Mayor‐Cucapah earthquakes, and the M 6.5 1992 Big Bear earthquake which was part of the Landers sequence (Figure ).…”