Numerical Simulation of Tremor Migration Triggered by Slow Slip and Rapid Tremor Reversals
Abstract
Recently discovered slow-slip events (SSE) and non-volcanic tremors have greatly enriched the spectrum of earthquake behavior. These phenomena offer a unique window into the mechanics of the bottom of the seismogenic zone of active faults. In an emergent view, this transition region has heterogeneous frictional properties, and is composed of frictionally unstable, velocity-weakening patches ("brittle asperities") embedded in a frictionally stable fault region. Tremor swarms are viewed as the collective response of brittle asperities interacting through transient aseismic slip in their surroundings. A hierarchy of migration patterns of tremors has now been observed in the Cascadia subduction zone, including large-scale along-strike tremor propagation at ~10km/day and rare swarms that propagate 10 times faster in the opposite direction ("rapid tremor reversals" or RTR). A cascade of brittle asperity failures mediated by transient creep is an appealing model to explain these migration patterns. We performed a quantitative study of this model through numerical simulations of heterogeneous rate-and-state faults under the quasi-dynamic approximation, solved by a Boundary Element Method. We conveniently generated SSEs, propagating at ~10 km/day, by adopting an ad hoc rate-and-state friction law with a transition from velocity-weakening to strengthening. Brittle asperities are defined as small patches of velocity-weakening friction with shorter characteristic slip distance Dc, larger friction parameters a and b or higher effective normal stress (sigma) than their surroundings. The state variable is governed by the "slip law", which allows conditionally stable behavior, i.e. a seismic response to fast enough perturbations. A collection of brittle asperities is distributed along the fault. We studied the effect of their size and spacing and of the contrast of a*sigma and Dc with respect to the background. We successfully simulated both observed migration patterns. The slow forward migration is obviously due to tremor triggering near the leading front of the propagating SSE pulse. Less trivially, our model also produces RTRs with similar characteristics as in Cascadia: spatially scattered swarms back-propagating at fast speed ~100 km/day. These RTRs are rare because they nucleate at the asperities with largest Dc. While our model reproduces key features of the complex spatio-temporal organization of tremors as observed in Cascadia and Japan, this comes at the cost of tuning some model parameters. Increasing a*sigma within the asperities increases the propagation distance reached by RTRs but also reduces their relative migration speed. We will report on results of a more comprehensive parametric study of the properties that control the propagation velocity and distance of RTRs.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2011
- Bibcode:
- 2011AGUFM.S33C..02L
- Keywords:
-
- 7290 SEISMOLOGY / Computational seismology;
- 8118 TECTONOPHYSICS / Dynamics and mechanics of faulting