Exploring Physical Controls on Subduction-Zone Slip Processes and Inverting Seismic Data for Dynamic Stress Changes

Abstract

The first goal of this dissertation is to investigate physical conditions and factors that control the two major types of slip phenomena, tsunami earthquakes and slow slip event (SSEs), along shallow subduction zones. The second goal is to develop an inversion method to obtain the coseismic dynamic stress evolution directly using seismic waveforms, instead of from inverted kinematic slip models. For the first goal, I utilize a recently developed dynamic earthquake simulator EQdyna, based on an explicit finite element method. In three coherent studies, I build shallow-dipping subduction zone models with various friction properties over the fault plane and different elastic properties in the hanging wall and footwall. The first study is based on a conceptual model to explore how heterogenous friction on a subduction plane, where asperities (strongly velocity weakening) are embedded in a conditionally stable zone (weakly velocity weakening), control the characteristics of tsunami earthquakes. It proves that the conceptual model works well for generating tsunami earthquakes with typical characteristics, such as slow rupture speed, long duration and high frequency energy depletion. In the second study, I design elastic and friction models to systematically compare their contributions towards the tsunami earthquake features. I find the heterogeneous friction model contributes to most of tsunami earthquake features, while the elastic property contributes to cascading failure of asperities and elevated tsunami hazard. In the third study, I use a heterogeneous friction model to explore the interactions between SSEs and earthquakes at shallow subduction zones over earthquake cycles. It is found that SSE activities are different preceding large and small earthquakes, which might provide a predictive reference to evaluate tsunami earthquake sizes. In addition to numerical forward modeling, seismic inversion can also help to study fault friction and stress property and the related earthquake rupture behaviors. My fourth study proposes a dynamic stress inversion method based on a fault-stress model to directly invert for coseismic dynamic stress evolution processes by fitting seismic data. This method provides an alternative to kinematic studies of earthquake sources using seismic data, with a potential of deciphering more physics from seismic recordings of earthquakes.