← Back to Gallery

Bayesian Dynamic Source Inversion of the 2004 Parkfield Earthquake: Insights From Linked 3D Dynamic Rupture and Afterslip Modeling Constrained by Gps and Strong Motions

The 2004 Mw 6.0 Parkfield earthquake is one of the best-observed events ever. Since then, computational capabilities and methods have tremendously advanced, enabling the illumination of the source process with unprecedented detail. We present a 3D Bayesian dynamic rupture finite-fault inversion of the 2004 Parkfield earthquake where the model parameters consist of the spatially variable prestress and friction parameters. We use a fast-velocity-weakening rate-and-state friction law to simulate the coseismic and postseismic rupture phases in the same modeling framework. Strong motion and GPS data, including three months of postseismic displacements, constrain the inversion. We use a parallel tempering Markov chain Monte Carlo approach to tackle the non-linear high-dimensional inversion problem.

The Bayesian approach provides an ensemble of models fitting the data from which we infer the mean and variability of various kinematic and dynamic parameters. We provide estimates of stress drop, fracture energy, and radiation efficiency, which are constrained by the data and the underlying physics. The coseismic rupture transitions from pulse-like to crack-like as it propagates towards the creeping section of the San Andreas fault. The pulse-like phase generates most of the seismic radiation while the crack-like rupture dominates the coseismic GPS recordings. We observe different rupture arrest mechanisms imprinting on the subsequent afterslip evolution. Afterslip concentrates within two major patches above the coseismic rupture zone separated by a barrier that also delays the coseismic rupture. Parts of the fault that already slipped coseismically continue to host afterslip. We find a gradual increase in afterslip rise time and identify different modes of afterslip. Our results provide new insights into earthquake rupture complexity and the transition from the coseismic to the postseismic phase.

Session: Learning Across Geological, Geophysical & Model-Derived Observations to Constrain Earthquake Behavior - II

Type: Oral

Room: Tubughnenq’ 5

Date: 5/1/2024

Presentation Time: 11:00 AM (local time)

Presenting Author: Nico Schliwa

Student Presenter: Yes

Additional Authors