Institute of Earth Sciences Seminar

Elías Rafn Heimisson (Stanford University)

Title: Simulating seismicity using rate-and-state friction: Application to the 2014 Bárðarbunga dike

Date: Wednesday, 29th August
Time: 12:00
Place: 3rd Floor meeting room, Askja

Abstract: 

It is well-accepted that stress changes in the crust trigger microseismicity that clusters in both time and space.  Many methods for detecting and locating seismic activity are established and widely used, however, the physical relationship between stress changes, and earthquake productivity remains poorly understood. This lack of understanding greatly limits the physical insight that researchers attain from seismic data and renders its interpretation mostly qualitative.  The first step in quantitatively linking seismicity rate and stress was taken by Dieterich 1994. He derived a constitutive relationship based on rate-and-state friction that describes the seismicity rate of a population of seismic sources as a function of stressing history. Observations validate the theory to a certain extent, but a further comparison is still needed, in particular, for complicated stressing histories. Dieterich's theory/model lends itself to practical applications through many simplifying assumptions, the impact of which has not been clear. In spite of mathematical simplicity, many consider the derivation of the theory abstruse, which makes this modeling approach less accessible than it otherwise could be.

In this talk, I review the main findings of Heimisson and Segall 2018, where we revisited the Dieterich theory. The model is rederived in a more straightforward manner and we assess many important assumptions. The principal result of this study is a modified Dieterich model, were the seismicity rate and the cumulative number of events are expressed in terms of a single integral that depends on the stressing history. This representation provides more insight and greater numerical efficiency over previous formulations of the theory. I demonstrate an application of the new modified theory by simulating the dike induced seismicity of the 2014 Bárðarbunga dike. I introduce a three-step inversion scheme, which combines InSAR, GPS data, and seismic observations, to constrain both time-dependent and independent fields that control the opening of a physics-based dike model. In the final step of the inversion, the dike induced seismicity is simulated based on an optimization using MCMC sampling of parameters that characterize fault orientation and frictional behavior of each population of seismic sources. The model shows good agreement with the space-time characteristics of the observed seismicity. The results offer a much-needed validation of the Dieterich theory for complicated stressing histories and are consistent with the dike induced seismicity being controlled by mechanical stress perturbations on preexisting faults.