Modeling Afterslip of the 2010 Mw 7.8 Mentawai Tsunami Earthquake Using Stress-driven Rate-and-State Frictional Slip

Abstract:

The 25 October 2010 M_w 7.8 Mentawai earthquake is a tsunami earthquake that occurred near-trench and produced much larger tsunami with respect to its magnitude. Following this tsunami earthquake, rapid and substantial postseismic deformation has occurred, and has been recorded by the Sumatran GPS Array (SuGAr). The amount of the postseismic deformation comprised ~30% of the displacement measured in the day of the earthquake [Hill et al., 2012]. Up to the present, the accumulated postseismic displacements have largely exceeded the coseismic displacements at a few nearby SuGAr stations. This large postseismic deformation cannot be explained by viscoelastic relaxation, which we estimate generated <1 cm displacements for this extremely shallow event. Therefore, the majority of the postseismic deformation can be explained only by afterslip.

The ratio of horizontal to vertical afterslip displacements has decreased over time, suggesting that afterslip occurred likely downdip of the coseismic rupture and may have moved progressively farther downdip. However, because SuGAr has limited spatial resolution at shallow depths, it is difficult to model the exact location and evolution of afterslip using GPS data and kinematic inversion only. Therefore, we use rate-and-state frictional slip models driven by coseismic stress changes to model the afterslip of this tsunami earthquake.

To model afterslip with velocity-strengthening characteristics, we divide the slab interface into small patches and assign them with frictional properties including a-b, confining pressure, characteristic slip distance, and reference velocity. By comparing the model predictions with the time series data in the Simulated Annealing framework, we explore the range and spatial variations of the frictional parameters. We also discuss the effects of using different coseismic slip distributions and layered media. Using this physics-based modeling technique, we are able to fit the postseismic time series very well, and the results indicate afterslip indeed occurred downdip of the coseismic patch and moved progressively farther downdip. We demonstrate that stress-driven frictional slip models are particularly useful for resolving low-resolution regions in kinematic inversion models.