Towards inclusion of dynamic slip features in stochastic models for probabilistic (tsunami) hazard analysis.

Abstract:

Stochastic slip modelling based on general scaling features with uniform slip probability over the fault plane is commonly employed in tsunami and seismic hazard. However, dynamic rupture effects driven by specific fault geometry and frictional conditions can potentially control the slip probability. Unfortunately dynamic simulations can be computationally intensive, preventing their extensive use for hazard analysis. The aim of this study is to produce a stochastic model that incorporates slip features observed in dynamic simulations.

Taking a Tohoku-like fault as a case study, numerous 2d spectral element dynamic simulations are performed using a variety of pre-stress distributions. Comparing the slip distributions generated from these simulations to traditional stochastic slip models we find that the stochastic models generally under represent slip near the free surface. This is an important feature in tsunami hazard with very large slip at shallow depth observed for the 2011 Tohoku earthquake. To incorporate dynamic features in the stochastic modeling we generate a depth dependent “transfer function” based on comparisons between the dynamic and stochastic models. Assuming that the differences between stochastic and dynamic slip distributions are predominantly depth dependent and not along strike, the transfer function is then applied to stochastic source models over a 3d geometry of the Tohoku fault. Comparing maximum tsunami wave height along the Japanese coast using a traditional stochastic model and one modified by the transfer function we find that the inclusion of the transfer function leads to the occurrence of more extreme events.

Applying this function to the traditional stochastic slip distribution as a depth-dependent PDF for the slip may allow for an approximated but efficient incorporation of regionally specific dynamic features in a modified source model, to be used specifically when a significant number of slip scenarios need to be produced, e.g. for probabilistic tsunami hazard analysis (PTHA).

This research is funded by the European Union's Seventh Framework Programme (FP7/2007-2013) under grant agreement n° 603839 (Project ASTARTE - Assessment, Strategy and Risk Reduction for Tsunamis in Europe).