Unified source model of ordinary earthquake and slow earthquake

Wednesday, 24 February 2016: 10:25 AM
Suguru Yabe and Satoshi Ide, University of Tokyo, Bunkyo-ku, Japan
Abstract:
On the plate interface in subduction zones, slip occurs, as either fast ordinary earthquake or slow earthquake. Although ordinary earthquakes and slow earthquakes occur side by side in the dip direction, they have different characteristics. To explain those seismic phenomena, several source models are suggested. Ide and Aochi (2005) suggested a hierarchical source model for ordinary earthquakes, in which slip-weakening distance is linear function of patch size. Ando et al. (2012) suggested source models for slow earthquake, in which seismic small patches are embedded on ductile background, to explain the constant moment rate of slow earthquake and slow propagation of slip with energetic tremors. In this study, we present two types of seismic source models, to explain fast and slow earthquakes as different phases of one mechanical system.

We first construct a 2D (line fault) model with rate and state friction law, and check how seismic patch distribution and stiffness of plate loading affects the slip behavior of the fault. The fault is composed of positive and negative A-B patches with uniform characteristic distance, and loaded homogeneously at a constant rate. During the inter-seismic period, velocity-weakening (VW) patches are locked and also suppress slip in the velocity-strengthening (VS) patches. Thus the VS patches accumulate slip deficit due to the pulling-back force. We confirm that larger stiffness will result in smaller slip deficit of VS patches. Higher density and smaller gap of VW patches lead to higher strain-energy accumulation. During the coseismic period, VS patches slip seismically together with VW patches, if sufficiently large strain energy is released. The slip behavior of VS patches changes rapidly from seismic to aseismic with decrease of the VW patch density. We also confirm that slip-weakening distance, which is calculated from fault-averaged stress and slip, increases with decrease of the VW patch density. This implies that scale dependency of slip-weakening distance in seismological observation will be explained by fractal distributions of VW patches.

As a next step, we are constructing a planer fault model. We will investigate how fault slip behavior depends on patch distributions and other parameters.