FEM Modeling of Lithospheric Rheology of Taiwan from GPS and PSInSAR Observations of Postseismic Deformation of the 1999 Chi-Chi Earthquake

Abstract:

Postseismic deformation following large earthquakes can persist from years to decades. Since the 1999 Chi-Chi earthquake (Mw=7.6), more than a decade of GPS records reveal postseismic deformation across Taiwan and have been used for afterslip and lithospheric rheology studies. Previous researches show that this earthquake has experienced rapid afterslip on deeper part of Chelungpu fault with slip rates decreased within years [Hsu et al., 2002 and 2007]. Other postseismic deformation signals were modeled by assumed Maxwell viscoelastic relaxation with heterogeneous lower crust and upper mantle [Rousset et al., 2012; Tang et al., 2015], which suggest that the pattern of vertical displacements is more sensitive to the geometry of heterogeneous viscoelastic lithospheric structure. To improve the spatial coverage of observation, this study processed both ERS and Envisat SAR images from 1999-2008 with PSInSAR techniques, and the result provide a better illustration of Line-Of-Sight (LOS) deformation field which is near to vertical. Both GPS and PSInSAR results reveal that the surface displacement rates (horizontal or vertical?) of eastern Taiwan are higher than interseismic rates after a decade, and this feature may reveal the importance of low viscosity zone beneath central range of Taiwan.

With the benefits from GPS and PSInSAR data, we intend to test other posssible viscosity and the geometry of viscoelastic structure. We adopt coseismic fault geometry and slip distribution of the Chi-Chi earthquake based on previous studies, and build a set of 2D rheological models with an elastic upper-crust layer overlain a viscoelastic lower-crust layer and a viscoelastic upper mantle. The depths of the two layer boundaries are determined according to subsurface tectonic and velocity structures inferred by previous literatures. We employ the finite element method (FEM), Pylith, to estimate the postseismic surface deformation corresponding to different viscosities. Benefit from the flexibility of FEM mesh building, we are able to model surface viscoelastic deformation corresponding to different geometry of viscoelastic structures, and the results are compared with our geodetic measurements for determining the best-fit rheological model.