HR: 1330h
AN: S42A-0141    [PDF]
TI: Spherical Viscoelastic Finite Element Model for Cascadia Interseismic Deformation
AU: * He, J
EM: jhe@nrcan.gc.ca
AF: Pacific Geoscience Centre, Geological Survey of Canada, 9860 West Saanich Road, Sidney, BC V8L 4B2 Canada
AU: Wang, K
EM: kwang@nrcan.gc.ca
AF: Pacific Geoscience Centre, Geological Survey of Canada, 9860 West Saanich Road, Sidney, BC V8L 4B2 Canada
AU: Wang, K
EM: kwang@nrcan.gc.ca
AF: School of Earth and Ocean Sciences, University of Victoria, 3800 Finnerty Road, Victoria, BC V8P 5C2 Canada
AU: Dragert, H
EM: hdragert@nrcan.gc.ca
AF: Pacific Geoscience Centre, Geological Survey of Canada, 9860 West Saanich Road, Sidney, BC V8L 4B2 Canada
AU: Dragert, H
EM: hdragert@nrcan.gc.ca
AF: School of Earth and Ocean Sciences, University of Victoria, 3800 Finnerty Road, Victoria, BC V8P 5C2 Canada
AU: Miller, M M
EM: meghan@geology.cwu.edu
AF: Dept.of Geological Sciences, Central Washington University, 400 East 8th Avenue, Ellensburg, WA 98926 United States
AB: We have developed a 3-D spherical viscoelastic finite element model for the Cascadia subduction zone to study temporal and spatial variations of interseismic deformation. Previous 3-D viscoelastic finite element models of subduction zone earthquake cycles all use the Cartesian system, with the surface of the earth map-projected on to a horizontal plane. For earthquakes that rupture very long plate-boundary segments, such as the 1700 Cascadia, 1960 Chile, and 1964 Alaska great earthquakes, the Cartesian approach is inconvenient and less accurate. 3-D analytical solutions take into account the spherical geometry of the earth but have difficulty dealing with realistic plate boundary structure. For the new spherical finite element model, we use 27-node tri-quadratic isoparametric element. The resultant large sparse matrix system is solved by the stabilized bi-conjugate gradient method with ILUT preconditioning of fill-in level 6. Our experience suggests that lower order elements in the spherical system would result in unacceptable numerical errors unless one set of mesh lines is strictly radial. For the great Cascadia earthquake, we employ a smooth coseismic rupture model inferred from thermal data and results of tsunami models of the 1700 event, but we test different slip distances. For interseismic deformation, we use the conventional backslip approach. The contemporary deformation of the Cascadia margin consists of interseismic strain accumulation and a geological secular motion that can be described by a rotation of the forearc relative to North America. To isolate the interseismic deformation, we remove the secular motion from both the model formulation and geodetic data. The model predicts decreasing margin-normal shortening rates throughout the interseismic period as a result of stress relaxation in the viscoelastic mantle. The rate of decrease depends on the assumed mantle viscosity. With a viscosity of 10$^{19}$ Pa s, model surface deformation at 300 years after the great earthquake agrees with geodetically observed contemporary deformation very well. The model also confirms the previous finding based on a Cartesian model that an inland region continues to move seaward several decades after the great earthquake.
DE: 4255 Numerical modeling
DE: 7200 SEISMOLOGY
DE: 8100 TECTONOPHYSICS
SC: Seismology [S]
MN: 2003 Fall Meeting