HR: 1340h
AN: S43B-1317    [Abstracts]
TI: ADER-DG Seismic Wave Propagation on Unstructured Quadrilateral Meshes
AU: * Castro, C E
EM: cristobal.castro@geophysik.uni-muenchen.de
AF: Munich University, Theresienstr. 41, Munich, 80333, Germany
AU: de la Puente, J
EM: jdelapuente@geophysik.uni-muenchen.de
AF: Munich University, Theresienstr. 41, Munich, 80333, Germany
AU: Käser, M
EM: martin.kaeser@geophysik.uni-muenchen.de
AF: Munich University, Theresienstr. 41, Munich, 80333, Germany
AU: Dumbser, M
EM: iagmidu@iag.uni-stuttgart.de
AF: University of Trento, via Messiano 77, Trento, 38050, Italy
AB: The ADER-DG method for seismic wave propagation has been developed in the recent years for unstructured triangular and tetrahedral meshes. In the present work we apply the same numerical approach to unstructured quadrilateral meshes instead. As the physical domains where we solve the seismic wave propagation equations are in general very complex, we are forced to use unstructured meshes with deformed elements. The geometrical flexibility of triangular and tetrahedral meshes are already well known; nevertheless we consider unstructured quadrilateral meshes as they are computationally less expensive when solving the governing equations and also enable easier comparisons with other existing numerical methods based on this mesh topology. Considering this approach we can reduce the computational time because we need less quadrilateral elements than triangular elements to discretize a particular physical domain considering the same edge lengths. Furthermore, the quadrilateral elements are bigger than the triangular elements and thus computations remain stable for larger time steps. Another advantage is that for quadrilateral meshes we can use a nodal basis based upon Gauss-Lobatto-Legendre integration points, instead of a modal basis, which further reduces the computational costs. Finally, applying the ADER-DG method on unstructured quadrilateral meshes we can compare the results and performance with the well-established Spectral Element Method utilizing exactly the same mesh. A further goal, currently under development, is to carry out simulations on hybrid meshes. The aim it to mesh the geometrically complex parts of a domain with triangles or tetrahedral while the simpler parts are meshed with the more efficient quadrilateral or hexahedral meshes, both using the same high-order ADER-DG method. As a consequence, for realistic setups, the computational costs could be optimized without compromising both the fidelity to the model and the accuracy of the numerical solution.
DE: 0500 COMPUTATIONAL GEOPHYSICS (3200, 3252, 7833)
DE: 0545 Modeling (4255)
DE: 0560 Numerical solutions (4255)
DE: 3285 Wave propagation (0689, 2487, 4275, 4455, 6934)
DE: 7290 Computational seismology
SC: Seismology [S]
MN: 2007 Fall Meeting