|
for Seismic Wave Propagation on Regional and Global Scales Gunnar Jahnke Department of Earth and Environmental Sciences,
|
 |
|
for Seismic Wave Propagation on Regional and Global Scales Gunnar Jahnke Department of Earth and Environmental Sciences,
|
|
Last update: 23. June 2009
|
Numerical Methods for Seismic Wave Propagation on Regional and Global Scales Regional modelling: 
Simulation of regional 3D models is done for fault zones, especially to investigate so calledTrapped Waves which are very sensible on fault structure at depth (for example seismograms see here). The detailed structure of a fault zone (FZ) plays an important role in problems related to fault mechanics, dynamic rupture, wave propagation and seismic hazard. FZs are thought to consist of a O(10-100)m wide region of decreased seismic velocity. The small spatial scales involved make such structures difficult to image with ray-theoretical methods (tomography). However, when seismic sources are located at or near such low-velocity structure, energy is trapped inside the FZ leading to dispersive wave trains carrying information on the FZ structure. These waves can travel many kilometers inside the fault before reaching the surface and are therefore strongly altered by its properties. Candidate trapped waves have been observed above several active faults. Inversion algorithms exist which explain these observations in terms of planar fault zone structures. However, at present it is not clear how reliable these estimates are as the effects of (even small) 3-D variations on trapped waves are not well understood. The goal of this study is to distinguish relevant and irrelevant effects on FZ waves due to 3-D structure. To achieve this we perform numerical calculations of various FZ geometries and analyze the waveforms, spectra and envelopes of synthetic seismograms. The main results are that (1) moderate changes of the shape of FZ or (2) small scale heterogeneities or (3) depth-dependent properties do not strongly affect the observed FZ waves. In contrast, strong effects are to be expected from (4) discontinuous FZ structure (e.g. offsets), which may at some point allow imaging such features at depth. The used method is a 3D high-order staggered-grid finite-difference scheme. The current Fault zone modelling is done in collaboration with my supervisor Prof. Heiner Igel (Geophysics Section, of the Department of Earth and Environmental Sciences, Ludwig - Maximilians Universität München) and Yehuda Ben-Zion (Department of Earth Science, University of Southern California, Los Angeles). |
Global modelling:
The Simulations are done on high performance computers such as the Hitachi SR8000 Computer at the Leibniz Rechenzentrum München. The implementation is done with Fortran90 and MPI. |
|
(together with Tarje Nissen-Meyer and Markus Treml)   In
order to simulate teleseismic wavefields entering regional structures,
a combination of regional 3d modeling in a spherical section and global
axi-symmetric modeling was implemented (see Figure on the right and this
poster
presentation). An application for that is e.g. modeling teleseismic
registrations on top of a regional plume structure. This method is capable
of simulating high frequencies, since for large parts of the model space
the efficient axi-symmetric wave propagation is used. Full 3D simulation
is only done in a regional part, defined by a spherical section. At each
time step of the computation, velocities and stresses have to be passed
from the axi symmetric to the spherical section. Restrictions of this method
are (1) source and background model have to be axi-symmetric and (2) the
3D section has to be at the surface, since transition of an abitrary 3D
wave field back to the axi symmetric section is not possible. The figure
on the left shows a cross section along the great circle path for a plume
like low velocity structure (vertical cylinder). The axi symmetric wave
field (SH-component) has already entered the 3D section. The wave field
is diffracted and the wave lengths are shortened within the cylinder due
to the lower velocity inside. |
|
Tutorial: Understanding seismic radiation patterns Java Applet: Interactive Wave Propagation (experimental)
The
global seismic modelling is done for models and sources with axi symmetry.
That means the model space is parameterized in radial and one angular (theta)
direction and the grid has the shape of a 2D circle. To obtain the corresponding
3D image of the earth, one has to rotate the circle along the theta=0/pi
axis (in phi direction). 
higher
frequencies. The axi symmetry in the algorithm restricts one to axi symmetric
models and sources. Thus things like point scatterers or double couple
sources can not be modeled. However, for many teleseismic questions it
is far more important to be able to simulate realistic frequencies (e.g.
~1Hz for body waves) than variations of the seismic parameters along all
three spartial directions. Some of these questions are: diffraction at
the CMB, teleseismic effects of subduction zones and scattering within
the mantle due to small scale inhomogeneities.
