SEISMIC SOURCE ESTIMATION BY FULL WAVE FORM INVERSION
TITLE:
SEISMIC SOURCE ESTIMATION BY FULL WAVE FORM INVERSION
DATE:
Friday, Apr 20th, 2012
TIME:
3:30 PM
LOCATION:
GMCS 214
SPEAKER:
Anders Petersson.
Center for Applied Scientific Computing.
Lawrence Livermore National Laboratory.
ABSTRACT:
We discuss the inverse problem of determining the source parameters of a small seismic event
(location, mechanism, start time, frequency), such that the wave form misfit between seismographic
recordings and simulated ground motions is minimized. Our approach is based on direct numerical
simulations of the elastic wave equation, allowing for complex heterogeneous material models and
realistic topography. A non-linear conjugated gradient approach is applied to solve the inverse
problem, where the gradient of the misfit (with respect to the source parameters) is calculated from
the numerical solution of an adjoint wave equation. Numerical experiments on simple 2-D models
illustrate the importance of scaling the source parameters before applying the conjugated gradient
iteration, preferably using the Hessian. A procedure based on arrival times is used to generate an
initial guess for the source parameters. For the cases considered here, the conjugate gradient
iteration often converges in 20-50 iterations.
Solving the inverse problem requires of the order O(100) numerical solutions of the elastic wave
equation. For 3-D models, such problems can only be solved on large parallel machines. We will
present the capabilities of our parallel open source code WPP, which was designed to solve seismic
wave propagation problems on the regional scale. A higher order accurate scheme is currently being
implemented to improve the frequency resolution and efficiency of the method. These enhancements
will be important for solving the three-dimensional inverse problem, for example in geothermal
applications where there is interest in using micro seismicity for imaging the geometry of a
fractured network.
This work performed under the auspices of the U.S. Department of Energy by Lawrence Livermore
National Laboratory under Contract DE-AC52-07NA27344. This is contribution LLNL-ABS-523199.
HOST:
Dr. Jose Castillo.
DOWNLOAD: