Time Reversal Project: Parkfield and SPICE benchmark

Lianjie Huang, Paul Johnson, Carène Larmat, Simon Kremers, Heiner Igel

Last Update: May 23 2008

Goal: Understand the potential of time reversal for (finite) source inversion using both the synthetic SPICE Tottori benchmark data (www.spice-rtn.org) and observed data from the Parkfield main shock (local data).

Approach/methodology: Time reversal is carried out with 3D forward modelling tools (viscoelastic, anisotropic) implemented on the Los Alamos (and Munich) clusters, these codes are

• Cartesian spectral elements (code)

• Spherical section spectral elements (code)

• Spherical section finite differences (code)

• Cartesian 2D code for testing (code)

• Cartesian finite difference (conversion from spherical section)

Progress report (End May 2008):

• Implementation of codes on mimaster

• Setup of simulations for Tottori and Parkfield geometries

• Testing of forward and reverse simulations

• Development of visualization programs for snapshots

• Programs to prepare adjoint seismograms

Problems:

• SEM programs not always stable when changing configuration, or processor layout (but no good debugging tool available on mimaster)

• Stability problem with PML for spherical section

Suggested Fixes

• Implementation of same code structure on Munich cluster, checking setup for time reversal configurations with Andreas Fichtner

Remaining tasks

• Initialization of 3D Parkfield model(s), it needs to be decided which model is most suitable

• Characterization of focusing region in space and time in an automated way

• Projections to orthogonal surfaces/fault plane

• Summing up of activated moments (stresses) -> Magnitude

• Extraction of rupture speed, source duration, source time function, asperity location

• Pre-processing of Parkfield data

• Time reversing Parkfield data

Presentations

• Progress report

• Uncertainties

• Talk on Wave Propagation

• Talk on Rotational Seismology