HR: 1330h
AN: S42B-0171    [PDF]
TI: Quick Finite-Fault Inversion and Strong Motion Prediction: Feasibility, Process, and Developments
AU: Wald, D J
EM: wald@usgs.gov
AF: U.S. Geological Survey, P.O. Box 25046 MS-966, Denver, CO 80225 United States
AU: * Ji, C
EM: jichen@gps.caltech.edu
AF: Seismological Laboratory, Caltech MS 252-21, Pasadena, CA 91125 United States
AU: Helmberger, D V
EM: helm@gps.caltech.edu
AF: Seismological Laboratory, Caltech MS 252-21, Pasadena, CA 91125 United States
AB: As part of a larger, concerted effort to rapidly assess the impact of major earthquakes globally and to provide input into other post-earthquake analyses (e.g., ground motions, stress changes, tectonic implications, etc.), we are developing and testing a new system to automatically determine finite-fault characteristics by inversion of teleseismic body waveforms. Although initially constrained from teleseismic data alone, which currently are the only rapid and reliable available data after major events worldwide, additional data will be used (either directly or as constraints) as they become available, including aftershock locations, geodetic displacements, and regional waveforms. Our ultimate goal is to provide well-constrained, estimated peak ground motions for input into a global ShakeMap, which in turn will allow automatic, reliable estimates of losses and overall impact. To evaluate the feasibility of such a system, we inverted for slip variations on a single plane for the 1999 ChiChi (M7.6) earthquake using only teleseismic data and then compared the 2-sec peak ground velocity map predicted by this model with the observations (constrained by over 400 stations). Even though the fault geometry of this earthquake was more complex than our simple approximation, as would be the case for our initial, automated solution, the predicted map matches the overall observed amplitude variations. While successful in that case, a fully automated system requires overcoming additional, significant hurdles, many of which we are discovering and addressing as we develop the processing system. One such hurdle addressed was the choice of the causative rupture plane from the two nodal planes of a moment tensor solution. We simply conduct finite fault inversions on the two planes simultaneously, and select the solution with a smaller error function. This approach worked well for recent the 2003 Carlsberg Ridge earthquake (M7.6), for which the inverted result matches the trend and extent of its aftershock sequence. Among the more challenging hurdles is compensating for travel time anomalies within an automated system. Using the 2002 Denali earthquake as an example, we show the importance of this correction and how we calibrated the teleseismic path effects with proximal fore- or aftershocks. We also show how an improved fault geometry based on a high-resolution DEM map enhanced the solution. Finally, we discuss ongoing developments, further results, and plans for this system.
DE: 7203 Body wave propagation
DE: 7212 Earthquake ground motions and engineering
DE: 7215 Earthquake parameters
DE: 7223 Seismic hazard assessment and prediction
SC: Seismology [S]
MN: 2003 Fall Meeting