Poster #008, SCEC Community Models (CXM)

CANVAS: An Adjoint Waveform Tomography Model of California and Nevada

Claire D. Doody, Arthur J. Rodgers, Christian Boehm, Michael Afanasiev, Lion Krischer, Andrea Chiang, & Nathan Simmons
Poster Image: 

Poster Presentation

2021 SCEC Annual Meeting, Poster #008, SCEC Contribution #11213 VIEW PDF
Full waveform inversion models are instrumental to elucidating earthquake processes in seismically active places like California. In an effort to better model waveforms in California, we present the California-Nevada Adjoint Simulations (CANVAS) model, an adjoint tomography model spanning California and Nevada. The CANVAS model includes data from 102 events ranging from magnitude 4.5 to 6.5 over the last 20 years. Each of the 102 events contains data from 150-475 stations, creating a dense path coverage over our domain. Data processing, misfit calculations, and iterations are performed by Salvus. We use the SPiRaL 3D model (Simmons et al., 2021) as our starting model for running multi-scale ...inversions of radially anisotropic Vp and Vs, and density. Iterations begin at 30-100 seconds period, then we decrease the minimum period to 25 seconds and then 20 seconds once the model has converged at each period band. For each model to converge, we iterate 17-30+ times per period band; over the course of these iterations. We use the time-frequency phase misfit function (Fichtner et al., 2009b); the total misfit is reduced by 35-50% per period band. The resulting model (work still in progress) shows wavespeed anomalies of up to 20% compared to our starting model, and more accurate waveform fits to the observed data. We also show results from a different starting model, WUS_AWT (Rodgers et al., this meeting), to investigate how different starting models influence the result of our inversions. The model inverted using WUS_AWT as our starting model shows similar reductions in misfit, wavespeed anomalies, and general structure of the inverted model.

Abstract for the 2021 Southern California Earthquake Center Annual Meeting, to be held online, September 12-17 2021. This work was supported by LLNL Laboratory Directed Research and Development project 20-ERD-008. This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344. LLNL-ABS-825538
SHOW MORE

Citation: Doody, C. D., Rodgers, A. J., Boehm, C., Afanasiev, M., Krischer, L., Chiang, A., & Simmons, N. (2021, 08). CANVAS: An Adjoint Waveform Tomography Model of California and Nevada. Poster Presentation at 2021 SCEC Annual Meeting.
Keywords: Adjoint Tomography, Full Waveform Inversion
Relevant SCEC Themes:
Community Models
Data-Intensive Computing