Short-Period, Anelastic and Anisotropic, Waveform-Based 3D Middle East Model to Improve Nuclear Explosion Monitoring
Abstract
Initial work improving the seismic waveform predictive capabilities for the Middle East adjoint tomography model demonstrated that enhancements were possible up to 15 seconds. These results at longer periods are in agreement with previous studies of the region. Full waveforms and adjoint techniques were used to identify deficiencies in the wave speed model. Initially, to progress to a minimum period of 15 seconds, transverse isotropy with a vertical symmetry axis was used in the mantle to stabilize the inversion. Travel times of all types of seismic waves showed the largest variance reduction, whereas there was minimal improvement of amplitude variations. Moreover, use of 3D wave speed models in source studies showed a dramatic improvement in the percent double couple of moment tensors over the entire data set. Future work will continue to improve the seismic waveform predictive capability of the Middle East adjoint tomography model. We intend to reduce the minimum period as much as possible allowing for use of shorter period arrivals, smaller sized sources, and data with smaller signal to noise ratios. To reach shorter periods, the 3D anelastic attenuation variations will become increasingly important and will be incorporated into the Middle East model. At shorter periods, knowledge of the 3D anelastic attenuation structure will improve the fit to the amplitudes. Incorporation of 3D anisotropy will also be essential as the minimum period continues to decrease. We will relax the current constraint of a vertical symmetry axis for transverse isotropy in an attempt to recover the anisotropy in the Asian continental lithosphere and surrounding microplates. Improved methods for routinely computing seismic waveforms at shorter periods will require larger computational and clustered systems, and improved, highly optimized algorithms and routines to reduce unnecessary computation.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 30, 2014
- Accession Number
- AD1001058
Entities
People
- Arthur Rodgers
- Brian M. Covellone
- Brian Savage
- Christina Morency
- Jeroen Tromp
Organizations
- University of Rhode Island