SCEC Award Number 11189 View PDF
Proposal Category Individual Proposal (Integration and Theory)
Proposal Title The Dependence of Coda on Anisotropic Scattering and Intrinsic Attenuation
Investigator(s)
Name Organization
Paul Davis University of California, Los Angeles
Other Participants
SCEC Priorities A10 SCEC Groups Seismology, GMP
Report Due Date 02/29/2012 Date Report Submitted N/A
Project Abstract
 Role of Seismic Scattering and the Development of Coda
Seismic scattering has been analyzed at CI stations with a view to examining how lithospheric structure contributes to coda (Davis et al 2011). A full-space scattering model developed by Hoshiba (1991) has been applied to coda from stations from the CI network and compared with results from the half-space layer model developed by Yoshimoto (2000). A comparison has been made between data from Peru, Mexico and California to assess the role of earthquake depth and lithospheric structure on inferred properties such as scattering, albedo, and mean free path. The methods seek to separate scattering from intrinsic attenuation. Intrinsic attenuationis significantly less frequency dependent than scattering attenuation and may even be frequency-independent. Scattering parameters at each station can be used to predict the stochastic part of coda.
Data from the Long Beach (~5000 station) seismic array was analyzed to separate deterministic forward scattered signals from stochastic ones generated by random scatterers. The highly coherent wavefield of the deterministic part of the coda was quantified in terms of entropy that drops at the arrival of the S waves and increases when the wavefield becomes incoherent, while the coda remains high. Variation of entropy on a high resolution network has been used to test scattering models.
Intellectual Merit Hazard from strong ground motion requires understanding coda. Models of coda are based on a combination of deterministic signals from layered structures followed by random forward and backward scattering. By using data from a densely sampled ( unaliased ) network in Long Beach we use entropy to separate the deterministic and stochastic parts of the coda for comparison with modeling.
Broader Impacts Trained a graduate student (originally form UNAM).
Exemplary Figure Figure 4 Entropy-energy analysis. a) Shows the map and classification of the stations. Empty circles show the central stations; black stations indicate the neighbor stations in a radius of 250m; gray circles are station not used in this computation. b) shows the average energy computed using Eq. 3; c) is the energy of the window; and d) is a sample seismogram randomly chosen from the network.