HR: 0830h
AN: S11E-0348    [PDF]
TI: Imaging of 3-D Small-Scale Heterogeneities Around the Nagamachi-Rifu Fault in Northeast Japan by F-K and Polarization Analyses
AU: * Taira, T
EM: taka@noreply.ep.sci.hokudai.ac.jp
AF: Division of Earth and Planetary Sciences, Graduate School of Science, Hokkaido University, North 10 West 8, Kita-ku, Sapporo, 060-0810 Japan
AU: Yomogida, K
EM: yomo@ep.sci.hokudai.ac.jp
AF: Division of Earth and Planetary Sciences, Graduate School of Science, Hokkaido University, North 10 West 8, Kita-ku, Sapporo, 060-0810 Japan
AU: Kuwahara, Y
EM: y-kuwahara@aist.go.jp
AF: Geological Survey of Japan, National Institute of Advanced Industrial Science and Technology, AIST Tsukuba Central 7, 1-1-1 Higashi, Tsukuba, 305-8567 Japan
AU: Imanishi, K
EM: imani@ni.aist.go.jp
AF: Geological Survey of Japan, National Institute of Advanced Industrial Science and Technology, AIST Tsukuba Central 7, 1-1-1 Higashi, Tsukuba, 305-8567 Japan
AU: Ito, H
EM: hisao.itou@aist.go.jp
AF: Geological Survey of Japan, National Institute of Advanced Industrial Science and Technology, AIST Tsukuba Central 7, 1-1-1 Higashi, Tsukuba, 305-8567 Japan
AB: Seismic array observation enables us to estimate spatial distribution of heterogeneities in a deterministic manner from coherent arrivals in coda waves. We shall propose a new imaging method using F-K analysis together with polarization analysis based on an auto-regression (AR) model by a dense three-component seismic array. The spatial distribution of small-scale heterogeneities was estimated by the following two steps: (1) identifying the scattering mode, (2) correcting each effect of source, site, and propagation terms. (1) In order to determine the scattering mode of incoming waves for an array, we estimated the slowness vector and the direction of the maximum polarization of each distinct arrival phase in coda using F-K and polarization analyses, respectively. We applied an AR model in order to estimate F-K power spectra and covariance matrixes in the time-frequency domain of high resolution. We identified the character of each phase as a scattered wave from the angle between the slowness vectors and the polarization directions estimated above. Since the sources were explosions, we could determine whether a given phase is P-P or P-S scattered wave. (2) Amplitudes of the observed seismograms are affected by source, station, and propagation effects, so it is necessary to correct these effects before estimating the reliable values of image relative scattering coefficient. We used the coda-normalization method for correcting source and station effects. On one hand, the correction of the propagation effect (amplitude recovery) has assumed the least-square fit of an a priori of attenuation factor in previous studies. In order to remove this assumption, we adopted Akaike's Information Criterion (AIC) to select the optimal one among varians kinds of expressions of attenuation factor. We applied the above imaging methods to the seismograms recorded by three dense three-component seismic arrays around the Nagamachi-Rifu fault in Northeast Japan operated by the Research Group for Deep Structure of Nagamachi-Rifu Fault and GSJ, AIST. We could identify clear scattered waves, particularly in latter part with a lapse time greater than 8 sec after the above correction processes. In order to image the spatial distribution of their responsible scatterers from the arrival times of these phases, we adopted a 3-D seismic velocity structure estimated by Nakajima (2003). As a result, the following remarkably distribution of scatterers around the Nagamachi-Rifu fault was revealed: (1) A cluster of scatterers dipping to the west is located at depth less than 10 km. These scatterers may correspond to the deep extension of the Nagamachi-Rifu fault observed at the surface. (2) Localized P-S scatterers are clearly identified at depths around 8 km in the north-west of these arrays.
DE: 7200 SEISMOLOGY
DE: 8010 Fractures and faults
DE: 8015 Local crustal structure
SC: Seismology [S]
MN: 2003 Fall Meeting