資源探査の研究動向と大規模データプロセッシング

[1K0101-07-06] Earthquake rupture parameter estimation using Full Waveform Inversion

○Iktae Jang¹, Hitoshi Mikada¹, Junichi Takekawa¹ (1. Kyoto University)

司会：久保大樹（京都大学）

キーワード：Full Waveform Inversion, Rupture Parameters, Numerical Experiment

To secure the safety from an earthquake, it is necessary to promptly establish a proper measure for the disaster mitigation for a hazard estimated as accurately as possible. Early warning systems utilizing earthquake signals have been generally used worldwide. We can obtain information about ground shaking and the earthquake source mechanism such as orientation, dip and rake angles, and the earthquake rupture propagation as source time function through the earthquake signals. Several recent studies have proposed to estimate the physical mechanism of the earthquake source, and one of the representative methods is the centroid-moment tensor (CMT) solution. However, long-term observation of aftershocks' spatial distribution is required using the CMT solution to identify the fault plane that causes the fault motion among two nodal planes. Since it is evident that promptness is an issue for an earthquake countermeasure, we need to think about a novel way.

Earthquake waveforms caused by two nodal planes perpendicular to each other become different, even though the CMT solutions for the two planes could be the same. The analysis of full waveforms acquired for an earthquake could be utilized to identify which nodal plane has caused the fault motion. This study applies the full waveform inversion (FWI) to estimate one of the fault parameters, the rake angle. We conduct three-dimensional numerical experiments based on the staggered-grid finite-difference scheme to obtain synthetic data and investigate the applicability of FWI evaluating rake angle and apply the FWI process as a first step. The numerical results show that the rake angle and source time function can be estimated with sufficient accuracy using full-waveform information. In future work, we will develop a new method for estimating both parameters simultaneously.