第2回 CRESTセミナー

1.日 時 平成25年5月30日(木)9:30 - 11:00

2.場 所 筑波大学 計算科学研究センター 会議室A

3.プログラム

【講演タイトル】

Detection of Hidden Earthquake Signals Using a Matched Filter Technique

【講演者】

Bogdan Enescu (筑波大学 生命環境系)

http://www.geol.tsukuba.ac.jp/~benescu/

【講演概要】

Large earthquakes are immediately followed by numerous aftershocks; however, many of them are missing in existing earthquake catalogs, mainly because they are masked by high-amplitude, larger-period mainshock waves and their seismic phase arrivals are overlapping (Kagan, 2004). Recovering these missing ‘early’ aftershocks is essential for understanding the physical mechanisms of earthquake interaction (e.g., Dieterich, 1994), tracking the post-seismic deformation around the mainshock rupture zone and in seismic hazard studies.

To overcome earthquake catalog limitations, recent studies (e.g., Enescu et al., 2009; Peng and Zhao, 2009, Kato et al., 2012; Marsan and Enescu, 2012) have used high-pass filtered continuous waveform data to detect as many early aftershocks as possible. A particularly effective method is the so-called “Matched Filter Technique” (MFT): seismograms of already catalogued earthquakes are used as “templates” which are cross-correlated with continuous waveform data for detecting similar nearby events (Peng and Zhao, 2009). The method has been successfully applied to study the aftershock activity of the 2008 M7.2 Iwate-Miyagi earthquake (Enescu et al., 2010) and the 2011 M9.0 Tohoku-oki earthquake (Lengline et al., 2012).

While the detection of aftershocks at relatively short times after large earthquakes was so far the main target of MFT analysis, there is strong interest in applying the same approach for longer time periods and larger areas to obtain more complete earthquake catalogs. The high-resolution earthquake datasets will help the detection of aseismic transients that may precede the occurrence of major earthquakes (e.g., Kato et al., 2012).

However, the computation times (order of days) required for applying MFT to relatively small-seize problems, prohibits the larger-scale use of the method. In order to speed-up the processing, I would like to propose a joint collaboration for implementing the MFT codes on parallel machines and applying them to seismic data in Japan.

Comments are closed.