Reliability evaluation by initial model randomization for refraction traveltime tomography

• Shiraishi Kazuya

  JGI, Inc.

• Abe Susumu

  JGI, Inc.

• Iwasaki Takaya

  Earthquake Research Institute, The University of Tokyo

• Saito Hideo

  JGI, Inc.

• Sato Hiroshi

  Earthquake Research Institute, The University of Tokyo

• Koshiya Shin

  Civil and Environmental Engineering, Iwate University

• Kato Naoko

  Earthquake Research Institute, The University of Tokyo

• Arai Ryuta

  Earthquake Research Institute, The University of Tokyo

• Kawanaka Taku

  JGI, Inc.

Bibliographic Information

Other Title

• 屈折初動走時トモグラフィ解析における初期モデルランダム化による解の信頼性評価
• クッセツ ショドウソウジ トモグラフィ カイセキ ニ オケル ショキ モデル ランダムカ ニ ヨル カイ ノ シンライセイ ヒョウカ

Abstract

 Tomographic inversion is a useful technique for direct estimation of physical properties in the earth. In the seismic survey from basin-scale to crustal-scale by deployment of the wide-angle survey line with dense seismic sources and receivers, the refraction traveltime tomography is an important approach to build a velocity model in the survey area. However, it is necessary to evaluate the reliability of the estimated velocity model which is solved as a nonlinear inverse problem. We adopted the Monte Carlo uncertainty analysis by initial model randomization for our traveltime tomography. This method provides a statistical reliability evaluation of the tomography result with a velocity model by averaging the tomographic results from hundreds of initial models and a spatial distribution of error range described by standard deviation of them. The velocity model with the reliability can be used in further studies such as a fullwave tomography, a pre-stack migration, and so on.<br>  In this paper, we show an example of numerical simulation, and a case study of field survey data. The seismic survey data was acquired on a 40 km-long line with dense sources and receivers in a lowland area where active faults exist. The reconstructed velocity structure from 500 random initial models shows reasonable spatial variation with high reliability above 3-4 km depth, which is comparable with the seismic section by reflection survey. The distribution pattern of the deviation implicitly indicates the structural character such as faults and velocity boundaries. Convergences of the deviation and the velocity update show that 200-300 initial models are required for a stable analysis. A checkerboard resolution test is a helpful way to evaluate spatial resolution of the tomography.<br>

Journal

• BUTSURI-TANSA(Geophysical Exploration)

  BUTSURI-TANSA(Geophysical Exploration) 63 (4), 345-356, 2010

  The Society of Exploration Geophysicists of Japan

Keywords

• Refraction traveltime tomography
• reliability evaluation
• initial model randomization
• Monte Carlo uncertainty analysis
• checkerboard resolution test

Details

• CRID: 1390001206496504192
• NII Article ID: 10029437044
• NII Book ID: AN10028069
• DOI: 10.3124/segj.63.345
• ISSN: 18814824
• ISSN: 09127984
• NDL BIB ID: 10972440
• Web Site: https://ndlsearch.ndl.go.jp/books/R000000004-I10972440
• Web Site: https://www.jstage.jst.go.jp/article/segj/63/4/63_345/_pdf
• Text Lang: ja
• Data Source:
  • JaLC
  • NDL
  • Crossref
  • CiNii Articles
  • KAKEN
• Abstract License Flag: Disallowed