Computerized identification of stress tensors determined from heterogeneous fault-slip data by combining the multiple inverse method and k-means clustering

Makoto Otsubo, Katsushi Sato, Atsushi Yamaji

doi:10.1016/j.jsg.2006.03.008

Computerized identification of stress tensors determined from heterogeneous fault-slip data by combining the multiple inverse method and k-means clustering

DOI Web Site Web Site 被引用文献13件

書誌事項

公開日: 2006-06

権利情報

https://www.elsevier.com/tdm/userlicense/1.0/
https://www.elsevier.com/legal/tdmrep-license

DOI

10.1016/j.jsg.2006.03.008

公開者: Elsevier BV

この論文をさがす

CiNii Books

説明

The multiple inverse method is a resampling technique that can separate stresses from heterogeneous fault-slip data. Numerous optimal stresses are determined for each extracted subset of data, and the clusters of these stresses are thought to represent significant solutions. Hitherto, the clusters have had to be visually recognized on stereonets. This study computerized the identification of the clusters by using the k-means clustering technique. We tested the technique using artificial datasets with known solutions. As a result, it was found that the present method detected objectively the correct solutions. In addition, the spread of each cluster was evaluated to indicate the confidence levels of the identified stresses that were represented by the cluster centers.