SEISMOLOGY AND GEOLOGY ›› 2022, Vol. 44 ›› Issue (6): 1634-1647.DOI: 10.3969/j.issn.0253-4967.2022.06.016

• Research paper • Previous Articles     Next Articles


ZHANG Li-juan1)(), WAN Yong-ge1,2),*(), WANG Fu-chang1), JIN Zhi-tong1,2), CUI Hua-wei3,2)   

  1. 1)The Institute of Disaster Prevention, Sanhe, Hebei 065201, China
    2)Hebei Key Laboratary of Earthquake Dynamics, Sanhe, Hebei 065201, China
    3)Shandong Earthquake Agency, Jinan 250102, China
  • Received:2021-11-24 Revised:2022-01-06 Online:2022-12-20 Published:2023-01-21
  • Contact: WAN Yong-ge


张丽娟1)(), 万永革1,2),*(), 王福昌1), 靳志同1,2), 崔华伟3,2)   

  1. 1)防灾科技学院, 三河 065201
    2)河北省地震动力学重点实验室, 三河 065201
    3)山东省地震局, 济南 250102
  • 通讯作者: 万永革
  • 作者简介:张丽娟, 女, 1983年生, 2008年于天津大学获基础数学专业硕士学位, 教授, 主要从事数值模拟、 地球动力学等方面研究工作, E-mail:
  • 基金资助:


The rupture process of earthquake generally involves multiple fault activities. The seismogenic fault is generally not a single fault plane, but a combination of multiple fault planes. Based on the principle that clustered small earthquakes often occur near the fault plane, and assuming that the hypocenters obey three-dimensional normal distribution around the center of the sub-fault planes, the three-dimensional spatial structure of the Yangbi earthquake fault in Yunnan Province is estimated based on the fuzzy clustering algorithm. The results in this paper are estimated from the perspective of data analysis. The results will be more accurate if the comprehensive analysis can be carried out in combination with geological, geophysical exploration and other means. The fuzzy clustering analysis is mainly carried out for regions with dense seismic source data. Because the program compiled by this method runs fast on an ordinary computer and can be calculated many times in a short time, the best result can be obtained. In this study, the shape of fault zone can be quickly calculated and analyzed, the shape and spatial distribution of branch fault zone is roughly consistent with the seismic distribution, which verifies that this method has certain predictive effect and application value.
Firstly, GK(Gustafson, Kessel)fuzzy clustering method is used to obtain the partition matrix for all sub classes of hypocenter, then the outliers are removed by using the partition matrix and appropriate threshold, and the subclasses containing fault planes are extracted. Finally, the parameters of each fault plane(including position, strike and dip)with 95%confidence level are determined. It is inferred from the results that the hypocenters are distributed along the fault zone almost parallel to the Weixi-Qiaohou Fault and gradually divided into three fault branches to southeast direction. The east branch dips to southwest, which is the main fault, corresponding to two sub fault planes, with strike of 134.22°, 132.65°and dip angle of 87.14°, 81.96°, respectively; the west branch nearly parallels to the east branch with strike and dip of 129.45°and 74.77°, respectively. Except for the three main faults, a blind fault near the Weixi Qiaohou fault zone is identified in this study, with a strike of 235.66°and dip of 66.30°. In this study, we determined the fault structure of the Yunnan Yangbi earthquake sequence by fuzzy clustering algorithm, which is independent of other methods by using seismic wave data, geodetic data and geological data. It is of significance for tectonic and geodynamic studies.
This data analysis algorithm can be applied to the shape analysis and prediction of fault zone by a large number of such source data. In consideration of earthquake prediction and earthquake disaster assessment, the knowledge of fault network structure in the vicinity of large earthquakes will also help to test different assumptions about stress transfer effects.

Key words: fuzzy clustering, geometry of seismogenic faults, outlier data, principal component analysis, 2021 Yangbi earthquake sequence


地震的破裂过程一般涉及多个断层面, 发震断层面一般并非单一断层, 而是由多个断层面相互组合形成。文中基于成丛小地震在断层面附近发生的原则, 假定所有震源点以断层面为中心, 服从三维正态分布, 利用模糊聚类算法给出了2021年云南漾濞地震断层的三维空间结构。首先, 采用GK(Gustafson, Kessel)模糊聚类算法对全部震源点目录进行处理, 得到矩阵的划分结果; 之后, 利用矩阵划分结果及给定的阈值剔除离群震源点, 提取含有断层面的子类; 最后, 在95%置信水平下确定断层面分布的矩形断层区域的精确空间位置、 倾向角及走向角等参数。由所得结果推测, 震源点沿与维西-乔后断裂带近平行的断裂带分布, 且向SE逐渐分为3个断裂分支。东侧分支大体倾向SWW, 为主断裂, 对应2个子断层平面, 走向分别为134.22°、 132.65°, 倾角分别为87.14°、 81.96°; 西侧分支与东侧分支大体平行, 走向为129.45°, 倾角为74.77°。除3条主断裂外, 还识别出一条走向为235.66°、 倾角为66.30°的隐伏断层, 位于维西-乔后断裂带附近区域。文中基于模糊聚类算法确定云南漾濞地震的断裂带形状分布, 对地震构造和地震动力学研究具有重要意义。

关键词: 模糊聚类, 断裂带形状, 离群点数据, 主成分分析, 2021年漾濞地震序列

CLC Number: