| CAP方法反演震源机制的误差分析:以胶东半岛两次显著中等地震为例 |
| |
| 引用本文： | 郑建常,林眉,王鹏,徐长朋.CAP方法反演震源机制的误差分析:以胶东半岛两次显著中等地震为例[J].地球物理学报,2015,58(2):453-462. |
| |
| 作者姓名： | 郑建常 林眉 王鹏 徐长朋 |
| |
| 作者单位： | 山东省地震局, 济南 250014 |
| |
| 基金项目： | 国家科技支撑计划项目(2012BAK19B04-01-05);山东省自然科学基金(ZR2012DQ006);中国地震局监测预报司震情跟踪工作专项(2014020103)资助 |
| |
| 摘 要： | 利用区域波形数据使用CAP方法反演中强地震的震源机制正逐渐得到广泛应用.本文以胶东半岛近期发生的两次显著中等地震为例,讨论了使用CAP方法反演震源机制时的误差估计,展示了反演结果的不确定性分析过程.2013年11月23日和2014年1月7日在山东莱州和乳山分别发生了M4.6和M4.3级中等地震,两次事件均造成了较大影响.我们基于CAP方法,使用自助抽样(bootstrap)技术多次重复反演过程,得到大样本量的震源机制解数据;基于这些数据,使用粒子群算法和聚类分析技术给出了优化解,估计了震源机制解的误差范围,并利用震源机制解的P、T轴给出了震源球上的概率密度分布.
|
| 关 键 词： | 莱州地震 乳山地震 波形反演 聚类分析 不确定性 |
| 收稿时间： | 2014-02-27 |
Error analysis for focal mechanisms from CAP method inversion: An example of 2 moderate earthquakes in Jiaodong Peninsula |
| |
| ZHENG Jian-Chang;LIN Mei;WANG Peng;XU Chang-Peng.Error analysis for focal mechanisms from CAP method inversion: An example of 2 moderate earthquakes in Jiaodong Peninsula[J].Chinese Journal of Geophysics,2015,58(2):453-462. |
| |
| Authors: | ZHENG Jian-Chang;LIN Mei;WANG Peng;XU Chang-Peng |
| |
| Institution: | Earthquake Administration of Shandong Province, Ji'nan 250014, China |
| |
| Abstract: | As an effective focal mechanism inversion method for regional earthquakes, CAP (Cut and Paste) is widely used in China in recent years. Its quality and error level need to be evaluated for such solutions are increasingly retrieved. On the other hand, when using the CAP method, those phases which fitted well are usually chosen for inversion, and the other phases which are thought 'bad’ or fitted not so good are ignored. It has been found that different station combinations will lead to varied results with unneglectable discrepancies. Objectively speaking, in a scientific perspective, this artificially selected process will increase uncertainties in final inversion results, especially under the present instrument status in China. Furthermore, because the grid search scheme used in the CAP method which is not evenly distributed on focal sphere, we can not give a convincing proof to illustrate that whether the cause of badly-fitted waveforms comes from data error or from un-sufficient searched solutions.#br#Two earthquakes of M4.6 and M4.3 occurred in Shandong Peninsula on 23 November 2013 and 7 January 2014, respectively. The former is the largest event in the Shandong area since 1995. For convenience, we label the former event as the Laizhou earthquake, and the latter as the Rushan earthquake, according to their epicenters. Taking these two events for example, this paper discusses error estimation for focal mechanism inversion using the CAP method. The paper also presents an uncertainty analysis process for the inversion results.#br#Briefly, a bootstrap technique is adopted, waveforms are randomly sampled with equal probability from origin dataset, and then used as data for a repeating inversion procedure. After a large number of inversions, e.g., 1000 times, we finally get bootstrap results consisting of 1000 focal mechanisms.#br#Based on these focal mechanisms, we conduct the following work. (1) We employ a PSO (particle swarm optimization) algorithm to search a solution, of which the Kagan angle is minimal to all the double couple models, and use its standard deviation as the inversion results uncertainty range. The uncertainty of solutions for the Laizhou event is ±23.7°; for Rushan event, is ±6.4°. These two solutions can be evaluated as ‘A’ level according to the Hardebeck's indicator of mechanisms quality.#br#(2) A clustering analysis is used for bootstrap results. For the Rushan event, the clustering center is coincident to the PSO optimized solution; for the Laizhou event, several centers are found in clustering, despite isolate solutions. There are two clustering centers, of which the corresponding data proportion is about 98.7 percent.#br#(3) Projecting P and T axes of bootstrap results onto a focal sphere, calculating its probability density, we get the probability density distribution of focal mechanisms on the focal sphere. Then we can give the confidence interval on different levels for mechanism solutions.#br#The method demonstrated in this paper is not confined to achieving more accurate focal mechanism and obtaining rational inversion error, while it can also be used to exclude isolate and incorrect solutions effectively, and avoid the effect of data from stations with larger disturbances. Therefore, this method can be used to invert focal mechanisms automatically immediately after moderate earthquakes occur. Sustaining by powerful computational capabilities, we can get more accurate and reliable focal mechanism results without manual work. |
| |
| Keywords: | Laizhou earthquake Rushan earthquake Waveform inversion Clustering analysis Uncertainty |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《地球物理学报》浏览原始摘要信息 |
| 点击此处可从《地球物理学报》下载免费的PDF全文 |
|
