联合概率法在合并相邻台网地震目录中的应用:以2014年鲁甸序列为例

A joint probabilistic approach for merging earthquake catalogs of two neighboring seismic networks:An example of the 2014 Ludian sequence catalog

中国目前实行的区域地震台网独立运行机制,使得在相邻不同台网的交界地区可能存在多个版本的地震目录和震相观测报告,影响了地震活动性分析与研究.为此,本文提出了一种基于联合概率的方法,可标明两个或多个相邻台网目录中相同的事件,合并它们的震相数据开展重新定位,并重构不同台网交界地区的统一地震目录.该方法的思路与分析步骤是:首先,计算获得不同台网之间具有最小发震时刻差异的两两地震的时空强差异分布,查找并剔除独立地震,计算事件合并的联合概率;其次,基于联合概率分析合并不同台网的地震目录和震相观测报告,对合并事件进行重新定位和定位误差分析,并基于G-R关系检验重构目录的完整性.本文以2014年鲁甸地震序列为例的初步应用结果显示,震相合并之后的地震定位精度相比之前单个台网的结果,特别是相比四川台网的目录,定位精度提高非常显著,合并后的目录与之前相对完整的云南目录接近,但相比由两个台网目录简单拼凑而成的目录更加准确.此外,研究还发现在目录合并过程中,对于4级以上的中强震,应选择MS而不是以ML震级标度;震相合并后被复用台站记录的到时信息可用于检测不同台网间的震相拾取是否存在系统偏差.本文提出的方法使得在相邻不同台网的过渡区形成一个统一且尽可能准确可靠的地震目录成为可能.

The current independent operating mechanism of regional seismic networks in China makes it possible that there are multiple versions of earthquake catalogs and seismic phase observations at the border areas of adjacent networks,which affects the analysis of seismicity.To solve this problem,we propose a joint probabilistic approach to find out the same events among seismic networks,merging their phase data and relocating the events,and re-yielding the unified catalog in the border region of adjacent networks.The procedures of this approach are presented below.First,we calculate the temporal,spatial and magnitude differences by pairwise comparison of the events among various networks which have the minimum occurrence time difference to obtain the difference distribution of the three source parameters.Second,we find the outlier and ticking out the isolate events.Third,the joint probability is evaluated from the distributions after these steps.Finally,we merge catalogs and phase reports based on the joint probabilistic analysis,relocate and analyze errors.The completeness of the re-yielded catalog can be checked by G-R relationship.We apply this method to the Ludian earthquake sequence.The results show that the phasemerged hypocenter precision is improved significantly compared with that before they are merged,particularly compared with the Sichuan catalog.The merged catalog is similar to the relatively complete Yunnan′s catalog,and more precise than that from simple combination of two catalogs.Meanwhile,in the process of catalog merging,M_S rather than M_L magnitude type should be chosen for M_L≥4.The arrival time information reported by the repeatedly used stations in the merged events can be employed to detect whether there is system bias between different networks.The method presented in this paper makes it possible to form a unified and reliable seismic catalog in the transition region of adjacent networks.