【摘要】 重点介绍禁核试北京国家数据中心研究的自动检测地震事件筛选方法。通过分析自动检测事件关联台站分布的合理性、关联台站的类型、关联信号参数的可靠性、多频带平均能量比等方法,判断自动检测事件的真伪。以国际数据中心审核公报事件为参考,研究的筛选方法可有效筛除43%自动检测虚假事件,使得自动检测误检率降低13%,并保持较低的误筛选率。此筛选方法可对国际数据中心以及禁核试北京国家数据中心不同处理平台的自动检测事件进行筛选分析。更多还原

【Abstract】 The methods for screening automatically-generated seismic events which developed by the CTBT China Beijing National Data Centre are introduced in this paper.The screening principles are based on analyzing the regional distribution rationality,the types of stations within the event,the parameter availability of signals related to the event as well as the ratio of reference data’s and signal’s mean energy getting from multi-band filtered waveform.The Reviewed Event Bulletins(REBs) of the IDC are taken as the ground-truth reference.Using the methods,about 43%false events can be screened out,reduce the false rate of the auto-matically-detection by 13%,with only few true events miss-classified.The methods can be widely used for different systems at IDC and CNNDC or some other similar systems.更多还原