摘要
本文采用基于深度学习的地震检测算法对山西预警台网记录的连续波形,开展离线的震相自动拾取工作,从连续波形检测、到时拾取、震相关联,地震定位和震级测定,最终形成完整的地震目录,即自动编目结果。将自动编目形成的地震参数从发震时刻、震中位置和震级这3个方面,与中国地震台网中心统一编目结果相比较,对比结果显示:发震时刻偏差平均值为-0.1 s,震中位置偏差平均值为3.9 km,震级偏差平均值为0.2级,偏差结果不大,说明了预警台网自动识别的地震定位精度高、与台网中心编目的结果有很好的一致性,下一步可纳入预警台网日常工作。
This article uses a deep learning based earthquake detection algorithm to automatically pick up seismic phases offline from the continuous waveforms recorded by the Shanxi Early Warning Network.From continuous waveform detection,arrival time picking,earthquake correlation,earthquake positioning and magnitude determination,a complete earthquake catalog is ultimately formed,which is the result of automatic cataloging.The earthquake parameters generated by the automatic cataloguing are compared with the unified cataloguing results of the China Earthquake Networks Centerfrom the three aspects of earthquake occurrence time,epicenter location and magnitude.The comparison results show that the average deviation of the earthquake occurrence time is-0.1 s,the average deviation of the epicenter position is 3.9 km,and the average deviation of the magnitude is 0.2.The deviation results are not significant,indicating that the automatic identification of earthquake positioning by the early warning network has high accuracy and good consistency with the results of the network center catalog.The next step can be included in the daily work of the early warning network.
作者
梁向军
刘林飞
刘雪娇
闫晓美
Liang Xiangjun;Liu Linfei;Liu Xuejiao;Yan Xiaomei(Shanxi Earthquake Administration,Taiyuan,China;Taiyuan Continental Rift Dynamics National Field Scientific Observation and Research Station,Taiyuan,China)
出处
《科学技术创新》
2024年第6期71-74,共4页
Scientific and Technological Innovation
基金
山西省地震局科研项目(SBK-2321)。
关键词
震相
自动识别
山西
预警台网
seismic phase
automatic identification
Shanxi
early warning network
