实时智能地震处理系统研发及其在2021年云南漾濞M_(S)6.4地震中的应用

Development of a real-time intelligent seismic processing system and its application in the 2021 Yunnan Yangbi M_(S)6.4 earthquake

本文介绍了一套基于人工智能的实时地震处理系统,该系统采用深度学习方法检测地震,具备地震检测、到时拾取、震相关联、地震定位和震级测定功能,可实时和离线处理连续地震波形数据和密集地震序列,产出的地震目录定位精度高、完备性和一致性好.以云南漾濞M_(S)6.4地震序列为例,可在震后2~4 min内产出地震目录,检测出的地震数量比人工多2~3倍,定位精度与人工处理结果相当,震级测量误差小,震级处理下限可达M L0.0级左右.该系统不仅可应用于省级区域地震台网,也可用于密集地震台阵,水库、矿山和火山等专用台网的实时处理.基于该系统产出的结果,可进一步计算b值变化,分析地震时空演化特征,开展序列实时精定位,为地震应急、地震预测预报和相关科学研究提供重要数据.该系统对于优化我国地震数据自动处理流程,提升地震编目工作的自动化水平具有重要意义.

This paper introduces a real-time seismic processing system based on artificial intelligence.The system uses the deep learning method to detect earthquakes.It includes earthquake detection,phase arrival picking,association,earthquake location and magnitude measurement.It can process continuous seismic waveform data and dense seismic sequences in real time and off-line modes.The obtained earthquake catalog shows significant improvement in location accuracy,completeness and consistency compared to manually determined catalog.For the 2021 Yangbi M_(S)6.4 earthquake sequence in Yunnan Province,the system can produce earthquake catalog within 2~4 minutes after the earthquake.The number of detected earthquakes is 2~3 times more than those of obtained from manual processing.The location accuracy is equivalent to the results of manual processing.The uncertainty of the estimated magnitudes is small and the lower limit of magnitude can reach~M L0.0.The system can be used not only for regional seismic network,but also for real-time data process of seismic network with different purposes,such as dense seismic observations in reservoir,mining and volcano areas.Based on the outputs of the system,we can further calculate the change of b-value,analyze the temporal and spatial evolution characteristics of earthquakes,and carry out real-time precise earthquake sequence relocation,to provide important data for earthquake emergency,earthquake prediction and related scientific research.The system is of great significance to optimize the automatic processing flowchart of seismic data and improve the automation level of seismic cataloging in China.