基于中国地震科学台阵资料及内蒙古、甘肃地震台网资料，采用结合台阵策略的震相拾取深度学习方法APP，开展内蒙古地震监测能力薄弱区——阿拉善右旗拾震能力研究。研究结果显示，APP方法检测到了人工目录中97.8%的地震，地震拾取总数为人工目录地震数的22倍。经tomoDD方法定位后，地震深度分布较符合内蒙古西部的地质构造特征。对震源深度与断裂位置间相关性的初步分析显示，深度随纬度变化中有5条深度“集中条带”与研究区7条断裂的位置相对应，深度随经度变化中有4条深度“集中条带”与研究区7条断裂的位置相对应。分析认为，APP拾取方法在实际地震资料应用中展示出较强的泛化能力，可为增强固定地震台网对于监测能力薄弱地区微震的识别能力，以及优化地震台网布局、提高监测能力薄弱地区的地震监测水平等提供参考。
Based on the data of the Himalayan array and seismic networks in Inner Mongolia and Gansu Province, this paper studies the seismic pickup capability in Alxa right banner, which is a weak earthquake monitoring area of Inner Mongolia, using a deep learning method based APP combined with the array strategy of seismic phase picking. The results show that the APP method detects 97.8% of the earthquakes in the artificial catalog, and gives a total number of earthquakes that is 22 times that of the artificial catalog. Earthquakes are relocated using tomoDD and the distribution of relocated earthquake depth is more consistent with the characteristics of geological structure in the west of Inner Mongolia. Preliminary analysis of the correlation between focal depth and fault location shows that there are five depth concentrated belts in the variation of depth with latitude correspond to the positions of seven faults in the study area, and four depth concentrated belts in the variation of depth with longitude correspond to the positions of seven faults in the study area. The analysis shows that the APP pickup method shows a strong generalization ability in the application of the actual seismic data, and provides references for enhancing the ability of fixed networks to identify microseisms in weak earthquake monitoring areas, optimizing the configuration of seismic networks, and improving the seismic monitoring level in the areas with weak monitoring ability.
2020,41(4): 90-99 收稿日期：2020-04-26
DOI：10.3969/j.issn.1003-3246.2020.04.013
基金项目：中国地震局地震科技星火计划（项目编号：XH18012)；内蒙古自治区地震局局长基金课题重点项目（项目编号：2020TM05）
作者简介：翟浩(1990-),本科,主要从事大震速报、地震编目等地震监测研究工作。E-mail:496831965@qq.com刘芳(1963-),本科,高级工程师(正研级),主要从事地震监测技术、数字地震资料应用及矿山微震监测等研究工作。E-mail:lfnm88@163.com
参考文献：
邓起东. 新编中国活动构造图[C]. 中国地震学会. 中国地震学会第八次学术大会论文摘要集. 北京:中国地震学会,2000:65.
侯金欣. 利用模板匹配方法研究稀疏台站地区地震活动性[D]. 北京:中国地震局地球物理研究所,2016.
蒋一然,宁杰远. 基于支持向量机的地震体波震相自动识别及到时自动拾取[J]. 地球物理学报,2019,62(1):361-373.
刘芳,蒋一然,宁杰远,等. 结合台阵策略的震相拾取深度学习方法[J]. 科学通报,2020,65(11):1 016-1 026.
王雨. 山西断陷带的地震分布及地壳波速结构[D]. 北京:北京大学,2019.
于子叶,储日升,盛敏汉. 深度神经网络拾取地震P和S波到时[J]. 地球物理学报,2018,61(12):4 873-4 886.
赵明,陈石,Yuen D. 基于深度学习卷积神经网络的地震波形自动分类与识别[J]. 地球物理学报,2019a,62(1):374-382.
赵明,陈石,房立华,等. 基于U形卷积神经网络的震相识别与到时拾取方法研究[J]. 地球物理学报,2019b,62(8):
3 034-3 042.
Perol T, Gharbi M, Denolle M. Convolutional neural network for earthquake detection and location[J]. Sci Adv, 2018, 4(2):e1700578.
Ross Z E, Meier M A, Hauksson E. P wave arrival picking and first-motion polarity determination with deep learning[J]. J Geophys Res:Solid Earth, 2018, 123(6):5 120-5 129.
Woollan J, Rietbrock A, Bueno A, et al. Convolutional neural network for seismic phase classification, performance demonstration over a local seismic network[J]. Seismol Res Lett, 2019, 90(2A):491-502.
Zhang H J, Thurber C H. Double-difference tomography:the method and its application to the Hayward Fault, California[J]. Bull Seismol Soc Am, 2003, 93(5):1 875-1 889, doi:10.1785/0120020190.
Zhang H J, Thurber C H. Development and applications of double-difference seismic tomography[J]. Pure Appl Geophys, 2006, 163(2/3):373-403.
Zhou Y J, Yue H, Kong Q K, et al. Hybrid event detection and phase-picking algorithm using convolutional and recurrent neural networks[J]. Seismol Res Lett, 2019, 90(3):1 079-1 087.
Zhu L J, Peng Z G, McClellan J, et al. Deep learning for seismic phase detection and picking in the aftershock zone of 2008 MW 7.9 Wenchuan Earthquake[J]. Phys Earth Planet Inter, 2019a, 293:106261.
Zhu W Q, Beroza G C. PhaseNet:A deep-neural-network-based seismic arrival-time picking method[J]. Geophy J Int, 2019b, 216(1):261-273.
