An automatic S phase picking algorithm for local earthquake events
-
摘要: 基于特征值分解方法，本文讨论了一种适用于地方震事件S波震相到时拾取的自动处理算法。该算法计算参数少、简便快捷、易于实现，通过选用七个不同长度的时间窗，有效地减小了窗长选择不合理所引起的震相拾取误差。利用福建地震台网记录的9 855条三分向波形记录进行测试，结果表明：本文方法的S波平均拾取偏差为（0.003±1.34） s，其中79.6%的记录拾取偏差小于0.5 s，4.1%的记录拾取偏差超过2.0 s，说明本文方法能够满足日常工作基本需求。综上分析认为，波形记录质量是影响拾取算法结果精度的最主要因素，信噪比较高的记录，其S波到时拾取偏差显著优于信噪比较低的记录，对信噪比较低的部分记录进行带通滤波预处理后，S波震相拾取精度也有所提升。Abstract: Base on eigenvalue decomposition calculation, we discussed an automatic S phase picking algorithm suitable for local earthquake events. The algorithm needs few parameters, less computer resources and is easy to realize. By choosing seven time windows with different time length, it effectively reduce the S phase pick errors that caused by unreasonable time window selection. By taking 9 855 three-component records of Fujian seismic monitoring network, we test the applicability and accuracy of this algorithm. The results show that average S phase picking error by this method is （0.003±1.34） s, 79.6% of them are within 0.5 s, 4.1% of them are greater than 2.0 s, which indicates that the algorithm can meet the demands for daily S phase picking. All above-mentioned suggests that the quality of records is the most important factor that affects the accuracy of picking results, the S phase picking errors for high signal-noise ratio records are usually smaller than those for low signal-noise ratio records, and the accuracy of S phase picking for some low signal-noise ratio records will be improved after a band-pass filter preprocess.
-
Key words:
- eigenvalue decomposition /
- S wave /
- seismic phase detection /
- real-time process /
- signal-noise ratio
-
图 1 本文S波拾取方法算例示意图
事件发震时刻为2015-01-02 00:33:56.89，震中距为3.8 km的GTSK台站记录，窗长为0.2 s。图（a−c）为三分向波形记录；图（d）为特征值时程；图（e）为特征函数f （t）；图（f）为峰度系数时程K（t）；图（g）为求取差分后特征函数ΔK（t）
Figure 1. An example of S phase picking by using the algorithm of this study
The origin time of the event is 00:33:56.89 on 2 January 2015. The wavforms were recorded by the station GTSK with epicentral distance 3.8 km，and window length is taken as 0.2 s. Figs. （a） to （c） are three-component seismic records；Fig. （d） shows the history of three eigenvalues；Fig. （e） is the characteristic function f （t）；Fig. （f） shows time history of Kurtosis coefficient K（t），and Fig. （g） shows the differential characteristic function ΔK（t）
图 2 本文所用台站记录随震中距和震级分布
Figure 2. Epicentral distance and magnitude distribution for earthquake records used in this study
图 4 应用本文方法的S波震相到时拾取偏差统计
图（a）为各记录S震相拾取偏差散点图，图（b）和（c）为S震相拾取偏差统计直方图
Figure 4. Statistics of S phase arrival time pick errors by using our method
Fig. （a） gives S phase picking error for each record，Figs. （b） and （c） show the histogram of the phase picking error
图 5 分区域S波震相到时拾取偏差结果统计
（a） 仙游震群；（b） 台湾海峡南部震群；（c） 其它事件
Figure 5. Statistics of S phase arrival time picking error in different regions
（a） Xianyou sequence；（b） Southern Taiwan Strait sequence；（c） Other events
图 6 S波震相到时拾取偏差较大的波形记录
（a） 低信噪比记录结果；（b） 图（a）中记录1—20 Hz带通滤波后的结果；（c） 多事件叠加记录结果；（d） EW向异常记录结果
Figure 6. Waveforms with large S phase arrival time picking error
（a） The result for a low SNR record；（b） The result for the record in Fig. （a） after 1−20 Hz band-pass filtering；（c） The result for a multi-event record；（d） The result for an abnormal record in EW component
图 7 地震记录信噪比的分布
（a） 拾取偏差大于2.0 s的406条记录；（b） 拾取偏差小于2.0 s的9449条记录
Figure 7. SNR distribution of seismic records
（a） 406 records with S phase picking errors larger than ±2.0 s；（b） 9449 records with S phase picking errors less than ±2.0 s
图 8 不同信噪比记录S波震相拾取偏差分布
（a） 信噪比小于2.0的记录；（b） 信噪比大于2.0的记录
Figure 8. Pi chart of S phase detection error for different SNR records
（a） Records with SNR less than 2.0；（b） Records with SNR larger than 2.0
图 9 应用本文方法所得的S－P到时差与人工拾取结果对比及其信噪比分布
（a） 所有记录；（b） 仙游震群；（c） 台湾海峡震群；（d） 其它事件
Figure 9. Comparison of the arrival time differences for S−P by our method with those by manual picking and SNR distributions
（a） All records；（b） Xianyou sequence；（c） Taiwan Strait sequence；（d） Other events
图 10 图6a中信噪比较低记录带通滤波前（a）、后（b）的S波到时拾取
各子图意思同图1，事件发震时间为2015-01-08 05:48:44.56，台站ZPCH的震中距为99.8 km，窗长为1.0 s
Figure 10. S phase picking for a low SNR record in Fig. 6a before （a） and after （b） band-pass filtering
Each subfigure has the same meaning as Fig. 1. The event occurred at 05:48:44.56 on 8 January 2015，which was recorded by the station ZPCH with epicentral distance 99.8 km. And time window length is taken as 1.0 s
表 1 不同信噪比记录的S波到时拾取偏差统计
Table 1. Statistic on S phase picking error for different SNR records
SNR 记录数量 偏差均值/s 偏差中值/s 偏差标准差/s ＜2 2 953 0.135 0.040 1.92 2—5 6 416 −0.051 0.050 0.99 ≥5 486 −0.202 0.040 0.96 
下载: 导出CSV
-
[1] 李安,杨建思,彭朝勇,郑钰,刘莎. 2020. 基于样本增强的卷积神经网络震相拾取方法[J]. 地震学报,42(2):163–176. doi: 10.11939/jass.20190070 [2] Li A,Yang J S,Peng C Y,Zheng Y,Liu S. 2020. Seismic phase identification using the convolutional neural networks based on sample enhancement[J]. Acta Seismologica Sinica,42(2):163–176 (in Chinese). doi: 10.11939/jass.20190070（inChinese） [3] 刘希强,周蕙兰,郑治真,沈萍,杨选辉,马延路. 1998. 基于小波包变换的弱震相识别方法[J]. 地震学报,20(4):373–380. doi: 10.3321/j.issn:0253-3782.1998.04.005 [4] Liu X Q,Zhou H L,Zheng Z Z,Shen P,Yang X H,Ma Y L. 1998. Identification method of weak seismic phases on the basis of wavelet packet transform[J]. Acta Seismologica Sinica,20(4):373–380 (in Chinese). [5] 刘希强,周蕙兰,沈萍,杨选辉,马延路,李红. 2000. 用于三分向记录震相识别的小波变换方法[J]. 地震学报,22(2):125–131. doi: 10.3321/j.issn:0253-3782.2000.02.002 [6] Liu X Q,Zhou H L,Shen P,Yang X H,Ma Y L,Li H. 2000. Identification method of seismic phase in three-component seismograms on the basis of wavelet transform[J]. Acta Seismologica Sinica,22(2):125–131 (in Chinese). [7] 马强. 2008. 地震预警技术研究及应用[D]. 哈尔滨: 中国地震局工程力学研究所: 14–39. [8] Ma Q. 2008. Study and Application on Earthquake Early Warning[D]. Harbin: China Earthquake Administration, Institute of Engineering Mechanics: 14–39 （in Chinese）. [9] 马强,金星,李山有,陈绯雯,廖诗荣,韦永祥. 2013. 用于地震预警的P波震相到时自动拾取[J]. 地球物理学报,56(7):2313–2321. doi: 10.6038/cjg20130718 [10] Ma Q,Jin X,Li S Y,Chen F W,Liao S R,Wei Y X. 2013. Automatic P-arrival detection for earthquake early warning[J]. Chinese Journal of Geophysics,56(7):2313–2321 (in Chinese). doi: 10.6038/cjg20130718（inChinese） [11] 赵明,陈石,房立华,Yuen D A. 2019. 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,62(8):3034–3042. doi: 10.6038/cjg2019M0495 [12] Zhao M,Chen S,Fang L H,Yuen D A. 2019. Earthquake phase arrival auto-picking based on U-shaped convolutional neural network[J]. Chinese Journal of Geophysics,62(8):3034–3042 (in Chinese). [13] Akaike H. 1974. A new look at the statistical model identification[J]. IEEE Trans Automat Control,19(6):716–723. doi: 10.1109/TAC.1974.1100705 [14] Allen R. 1982. Automatic phase pickers:Their present use and future prospects[J]. Bull Seismol Soc Am,72(6B):S225–S242. doi: 10.1785/BSSA07206B0225 [15] Amoroso O,Maercklin N,Zollo A. 2012. S-wave identification by polarization filtering and waveform coherence analyses[J]. Bull Seismol Soc Am,102(2):854–861. doi: 10.1785/0120110140 [16] Anant K S,Dowla F U. 1997. Wavelet transform methods for phase identification in three-component seismograms[J]. Bull Seismol Soc Am,87(6):1598–1612. [17] Baer M,Kradolfer U. 1987. An automatic phase picker for local and teleseismic events[J]. Bull Seismol Soc Am,77(4):1437–1445. doi: 10.1785/BSSA0770041437 [18] Chu C K P,Mendel J M. 1994. First break refraction event picking using fuzzy logic systems[J]. IEEE Trans Fuzzy Syst,2(4):255–266. doi: 10.1109/91.324805 [19] Cichowicz A. 1993. An automatic S-phase picker[J]. Bull Seismol Soc Am,83(1):180–189. [20] Dai H C,MacBeth C. 1995. Automatic picking of seismic arrivals in local earthquake data using an artificial neural network[J]. Geophys J Int,120(3):758–774. doi: 10.1111/j.1365-246X.1995.tb01851.x [21] Flinn E A. 1965. Signal analysis using rectilinearity and direction of particle motion[J]. Proc IEEE,53(12):1874–1876. doi: 10.1109/PROC.1965.4462 [22] Jurkevics A. 1988. Polarization analysis of three-component array data[J]. Bull Seismol Soc Am,78(5):1725–1743. [23] Küperkoch L,Meier T,Brüstle A,Lee J,Friederich W. 2012. Automated determination of S-phase arrival times using autoregressive prediction:Application to local and regional distances[J]. Geophys J Int,188(2):687–702. doi: 10.1111/j.1365-246X.2011.05292.x [24] Kong Q K,Trugman D T,Ross Z E,Bianco M J,Meade B J,Gerstoft P. 2019. Machine learning in seismology:Turning data into insights[J]. Seismol Res Lett,90(1):3–14. doi: 10.1785/0220180259 [25] Lois A,Sokos E,Martakis N,Paraskevopoulos P,Tselentis G A. 2013. A new automatic S-onset detection technique:Application in local earthquake data[J]. Geophysics,78(1):KS1–KS11. doi: 10.1190/GEO2012-0050.1 [26] Lomax A,Satriano C,Vassallo M. 2012. Automatic picker developments and optimization:FilterPicker:A robust,broadband picker for real-time seismic monitoring and earthquake early warning[J]. Seismol Res Lett,83(3):531–540. doi: 10.1785/gssrl.83.3.531 [27] Perol T,Gharbi M,Denolle M. 2018. Convolutional neural network for earthquake detection and location[J]. Sci Adv,4(2):e1700578. doi: 10.1126/sciadv.1700578 [28] Ross Z E,Meier M A,Hauksson E,Heaton T H. 2018a. Generalized seismic phase detection with deep learning[J]. Bull Seismol Soc Am,108(5A):2894–2901. doi: 10.1785/0120180080 [29] Ross Z E,Meier M A,Hauksson E. 2018b. P wave arrival picking and first-motion polarity determination with deep learning[J]. J Geophys Res:Solid Earth,123(6):5120–5129. doi: 10.1029/2017JB015251 [30] Ruud B O,Husebye E S. 1992. A new three-component detector and automatic single-station bulletin production[J]. Bull Seismol Soc Am,82(1):221–237. [31] Saragiotis C D,Hadjileontiadis L J,Panas S M. 2002. PAI-S/K:A robust automatic seismic P phase arrival identification scheme[J]. IEEE Trans Geosci Remote Sens,40(6):1395–1404. doi: 10.1109/TGRS.2002.800438 [32] Vidale J E. 1986. Complex polarization analysis of particle motion[J]. Bull Seismol Soc Am,76(5):1393–1405. [33] Zhu L J, Peng Z G, McClellan J. 2018. Deep learning for seismic event detection of earthquake aftershocks[C]//52nd Asilomar Conference on Signals, Systems, and Computers. Pacific Grove: IEEE: 1121–1125. [34] Zhu W Q,Beroza G C. 2018. PhaseNet:A deep-neural-network-based seismic arrival-time picking method[J]. Geophys Int J,216(1):261–273. doi: 10.1093/gji/ggy42 -
查看大图
计量
- 文章访问数: 78
- HTML全文浏览量: 18
- PDF下载量: 20
- 被引次数: 0