A comparative study on three methods of onset-time determination for teleseismic P arrivals and parameters optimization
-
摘要: 分析了自回归赤池信息准则（AR-AIC）、高阶统计量（HOS）和累积和（CUSUM）等3种到时拾取方法中参数对远震P波到时估计的影响，以450个远震P波信号为样本集，参考人工拾取到时，以网格搜索方式确定了每种方法的最优参数。之后重新选取信噪比处于[2，20]区间的100个远震P波信号，用确定最优参数后的3种到时拾取方法估计其P波初至时间，并比较了3种方法对低信噪比远震P波的拾取准确度。结果表明，AR-AIC方法和CUSUM方法对低信噪比远震P波的拾取准确度要优于HOS方法，CUSUM方法的计算速度最快，HOS方法由于其原理的限制更适用于信噪比较大、初动较尖锐的信号。
-
关键词:
- 到时拾取 /
- 自回归赤池信息准则（AR-AIC）方法 /
- 高阶统计量（HOS）方法 /
- 累积和（CUSUM）方法
Abstract: This paper analyzed the effect of parameters on estimation of onset times for teleseismic P waves by AR-AIC （auto-regression Akaike information criterion）, HOS （higher order statistics） and CUSUM （cumulative sums） methods. The parameters of each method were optimized in the way of grid searching by taking 450 teleseismic P phases as sample set and the manual picked arrival times as reference. And then 100 teleseismic P phases were reselected with SNR between 2 and 20, and the onset times were estimated by the three methods with optimized parameters, so as to evaluate the accuracy of estimation for low-SNR teleseismic P phases. The results show that the accuracy of estimated phase arrival times by the AR-AIC and CUSUM methods is better than by the HOS method, the calculation speed by the CUSUM method is the fastest, and the HOS method is better applied to high-SNR and sharp onset signals.-
Keywords:
- automated picking /
- AR-AIC method /
- HOS method /
- CUSUM method
-
-
![]()
图 1 AR-AIC模型示意图
N为序列的长度，M为拟合的模型阶数，k为噪声与信号序列的分离点
Figure 1. The schematic diagram of AR-AIC model
N is the length of the sequence，M is the order of the AR-AIC model，and k is the divided point of noise and signal parts
![]()
图 2 实际地震的垂向记录及基于3种方法的到时拾取结果
（a） 实际地震信号的垂向记录；（b） 1—2 Hz滤波后的地震波形；（c） AR-AIC方法自动拾取的到时；（d） HOS方法自动拾取的到时；（e） CUSUM方法自动拾取的到时
Figure 2. The vertical component record of a real earthquake and the results of arrival time picked by the three estimation methods
（a） The vertical component record of a real earthquake；（b） The seismic waveform after 1—2 Hz bandpass filtering；（c） The estimated result with AR-AIC method；（d） The estimated result with HOS method；（e） The estimated result with CUSUM method
![]()
图 3 信噪比为31.1时基于HOS方法的P波信号拾取到时过程
（a） 0.75—1.5 Hz滤波后的地震波形；（b） 峰度曲线；（c） 峰度-AIC曲线
Figure 3. The process of onset time estimation using HOS method for P signal with SNR 31.1
（a) The seismic wave after 0.75-1.5 Hz bandpass filtering；（b） The kurtosis curve；（c） The kur-AIC curve
![]()
图 4 信噪比为4.5时基于HOS方法的P波信号拾取到时过程
（a） 0.75—1.5 Hz滤波后的地震波形；（b） 峰度曲线；（c） 峰度-AIC曲线
Figure 4. The process of onset time estimation using HOS method for P signal with SNR 4.5
（a) The seismic wave after 0.75-1.5 Hz bandpass filtering；（b） The kurtosis curve；（c） The kur-AIC curve
![]()
图 5 基于AR-AIC方法（a），HOS方法（b）和CUSUM方法（c）的自动拾取到时与人工到时误差分布及其比较（d）
Figure 5. Error between auto-onset times and manual onset times based on the methods AR-AIC （a），HOS （b） and CUSUM （c） and their comparison （d）
![]()
图 6 3种方法拾取到时的误差与信噪比的关系图
Figure 6. The relationship between errors in onset time estimation and SNR by using the three methods
表 1 3种到时拾取方法中待确定的参数
Table 1 The parameters used in the three methods for determining arrival times
方法 待确定参数 AR-AIC方法 拾取时窗长度T，噪声段长度Tn，信号段长度Ts，模型的阶数M HOS方法 拾取时窗长度T，滑移窗长度W，一阶导数门限系数α CUSUM方法 拾取时窗长度T 
下载: 导出CSV
表 2 用于合成地震波的3种到时拾取方法的参数取值组合
Table 2 The combination of parameter selection of the three methods for arrival times picking in synthesizing seismic wave
序号 AR-AIC方法 HOS方法 CUSUM方法 T Tn Ts M T W α T 组合① 20 2 3 17 30 9 0.36 22 组合② 20 3 4 17 26 7 0.36 12 组合③ 30 3 4 8 24 4 0.24 32
下载: 导出CSV
表 3 不同参数取值时自动到时与人工到时的误差
Table 3 The difference between auto-onset times and manual onset times with different parameter combination selection
SNR AR-AIC HOS CUSUM 组合① 组合② 组合③ 组合① 组合② 组合③ 组合① 组合② 组合③ 31.1 −0.03 −0.02 0 0.12 0.09 0.11 0.03 0.05 0.03 19.3 −0.02 −0.05 0.06 0.20 0.19 0.18 0.06 0.08 0.07 9.6 0.09 0.09 0.10 0.33 0.43 0.28 0.13 0.15 −0.34 4.5 0.16 0.22 0.31 0.46 0.57 0.56 −0.39 0.15 −0.39 2.4 0.19 0.18 −0.39 0.44 0.66 0.63 −0.39 0.18 −0.39
下载: 导出CSV
表 4 3种方法所得到的到时拾取误差统计
Table 4 Error statistics of phase onset time estimation by using the three methods
到时拾取方法 自动拾取与人工拾取的误差处于不同区间所占比例 误差均值/s 误差标准差/s 误差<0.1 s 误差<0.3 s 误差<0.5 s 误差<1 s AR-AIC方法 36% 79% 93% 100% 0.19 0.15 HOS方法 13% 59% 87% 97% 0.31 0.28 CUSUM方法 28% 82% 95% 100% 0.20 0.17
下载: 导出CSV
-
何燕,靳平,肖卫国,王红春. 2014. 迭代累积平方和算法与自回归赤池准则算法在地震信号到时估算中的比较研究[J]. 地震学报,36(2):209–219. He Y,Jin P,Xiao W G,Wang H C. 2014. A comparative study between ICSS and AR-AIC algorithms on onset time estimation for seismic signals[J]. Acta Seismologica Sinica,36(2):209–219 (in Chinese)
靳平,张诚鎏,沈旭峰,王红春,潘常周,严峰,王电源. 2014. 基于信号整体与局部特征的地震数据自动处理方法研究[J]. 地震学报,36(3):464–479 Jin P,Zhang C L,Shen X F,Wang H C,Pan C Z,Yan F,Wang D Y. 2014. A novel technique for automatic seismic data processing using both integral and local features of seismograms[J]. Acta Seismologica Sinica,36(3):464–479 (in Chinese)
林凡生, 邹志辉. 2016. 基于STA/LTA与多道互相关结合的地震波初至自动拾取方法[G]//国家安全地球物理丛书（十二）: 地球物理与信息感知. 西安: 西安地图出版社: 74–80. Lin F, Zou Z H. 2016. Automatic seismic first-arrival picking method based on the combination of STA/LTA and multi-channel cross correlation[G]//National Security Geophysics Series （12）: Geophysics and Information Perception. Xi’an: Xi’an Cartogra-phic Publishing House: 74–80 （in Chinese）.
盛冠群,李振春,王维波,高澜. 2015. 基于小波分解与高阶统计量的微地震初至拾取方法研究[J]. 石油物探,54(4):388–395 Sheng G Q,Li Z C,Wang W B,Gao L. 2015. A new automatic detection method of microseismic events based on wavelet decomposition and high-order statistics[J]. Geophysical Prospecting for Petroleum,54(4):388–395 (in Chinese)
谭玉阳,于静,冯刚,何川. 2016. 微地震事件初至拾取SLPEA算法[J]. 地球物理学报,59(1):185–196 Tan Y Y,Yu J,Feng G,He C. 2016. Arrival picking of microseismic events using the SLPEA algorithm[J]. Chinese Journal of Geophysics,59(1):185–196 (in Chinese)
田优平,赵爱华. 2016. 基于小波包和峰度赤池信息量准则的P波震相自动识别方法[J]. 地震学报,38(1):71–85 Tian Y P,Zhao A H. 2016. Automatic identification of P-phase based on wavelet packet and kurtosis-AIC method[J]. Acta Seismologica Sinica,38(1):71–85 (in Chinese)
王海军,靳平,刘贵忠,王晓明. 2003. 区域震相初至估计[J]. 西北地震学报,25(4):298–303 Wang H J,Jin P,Liu G Z,Wang X M. 2003. Accurate estimation for arrival time of seismic wave[J]. Northwestern Seismologi-cal Journal,25(4):298–303 (in Chinese)
岳龙,刘怀山,刘凯,徐秀刚,邢磊. 2015. 基于时频分析的初至拾取方法研究[J]. 石油物探,54(5):508–520 Yue L,Liu H S,Liu K,Xu X G,Xing L. 2015. First-break picking based on time-frequency analysis[J]. Geophysical Prospecting for Petroleum,54(5):508–520 (in Chinese)
曾富英,李敏锋,申维. 2002. 地震波初至拾取的分形研究[J]. 现代地质,16(2):209–213 Zeng F Y,Li M F,Shen W. 2002. The fractal study on detecting arrival time of the first break[J]. Geoscience,16(2):209–213 (in Chinese)
周银兴. 2009. 微震事件检测及震相自动识别研究[D]. 北京: 中国地震局地震预测研究所: 49–51. Zhou Y X. 2009. Research on the Micro-Earthquake Detection and Seismic Phase Automatic Identification[D]. Beijing: Institute of Earthquake Prediction, China Earthquake Administration: 49–51 （in Chinese）.
Akaike H. 1973. Information theory and an extension of the maximum likelihood principle[C]//Proceedings of the 2nd International Symposium on Information Theory. Budapest: Akadémiai Kiadó: 267–281.
Allen R V. 1978. Automatic earthquake recognition and timing from single traces[J]. Bull Seismol Soc Am,68(5):1521–1532
Allen R V. 1982. Automatic phase pickers:Their present use and future prospects[J]. Bull Seismol Soc Am,72(6B):S225–S242
Baer M,Kradolfer U. 1987. An automatic phase picker for local and teleseismic events[J]. Bull Seismol Soc Am,77(4):1437–1445
Der Z A,Shumway R H. 1999. Phase onset time estimation at regional distances using the CUSUM algorithm[J]. Phys Earth Planet Inter,113(1/4):227–246
Der Z A,Shumway R H. 2003. Automatic interpretation of regional short period seismic signals using CUSUM-SA algorithms[J]. Lecture Notes in Earth Sciences,98:41–59
Greg W B, David J B, Jerry A C. 2001. IDC Processing of Seismic, Hydroacoustic, and Infrasonic Data[R]. Reston, Virginia: Science Applications International Corporation Research, SAIC-99/3023: 22–26.
Inclán C,Tiao G C. 1994. Use of cumulative sums of squares for retrospective detection of changes of variance[J]. J Am Stat Associat,89(427):913–923
Jin P,Zhang C L,Shen X F,Wang H C,Pan C Z,Lu N,Xu X. 2014. A novel technique for automatic seismic data processing using both integral and local feature of seismograms[J]. Earthquake Science,27(3):337–349
Jin P. 2015. Toward high-confidence and full automation of seismic data processing for CTBT monitoring[C/OL]//Science and Technology 2015 Conference. [2015-12-18]. http://www.ctbto.org/fileadmin/user_upload/SnT2015/SnT2015_Orals/T3.3-O8.pdf.
Jin P,Pan C Z,Zhang C L,Shen X F,Wang H C,Lu N. 2015. A novel progressive signal association algorithm for detecting teleseismic/network-outside events using regional seismic networks[J]. Geophys J Int,201(3):1950–1960
Jurkevics A. 1988. Polarization analysis of three-component array data[J]. Bull Seismol Soc Am,78(5):1725–1743
Kedrov O K,Ovtchinnikov V M. 1990. An on-line analysis system for three-component seismic data:Method and preliminary results[J]. Bull Seismol Soc Am,80(6B):2053–2071
Küperkoch L,Meier T,Lee J,Friederich W,EGELADOS Working Group. 2010. Automated determination of P-phase arrival times at regional and local distances using higher order statistics[J]. Geophys J Int,181(2):1159–1170
Küperkoch L, Meier T, Diehl T. 2012. Chapter 16: Automated event and phase identification [G]//Bormann P ed. New Manual of Seismological Observatory Practice 2 （NMSOP-2）. Potsdam: IASPEI, GFZ German Research Centre for Geosciences: 17–23.
Maeda N. 1985. A method for reading and checking phase time in auto-processing system of seismic wave data[J]. Zisin,38(3):365–380
Nippress S E J,Rietbrock A,Heath A E. 2010. Optimized automatic pickers:Application to the ANCORP data set[J]. Geophys J Int,181(2):911–925
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
Sleeman R,van Eck T. 1999. Robust automatic P-phase picking:An on-line implementation in the analysis of broadband seismogram recordings[J]. Phys Earth Planet Inter,113(1/4):265–275
计量
- 文章访问数: 1179
- HTML全文浏览量: 516
- PDF下载量: 62
