地震研究

以声谱图像模式识别为基本思想,改进了基于地震波波谱包络分布特征的初至震相检测方法。改进后的算法,首先将波形记录特征转换为谱能量分布,以台站固有背景噪声的时频特征为基准,提取出记录信号的时频包络特征作为事件触发检测的目标函数。为降低主震尾波对检测后续余震的影响,采用非线性标度缩放因子对大于台站背景噪声水平的能量进行归一化处理。通过2组密集余震序列的实测结果表明:相对STA/LTA算法自动检测结果,新方法正确触发率提高约19.6%,漏触发率则降低了22.6%,表现出更强的抗干扰能力和触发效率,且运行速度快,能够满足地震实时处理系统的技术要求。

Based on the basic idea of acoustic spectrum pattern recognition,we have improved the first arrival seismic phase detection method based on the envelope characteristic of seismic wave spectrum. The improved algorithm is based on the time-frequency characteristics of the inherent ambient noise of station as the criterion of the event triggered detection,and the waveform record feature is converted to spectral energy distribution,and the scaling factor is used to normalize the energy that is larger than the ambient noise level of the station,in order to reduce the influence of the main shock wave on detecting subsequent aftershocks. Through the measurement of two sequences of dense aftershock,the results show that the method presented in this paper,the correct trigger rate is improved by about 19.6%,and the leakage trigger rate is reduced by 22.6%,compared with the automatic detection result of STA/LTA algorithm. It shows stronger anti-interference ability and trigger efficiency,and it is similar to the STA/LTA algorithm,which is running fast and can meet the requirements of the real-time seismic processing system.

引言
1 研究方法
2 实测结果
3 结论与讨论

图1 2014年9月宝应台背景噪声的加速度功率谱概率密度分布结果<br/>Fig.1 The probability density function of power spectrum density of ambient noise at the Baoying station on September 2014

图1 2014年9月宝应台背景噪声的加速度功率谱概率密度分布结果
Fig.1 The probability density function of power spectrum density of ambient noise at the Baoying station on September 2014

图2 宝应台震后1 h的原始波形(a)、时频能量分布及包络特征曲线(b)及nij=-140 dB时的时频能量分布(c)<br/>Fig.2 The original waveform in one hour after the earthquake recorded by the Baoying station(a), its time-frequency energy distribution and the envelope feature curve(b),and the time-frequency energy distribution results while nij=-140dB(c)

图2 宝应台震后1 h的原始波形(a)、时频能量分布及包络特征曲线(b)及nij=-140 dB时的时频能量分布(c)
Fig.2 The original waveform in one hour after the earthquake recorded by the Baoying station(a), its time-frequency energy distribution and the envelope feature curve(b),and the time-frequency energy distribution results while nij=-140dB(c)

图3 2014年9月石场台背景噪声的加速度功率谱概率密度<br/>Fig.3 The probability density function of power spectrum density of ambient noise at the Shichang station on September 2014

图3 2014年9月石场台背景噪声的加速度功率谱概率密度
Fig.3 The probability density function of power spectrum density of ambient noise at the Shichang station on September 2014

表1 江苏高邮M4.9地震序列的触发检测结果对比<br/>Tab.1 Comparison of the triggered detection results by different methods at the Gaoyou M4.9 earthquake sequence in Jiangsu

表1 江苏高邮M4.9地震序列的触发检测结果对比
Tab.1 Comparison of the triggered detection results by different methods at the Gaoyou M4.9 earthquake sequence in Jiangsu

图4 石场台震后1 h的原始波形(a)和时频能量分布(b)<br/>Fig.4 The original waveform in one hour after the earthquake recorded by the Shichang station(a), and its time-frequency energy distribution and the envelope feature curve(b)

图4 石场台震后1 h的原始波形(a)和时频能量分布(b)
Fig.4 The original waveform in one hour after the earthquake recorded by the Shichang station(a), and its time-frequency energy distribution and the envelope feature curve(b)

图5 两种算法自动计算时长为1 h的连续记录数据的计算耗时对比<br/>Fig.5 The time-consuming comparison of two algorithms to automatically process the time length of 1 hour continuous recording data in this study

图5 两种算法自动计算时长为1 h的连续记录数据的计算耗时对比
Fig.5 The time-consuming comparison of two algorithms to automatically process the time length of 1 hour continuous recording data in this study

黄媛,吴建平,张天中.2008.汶川8.0级大地震及其余震序列重定位研究[J].中国科学:地球科学,38(10):1242-1240.
刘希强,周蕙兰,沈萍,等.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
王俊,徐戈,孙业君.2009.江苏省区域地表背景噪声特性的分析[J].地震研究,32(2):155-161.
王俊,郑定昌,詹小艳,等.2013.基于背景噪声互相关格林函数的单台时间误差估计[J].地震学报,35(6):888-901
王勤彩,陈章立,郑斯华.2009.汶川大地震余震序列震源机制的空间分段特征[J].科学通报,54(16):2348-2354
朱艾斓,徐锡伟,刁桂荃,等.2008.汶川MS8.0地震部分余震重新定位及地震构造初步分析[J].地震地质,30(3):759-767.
Allen R.1978.Automatic earthquake recognition and timing from single trace[J].BSSA,68(5):1521-I532.
Bormann P.2012.New Manual of Seismological Observatory Practice 2(NMSOP-2),Potsdam:Deutsches GeoForschungsZentrum GFZ,pp.1-65,doi:http://doi.org/10.2312/GFZ.NMSOP-2_ch8
Joswig M.1990.Pattern recognition for earthquake detection[J].BSSA,80(1):170-186.
Joswig M.1993.Single-trace detection and array-wide confidence association of local earthquakes and explosions[J].Computers and Geosciences,19(2):207-221.
Leonard M,Kennett B L N.1999.Multi-component autoregressive techniques for the analysis of seismograms[J].Phys Earth Planet Interi,113(1-4):247-264.
Leonard M.2000.Comparison of manual and automatic onset time picking[J].BSSA,90(6):1384-1390.
Massa M,Ferretti G,Spallarossa D.2006.Improving automatic location procedure by waveform similarity analysis:An application in the South Western Alps(Italy)[J].Phys Earth Planet lnteri,154(1):18-29.
McNamara D E,Boaz R I.2005.Seismic Noise Analysis System Using Power Spectral Density Probability Density Functions:A Stand-Alone Software Package[R].Virginia U S Geological Survey.
Patane D,Ferrari F,Ferrucci F.1999.First application of ASDP software:A case study at Mt.Etna volcano and in the Acri region(Southern Italy)[J].Phys Earth Planet lnteri,113(1-4):75-88.
Sleeman R,Van E,Torild.1999.Robust automatic P-phase picking:An on-line implementation in the analysis of broadband seismogram recordings[J].Phys Earth Planet Interi,113(1-4):265-275.
Stevenson P R.1976.Microearthquakes Flathead Lake,Montana:A study using automatic earthquake processing[J].BSSA,66(1):61-79.
Wang J,Teng T L.1995.Artificial neural network-based seismic detector[J].BSSA,85(1):308-319.
Withers M,Aster R,Young C,Beiriger J,et al.1998.A comparison of select trigger algorithms for automated global seismic phase and event detection[J].BSSA,88(1):95-106.

备注

引言

1 研究方法

2 实测结果

3 结论与讨论

期刊信息