地震台阵上信号方位角和慢度的时、频域估计方法比较
|
摘要
地震台阵常用频率-波数分析法来估计地震信号的后方位角和慢度。尽管有关的算法是众所周知的,但不同的实现在某些情况下可能导致不同的结果。如应用在宽频带的澳大利亚WRA台阵的记录时,标准的f-k分析方法往往给出不正确的结果。本文发现错误是由FFT的频谱泄漏效应引起的,如在进行FFT之前对原始数据进行高通滤波则可以有效地避免这样的错误。进一步对在时域中直接估算信号方位角和慢度的一种算法进行了分析,比较发现时域方法的计算速度和精度都不低于频域方法,且在某些低信噪比的情况下,前者可以给出更可靠的结果。
Frequency wave-number analysis(f-k analysis) is often taken to estimate the slowness and azimuth of sismic signal in seismic array.There are different algorithms of f-k analysis and in some special cases they lead to quite different results.For example,as being applied to the WRA array of Australia which is different from many other short period arrays by employing broadband seismic recordings at its substations,the standard f-k algorithm usually yields incorrect estimations of azimuth and slowness.It is found in this paper that the error is caused by the spectral leakage effect of FFT and may be avoided by applying a highpass filter to raw data before fk.After comparing the time domain approach of estimating azimuth and slowness in the f-k method with the frequency domain,it is shown that the time domain approach actually is not less accurate or efficient than the improved frequency domain approach,and in the case of signal-to-noise ratio is very low,the results obtained from the time domain approach are more reliable.
Frequency wave-number analysis(f-k analysis) is often taken to estimate the slowness and azimuth of sismic signal in seismic array.There are different algorithms of f-k analysis and in some special cases they lead to quite different results.For example,as being applied to the WRA array of Australia which is different from many other short period arrays by employing broadband seismic recordings at its substations,the standard f-k algorithm usually yields incorrect estimations of azimuth and slowness.It is found in this paper that the error is caused by the spectral leakage effect of FFT and may be avoided by applying a highpass filter to raw data before fk.After comparing the time domain approach of estimating azimuth and slowness in the f-k method with the frequency domain,it is shown that the time domain approach actually is not less accurate or efficient than the improved frequency domain approach,and in the case of signal-to-noise ratio is very low,the results obtained from the time domain approach are more reliable.
引文
[1]Capon J.High-resolution frequency-wavenumber spectrumanalysis[J].Proc.IEEE 57,1969,1408-1418.
[2]Kvaerna T,DJ Doornbos.Anintegrated approachto slownessanalysis with arrays and three-component array data[A].in:Semiannual Technical Summary,1 October 1985-31 March1986.Kjeller,Norway:Sci.Rept.No.2-85/86,NORSAR,1986.
[3]Kvaerna T,F Ringdal.Stability of various F-K esti mationtechniques[A].in:Semiannual Technical Summary,1 April1986-30 Seotember 1986,Sci.Rept.No.1-86/87,Kjeller,Norway:NORSAR,1986.29-40.
[4]Jin Wang.A Scheme for Initial Beam Deployment for the In-ternational Monitoring System Arrays[J].Pure and AppliedGeophysics,2002,I59:1005-1020.
[5]Peter Bormann.New Manual of Seismological ObservatoryPractice(Volume 1)[M].Potsdam:Geo Forschungs Zen-trum,2002.
[2]Kvaerna T,DJ Doornbos.Anintegrated approachto slownessanalysis with arrays and three-component array data[A].in:Semiannual Technical Summary,1 October 1985-31 March1986.Kjeller,Norway:Sci.Rept.No.2-85/86,NORSAR,1986.
[3]Kvaerna T,F Ringdal.Stability of various F-K esti mationtechniques[A].in:Semiannual Technical Summary,1 April1986-30 Seotember 1986,Sci.Rept.No.1-86/87,Kjeller,Norway:NORSAR,1986.29-40.
[4]Jin Wang.A Scheme for Initial Beam Deployment for the In-ternational Monitoring System Arrays[J].Pure and AppliedGeophysics,2002,I59:1005-1020.
[5]Peter Bormann.New Manual of Seismological ObservatoryPractice(Volume 1)[M].Potsdam:Geo Forschungs Zen-trum,2002.
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心