自相关和褶积法单频干扰识别与消除

Identification and elimination of monofrequency interference on seismic data by an autocorrelation and convolution algorithm.

(1.成都理工大学,四川成都610059;2.CNPC东方地球物理公司物探技术研究中心,河北涿州072750)

Chengdu University of Technology,Chengdu 610059,China

在地震数据采集过程中会出现从浅层到深层频率、相位和振幅基本保持不变的50Hz单频干扰。常规时间域压制噪声的方法虽然能够压制这种干扰,但在实际海量地震数据的运算中,耗费时间长,计算效率低。基于正、余弦函数的褶积和自相关运算结果是同频率的正、余弦函数,则单频干扰的褶积和自相关运算结果也是同频率的正、余弦函数。因此,可以通过褶积和自相关运算来确定单频干扰。自相关和褶积法利用地震数据的自相关函数和褶积函数,计算单频干扰的自相关函数和褶积函数,运用正、余弦函数自适应减算法快速估算单频干扰。由于频率和相位与单频干扰呈非线性函数关系,因此使用任何估算方法都非常耗时。运用自相关和褶积法识别与消除单频干扰不需要估算单频干扰的频率、相位和振幅,因此可以高效、快速地估算出单频干扰,达到消除单频干扰的目的。该方法的最大优点是在快速、高效地消除单频干扰的同时不损害单频干扰频率分量附近有效信号的频率成分,提高了单频干扰频率分量附近数据的信噪比。合成数据和实际数据试算结果表明，该方法有效且可行。

During field seismic data acquisition,there is a strong monofrequency interference (MFI) around 50Hz in seismic records.The frequency,phase and amplitude of MFI can be considered basically invariable from shallow to deep layer.Traditional time-domain suppression methods can suppress this interference,but their computational efficiency is very low,and operational cost is very high,so it is not very effective for the real huge seismic data.The autocorrelation and convolution operation results of a sine-cosine function are also the sine-cosine functions with the same frequencies,so the autocorrelation and convolution operation results of a MFI are also the sinecosine functions of the same frequencies with the MFI.Thus,the MFI can be estimated by the autocorrelation and convolution operations for seismic records.For the autocorrelation and convolution algorithm (ACA),invoking the autocorrelation and the convolution function of a seismic record,those of the MFI are computed,and the MFI is quickly estimated by a sine-cosine function self-adaptive subtraction algorithm (SCFASA).Since a MFI has a nonlinear function relation with the frequency and phase,the computation of any parameter estimation algorithms takes long time.However,the identification and elimination of the MFI by the ACA would not need to estimate its amplitude,frequency and phase,so which can be utilized to estimate an MFI quickly and efficiently for the purpose of eliminating it.The greatest advantage of the ACA is that only interference with MFI component is quickly and effectively eliminated and its neighboring useful signal will not be damaged,so the signal-to-noise ratio has been improved in the vicinity of a MFI frequency component.The synthetic and real data examples illustrate that the method is feasible and effective.