论文部分内容阅读
由于信号传播过程中的畸变、变弱和分辨率的降低,从远震源测得的信号中信息成分常常被隐没。在大地中传播的地震波能量,可用常Q值模型(与频率无关)来近似衰减和波散等传播效应。对于传播着的平面波,由该模型可以导出一个空间衰减因子,它是频率的无界函数。因此,常Q值滤波器的宽带反演是不存在的。当震源与接收器间距离固定时,传播路径效应可在地震频带内用反演平面波传播的反褶积予以消除。传播反演通过时间反转(用-Q代替Q)实现,从而改变吸收以增加复波数。通常所测得的地震道包含着由深部变化而返回的信息。具有不同衰减量的波互相干涉使得反演过程复杂化。从伴有反向传播的多个平面波的迭合出发,提出了一个削弱大地滤波的通用反演方法。为了说明衰减随着深度增大,所以平面波反滤波是时变的。这一时变反滤波具有简单的傅里叶积分表达式,其中波数是复数,并包含着传播方向(因而随着距离增大而增大,而不是衰减)。为了控制子波边瓣,道内使用了一个频率域的窗函数(Hanning窗)。业已证明,这种两步法平面波反褶积比常规反褶积方法优越。野外资料试验表明这种方法对于消除VSP(垂直地震剖面)和地面测量中的衰减效应是有效的。在地震频带内估算了相位畸变,并降低了同相轴间的干涉。对于时间-带宽(传播时间信号的带宽)积小于有效Q值的同相轴,这种反演几乎是严格的。对于时间带宽大于Qeff的深部同相轴,出现巨大的子波边瓣。对谱的边缘进行削尖可以部分地压制子波边瓣。但浅层反射的巨大子波边瓣限制了带宽,在较深的同相轴上,带宽可以恢复到约2Qeff/t,t是到该同相轴的传播时间。反演算法的改进(例,用维纳滤波代替Hanning窗以控制边瓣)很可能改善我们的结果,但在许多情况下既使少量的测量噪声都会把反射序列限制在小于2倍有效Q值的时间-带宽乘积之内。
Due to the distortion, weakening and resolution of signal propagation, the information components in signals measured from teleseismic source are often hidden. The seismic wave energy propagating in the earth can be approximated by the constant Q-value model (independent of frequency) for propagation effects such as attenuation and dispersion. For propagating plane waves, a spatial attenuation factor can be derived from the model, which is an unbounded function of frequency. Therefore, there is no broadband inversion of constant Q value filter. When the distance between the source and the receiver is fixed, the propagation path effect can be canceled by the deconvolution of inverse plane wave propagation in the seismic frequency band. Propagation inversion is achieved by reversing the time (replacing Q with -Q), thereby changing the absorption to increase the number of complex waves. The commonly measured traces contain information that is returned as a result of deep changes. Waves with different amounts of attenuation interfere with each other and complicate the inversion process. Starting from the combination of multiple plane waves with back propagation, a general inversion method of weakening the ground filter is proposed. To illustrate the attenuation increases with depth, the plane wave inverse filtering is time-varying. This time-varying inverse filter has a simple Fourier-integral expression where the wave number is complex and contains the direction of propagation (thus increasing rather than attenuating as the distance increases). In order to control the wavelet flank, the window uses a window function in the frequency domain (Hanning window). It has been shown that this two-step planar wave deconvolution is superior to conventional deconvolution methods. Field data experiments show that this method is effective in eliminating the attenuation effects of VSP (vertical seismic profile) and ground measurements. The phase distortion was estimated in the seismic frequency band and the inter-event interference was reduced. This inversion is almost exact for the time-bandwidth (bandwidth of propagation time signal) product that is less than the effective Q-value of the in-phase axis. For deep events whose time bandwidth is larger than Qeff, a huge wavelet flank appears. Sharpening the edge of the spectrum can partially suppress the wavelet flank. However, the large sub-wave lobe reflected by the shallow layer limits the bandwidth. On the deeper axis, the bandwidth can be restored to about 2Qeff / t, where t is the propagation time to this phase. Improvements to the inversion algorithm (eg Wiener filtering instead of Hanning window to control the sidelobe) may improve our results, but in many cases even a small amount of measurement noise will limit the reflection sequence to less than 2 times the effective Q The time-bandwidth product is within.