Study On Passive Seismic Interferometry And Primary Estimation

Seismic interferometry is developed in recent years, and quickly becomes a hot direction focused by many scholars. Seismic interferometry can reconstruct the wave-field which propagates between two receivers, to synthesis the new seismic wave-field, which is called the virtual-shot recording. One of the two receivers is regarded as a virtual source, another as a new receiver of the virtual-shot recording. Seismic interferometry can extract the effective signals from the chaos background noise, and obtain the related information about the underground geological structure. In this paper, we study the acoustic and elastic passive seismic interferometry and primary estimation from passive seismic data respectively. Firstly, the cross-correlation algorithm of acoustic seismic interferometry is derived in detail. On this basis, we also derive the deconvolution algorithm of seismic interferometry. After that, the characteristics of the transient and noise passive seismic data are analyzed and compared, and reconstruct the wave-field of the passive seismic recording with the cross-correlation and deconvolution algorithm respectively to synthesize the virtual-shot recording. Comparing with the control source recording, it demonstrates the accuracy of the virtual-shot recording. Meanwhile, we can find that the resolution of the deconvolution algorithm is higher than the cross-correlation algorithm.Secondly, the paper proposes passive seismic interferometry in Linear Radon domain. As in the time-space domain, seismic interferometry is implemented through stacking the interferometry of the full passive seismic data to synthesize the virtual-shot recording. But, the contribution for synthesizing the virtual-shot recording mainly comes from the stationary-phase point. That is the interferometric effects of the vast majority of passive seismic data are counteracted with each other, and the main contribution for synthesizing the virtual-shot recording comes from a small part of passive seismic data. Through analyzing, we can find that the effective seismic response received by the two receivers comes from the same source. Then the response received by the two receivers should have the same ray parameter. According to the characteristics of the linear Radon transform, the paper makes the passive seismic data in time-space domain transform into the Linear Radon domain, and makes the passive seismic data with the same ray parameter correlate with each other in the Linear Radon domain. The virtual-shot recording is synthesized successfully. On this basis, we shrink the scanned aperture for comparing and analyzing the image effectiveness of the synthetic virtual-shot recording. It can be fund that the effective responses in the synthetic virtual-shot recording are still reconstructed better in the Linear Radon domain when the scanned aperture is small. That further demonstrates the contribution for synthesizing the virtual-shot recording mainly comes from the stationary-phase point. Meanwhile, the limited passive seismic data in the Linear Radon domain can reconstruct the virtual-shot recording better. In this way, the computational efficiency of seismic interferometry is improved greatly.Thirdly, the synthetic virtual-shot recording of the acoustic passive seismic data contains the primaries and the surface-related multiples. The paper studies estimating primaries from the acoustic passive seismic data based on the theory of estimating primaries by sparse inversion from the control sources. The paper uses the convex optimization method with L1-norm constraint to solve the problem of estimating primaries from the acoustic transient passive seismic data, which avoids using a time-window during the traditional process to prevent the inversion from getting into the local optimization situations. In the inversion, the paper combines the sparse transform with the convex optimization method with L1-norm constraint, which improves the accuracy of the estimated primaries. The synthetic virtual-shot recording of the acoustic noise passive seismic data also has much surface-related multiples. The paper proposes the method of estimating primaries from the noise passive seismic data based on the theory of the transient passive seismic data. The matrix data of the acoustic noise passive seismic data which is similar with the transient one, can also be expressed with the transmission direct wave arrivals and the surface-related multiples caused by the transmission direct wave arrivals. It assumes that the energy of the transmission direct wave arrivals are small. Then the objective function of estimating primaries from the acoustic noise passive seismic data can be obtained. As the noise passive seismic data do not have the sparsity, the traditional method and the L1-norm constraint method are not suitable for solving the problem of estimating primaries from the noise passive seismic data. According to the characteristics of the noise passive seismic data, the paper introduces the LSQR algorithm based on the principle of minimum energy to estimate primaries from the acoustic noise passive seismic data. It also gets a good results.At last, the study for the passive seismic data is generalized to the elastic wave by the acoustic wave. The elastic reciprocal theory and the elastic seismic interferometry including the transient and noise sources are derived respectively. Then, the theory model is used to simulate the elastic transient passive seismic recording. It can be found that the elastic passive seismic data contains not only P-wave, but also S-wave. Except that, it also includes many converted waves, which make the passive seismic recording become very complex. On the contrary, the effective information becomes rich. We reconstruct the wave-field of the elastic passive seismic data by the cross-correlation algorithm of seismic interferometry. The reflecting responses of P-wave and S-wave in the synthetic elastic virtual-shot recording are reconstructed better. Comparing with the control-shot recording with the free surface, it demonstrates the accuracy of the elastic virtual-shot recording.The elastic virtual-shot recording not only contains the primaries and the surface-related multiples, but also includes much converted waves The elastic passive seismic data, similar with the acoustic one, can be divided into the transmission direct wave, the surface-related multiples caused by the transmission direct wave and the converted waves aroused by the surface-related multiples. Through introducing the method of estimating primaries from the acoustic passive seismic data to deal with the elastic one, we propose the method of estimating primaries from the elastic passive data. The method of solving the optimistic problem with L2-norm constraint and L1-norm constraint are used to estimating primaries from the elastic passive seismic data respectively. Comparing with the elastic control-shot recording under the absorb-surface condition, the estimated elastic primaries not only contain the primaries, but also include the converted waves. Through analyzing, we can find that the surface-related multiples caused by the transmission direct wave arrivals generate the converted waves at the free surface or interface. During estimating the elastic primaries, the converted waves caused by the surface-related multiples can’t be suppressed, so the estimated elastic primaries will contain the converted waves.