| The proposition and application of seismic interferomery promote the study ofseismic wave propagation. This method can get new response by correlating originalresponse received at two different locations. The new response includes two parts, thecausal part and anti-causal part. The causal part is very similar with the response thatare received at one receiver and activated at the other. Seismic interferometry includespassive source seismic interferometry and control source seismic interferometry.Passive source seismic interferometry can make the noise signal become useful. Withthis method, the useful information can be extracted from the noise signal or the codewave data. It should be noted that there is no information generated in this process,but the information hidden in the response can be extracted from the original response.Then, the further process can be done.The controlled source seismic interferometry can generate new seismic data from theoriginal data. Specifically speaking, the source can be redatumed to the receiverlocation. The geometry can get closer to target zone. So the imaging precision can beimproved dramatically. At the meantime, since the geophone becomes the virtualsource, the illumination area can be enlarged. In practice, it is very hard to collect zerooffset data. Using seismic interferometry can eliminate the effect of offset andgenerate the zero offset data. Also, based on seismic interferometry, many methodsare proposed or improved.Firstly, this paper introduces the basic theory of seismic interferometry from1D directwave interferometry. Then, through2D direct wave seismic interferometry, we studythe basic requirement of seismic interferometry. We investigate seismic interferometrywith impulsive source and noise source. Finally, we study the reflect waveinterferoemtry; investigate the theory of reflect wave interferometry; derive the basicseismic interferometry equation.Through numerical modeling, we study the feasibility and imaging factors of passiveseismic interferometry. The results show that passive seismic interferometry imaging is feasible. If the passive sources distribute widely and the medium around sources isstable, the imaging will be good. Also, the source wave must be uncorrelated. Thelonger the source wave is, the better the imaging is. Finally, if the record time is longenough, the imaging will be good.For controlled source seismic interferometry, we study its use in time-space domain.We study the transform of VSP-SSP, VSP-SWP and SSP-SSP. Through seismicintererometry, the geometry can get closer to the target zone; also, using this methodcan eliminate the effect of offset; improve the imaging precision.When the controlled source seismic interferometry is applied in time-space domain,many spurious events will appear in the virtual source profile. If the original data isincomplete, the virtual profile will contain more spurious events. Also, noise signalwill reduce the quality of virtual profile. Therefore, we combine the Radon transformwith seismic interferometry, propose Radon domain seismic interferometry. Firstly,we combine qtransform with seismic interferometry. Since the qtransformcan be used in data interpolation, qdomain seismic interferometry can suppressthe effect of missing data. Also, the spurious events can be reduced, because offiltering. In this paper, we test this method both on synthetic data and real data.Finally, we propose linear Radon domain multidimensional deconvolution seismicinterferometry, when multidimensional deconvolution seismic interferometry, theinversion may unstable due to irregularity of data. When transform data to linearRadon domain, the data become regular. So the inversion will become stable. At themeantime, this method can suppress the effect of noise; reduce the spurious eventsand improve the virtual source profile. Through the test on synthetic data, this methodcan generate the virtual source profile accurately. Besides, this method is lesssensitive to noise. Even if the SNR is very low, the virtual source profile also hasgood precision. We also test this method on real data. The result is pretty good. |
