【作者】 许卓；

【导师】 韩立国； Christopher Juhlin；

【作者基本信息】 吉林大学 ， 地球探测与信息技术， 2012， 博士

【摘要】 地震干涉法的提出及其在野外实际数据上的应用，极大的丰富了对地震波传播规律的认识和研究，被认为是地球物理发展中的一个重大进展。地震干涉法的适用范围从一维的层状均匀声波介质推广到三维弹性介质、任意非均匀介质或各向异性介质，甚至可被推广到运动的介质以及衰减介质。地震干涉法的基本思想是从混沌无序的地震信号中提取出有用信号，通过对记录到的地震信号进行干涉，得到新的地震信号，新形成的地震信号不仅包含了原始地震信号的特性，而且能反映出原始信号所不具有的某些重要特征。地震波干涉方法分为主动源地震干涉和被动源地震干涉，其震源可以是确定性震源也可以是噪声震源。主动源地震干涉法为地震勘探提供了一种新的处理方法，可以用于多种数据类型。以垂直地震剖面（VSP）数据类型的应用为代表，通过地震干涉法可以将VSP数据转换成其他类型的数据，如单井地震剖面（SWP）数据，地表地震剖面（SSP）数据等。被动源地震干涉法用于从野外地震噪声中提取出有用信号来获取地震波穿过介质的地质特性并推断地下地质构造的特征。本文中，我们分别就地震干涉法在主动源地震数据和被动源地震数据上的应用进行深入的研究和讨论。首先阐述了地震干涉法的基本原理，以互易定理为基础，按照无损任意非均匀介质的假设，推导准确的声学和弹性动力学格林函数的表述。而重构的格林函数中不仅包含两个检波点之间的直达波场，也包含所有一次反射和多次散射。对于地震干涉提取格林函数的表述是基于用于相关处理中的检波器是被一个封闭的震源面所包围的这样一个假设。一般情况下，往往不能满足地震干涉应用中所需的震源分布条件，因此文中通过数值模拟实验调查当理论条件不能被满足，且震源位于地下深部位置处时，震源分布特性对重构的反射响应的影响进行讨论。在地震勘探的实际应用中，通常不能满足理论上所需的连续震源分布，但在干涉计算过程中，震源间隔应尽量布设的密集一些，以防止空间假频的产生。当震源只集中分布于地下局部区域内时，重构的道集中只有部分反射同相轴能被恢复出来，且大大降低了重构响应的信噪比。一般而言，可以利用地震干涉法从地下的透射响应中重构出有效的反射信息，并对重构的反射信息进一步应用偏移处理，偏移成像可以改善重构共炮点道集中反射信息的不完整性，获得令人满意的结果。文中对地震干涉法在主动源数据上的应用进行了深入系统的研究，主要针对的是VSP型数据。当常规地震勘探数据处理方法不能对上覆盖层复杂变化的地震资料获得满意的成像效果时，可以通过虚拟震源方法将VSP数据转换成SWP数据，变换之后，原来位于地表的震源被重建到井下位置处，震源-检波器的井下排列离地质目标体更近，且可以有效的避免近地表变化和复杂上覆盖对数据的影响，不需要知道上覆盖层的复杂速度模型，为处理这种复杂地震资料提供了一种有效途径。本文主要利用虚拟震源方法探测地下陡构造地质体，通过数值模拟发现地质体越陡峭，获得的成像效果越好，并且虚拟震源方法可以自动校正静校正量。虚拟震源方法对于弹性波数据的应用也是有效的，能够获得合理的成像结果。此外，模拟实验表明，虚拟震源方法应用于稀疏观测系统仍然是非常有效的，尽管随着震源及检波器间隔的增大，信噪比随之下降，但是仍能构建出正确的成像结果。标准VSP数据的成像精度虽然很高，但是成像范围却很有限，仅能对从第一个检波器开始向下的一个三角锥形区域内的地质体进行成像。为了扩展地下地质构造的成像范围，可以利用地震干涉中的相关算法将VSP数据重构为虚拟SSP数据，并对虚拟SSP数据进行成像处理。由此得到的结果中，地下地质构造的成像范围等同于具有相同震源覆盖范围的常规地表地震勘探中获得的成像范围。此外，该变换校正了井中的静校正量，并且不需要知道井中震源激发时间，震源以及检波器的位置信息。文中利用VSP-SSP相关变换将正演模拟的VSP数据转换成虚拟SSP数据，并对重构的地表地震数据进行常规处理，获得地下地质构造的成像。从结果中看出，和标准VSP数据的成像范围相比，地下的成像范围被大大扩展了。通过数值模拟测试了观测系统对VSP-SSP相关变换方法所得结果的影响。当采用井下不同分布范围上局部检波器所获得的响应时，应用VSP-SSP相关变换方法重构的SSP道集受到不同程度的影响，部分反射同相轴不能被正确的重构出来，道集的信噪比也降低了。但是采用稀疏的观测系统时，VSP-SSP相关变换方法依然能够重构出质量较好的SSP道集。当需要对采集自同一地区的VSP和SSP数据进行整合时，可以利用VSP-SSP褶积变换将VSP数据转换成虚拟SSP数据。形成的虚拟SSP数据可以和实际SSP数据进行对比分析，用于确定地下反射体的岩性。而且在该变换过程中也不需要知道速度模型和VSP测量中检波器的位置信息。我们利用VSP-SSP褶积变换将正演模拟的VSP数据转换成虚拟SSP数据，并对重构的地表地震数据进行常规处理，获得地下地质构造的成像。与通过相关算法从VSP数据中重构出来的虚拟SSP数据相比，褶积算法重构的虚拟SSP道集的信噪比更高，对地下较深处反射体的分辨率更好，但是成像范围却很有限。同样通过数值模拟测试了观测系统对VSP-SSP褶积变换方法所得结果的影响。采用仅分布于井下局部范围上检波器所获得的响应时，应用VSP-SSP褶积变换方法重构的SSP道集受到不同程度的影响，位于浅层的反射同相轴不能被重构出来，但依然保持了较高的信噪比。对于稀疏的观测系统，VSP-SSP褶积变换方法依然能够重构出信噪比较高的虚拟SSP道集。将VSP-SSP相关变换和VSP-SSP褶积变换相联合，能够将两种变换方法的优点结合到一起，对构建出的虚拟SSP道集进行常规成像处理，所得结果的成像范围和相关算法中的一样，大大扩展了地下地质构造的成像范围，而成像精度和褶积算法中的一样，对来自于深部的地质体也能分辨出来，且具有较高的信噪比。被动源地震数据主要采集自地下的微震，记录的背景噪声信号通常能量较弱，信噪比较低。直接对被动源数据进行成像处理，研究储层的地质构造有一定的技术难度。利用地震干涉法从背景噪声数据中提取出新的地震数据，能够把无序的噪声变成有用的信号，将重构的面波响应和反射波信息用于推断地下浅层和深部的地质构造特征。本文通过数值模拟和野外实际数据深入研究了地震干涉法在被动源数据上的应用。首先专门针对被动源地震数据的特点，设计了适合的正演模拟方案，并模拟生成了被动源地震数据。然后对采集自德国Ketzin地区的CO2冲注试验区记录时长将近25小时的背景噪声进行被动源地震干涉技术的应用研究。根据该工区内两条地震测线上所有可用检波器接收到的背景噪声记录，重构出共炮点道集。在重构的道集中可以观察到面波、折射波和反射波信息，与该地区主动源地震测量产生的共炮点道集相比较，验证了虚拟共炮点道集中重构的面波响应和反射波响应的有效性。通过产生共偏移距叠加道集提高虚拟共炮点道集的信噪比。利用重构的面波响应的分散特性估计出面波分散曲线，通过分散曲线的反演获得该地区地下的1D和2D剪切波速度剖面，获得的速度模型与根据该地区主动源数据的旅行时层析反演得到的速度模型有很好的一致性，但是从被动源数据中获得的速度模型对地表附近的浅层具有更高的分辨率。此外，与该地区其他主动源资料中获得的速度剖面相比，也有很好的匹配性。对重构的反射信息进行处理，产生的成像结果能够反映地下地质构造的特征，与对相同测线记录的主动源数据进行处理得到的成像结果有很好的一致性，为被动源地震勘探的实际应用提供了有力的技术支持。更多还原

【Abstract】 Seismic interferometry method and its application in the field data, greatly enrichedthe understanding and study of seismic wave propagation, it is considered to be a majoradvance in the development of geophysics. The scope of application of seismicinterferometry method is from the one-dimensinal layered acoustic medium to thethree-dimensional elastic medium, arbitrary inhomogeneous media or anisotropic media,and even can be extended to the moving media and attenuation media. The basic idea ofseismic interferometry method is retrieval of useful signal from the chaotic disorderseismic signal, new seismic signal is generated by the interference of recorded seismicsignal, the new seismic signal not only include the properties of original seismic signal, butalso retrieve some important features that the original signal does not have. Seismicinterferometry method can be used in both passive seismic measurements and controlledsource seismic measurements, the source can be certain source or noise source. Activesource seismic interferometry provide a new approach for seismic exploration, it can beused for a variety of data types. Take application of vertical seismic profile (VSP) data asrepresentative, VSP data can be transferred into the other data type by seismicinterferometry method, such as single seismic profile (SWP) data, surface seismic profile(SSP). Passive source seismic interferometry method applied to retrieval of useful signalfrom ambient seismic noise to obtain geological properties of medium, and then infer thecharacteristics of subsurface structure. In this paper, we respectively study and discussionthe application of seismic interferometry method for active source seismic data and passivesource seismic data.First of all, describe the basic principle of seismic interferometry methods, based onreciprocity theorems, underlying assumption of inhomogeneous, lossless medium, acousticand elastic Green’s function representations can be derived. The reconstructed Green’sfunction not only include the direct wave between two receivers, but also include primaryreflection and multiple scattering. Retrieved Green’s function by seismic interferometry isbased on the assumption that receivers are surrounded by a closed source surface. Undernormao circumstances, the source distribution conditions required by seismicinterferometry can not be met. So in this paper, we investigate that when the circumstancethat theory condition can not be met, the sources distributed in the subsurface, how the properties of source distribution effect on the retrieval of reflected waves. In the practicalapplication of seismic exploration, it always hard to meet the theoretically required for acontinuois source distribution, but during the calculation, in order to avoid aliasing, thesource interval should be set intensively. When the source only distributed in the localsubsurface area, only part of reflected event can be reconstructed in the retrieved gathers,the signal to noise ration is decreased seriously. In general, retrieved reflection informationfrom subsurface transmission response by seismic interferometry method is very effective.Process the retrieved reflection information and then used to migration process, theresulted image can be improve the incomplete reflected waves in the retrieved commonsource gathers and obtain satisfactory results.In the paper, we study on application of seismic interferometry method for activesource seismic data, here mainly target at VSP data. When the conventional seismic dataprocessing methods failed to acquire the satisfactory imaging results of seismic datainclude overburden complexity, we could apply the virtual source method to convert thetraditional vertical seismic profile gathers into virtual single well profile gathers. After that,redatumed geometry of source-receiver arrays is closer to the structure target and belowthe complex overburden so that we can get the better image resolution of the target afterconventional process and avoid the effected by unknown velocities of complex overburden,provide an efficient way to handle this complex seismic data. In the paper, the virtualsource method applied to detect steep structure such as faults present in the subsurface.Though the synthetic data found that the steeper structure the better image results, andautomatically correct the statics. The virtual source method is still work well for elasticdata sets. Moreover, modeling experiment show that virtual source method is still effectivefor the sparse geometry, we could get the good imaging results. Although the signal tonoise ratio decreased as source interval and receiver interval increased, the image result iscorrect.Although the imaging accuracy of the standard VSP data is high, the imaging range isvery limited that is defined by a small triangle, with the triangle apex at the shallowestreceiver. In order expand the imaging range of subsurface structure, one can transformVSP data into virtual SSP data by correlation algorithm of seismic interferometry, and thenthe imaging process used to virtual SSP data. The imaging range of subsurface structurefrom the results has same area as that of the original VSP source distribution along thesurface. In addition, this transform eliminates well statics and the need to know the sourceor receiver position in the well and excitation time of source in the well. In this paper,VSP-SSP correlation transformation effectively transform synthetic VSP data into virtual SSP data, and then migrated by a standard surface seismic data processing to give ainterferometric image of subsurface structure. From the results, compared with standardVSP image, VSP-SSP correlation transformation is a huge increase in the subsurfaceillumination. Test on the effect of geometry on the VSP-SSP correlation transformation bynumerical simulation. When the receivers in the well distributed on local area,reconstructed SSP gathers by VSP-SSP correlation transformation with various degrees ofeffect, part of reflected event can not be reconstructed correctly, the signal to noise rationof gathers decreased. When the VSP-SSP correlation transformation applied to sparsegeometry, the reconstructed SSP gathers still has good quality. When integrate VSP andSSP data that recorded over the same area, one can transform VSP data into virtual SSPdata by VSP-SSP convolution transformation. Resulted virtual SSP data can be comparedwith actual SSP data, and used to determine the lithology of subsurface reflector body.During the transformation process, it is no need to know the location information ofreceivers in the well. We transform VSP data into virtual SSP data by VSP-SSPconvolution transformation, and then the conventional seismic processing used to virtualsurface gathers to obtain the image of subsurface structure. Compared with virtual SSPdata generated by correlation transformation, the signal to noise ratio of virtual SSP datagenerated by convolution transformation is higher, the resolution of deeper reflector isbetter, but the imaging range is limited. Test on the effect of sparse geometry on theVSP-SSP convolution transformation by numerical simulation. When the receivers in thewell distributed on local area, reconstructed SSP gathers by VSP-SSP convolutiontransformation with various degrees of effect, reflected event of shallow reflector can notbe reconstructed correctly, but the signal to noise ration of gathers is still high. When theVSP-SSP correlation transformation applied to sparse geometry, the reconstructed SSPgathers still has good quality. Combine VSP-SSP correlation transformation and VSP-SSPconvolution transformation, the advantages from two transformations combined together.Process the virtual SSP gathers by conventional image processing, the resulted imagingrange is same with it from correlation transformation and result in a huge increase in thesubsurface illumination, at the same time, the imaging accuracy is same with it fromconvolution transformation and deeper subsurface structure can be detected.Passive source seismic data collected from the underground microseismic, generally,recorded ambient noise signal has relatively weak energy and low signal to noise ratio. It ishard to directly process the passive seismic data and use it to study of geological structurein the reservoir. Retrieved new seismic data set from ambient noise by seismicinterferometry, the noise into useful signal. The retrieved surface waves and reflected waves can be used to infer geological properties in the shallow subsurface structure anddeeper subsurface sreucture. This paper study application of seismic interferometrymethod for passive source seismic data by synthetic data and ambient seismic noise datasets. First of all, design forward modeling scheme suit for passive source seismic data,and then generate synthetic passive seismic data. Then we apply the seismic interferometrytechnique to approximately25hours of recordings of ambient seismic noise at the Ketzinexperimental CO2sequestration site, Germany. Common source gathers were generatedfrom the ambient noise for all available receivers along two seismic lines bycross-correlation of noise records. The retrieved response includes surface waves, refractedwaves and reflected waves. Comparison with surface waves and reflected waves fromactive source seismic survey, confirming the validity of the passive data processing.Stacking of common offset gathers increases the signal to noise ratio of retrieved commonsource gathers. A surface wave dispersion curve could be estimated from the retrievedsource gathers. Inversion of the dispersion curve results in an average1D and2D S-waveprofile below the site that is consistent with velocity models from traveltime tomographyof active source data from the area, but has higher resolution for the shallow subsurface. Inaddition, compared with velocity profile from other previous active source data, velocityprofile from surface wave dispersion curves also show similarities. Data processing of theretrieved common source gathers results in a stacked section that is in reasonableagreement with that of an active source section from same survey lines. The stackedsection from passive seismic survey have the potential to provide information on thedeeper subsurface structure， provide a effective technical support for the practicalapplication of passive source seismic exploration.更多还原