【作者】 尚帅；

【导师】 韩立国；

【作者基本信息】 吉林大学 ， 地球探测与信息技术， 2014， 博士

【摘要】 地下介质的复杂性造成地震波在地下的传播是非稳定的，谱分解作为一种能够同时在时频域刻画地震信号，研究地震信号瞬时频率特征的技术被广泛应用。随着油气勘探程度的日益增大，地震勘探的目标由构造油气藏转为岩性油气藏，埋藏深、储层薄、物性差、非均质强等特性使得地震勘探的难度越来越大。针对这种情况，常规的谱分解技术由于其时频分辨率较低，已经不能满足高分辨率地震数据处理、解释的需求。本文在回顾常规谱分解方法的基础上，实现了两种高分辨率地震数据谱分解技术—基于能量谱重排的小波变换(Reassigned wavelet transform, RWT)和压缩小波变换(Synchrosqueezing wavelet transform, SWT)谱分解，这两种方法都是建立在小波变换的基础上，相对于传统的小波变换，其时频分辨率都得到了提高，把这两种方法引入地震数据处理和解释领域，取得了较好的效果。第一种高分辨率谱分解方法是基于能量谱重排的小波变换，该方法主要是针对小波变换后的时频能量，对其按照重新计算的位置（其局部重心）进行重排，从而达到提高时频分辨率的目的。文章详细阐述了这种方法的实现过程，并通过合成算例证实其相对于传统小波变换具有较高的时频分辨能力。利用该方法对叠后地震数据进行分频解释，体现出该方法的优势。从能谱重排小波变换的数学公式推导可以看到，虽然这种方法通过对小波变换结果进行一定的后续处理提高了其时频分辨能力，但它在数学上是不可逆的，这也就意味着如果想从这种方法得到的时频分析结果重构地震信号在数学上是不可实现的，这使得基于能量谱重排的小波变换在地震数据高分辨率处理方面的应用受限。第二种高分辨率谱分解方法是压缩小波变换方法，不同于能谱重排小波变换，该方法只沿频率方向对小波变换复数谱进行重排，在提高时频分辨率的同时保持了其数学上的可逆性，这就为其在地震数据处理和解释中的应用提供了广阔的天地。文章在详细阐述该方法原理的基础上，提出了基于该方法的地震随机噪声和面波压制技术，合成算例和实际数据证明了这种方法在地震去噪中的有效性。与此同时，考虑地震噪声的影响，基于地震数据结构的自相似性，提出了自适应非局部均值地震噪声压制方法，大大改善了地震数据去噪效果。针对复杂地区，由于地震波的传播和地下介质的吸收作用，造成地震资料分辨率降低，不能满足地震地质解释的精度。本文提出的基于压缩小波变换的提高分辨率技术，可以对深浅层地震剖面能量进行补偿，拓宽了地震资料的有效频带范围，提高了地震资料的分辨率，同时不破坏地震剖面同相轴的能量相对关系，有利于后续地质解释与岩性反演等工作，采用这种方法对实际地震资料进行处理，验证了方法的有效性。地震波在地下传播过程中，除了球面扩散、大地的滤波作用等造成地震波能量的衰减，当在含油气储层中传播时，也会造成地震波能量的强衰减和地震波频率成分的改变，而这种衰减现象往往伴随着地震波的频散，这一点已经被广大科研工作者证实。本文结合谱分解方法和AVO方程，提出一种新的频散属性，该属性与纵波速度频散属性一样可以在一定程度上刻画储层的含流体情况，可以作为一种储层烃类预测的手段。更多还原

【Abstract】 The propagation of seismic waves beneath the surface is non-stable because thecomplexity of subsurface. Spectral decomposition, which can describe the frequencycharacteristics varying with time of seismic signal, is widely used in geophysics. Withthe increase of oil/gas exploration, the exploration targets are shifting into litologicreservoirs, making it more and more difficult because of the complexity of reservoirs.For this reason, conventional spectral decomposition technology is unable to meet theseismic data processing and interpretation needs, and the development of new spectralanalysis technique with high resolution is imminent.We first review the previous research on spectral decomposition, and thenintroduce two new techniques for high resolution spectral decomposition of seismicdata, including reassigned wavelet transform (RWT) and synchrosqueezing wavelettransform (SWT), which both are improved versions of original wavelet transform,making it helpful to seismic data processing and interpretation.One kind of the new seismic spectral decomposition methods is reassignedwavelet transform based on energy reassigment, which assigning its energy to a newposition (its local gravity center). The paper shows how the reassigned wavelettransform works and we compare its resolution with wavelet transform by syntheticdata. We conduct this method on post-stack seismic profile and reveal hydrocarbonreservoir using low-frequency abnormity. Although the readability of time-frequencydistribution can be improved after energy reassignment, it is not invertible and thereconstruction of signal is difficult, which make its application in high-resolutionseismic processing limited.Another high-resolution spectral decomposition method is synchrosqueezingwavelet transform, which is different from the reassigned wavelet transform, focusingthe time-frequency representation by squeezing its value along the frequency axis andoffers an exact reconstruction formula for signals. So it can be widely used in seismicdata processing and interpretation. The theory of synchrosqueezing wavelet transformis stated in this paper, and then we proposed two new denoising algorithms based on SWT, synthetic and field examples show these methods work well. And so on,considering the impact of seismic noise, we also propose an adaptive nonlocal meansseismic denoising method, which is based on the self-similarity of seismic datastructure.Since the absorption of underground layers and seismic wave attenuation, theresolution of seismic data can not meet the accuracy of geophysical and geologicalinterpretation. This paper proposed a seismic frequency width extension method basedon SWT without destorying the energy relationship between relative phase axis,making it conducive to subsequent geological interpretation and lithology inversionwork. Real seismic data show our method performs well.In addition to spherical spreading, filtering effect of earth, etc. It is known whenseismic signal penetrates through hydrocarbon reservoirs, its amplitude will showstrong attenuation and its high frequency components will be attenuated more thanlow-frequency components. This phenomenon is often accompanied by seismic wavedispersion, which has been confirmed by the majority of scientists. Combiningspectral decomposition with avo theory, a new dispersion attribute is proposed in thispaper, which can describe the presence of fluid in reservoirs partly like P-wavevelocity dispersion property, and can be used as a method for hydrocarbon detection.更多还原