地震数据解释中小波去噪方法研究

发布时间：2018-07-05 11:10

本文选题：小波滤波 + 相干体算法　；参考：《吉林大学》2015年硕士论文

【摘要】：在大规模工业化的今天，社会对石油的需求空前强烈，对石油资源的勘探开发已经被提到了一个空前的重要地位。随着勘探工作的不断深入，，各种复杂的地质情况都会遇到，这就对勘探技术提出了更高的要求。一种比较节约的勘探方法就是采用人工地震的方式获取地层反射的地震数据，通过对地震数据的解释来勘探地下的地质结构。地层中的断层结构对于石油勘探具有重大意义，通过对断层的分析，地质勘探人员可以得出地下的油藏情况。相干体技术是当前对地震数据的断层解释的有效方法。它通过计算相邻的若干地震道的相似程度，来检测地震波同相轴的不连续性。这些计算得到的相似性的数值，按照他们的空间、时间排列，组合成了相干体数据。相干体消除了振幅波动给断层识别带来的不便，突出了断层给相邻道数据带来的不连续特征。对于识别断层具有极大的帮助。针对地震数据的特点，和相干体技术的优点，本文从改善相干体源数据的角度入手，进行了相关的研究工作。研究地震数据的特点，发现地震数据是一种含有正负值，并且均值为零的时间振幅波。这些地震波在相邻数据道的相似程度，决定了相干体值的大小。将地震道上的数据的相邻点做差，就可以得到一个同原数据波的相位近似相差的新的数据波。对由这种数据波组成的数据体进行相干体计算，可以得到与原来的地震数据相干体类似，但可以显示更多地质结构细节的相干体。使用一维小波对地震数据进行小波分解，剔除高频部分，得到更加平滑的地震道数据波形。这样的地震数据，在相邻道的相似性上，得到了提高，从而提高了相干体的质量。之所以对分解的地震数据直接剔除高频部分，而不是像使用小波处理图像那样进行阈值滤波，就是考虑到地震数据是波动性的数据，在过零点附近的波动变化快，若果采用阈值滤波就会产生阶梯效果。还有就是直接剔除高频部分，计算速度更高。对地震数据进行三维小波滤波，增强数据道之间的横向相似性，消除地震道间的噪声干扰。研究三维的地震数据小波滤波的时候，使用了两种不同的滤波方法，发现这两种方法各有优缺点。一种是，分别在三个方向上分别使用小波滤波，并进行重构。得到三个方向的滤波结果，再求其平均值作为最终的滤波结果。另外一种方法就是，传统的可分离变量的三维小波滤波。在剔除三个方向小波变换的得到的高频部分之后，再进行重构，得到滤波后的数据。这两种滤波方法都对于在此基础上计算的相干体又增强的作用。第一种方法得到的地震道数据与原始地震数据变化不大，所以对地质体中的较小的结构保留较好，但同时也保留了相应数量的噪声。第二种方法很好地提高了相邻道的相似性，对于较大的断层结构的保留也非常好，很适合用于识别大断层。鉴于本文在试验的时候选取了不同的地震数据，应用了不同的方法。地震数据A，适合使用一维的小波滤波方法，而地震数据体B适合使用三维的小波滤波方法。本文并没有给出一个普适的方法，只是给出了一些可供选择地方法。
[Abstract]:In today's large-scale industrialization, the demand for oil is unprecedentedly strong. The exploration and development of petroleum resources has been mentioned as an unprecedented important position. With the deepening of exploration work, all kinds of complex geological conditions will be encountered, which puts forward higher requirements for exploration technology. A relatively economical exploration method. It is to obtain the seismic data reflected by the artificial earthquake, and to explore the geological structure of the underground by interpreting the seismic data. The fault structure in the stratum is of great significance to the oil exploration. By analyzing the fault, the geological Explorer can get the oil reservoir under the ground.
Coherent body technique is an effective method of fault interpretation of seismic data. By calculating the similarity of several adjacent seismic channels, the discontinuity of the phase axis of seismic waves is detected. These calculated similarity values, according to their space and time arrangement, the coherent body data are synthesized. The coherent body eliminates the amplitude wave. The inconvenience caused by dynamic fault recognition highlights the discontinuity characteristics brought by faults to adjacent road data, which is of great help in identifying faults.
In view of the characteristics of seismic data and the advantages of coherence cube technology, this paper has done some related research work from the perspective of improving coherent source data.
The characteristics of seismic data are studied. It is found that the seismic data is a time amplitude wave with positive and negative values and the mean value is zero. The magnitude of the coherent body value is determined by the similarity of these seismic waves in the adjacent data channels. The difference of the adjacent points on the seismic data can be obtained by the approximate phase difference between the same data wave.
A new data wave. The coherence body calculation of the data body composed of this data wave can be similar to the original seismic data coherent body, but can display more details of the geological structure.
We use one dimensional wavelet to decompose the seismic data, eliminate the high frequency part, and get a more smooth waveform of the seismic trace data. This seismic data is improved on the similarity of adjacent channels, thus improving the quality of the coherent body. The reason why the number of decomposed earthquakes is eliminated directly, rather than the use of the wavelet In the same way, the threshold filtering means that the seismic data is fluctuant, and the fluctuation near the zero crossing is fast. If the threshold filter is used, the step effect will be produced.
The seismic data is filtered by three dimensional wavelet, and the transverse similarity between the data channels is enhanced and the noise interference between the seismic channels is eliminated. When the wavelet filtering of 3D seismic data is studied, two different filtering methods are used and the advantages and disadvantages of the two methods are found. One is to use wavelet filtering respectively in three directions, respectively. The filtering results of three directions are obtained, and the average value of the filter is obtained as the final filtering result. The other method is the three-dimensional wavelet filtering of the traditional separable variables. After eliminating the high frequency part of the three direction wavelet transform, the reconfiguration and the filtered data are obtained. The two filtering methods are all correct. The first method has little change of the seismic data and the original seismic data, so the smaller structure in the geological body is well preserved, but the corresponding number of noise is retained at the same time. The second methods can improve the similarity of the adjacent channels well, for the larger fault junction. The structure is also well preserved and suitable for identifying large faults.
In view of the selection of different seismic data in this paper, different methods are applied. The seismic data A is suitable for using one dimensional wavelet filtering method, and the seismic data body B is suitable for the use of three-dimensional wavelet filtering. This paper does not give a universal method, only a few alternative methods are given.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：P631.44

【参考文献】