陆地地震勘探环境噪声特性研究及检验

发布时间：2018-05-07 09:41

本文选题：地震勘探 + 环境噪声　；参考：《吉林大学》2015年硕士论文

【摘要】：陆地地震勘探是煤田和石油勘探中一种重要勘探方法。但是在野外采集过程中，地震资料不可避免的会混入各种噪声，使有效地震信号淹没在噪声中，令资料处理难度加大，也对后续数据的解释分析造成困扰，直接影响了勘探结果的准确性。因此，前人对于地震勘探的研究一直都集中在噪声的压制上。在地震勘探噪声消减领域，对于噪声的性质一直采用传统认识：线性平稳高斯。这是一种理想化的噪声性质，方便于为选择地震随机噪声压制方法提供理论依据。然而，在某些情况下基于此思想的去噪方法对实际噪声的压制效果并不理想。这就需要对实际噪声的特性有一个重新的认识，向这种理想化的传统认识提出质疑与挑战。环境噪声作为陆地地震勘探随机噪声中重要组成成分，目前还没有被科学系统的研究过。本文利用时频峰值滤波中对噪声高斯性的无偏估计理论，分别就混入高斯白噪和实际纯环境噪声记录的模拟地震数据进行去噪，通过去噪后的信噪比比较，认为实际噪声具有一定的非高斯特性。同样利用维纳滤波原理对于平稳过程的理论要求，进行了高斯白噪和实际纯环境噪声记录的去噪仿真实验，两者信噪比差异较大，同样认为噪声的非平稳特性存在，确定了噪声平稳性与高斯性检验的必要性。对四个测区：林带，沙漠，山地，黄土塬进行了平稳性和高斯性的检验。其中平稳检验方法采用基于Fokker-Planck方程的非参数的平稳性检验方法，由于考虑到目前滤波算法普遍采用时频域滤波方法或者针对短时窗内局部数据滤波，截取数据点数不多，因此对四个测区的短时噪声平稳性进行检验比较及差异分析。结果表明，各测区的噪声均具有非平稳性质，且各个测区间非平稳情况不同。为此进一步分析噪声局部平稳性情况，实验结果认为噪声是可以进行局部分段平稳研究的。应用四种常用的正态检验方法：SW检验，KS检验，，Lilliefors检验，JB检验进行蒙特卡罗实验，最后选取检验功效最优的SW检验方法作为实际噪声高斯性检验方法。对不同测区不同长度的高斯性进行比较。检验结果表明，非高斯情况随着沙漠，黄土塬，山地，林带依次减弱，但是当数据长度足够长时，四个测区的实际地震数据都显现出极大非高斯特性。本文的主要研究方向是对陆地地震勘探环境噪声的特性进行检验，其中特性主要包括平稳性与高斯性。通过系统研究不同地区不同长度纯环境噪声纪录，对陆地地震勘探环境噪声特性研究进行了补充，为今后噪声建模、依据噪声性质去噪提供理论依据。
[Abstract]:Land seismic exploration is an important exploration method in coalfield and petroleum exploration. However, in the process of field acquisition, the seismic data will inevitably be mixed with various noises, which will make the effective seismic signal submerged in the noise, make the data processing more difficult, and also cause trouble to the interpretation and analysis of the subsequent data. It directly affects the accuracy of exploration results. Therefore, previous researches on seismic exploration have been focused on noise suppression. In the field of noise abatement in seismic exploration, the character of noise has been traditionally recognized: linear stationary Gao Si. This is an idealized noise property, which is convenient to provide a theoretical basis for the selection of seismic random noise suppression methods. However, in some cases, the denoising method based on this idea is not effective in suppressing actual noise. Therefore, it is necessary to have a new understanding of the characteristics of actual noise and challenge this idealized traditional understanding. As an important component of random noise in terrestrial seismic exploration, environmental noise has not been studied by scientific system. In this paper, the unbiased estimation theory of noise Gao Si in time-frequency peak filtering is used to Denoise the simulated seismic data mixed with the actual pure ambient noise, and the signal-to-noise ratio (SNR) after denoising is compared. It is considered that the actual noise has some non-Gao Si characteristics. At the same time, according to the theoretical requirements of Wiener filtering principle for stationary process, the simulation experiments of Gao Si white noise and real pure ambient noise record are carried out. The signal-to-noise ratio between them is quite different, and the non-stationary characteristic of noise is also considered to exist. The necessity of noise smoothness and Gao Si test is determined. The stability and Gao Si of four survey areas: forest belt, desert, mountain and loess tableland were tested. Among them, the stationary test method is based on the Fokker-Planck equation. Considering that the current filtering algorithms generally use time-frequency filtering method or local data filtering in the short-time window, the number of truncated data points is not large. Therefore, the short-time noise stationarity of the four measuring areas is tested and compared and the differences are analyzed. The results show that the noise of each measuring area is nonstationary and the non-stationary condition is different in each interval. The experimental results show that the noise can be studied by local piecewise stationary analysis. In this paper, four normal test methods are used: KS test and Lilliefors test / JB test to carry out Monte Carlo experiment. Finally, the optimal SW test method is selected as the actual noise Gao Si test method. Gao Si of different length in different measuring area was compared. The results show that the non- situation weakens in turn with the desert, loess plateau, mountain and forest belt, but when the length of the data is long enough, the actual seismic data of the four measured areas show a great non- characteristic. The main research direction of this paper is to test the characteristics of environmental noise in terrestrial seismic exploration, and the characteristics mainly include stationarity and Gao Si. By systematically studying the pure environmental noise records of different regions and different lengths, this paper supplements the research on environmental noise characteristics of land seismic exploration, and provides theoretical basis for noise modeling and denoising based on noise properties in the future.
【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】：P631.4

【参考文献】