【作者】 王冬梅；

【导师】 李月；

【作者基本信息】 吉林大学 ， 通信与信息系统， 2016， 博士

【摘要】 陆地地震勘探是资源勘探的重要手段。面对复杂的地表和近地表地质条件、越来越深以及空间越来越复杂的勘探目标,给地震勘探提出更高的要求。在陆地地震勘探的整个过程(采集、处理、解释)中,地震资料处理是至关重要一环,特别是处理过程中随机噪声的压制效果直接影响着后续一系列处理,包括速度分析,偏移成像以及反演等。然而,野外采集地震数据时,检波器接收到的地震记录中含有大量随机噪声信息,特别是勘探目标比较复杂时,信号很弱而噪声很强,记录整体信噪比较低,如何在低信噪比条件下有效压制随机噪声对地震勘探而言就显得尤为重要。由于信噪比较低,信号能量较弱,为了能有效的压制随机噪声就需要对地震勘探随机噪声的时空特性有较系统的认识和研究,从而指导随机噪声压制方法选择、改进以及压制策略。目前,对于随机噪声特性还未进行过科学系统的研究,这直接影响地震记录信噪比的有效提高以及随机噪声消减技术的发展。有效的改进随机噪声消减技术是建立在对地震勘探随机噪声特性准确认识的基础上。因此,关于地震勘探随机噪声特性的研究具有重要的实际意义。本文主要针对陆地地震勘探随机噪声,对其时域、空间域以及时空域特性进行详细系统的研究。主要对随机噪声的高斯性、平稳性、线性特性、功率谱特性和相似性进行综合分析。对比不同采集环境(草原、沙漠、山地、林带)下,随机噪声时空分布特性的差异,分析采集环境与随机噪声时空特性的关系。研究中,分析采集环境复杂程度和噪声时长等因素对于随机噪声特性的影响。最后,结合随机噪声生成机制和采集环境的特点,对研究结果的合理性进行综合论证。具体研究成果如下:将高阶统计量同Shapiro-Wilk检验法相结合来判断地震勘探随机噪声的高斯性。首先对地震勘探随机噪声的峰度值和偏度值高阶统计量的特性进行分析,发现随机噪声在高阶统计量分布方面同高斯随机过程之间存在明显差异。在此基础上,应用Shapiro-Wilk检验法对随机噪声的高斯性进行准确的判断。结果表明,随机噪声的高斯性受到采集环境特点的影响。采集环境越简单,随机噪声的非高斯性越明显;采集环境越复杂,随机噪声的非高斯性越弱。通过对不同地区随机噪声记录的高斯性检验结果进行对比分析,发现非高斯随机噪声记录占比普遍很高。因此,陆地地震勘探随机噪声应该被认为是非高斯随机过程。本文首次将基于时频分析和替代数据的平稳性检验方法引入到陆地地震勘探随机噪声的平稳性判断中,并对不同时长随机噪声的平稳性进行判断。检验方法的基本思想是通过比较随机噪声序列同替代数据在时频特性上的差异,从而对随机噪声序列的平稳性进行判断分析。通过对不同地区随机噪声的平稳性进行统计分析,研究结果表明,陆地地震勘探随机噪声不是平稳随机过程,其特性随着随机噪声时长和采集环境特点发生变化。复杂环境中的随机噪声非平稳性较强,而简单环境中的随机噪声平稳性较好。同时,对短时长和长时长随机噪声的平稳性进行对比分析,结果显示,短时长随机噪声的平稳性要好于长时长随机噪声的平稳性。后续实验结果与陆地地震勘探随机噪声平稳性结论相一致。选用Thomson谱估计方法对陆地地震勘探随机噪声的功率谱特性进行研究。传统上,通常认为陆地地震勘探随机噪声在时空域是满足白噪特性的。研究结果表明,陆地地震勘探从功率谱特性上看并不是白噪过程。随机噪声的功率谱特性随着采集环境特点的不同而发生变化,对于沙漠地区、草原地区等简单的地理环境而言,陆地地震勘探随机噪声的能量主要集中在低频处,可以认为是低频、窄带色噪声;对于林带地区、山地地区等复杂的地理环境而言,随机噪声中高频成分相对丰富,频带较宽,可以近似认为是宽频色噪声。同时,对陆地地震勘探随机噪声的时空相似性进行科学分析。研究结果表明,陆地地震勘探随机噪声的相似性主要受采集环境特点的影响。对于简单采集环境而言,随机噪声表现出一定的相似性;而对于复杂采集环境而言,随机噪声的时空相似性较差。在此基础上,本文结合采集环境特点和随机噪声平稳性研究结论,对随机噪声相似性统计结果的合理性进行必要的论证。采用延时矢量方差法和量化延时矢量方差方法,对陆地地震勘探随机噪声的线性特性进行研究,并对不同地区随机噪声的线性特性进行综合判断。研究结果表明,随机噪声不能简单的归为线性或非线性随机过程。随机噪声的线性特性同采集环境的特点密切相关。当采集环境相对简单时,随机噪声的线性特性较好,可以近似认为是线性随机过程;当采集环境复杂时,随机噪声中有非线性成分存在,简单的线性模型已经无法表征出随机噪声的特性,这时随机噪声应该被认为是非线性随机过程的产物。本文对陆地地震勘探随机噪声的高斯性、平稳性、线性、功率谱特性和相似性进行科学系统的研究。在研究中,对不同典型地理环境下采集到的随机噪声的特性进行判断,同时对不同地区随机噪声在特性方面的差异进行对比分析,得到关于陆地地震勘探随机噪声特性的总体认识。研究结果表明,传统上对于随机噪声特性的认识有一定的片面性,存在一定的局限性。随机噪声的特性是同采集环境密切相关的,不同时长随机噪声在特性方面也存在一定的差异。在此基础上,通过一系列的理论分析和正演实验,论证本文研究结果的正确性和可靠性。综上,本文的研究成果是对陆地地震勘探随机噪声特性的一种新认识,是对传统理论的一种有益的补充,为改进随机噪声消减技术和随机噪声建模领域的研究提供理论依据,具有较好的实际意义和应用前景。更多还原

【Abstract】 Land-seismic-exploration is one of the most important methods for the mineral prospecting. Due to the complexity of surface and near-surface conditions, we need to improve the seismic-exploration technology to detect the resources in deep and irregularlayers. In the procedures of seismic-exploration(acquisition, processing and interpretation), seismic data processing is very important. Especially, suppression effect of random noise impacts on the subsequent series of processes, including velocity analysis,migration imaging and inversion. However, when the seismic data is collected in the field,the detectors receive the seismic record contains a lot of random noise information,especially when the exploration target is more complex, the signal is very weak and the noise is strong, the overall record has low signal noise ratio. How to suppress random noise under the condition of low SNR is particularly important for seismic exploration.Due to the low signal noise ratio, the signal energy is weak. The systematic research on spatiotemporal characteristics of the land-seismic-exploration random noise can effectively suppress random noise to guide the random noise suppression method’s selection, improvement and suppression strategies. Until now, the characteristics of the random noise have not been deeply studied. It has been the obstacles for the increasing signal noise ratio of seismic records and the the improving of the seismic-exploration random noise attenuation methods. For improving the noise attenuation techniques, the proper understandings about the properties of the seismic-exploration random noise are acquired. From this viewpoint, the research about the characteristics of the seismic random noise has a great significance.In this dissertation, we mainly focus on the time domain, space domain and spatiotemporal characteristics of the seismic-exploration random noise. The Gaussianity,stationarity, linearity, power spectral density and similitude of the seismic-exploration random noise is investigated. We compare random noise spatiotemporal distribution characteristics in different acquisition environment(grassland, desert, hilly areas, forest belt), and analyze the relationship between the environment and the spatiotemporal characteristics of random noise. On this basis, the influence of the complexity of the environments and the noise duration on the spatiotemporal characteristics of the random noise is analyzed. Furthermore, we explain the accurateness of our findings by taking the noise generating mechanism and environment features into consideration.The main findings of this dissertation are summarized as follows:By applying the Shapiro-Wilk testing methods and the analyzing of the high-order moments, the Gaussianity of the seismic-exploration random noise is studied. In this study,the kurtosis and the skewness of the seismic-exploration random noise is investigated. Theresults show that the seismic-exploration random noise has a great difference from the Gaussian process. On this basis, we apply the Shapiro-Wilk testing methods to scientifically study the Gaussianity of the random noise. The results indicate that the Gaussianity of the random noise has a great relationship with the properties of the collected environments. The seismic-exploration random noise in a simple environment is always superior to the noise in a complex environment in terms of the Gaussianity. By analyzing the Gaussianity testing results, we obtain that the portions of the non-Gaussian noise data are very huge. It means that the seismic-exploration random noise is a non-Gaussian process.Our dissertation is one of the first attempts to investigate the stationarity of the random noise by applying the stationarity testing methods based on time-frequency(TF)analysis and surrogates. At the same time, we test the stationarity of the random noise with different duration. The basic idea of the testing methods is to assess the stationarity of the testing series by compare its TF features with those of its surrogates. We analyze the properties of the random noise in different environments. The results of our study suggest that the seismic-exploration random noise is not stationary. The properties of the noise change with the time-length of the noise and complexity of the acquired environments.The results indicate the stationarity of the noise in a simple environment is superior to that of the noise in a complex environment. Besides, we also compare the stationarity of the noise in different time-length. Based on this, it is shown that short noise is always more stationary than the long one. We also use some experiments to prove the correctness of our findings.In this dissertation, the Thomson’s PSD estimation method is used to analyze the power spectral density(PSD) of the seismic-exploration random noise. In this dissertation,we use Thomson’s PSD estimation method to investigate the spectral properties of the random noise. The results indicate that the noise is not a white noise process. The spectral of the noise is related to the features of acquired environments. The energy of the noise in a simple environment, e.g. desert and grassland, is concentrated in low frequencies. Thus,it can be seen as low-frequency-narrow-band color noise. However, the noise in a complex environment, e.g. forest belt and hilly areas, has more energy in high frequencies. In contrast, the noise in those environments is wide-band. It can be viewed as wide-band color noise. In addition, we also investigate the similitude of the seismic-exploration random noise. Our findings indicate the similitude of the noise changes with the properties of the acquired environment conditions. There is similitude existed in the random noise records from a simple environment. However, it is not obvious for the noise in a complex environment. On this basis, we combine the features of the acquired environment and the results of the stationarity analysis to give a reasonable explain to the findings in similitude analysis.In this dissertation, the Delay Vector Variance(DVV) and Quantified Delay VectorVariance(QDV) methods is used to investigate the linearity of the seismic-exploration random noise. We analyze the linearity of the noise data collected in the different environments. The results show that the seismic-exploration random noise cannot be simply classified as a linear process or a non-linear process. The linearity of the seismic-exploration random noise has a strong relation with the collected environment conditions. For a simple environment, the linearity of the random noise is great, and it can be seen as a linear process. However, for a complex environment, there are some non-linear components existing in the random noise. From this view, the seismic-exploration random noise in a complex environment should be considered as the production of a non-linear process.In this study, we scientifically investigate the Gaussianity, stationarity, linearity,power spectral density and similitude of the seismic-exploration random noise. We check the characteristics of the random noise collected in the different geology environments,and compare the corresponding results of the different noise data. On this basis, we obtain the general understandings about the spatiotemporal characteristics of the seismic-exploration random noise. The results indicate that the traditional understandings about the seismic-exploration random noise characteristics have their own limitations. The characteristics of the seismic-exploration random noise have strong relationship with the collected environment conditions. Moreover, the noise durations also have influence on the properties of the random noise. On this basis, we prove the correctness of our findings by the theoretical analysis and the proper experiments. In general, the findings of our dissertation provide a new understanding about the seismic-exploration random noise characteristics. The findings of our dissertation can also be used as the theoretical foundations for improving the denoising techniques and modeling research for the seismic-exploration random noise. In conclusion, our study has the great practical significance and application potentials.更多还原