【作者】 黄力军；

【导师】 赵宪生；

【作者基本信息】 成都理工大学 ， 应用地球物理， 2009， 硕士

【摘要】 地震子波的提取是地震反演中非常重要的一个方面。准确的地震子波估计对于高分辨率、高信噪比、高保真度的油气勘探数据处理有极为重要的意义。本文详细介绍了同态反滤波、广义线性反演、展开相位谱法的基本原理,并从褶积模型出发,建立地震记录模型,分别采用以上方法对地震子波的提取进行了一些理论和实际资料处理研究。同态反滤波法建立在地层反射系数序列是变化的,相邻道子波基本不变的基础上的,在剖面上选取时窗时,选择在信噪比和分辨率比较高和同相轴较一致、连续性较好的目的层附近选取一叠加段,然后再将时窗强行随机化,即对所选的同一叠加段中的各时窗起始时间点,人为加上一个随机数,使各时窗的起始点服从随机分布,相应时窗内所包含的反射系数序列也随之重新分布。这样就满足了反射系数序列为变化的,地震子波保持基本不变的条件。最后将这些时窗的对数谱进行叠加,使子波这个不变部分的对数谱能量增加,而反射系数序列这个变化部分的能量极大地减弱,从而达到提取子波的目的。为了提高所提取子波的精度,本文还采用了对数谱平均、子波的拾取与整形等处理方法,取得了好的效果。广义线性反演是用CDP叠加地震道反演求取地震子波的一种技术,其最大特点是误差不随深度而积累。本文实验了初值误差对反演结果的影响,在保证反射系数符号正确的前提下,即使对与真实模型偏离较大的初始模型,广义线性反演方法也能够使它收敛于真实模型。展开相位谱法是首先计算出子波振幅谱和相位谱,再用所求得的子波振幅谱和相位谱,按欧拉公式转换成复数,通过反付氏变换就得到了子波序列。它是在无任何相位特性假设的条件下,从多道原始地震记录中统计提取出来的。本文从不同角度研究以上方法的可行性及其效率问题,仿真和真实数据处理结果表明,同态反滤波、广义线性反演、展开相位谱法在不同分辨率、不同信噪比、不同相位子波合成的地震记录反演中效果明显。通过上述方法的应用,子波提取准确度和效率得到了显著提高,同时也预示着这些方法在地震信号处理领域有广阔的应用潜力。另外,本文还以川东南地区己有资料为基础,对物理模型采集的地震记录和实际野外采集的地震记录进行了子波提取实验,并利用提取的地震子波对地震资料进行了层位标定、反褶积、波阻抗反演及沉积相反演等处理。实验结果证明了本文介绍的子波反演方法的正确性和有效性。更多还原

【Abstract】 seismic wavelet extraction is one of the very important part of seismic inversion. Acurate estimation of seismic wavelet Is very important to hige-resolution, hige signal-noise ratio and high fidelity petroleum exploration and production data procession. This paper datailed Introductes the Basic Principle of homomorphic inverse filtering, generalized linear inversion and spreading phase spectrum method, and establed seismic record model based on convolution model, then makes theory and real data proceding research on seismic wavelet extraction throw the above methods. The homomorphic inverse filtering method is based on that the coefficient series is varied and the seismic neighbouing traces is on the whole variant. When selecting time-window on profile,select a stack segment over the high signal-noise ration and resolution, Alignment rather accord, and succession quite well target zone,then make the time-window compel Randomized, it means adding a random number on the every time start point of time-window of the selected same one stack segment, to make evey start point accorded random distribution, so reflection coefficient series on the evey time-window will redistribute accorded it. so it meet that the reflection proficient series is varied and the seismic wavelet keep invariant. The last,overlap the log-spectral of the time-window, to improv the energe of invaried log-spectral of wavelet, but the energe of varied reflection coefficient series was fully reduced, thereby come to the aim of wavelet extraction. To improve the accuration of wavelet extraction, this paper take the method of log-spectral averagement, wavelet pick-up and reform .generalized linear inversion is a technology of picking up seismic wavelet using CDP overlap seismic trace incersion, its most feature is the error can’t accumulate according the deepth. The paper test the effect of initial value error to the result inversion: under the premise of keeping the reflection coefficient sign correct, even to the initial model which deviate far from the real model, the method of generalized linear inversion can make it convergent to the real model.spreading phase spectrum method is to compute the wavelet amplitude spectrum and phase spectrum, and then use the wavelet amplitude spectrum and phase spectrum to transform to complex with Euler Formula, get the wavelet series through inverse Fourier Transform . It is statistically extracted from multi-trace raw seismic record without any phase character suppose.This paper research the feasibility and efficiency of the above methods from different apart. The result of treat to fidelity and real data study shows that homomorphic inverse filtering, generalized linear inversion and spreading phase spectrum method have obvious effect in the seismic wavelet records synthesed in different resolution, signal-noise ratio, phase wawelet inversion.Through the application of the above methods, the precision and efficiency of the seismic wavelet extraction gain obvious enhancement. And the same time, the achievements indicate the broad application future of these methods in the field of the seismic sign processing.Besides, based on the owened data about Southeast of the Sichuan, this paper make wavelet extraction test on the seimesic record gathered from physical model and virtually gathered from field. And make horizon calibration, deconvolution, Impedance Inversion and sedimentary facies inversion proceding on the seismic records using the extracted seismic wavelet.the results has demonstrated the corrcetion and the efficiecnt of the wavelet inversion methods introducted in the paper.更多还原