【作者】 李燕；

【导师】 钱荣毅； 徐凌；

【作者基本信息】 中国地质大学（北京） ， 地球探测与信息技术， 2009， 硕士

【摘要】 近地表结构复杂地区的地震资料的处理和解释是当前陆地地震勘探的一个主要任务,地震层析成像是一种地层重建和速度反演的地震数据处理方法,利用初至波层析技术构建近地表速度场可以从本质上消除近地表时间的畸变。本文主要做了如下工作:1、通过实验发现现有的正演算法不能适应复杂的介质速度分布模型以及较大的速度梯度变化;同时,在反演算法中,也存在着稳定性不高,对初始模型的依赖性较强等问题。通过比较,本次研究采用旅行时线性插值法(LTI, Linear Traveltime Interpolation) ,和迭代重建法(SIRT, Simultaneous Iterative Reconstruction Technique),实现了速度场的正演和反演。2、在反演过程中,发现最大炮检距的选取范围对反演效果影响很大,深层误差会影响表层速度的反演精度;同时,最大炮检距太小,会导致覆盖近地表地层的射线不足,也会影响反演精度。研究发现最大炮检距可以根据炮间距,在初始模型上正演时射线走的深度以及想要的反演深度来确定。3、为了验证层析反演算法的稳定性和收敛性,进行了理论模型实验;同时,引入了井深和随机误差。通过实验得出结论:①层析算法反演的速度模型趋势是正确的。②在测量和拾取准确的情况下,反演结果较好。③存在误差的情况下,模型收敛,效果略差。④模型反演中,井深比随机误差影响大。4、在结合LTI和SIRT算法实现层析静校正的过程中,首先,建立了三种初始模型,选取了西部地区西秋的一条二维测线的地震资料,通过比较,速度突变型速度模型反演效果最好;其次,找出影响反演效果的原因,对速度场进行了插值和平滑,插值方法是反距离平方法,平滑分别采用一维平滑和二维平滑,通过比较,插值和平滑后,速度场反演精度有明显提高。5、通过分析传统软件,它们通常采用等深和等高的方法提取等速面,这可能导致错误的静校正量,为了解决这一问题,改进了提取等速面的方法,使反演精度提高了。同时,为了消除等速面上的一些速度突变值,采用中值滤波和均值滤波的方法,平滑了等速面。6、从应用层析静校正剖面图来看,有改进的地方,也有变差的地方,造成此结果的原因有待进一步深入研究。本文通过速度模型和实际资料的反演结果表明,该方法能稳定、有效、快速地重建近地表的速度场分布,要很好地反演出此构造得依靠野外观测系统误差的减小,先验信息的利用,算法适应性高,参数的适当选择等。更多还原

【Abstract】 The seismic data processing and interpretation for the near-surface structure of complicated region is a major task of seismic exploration on land .Seismic tomography is a seismic data processing methods which reconstructs statum and inverse velocity. The use of the first break tomography to build near-surface velocity field can eliminate distortion of the time-distance curves. In this paper, following jobs have been done:1. The experiment shows that the existing ray tracing algorithm can not adapt to complicated velocity model as well as velocity model with large change in velocity gradient.At the same time, there are also some problems in the inversion algorithm ,such as low stability, the strong dependence of the initial model and so on. By comparison, this paper uses LTI(Linear Traveltime Interpolation), and SIRT(Simultaneous Iterative Reconstruction Technique) to realize the forward and inversion of the velocity field.2. In the inversion process,the selection of the largest offset has great effect on the inversion because the error of the deep statum will affect the accuracy of the near-surface velocity inversion. But if the largest offset is too small,it will cause the rays are not enough to cover near surface which will also affect the accuracy of inversion.Largest offset can be determined by the depth of the ray traveling in initial model and the required invesion depth3. In order to verify the stability and convergence of the tomographic inversion algorithm,theoretical model experiment is done,in which the depth error and randon error are introduced. The conclusions of the experiment are following:①tomography algorithm inversion trend is correct.②pick up in the accurate measurement and Inversion results is better with accurate measurement and pick-ups.③Error can lead to worse convergence of the velocity model.④In model inversion, the influence of the overall deviations is bigger than that of random error.4. In the realization of LTI and SIRT algorithm, first of all,three kinds of initial model are built,and the seismic data of a two-dimensional survey line is selected. By comparison,the inverstion result of the model is better. Second,the reasons for the effect of velocity inversion are identified; Interpolation and smoothing technique are used to improved the accuracy.5. Traditional softwares usually usd fixed depth to extract the isovelocity surface, this may lead to the wrong amount of static correction. In order to solve this problem, we improved the methods of extraction of isovelocity surface to improve the inversion accuracy . At the same time, in order to eliminate some of the velocity mutation of the isovelocity surface we use median filter and mean filter method to smooth the surface velocity6. There is improvement, but also a place to show signs of deterioration in static Correction profiles.The result in this outcome is expecting further study.In this paper, theoretical models and actual data inversion results show that the method can reconstruct near-surface distribution of the velocity field stably, effectively and rapidly . To perform well in this anti-structure, it is necessary to rely on errors reduction of field observation system ,use of first examination of information , the high adaptivity of the algorithm, and the appropriate choice of parameters.更多还原