【作者】 高磊；

【导师】 周熙襄；

【作者基本信息】 成都理工大学 ， 地球探测与信息技术， 2008， 博士

【摘要】 复杂近地表结构不仅影响地震勘探的激发和接收,使野外采集难以获得较高信噪比的原始资料,而且严重影响地震资料的成像和振幅保真处理,弄清复杂地区近地表速度结构,对于解决采集中的激发问题、资料处理中的静校正问题、波场延拓问题、振幅保真问题都具有十分重要的意义。多年来人们发展了多种表层调查方法,其中一维表层调查方法有浅层折射法、微测井法等,二维和三维表层调查方法有内插法、层析成像法、折射初至方法等,这些方法中的折射初至方法由于能够快速的建立表层模型、而且方法比较容易实现而得到广泛的应用。利用折射初至方法来建立表层模型的方法很多,其中扩展广义相遇法、延迟时法等是比较好的方法。论文在前人研究的基础上,总结了扩展广义相遇法和延迟时法的优缺点,提出了一种求取折射层速度的新方法-广义圆滑法,这种算法受地表影响小,求取的速度精度高;论文提出了一种迭代延迟时法,这种方法收敛速度快,求取的表层厚度准确,通过实际资料的计算表明,这是一种可以较好解决表层低速带结构的算法;论文还提出了一种折射层成像方法,这是一种利用地震资料直接对表层进行成像的算法,这种算法成像的结果可得到低速带结构,并且比较容易实现;综合广义圆滑法、迭代延迟时法和折射层成像方法的优点,最后提出了一种更加有效的算法—成像迭代延迟时法,使用这种方法,求取了多个地区的低速带结构,建立的地表模型,其计算结果较为理想。利用折射初至方法建立地表模型,首先要解决的就是初至波的自动拾取问题。针对初至波的自动拾取有很多计算方法,这些方法对于初至质量好的记录都有着不错的拾取效果,但是随着记录质量的降低,自动拾取的初至可信度越来越差。在复杂地表条件下,初至波往往受到干扰,变化较大,现有的初至波自动拾取方法的精度和速度都无法适应这种情况,在大多数情况下,初至波的拾取最终仍需要人工逐炮对自动拾取的结果进行修改。基于这样情况,文中首次提出了基于边缘检测和边界追踪技术的初至波自动拾取方法。基于上面方法的思想,又首次提出了单道边界检测和样条插值算法来完成初至波的自动拾取。这些方法在实际资料初至波的拾取中取得了令人满意的效果。随着人们环保意识的增强和可控震源技术本身的优势,越来越多的地区开始使用可控震源激发地震信号。但由于可控震源自身的原因,其记录中的初至波不如爆炸震源产生的初至波那样明显,经常会出现“双仞至”和“多初至”现象。这种现象使得可控震源初至很难准确自动拾取。可控震源初至波的自动拾取成为生产中迫切需要解决的一个问题。论文基于可控震源记录中初至的特点,首次提出了适合可控震源初至波自动拾取的方法。论文的主要贡献是:1.首次提出了基于边缘检测和边界追踪的初至波自动拾取方法:把地震记录道或道集认为是一幅图,利用图像处理中的边缘检测和边界追踪技术来拾取初至波。2.首次提出了单道边界检测和样条插值算法的初至波自动拾取方法:根据初至波在单道里面有能量分界的特性,只要能准确确定单道记录能量的分界就可以确定初至波的位置,然后通过样条插值算法把边界确定出来。3.首次提出了一套解决可控震源初至波自动拾取方法:研究了可控震源的初至波的特点,提出了一些解决方法,这些方法中主要是关于相关与能量的改进方法,并在此基础上,首次把全局寻优方法中的禁忌算法和模拟退火算法用到了可控震源初至拾取的问题中。4.提出了广义圆滑法:改进了广义相遇法求取折射层速度的方法,使在地表不规则的情况下也能求取比较准确的折射层速度。5.提出了迭代延迟时法:改进了延迟时法,使求取的表层的厚度更加准确。6.提出了折射层成像方法:在没有任何信息的情况下,能够直接对地震道进行处理,产生叠加剖面,对表层进行成像。7.提出了成像迭代延迟时法:综合了广义圆滑法、迭代延迟时法和折射层成像方法,提出了成像迭代延迟时法,可以求取比较准确的表层模型。更多还原

【Abstract】 The complex near-surface structure not only influences the exploding and receiving of seismic exploration, and it’s difficult to obtain the high-SNR original data in field collection, but also influences the imaging and amplitude fidelity of the seismic data. So it’s important to find out the near-surface velocity structure of complex area for solving the exploding problem, static correction problem in data processing, wave field extrapolation and amplitude fidelity.For the past many years, people have developed many kinds of methods for near-surface modeling. Among the 1-Dimensional investigative methods, there are shallow refraction and microseismogram log etc. And among 2-D and 3-D investigative methods, there are Interpolation, tomography, and first arrival refraction etc. Of these methods, first arrival refraction is widely used because it can build the general surface model quickly, and it’s not complicated to realize it. The representational methods to build surface model by first arrival refraction are EGRM (Extended Generalized Reciprocal Method) and Delayed Time.On the basis of former research, in this paper we sum up the relative merits of EGRM and Delayed Time, and advance a new method for seeking the velocity of refraction, which is little affected by the surface and has a high velocity precision; And we raise a new method named Iteration Delayed Time (IDT), which has a high velocity of convergence and can seek a exact layer thickness. It is a good arithmetic to obtain the structure of surface velocity zone by computing the actual data. Refracting Layer Imaging (RLI) is an arithmetic which can form a image of refracting layer from the seismic data directly, and the image is similar to the CMP stacked profile by reflecting method, but the former is the structure of low velocity layer and can be accomplished simply; To sum up the advantages of GRM, IDT and RLI, we raise a more effectual method—Imaging Iteration Delayed Time, by which we sought the structure of low velocity layer in several areas and built surface model, and got perfect computing results.It’s essential to pick up the first-arrival wave when building the surface model by first arrival refraction. There are many design procedures for picking up the first arrival wave automatically. Each method has a good effect for the data record which has a well first arrival quality. But with the reduction of the record quality, the reliability of first arrival wave by automatic pick is worse and worse. In the condition of complex surface, first arrival wave is interfered sometimes and changes too much, so the precision and speed of existing methods for picking up the first arrival wave automatically cannot adapt this condition. In most cases, the pick still to be done shot by shot with manpower. Base on this condition, in this paper a new method based on edge detection and boundary tracing is firstly introduced and based upon the idea of this method, the picking is accomplished firstly with the single track boundary detection and sample interpolation arithmetic. These methods have achieved well effects in the first arrival wave picking of actual data.Along with the increase of people’s environmental consciousness and the advantage of vibrators itself, more and more vibrators are used to explode seismic signal. But because of the reason of vibrators itself, of its data the first arrival wave is less obvious than the first arrival wave from the explosive source, and "dual-wave arrival" and "multi-wave arrival" are often appeared. Owing to this phenomenon, it’s hard to automatically pick up the first arrival of vibrators accurately. So it’s an urgent problem to pick up the first arrival of vibrators. In this paper based on the first arrival characteristic of vibrators, a method for picking up the first arrival wave automatically is introduced for the first time.The primary contributions of this paper are as follows:1. For the first time this paper proposes a method for picking up the first arrival wave automatically based on the edge detection and boundary tracing:we consider recording traces an image and pick up the first arrival using marginal checking and boundary tracing of image processing.2. For the first time this paper proposes a method for picking up the first arrival wave automatically based on arithmetic of single track boundary detection and sample interpolation.3. For the first time a suit of pick up the first arrival wave automatically for the vibrators is introduced. From the research of first arrival wave characteristic of vibrators, Some of these methods are concerned with related correlation and energy methods, and based on them the tabu search and simulated annealing are firstly applied in the first arrival picking of controlled source.4. The paper proposes general smoothing: Improving the method of GRM to inquire the velocity of refraction.Hence,it can obtain the exact velocity of refraction when earth’s surface is irregular.5. The paper proposes Iteration Delayed Time (IDT):By improving Delayed Time ,it can obtain more exact depth of near surface layer.6. The paper proposes Refracting Layer Imaging (RLI):Without any informations,it can handle recording traces and image the near surface layers.7. The paper proposes Imaging Iteration Delayed Time( IIDT): It synthesizes general smoothing and IDT and RLI,so it can increase the precision for seeking the velocity of refracting layer and thickness of surface layer.更多还原