1、 2248 单词， 4098 汉字 出处： Zhang S, Schuster G, Luo Y. Wave-equation reflection traveltime inversionC/2011 SEG Annual Meeting. Society of Exploration Geophysicists, 2011. 毕 业论 文 文献翻译 完成日期 2014 年 5 月 28 日院系名称： _地球物理与信息工程学院 专业名称： _勘查技术与工程 _ 学生姓名： 学 号： 指导教师： 1 Wave-equation Reflection Traveltime Inversion S
2、anzong Zhang, Gerard Schuster, King Abdullah University of Science and Technology,and Yi Luo, Saudi Aramco SUMMARY The main difficulty with iterative waveform inversion using agradient optimization method is that it tends to get stuck inlocal minima associated within the waveform misfit function.Thi
3、s is because the waveform misfitfunction is highly nonlinearwith respect to changes in the velocity model. To reducethis nonlinearity, we present a reflection traveltime tomographymethod based on the wave equation which enjoys a morequasi-linear relationship between the model and the data. Alocal cr
4、osscorrelation of the windowed downgoing direct waveand the upgoing reflection wave at the image point yields thelag time that maximizes the correlation. This lag time representsthe reflection traveltime residual that is back-projectedinto the earth model to update the velocity in the same way aswav
5、e-equation transmission traveltime inversion. No traveltimepicking is needed and no high-frequency approximationis assumed. The mathematical derivation and the numericalexamples are presented to partly demonstrate its efficiency androbustness. INTRODUCTION Prestack depth migration of 3D seismic data
6、 is the industrystandard for computing detailed estimates of the earths reflectivitydistribution. However, an accurate velocity modelis a precondition for imaging complex geological structures.To estimate this velocity model, there are three primary inversionmethods: migration velocity analysis (MVA
7、), traveltimeinversion, and full waveform inversion. For migration velocityanalysis (Symes and Kern, 1994； Sava and Biondi, 2004；Shen and Calandra, 2005), the optimal migration velocity isthe one that best flattens the reflection events in a common imagegather. For traveltime inversion (Dines and Ly
8、tle, 1979；Paulsson et al., 1985； Ivansson, 1985； Bishop et al., 1985；Lines, 1988), the traveltimes of refraction and reflection arrivalsare used to invert for smooth features of the velocitymodel, while full waveform inversion (Tarantola, 1986, 1987；Mora, 1987； Crase et al., 1992； Zhou et al., 1995；
9、 Pratt, 1998)inverts the waveform information for fine details of the earthmodel. A more detailed analysis shows that traveltime inversion isconstrained by a high-frequency approximation, and so it failsto invert for the earths velocity variations having nearly thesame wavelength or less than that o
10、f the source wavelet. 2 Consequently,the resolution of the velocity model constructed fromthe traveltimes is much less than that of full waveform inversion.The merit is that the traveltime misfit function (normedsquared error between observed and calculated traveltimes) isquasi-linear with respect t
11、o velocity perturbations so that anefficient velocity inversion can be achieved even if the startingmodel is far from the actual model (Luo and Schuster,1991a and 1991b； Zhou et al., 1995). Although very sensitiveto the choice of starting models or noisy amplitudes,full waveform inversion can someti
12、mes reconstruct a finely detailedestimation of the earth model. This is because there is nohigh-frequency assumption about the data, and almost all seismicevents are embedded in the misfit function. The problemwith full waveform inversion, however, is that its misfit function(normed squared error be
13、tween the observed and syntheticseismograms) can be highly nonlinear with respect to changesin the velocity model. In this case, a gradient method will tendto get stuck in a local minima if the starting model is far awayfrom the actual model. To exploit the strengths and ameliorate the weaknesses of
14、 bothray-based traveltime tomography and full waveform inversion,wave-equation-based traveltime inversion was developed to invertthe velocity model (Luo and Schuster, 1991a and 1991b；Zhou et al, 1995； Zhang and Wang, 2009； Leeuwen and Mulder,2010). This kind of inversion methods inverts traveltimeus
15、ing the gradient calculated from the wave equation. It isnot constrained by a high-frequency approximation and traveltimepicking is not necessary. Other important benefits area convergence rate that is somewhat insensitive to the startingmodel, a high degree of model resolution, and a robustness int
16、he presence of data noise. However, these traveltime inversionmethods are designed to invert transmission waves in seismicdata, and are not designed to invert the reflection traveltimes.Unlike refraction and direct waves, reflection waves can providemore velocity information about the deeper subsurf
17、ace formodel inversion. However, full waveform inversion of reflectionwave is difficult if the initial velocity model is far from thetruemodel. To overcome this limitation, this paper presents theextension of wave-equation transmission traveltime inversion(WTI) (Luo and Schuster, 1991a and 1991b) to
18、 wave-equationreflection traveltime inversion (WRTI). This paper is organized into three sections. The first sectiondescribes the basic theory of image-domain wave-equation reflectiontraveltime inversion. The second section shows numericalexamples to verify the effectiveness of this method.The last section draws some conclusions.
