基于初至波走时层析成像的Tikhonov正则化与梯度优化算法
详细信息 本馆镜像全文    |  推荐本文 | | 获取馆网全文
摘要
初至波走时层析成像是利用地震初至波走时和其传播的射线路径来反演地下介质速度的技术.该问题本质上是一个不适定问题,需要使用正则化方法并辅之以适当的最优化技巧.本文从数值优化的角度介绍了初至波走时层析成像的反演原理,建立了Tikhonov正则化层析成像反演模型并提出求解极小化问题的加权修正步长的梯度下降算法.该方法可以从速度模型的可行域中迭代找到一个最优解.数值试验表明,该方法是可行和有应用前景的.
With the development of seismic exploration,inversion and imaging become key issues because of complex structures.It is in urgent need to build accurate near-surface velocity models.Nowadays,three primary numerical methods are developed to acquire the velocity model,i.e.,stack velocity analysis,migration velocity analysis and tomography velocity analysis.For the near-surface seismic problem,the first two methods are not suitable because of insufficient fold numbers and reflections.So the tomography method has received much more attention.First-arrival traveltime seismic tomography refers to inversion of medium velocity using firstarrival seismic wave traveltimes and their ray paths.The first task is model parameterization which discretizes the stratigraphic model into many slowness units by gridding.Secondly,basedon the slowness units,the ray paths are analyzed by the shortest traveltime ray tracing.Then the traveltime equation is established to solve the velocity model.This is an ill-posed inverse problem.Proper regularization technique and optimization methods are required.Therefore,a Tikhonov regularization model with constraints on feasible set was established,and a gradient descent method with modified step sizes was also developed to obtain an optimized solution.Three different theoretical models were designed to test the new algorithm.The first is a horizontal layered model:there were three horizontal layers with the velocity of 600 m·s-1,1200m·s-1 and 2000m·s-1 from top to bottom in the real model.By random disturbance of the model,we got an initial velocity model which was far away from the real model.The inversion result shows that the new algorithm can converge to the true model quickly even with poor initial condition.This shows that the new algorithm is stable and fast in convergence.Comparison with the well-known conjugate gradient(CG)method indicates that this new algorithm requires less memory and has higher convergence speed than the traditional CG algorithm.Specifically,the memory used by the CG algorithm is 2times that of the new algorithm,and the running time of the CG algorithm is 1.82 times that of the new algorithm.The second is a graben velocity model.Again,by random disturbance to the model,we got an initial velocity model which was far away from the true model.The inversion results also show that the new algorithm can converge quickly even with poor initial conditions and obtain a satisfactory result.The third is a fault velocity model with irregular interfaces.The inversion results show that the new algorithm could converge quickly with poor initial conditions and obtain a satisfactory result even if the model is complex.Based on the theory of first-arrival traveltime seismic tomography,a Tikhonov regularization model with constraints on feasible set was established,and a gradient descent method with modified step sizes was also developed to obtain an optimized solution.The new method has three characters:(1)In forward modeling,the shortest traveltime ray tracing is suitable for complex models with highly changing velocities.Sources and receivers can be arranged arbitrarily.The running time has no correlation with the degree of complex structure of the model.And a 3D calculation is easy to perform.(2)Aiming at the ill-posed problem,aproper regularization technique was developed to make the solution converge stably.(3)This method can find an optimal solution from the feasible region of the velocity model by iterations.
引文
Bios P,Porte M L,Lavergne M,et al.1972.Well-to-well seismicmeasurements.Geophysics,37(3):471-480.
    Bishop T N,Bube K P,Cutler R T,et al.1985.Tomographicdetermination of velocity and depth in laterally varying media.Geophysics,50(6):903-923.
    Chang X,Lu M X,Liu Y K.1999.Error analysis and appraisals forthree general solutions in seismic tomography.Chinese Journalof Geophysics(in Chinese),42(5):695-701.
    Chang X,Liu Y K,Wang H,et al.2002.3-D tomographic staticcorrection.Geophysics,67(4):1275-1285.
    Cheng G,Ma Z T,Geng J H,et al.2002.A review on the growthof seismic tomography.Progress in Exploration Geophysics(inChinese),25(3):6-12.
    Cheng G.2004.Theory and applications of seismic reflectiontraveltime tomography[Ph.D.thesis](in Chinese).Shanhai:Tongji University.
    Clapp R G,Biondi B,Claerbout J F.2004.Incorporating geologicinformation into reflection tomography.Geophysics,69(2):533-546.
    Daily W D.1984.Underground oil-shale retort monitoring usinggeotomography.Geophysics,49(10):1701-1711.
    Dines K A,Lytle R J.1979.Computerized geophysical tomography.Proc.IEEE,67(7):1065-1073.
    Dyer B C,Worthington M H.1988.Seismic reflection tomography:a case study.First Break,6:354-366.
    Fomel S.2007.Shaping regularization in geophysical estimationproblems.Geophysics,72(2):R29-R36.
    Guo J R,Feng X,Wang J X,et al.2008.Study of shortest pathmethod of ray tracing algorithm.Journal of Jilin University(Earth Science Edition)(in Chinese),38(S1):72-75.
    Han X L,Yang C C,Ma S H,et al.2008.Static of tomographicinversion by first breaks in complex areas.Progress inGeophysics(in Chinese),23(2):475-483.
    He L,Zhang W,Zhang J.2013.3Dwave-ray traveltime tomography fornear surface imaging.83rd SEG Annual Conference,Houston,Texas,1749-1753.
    Jing Y H.2009.Seismic first break travel-time tomography and itsapplication in near-surface velocity model building[Ph.D.thesis](in Chinese).Xi′an:Chang′an University.
    Li L M,Luo S X,Zhao B.2000.Tomographic inversion of firstbreak in surface model.OilG eophysical Prospecting(in Chinese),35(5):559-564.
    Liu Y Z,Dong L G,Xia J J.2007.Regularization methods for firstarrival travel time tomography.Oil Geophysical Prospecting(in Chinese),42(6):682-698.
    Lu H Y,LiuY K,Chang X.2013.MSFM based travel timescalculation in complex near surface model.Chinese Journal ofGeophysics(in Chinese),56(9):3100-3108,doi:10.6038/cjg20130922
    McMechan G A,Harris J M,Anderson L M.1987.Cross-holetomography for strongly variable media with applications toscale model data.Bulletin of the Seismological Society ofAmerica,77(6):1945-1960.
    Somerstein S F,Berg M,Chang D,et al.1984.Radio-frequencygeotomography for remotely probing the interior of operatingmini and commercial-sized oil-shale retorts.Geophysics,49(9):1288-1300.
    Wang B,Braile L W.1995.Effective approaches to handling nonuniform data coverage problem for wide-aperture refraction/reflection profiling.The 65th Ann.International Meeting,SEG,Expanded abstracts,659-662.
    Wang W,Han B,Tang J P.2013.Regularization method withsparsity constraints for seismic waveform inversion.ChineseJournal of Geophysics(in Chinese),56(1):289-297,doi:10.6038/cjg20130130
    Wang Y F.2007.Computational Methods for Inverse Problems andTheir Applications(in Chinese).Beijing:Higher EducationPress.
    Wang Y F,Yagola A G,Yang C C.2012.Optimization andRegularization for Computational Inverse Problems and Applications.Berlin:Springer.
    Wang Y F,Stepanova I E,Titarenko V N,et al.2011.InverseProblems in Geophysics and Solution Methods(in Chinese).Beijing:Higher Education Press.
    White D J.1991.Two-dimensional seismic refraction tomography.Geophysical Journal,72(2):223-245.
    Woodward M J,Nichols D,Zdraveva O,et al.2008.A decade oftomography.Geophysics,73(5):VE5-VE11.
    Xiao T Y,Yu S G,Wang Y F.2003.Numerical Methods for theSolution of Inverse Problems(in Chinese).Beijing:SciencePress.
    Yang W C,Li Y M.1993.Applied Seismic Tomography(inChinese).Beijing:Geological Publishing House,
    Yuan Y X.1993.Numerical Methods for Nonlinear Programming(in Chinese).Shanghai:Shanghai Science and TechnologyPublisher.
    Zhang J,Toksoz M N.1998.Nonlinear refraction traveltimetomography.Geophysics,63(5):1726-1737.
    Zhang J Z.2004.First break tomography for near-surface layers inseismic exploration.Journal of Xiamen University(NaturalScience)(in Chinese),43(1):63-66,doi:10.3321/j.issn:0438-0479.2004.01.016.
    Zhang P,Liu H,Li Y M.2000.The situation and progress of raytracing method research.Progress in Geophysics(in Chinese),15(1):36-45,doi:10.3969/j.issn.1004-2903.2000.01.002.
    常旭,卢孟夏,刘伊克.1999.地震层析成像反演中3种广义解的误差分析与评价.地球物理学报,42(5):695-701.
    成谷,马在田,耿建华等.2002.地震层析成像发展回顾.勘探地球物理进展,25(3):6-12.
    成谷.2004.地震反射走时层析理论与应用研究[博士论文].上海:同济大学海洋与地球科学学院.
    郭继茹,冯晅,王俊祥等.2008.最佳路径射线追踪算法研究.吉林大学学报(地球科学版),38(增刊):72-75.
    韩晓丽,杨长春,麻三怀等.2008.复杂山区初至波层析反演静校正.地球物理学进展,23(2):475-483.
    景月红.2009.地震初至波走时层析成像与近地表速度建模[硕士论文].西安:长安大学地球探测与信息技术学院.
    李录明,罗省贤,赵波.2000.初至波表层模型层析反演.石油地球物理勘探,35(5):559-564.
    刘玉柱,董良国,夏建军.2007.初至波走时层析成像中的正则化方法.石油地球物理勘探,42(6):682-698.
    卢回忆,刘伊克,常旭.2013.基于MSFM的复杂近地表模型走时计算.地球物理学报,56(9):3100-3108,doi:10.6038/cjg20130922.
    王薇,韩波,唐锦萍.2013.地震波形反演的稀疏约束正则化方法.地球物理学报,56(1):289-297,doi:10.6038/cjg20130130.
    王彦飞.2007.反演问题的计算方法及其应用.北京:高等教育出版社.
    王彦飞,斯捷潘诺娃I E,提塔连科V N等.2011.地球物理数值反演问题.北京:高等教育出版社.
    肖庭延,于慎根,王彦飞.2003.反问题的数值解法.北京:科学出版社.
    杨文采,李幼铭.1993.应用地震层析成像.北京:地质出版社.
    袁亚湘.1993.非线性规划数值方法.上海:上海科技出版社.
    张建中.2004.近地表介质地震初至波层析成像.厦门大学学报,43(1):63-66,doi:10.3321/j.issn:0438-0479.2004.01.016.
    张钋,刘洪,李幼铭.2000.射线追踪方法的发展现状.地球物理学进展,15(1):36-45,doi:10.3969/j.issn.1004-2903.2000.01.002.

版权所有:© 2023 中国地质图书馆 中国地质调查局地学文献中心