基于GPU技术的叠前时间偏移及其在玛湖地区的应用
|
摘要
地震勘探是重要的地球物理方法之一,尤其在油气勘探中,地震勘探是目前最主要的方法。随着油气资源的开采,勘探难度也逐渐增大,精细化处理解释更是当前的要求。在地震资料处理中,偏移成像技术是早些年已经发展起来的技术,偏移算法的不断完善、精确,使得地下成像归位更接近真实构造。可是,复杂的偏移计算一直是与计算效率相矛盾的,如果想提高成像质量,是要以牺牲大量的计算时间为代价的。
本文研究了GPU叠前时间偏移的方法及应用,它是针对偏移算法的精确化和偏移效率的提高而实现的。所研究的GPU叠前时间偏移采用目前常用的克希霍夫积分偏移方法。主要进行了以下几个方面的工作:(1)推导了一种弯曲射线走时的克希霍夫叠前时间偏移的算法。由于克希霍夫积分偏移的精确性主要取决于地震波走时的计算,因此根据假设近似条件,地震波在VTI介质中以弯曲射线的形式传播,将精确计算时距方程高阶项得到的走时代入波动方程克希霍夫积分解中,来进行偏移处理。(2)推导出保幅克希霍夫积分偏移的保幅因子。根据振幅随入射角的衰减特性,推导了振幅补偿因子,对振幅进行补偿,以达到保幅处理的目的。(3)简要讨论了各向异性校正,它使复杂构造地区的成像更精确。(4)引入GPU技术以解决偏移计算效率问题。伴随着偏移算法的精确,其复杂性和计算量也成倍增长,为了解决计算效率的问题,引入GPU技术。(5)用实际资料的动力学特征和时间切片对比说明GPU偏移的结果是同CPU偏移的结果一样可靠的。(6)研究GPU叠前时间偏移的技术特色及效果分析。按照GPU叠前时间偏移的步骤,逐步分析该偏移方法的技术特色,研究其对偏移结果的影响,通过举例分析,得到偏移的效果及与其他偏移效果的对比分析。(7)应用实际地震资料处理。通过实际地震资料处理和对比分析,说明基于GPU技术的叠前时间偏移的效果和效率,并说明这种偏移方法完全适用于实际应用。
基于GPU技术的叠前时间偏移研究是一项侧重于应用的技术,目前,叠前时间偏移已经是常规的处理流程,因此,更精确的偏移方法,更省时的偏移效率对于地震数据处理流程更显重要。研究表明,本文所采用的技术和研究方法是切实可行的,在效果和效率上也能取得令人满意的结果。
Seismic exploration is one of the most important geophysical methods. Respecting to oil prospecting, seismic exploration is the main method. It consists of acquisition, processing, and interpretation. These flows are connected because the results of the former flow are the inputting data for the next. It is harder for exploration with the exploitation of oil. The requirement of accuracy of processing and interpretation is higher. In the flow of processing data, migration is the technique developed many years ago. With the development of arithmetic of migration, the imaging of subsurface could be described as real formation. However, the contradiction between accuracy and efficiency of migration is always there. The quality of imaging can be improved sacrificing a lot of time for calculation.
In this paper, it discussed a new technology, GPU pre-stack migration and its application. It can solve the difficulties of quality and efficiency. GPU pre-stack migration used normally Kirchhoff integral migration method. Works done are as follows: (1) It discussed arithmetic of ray-bending pre-stack Kirchhoff migration. This was an arithmetic that the wave spread from the VTI medium in ray-bending. It was calculated by results of Kirchhoff integral of wave equation, replacing the travel time by the result of time-offset equation in high order. (2)It deduced the factor of amplitude preservation. According to the amplitude variation with incidence, it compensated the amplitude in order to preserve it. In the calculation of this arithmetic, it used anisotropy correction to image the accurate formation. (3)It discussed the anisotropy correction in brief. The correction made out precise migration results. (4) GPU technology could improve the efficient. With the improvement of migration’s arithmetic, the problems of complex calculation came out. In order to solve this problem, in this paper, it brought in GPU technique. (5) It used real seismic data processing to prove the dependability of GPU method. (6) Studied the characteristics and results’analysis of GPU pre-stack migration. It analyzed the advantages of this method step by step, studying the consequence to the results. (7) It discussed the application in real seismic data processing. It would compare the new processing data to the old ones in order to illustrate the results and improvement of GPU based pre-stack time migration.
GPU pre-stack migration is a method which is good for practice. In this time, pre-stack migration is normal flow. Therefore, precise migration method and high efficient are important to processing. The technique discussing in this paper is reliable.
本文研究了GPU叠前时间偏移的方法及应用,它是针对偏移算法的精确化和偏移效率的提高而实现的。所研究的GPU叠前时间偏移采用目前常用的克希霍夫积分偏移方法。主要进行了以下几个方面的工作:(1)推导了一种弯曲射线走时的克希霍夫叠前时间偏移的算法。由于克希霍夫积分偏移的精确性主要取决于地震波走时的计算,因此根据假设近似条件,地震波在VTI介质中以弯曲射线的形式传播,将精确计算时距方程高阶项得到的走时代入波动方程克希霍夫积分解中,来进行偏移处理。(2)推导出保幅克希霍夫积分偏移的保幅因子。根据振幅随入射角的衰减特性,推导了振幅补偿因子,对振幅进行补偿,以达到保幅处理的目的。(3)简要讨论了各向异性校正,它使复杂构造地区的成像更精确。(4)引入GPU技术以解决偏移计算效率问题。伴随着偏移算法的精确,其复杂性和计算量也成倍增长,为了解决计算效率的问题,引入GPU技术。(5)用实际资料的动力学特征和时间切片对比说明GPU偏移的结果是同CPU偏移的结果一样可靠的。(6)研究GPU叠前时间偏移的技术特色及效果分析。按照GPU叠前时间偏移的步骤,逐步分析该偏移方法的技术特色,研究其对偏移结果的影响,通过举例分析,得到偏移的效果及与其他偏移效果的对比分析。(7)应用实际地震资料处理。通过实际地震资料处理和对比分析,说明基于GPU技术的叠前时间偏移的效果和效率,并说明这种偏移方法完全适用于实际应用。
基于GPU技术的叠前时间偏移研究是一项侧重于应用的技术,目前,叠前时间偏移已经是常规的处理流程,因此,更精确的偏移方法,更省时的偏移效率对于地震数据处理流程更显重要。研究表明,本文所采用的技术和研究方法是切实可行的,在效果和效率上也能取得令人满意的结果。
Seismic exploration is one of the most important geophysical methods. Respecting to oil prospecting, seismic exploration is the main method. It consists of acquisition, processing, and interpretation. These flows are connected because the results of the former flow are the inputting data for the next. It is harder for exploration with the exploitation of oil. The requirement of accuracy of processing and interpretation is higher. In the flow of processing data, migration is the technique developed many years ago. With the development of arithmetic of migration, the imaging of subsurface could be described as real formation. However, the contradiction between accuracy and efficiency of migration is always there. The quality of imaging can be improved sacrificing a lot of time for calculation.
In this paper, it discussed a new technology, GPU pre-stack migration and its application. It can solve the difficulties of quality and efficiency. GPU pre-stack migration used normally Kirchhoff integral migration method. Works done are as follows: (1) It discussed arithmetic of ray-bending pre-stack Kirchhoff migration. This was an arithmetic that the wave spread from the VTI medium in ray-bending. It was calculated by results of Kirchhoff integral of wave equation, replacing the travel time by the result of time-offset equation in high order. (2)It deduced the factor of amplitude preservation. According to the amplitude variation with incidence, it compensated the amplitude in order to preserve it. In the calculation of this arithmetic, it used anisotropy correction to image the accurate formation. (3)It discussed the anisotropy correction in brief. The correction made out precise migration results. (4) GPU technology could improve the efficient. With the improvement of migration’s arithmetic, the problems of complex calculation came out. In order to solve this problem, in this paper, it brought in GPU technique. (5) It used real seismic data processing to prove the dependability of GPU method. (6) Studied the characteristics and results’analysis of GPU pre-stack migration. It analyzed the advantages of this method step by step, studying the consequence to the results. (7) It discussed the application in real seismic data processing. It would compare the new processing data to the old ones in order to illustrate the results and improvement of GPU based pre-stack time migration.
GPU pre-stack migration is a method which is good for practice. In this time, pre-stack migration is normal flow. Therefore, precise migration method and high efficient are important to processing. The technique discussing in this paper is reliable.
引文
[1] Abma, R., Sun, J., and Bernitsas, N.，Antialiasing methods in Kirchhoff migration, Geophysics, 1999, 64, 1783~1792.
[2] Augustin A. Dubrulle. Numerical methods for the migration of constant-offset sections in homogeneous and horizontally layered media. Geophysics, Vol, 48, No. 9 (September 1983); P. 1195~1203.
[3] Bancroft. J C, Geiger H K, Margrave G F. The equivalent offset of prestack time migration. Geophysics, 1998, 63:2 042~2054.
[4] Baysal,E.，Kosloff, D. D., and Sherwood, J.W. C., Reverse time migration, Geophysics, 1983, 48, 1514~1524.
[5] Berryhill, J. R., Wave equation datuming before stack(short note), Geophysics, 1984, 49, 2064~2067.
[6] Chuan He, Chuanwen Sun, et al.，Prestack Kirchhoff Time Migration on High Performance Reconfigurable Computing Platform, 2005, 75th Ann. Internat. Mtg., Soc. Expl.Geophys. Expanded Abstracts: 1903~1905.
[7] Ehinger, A., Lailly, P., and Marfurt, s K. J., Green's function implementation of common-offset wave-equation migration, Geophysics, 1996, 61, 1813~1821.
[8] Ekren, B. O., and Ursin, B., True-amplitude frequency-wavenumber constant-offset migration, Geophysics, 1999, 64,915~924.
[9] Fei, T., Dellinger, J. A., Murphy, G. E., Hensley, J·L., and Gray, S. H·，Anisotropic true-amplitude migration, 68t" Annual Internat. Mtg., Soc, Expl, Geophysics, Expanded Abstracts, 1998, 1677~1679.
[10] Gazdag, J., Wave-equation migration with the phase-shift method, Geophysics, 1978, v.43, p. 1324~1351.[11』Hale, D., Dip-moveout by Fourier transform, Geophysics, 1984, 49,741~757.
[12] John D. Owens, Mike Houston, David Luebke et al. GPU Computing. Proceedings of the IEEE, 2008, 96: 5 8795~ 899.
[13] Judson, D. R., Schultz, P. S., and Sherwood, J.W. C., Equalizing the stacking velocities of dipping events via Devilish, Presented at the 48t" Ann. Internat. Mtg., Soc. Expl. Geophys. 1978. (Brochure published by Digicon Geophysical Corp.)
[14] Mayne, W. H.，1962, Common reflection point horizontal data stacking techniques, Geophysics, 1962, 27, 927~938.
[15] M .Harris, Mappiing computational concepts to GPUs. GPU Gems, 2005, 2: 493~508.
[16] Operto, M. S., Xu, S., and Lambare, G., Can we quantitatively image complex structures with rays. Geophysics, 2000, 65, 1223~1238.
[17] Ruben Martinez, Sugarland，TX(US) ChuanWen Sun, Katy,TX(US) 3D PRESTACK TIME MIGRATlON METHOD. United States Patent, Nov.30, 2004.
[18] Samuel H.Gray, John Etgen, Joe Dellinger et al.Seismic migration problems and solutions. Geophysics, Vol 66, No. 5(September-October 2001); P. 1622~1640.
[19] Stolt, R. H.，Migration by Fourier transform, Geophysics, 1978, v. 43, p. 23~48.
[20] Whitmore, N.D., Iterative depth migration by backward time propagation, 53rd Ann. Internat. Mtg., Soc. Expl.Geophys, Expanded Abstracts, 1983, 382~385.
[21]李肯立，彭俊杰，周仕勇.基于CUDA的Kirchhoff叠前时问偏移算法设计与实现.计算机应用研究.2009, 26(12): 4474~4477.
[22]李庆忠.走向精确勘探的道路——高分辩滤地震勘探系统工程剖析.北京:石油工业出版社，1993.
[23]刘财等编著.勘探地震资料处理新方法及新技术.北京:科学出版社，2006.
[24]刘奇琳，黄跃，唐建明等.波动方程叠前深度偏移的GPU技术.物探化探计算技术，2010, 32(4):386~391.
[25]李振春，张军华.地震数据处理方法.东营:石油大学出版社，2004.8
[26]李振春，朱绪峰，韩文功.真振幅偏移方法综述.勘探地球物理进展，2008,31 (1):11~15.
[27]雪冬，贾光华，全兆岐.Kirchhoff叠前时问偏移处理技术及应用.物探化探计算技术.2009,31(2):126~129.
[28]罗银河，刘江平，董桥梁.Kirchhoff弯曲射线叠前时问偏移及应用.天然气工业，2005, 25(8): 35~37.
[29]麻三怀，杨长春，孙福利等.克希霍夫叠前时间偏移技术在复杂构造带地震资料处理中的应用.地球物理学进展，2008, 23(3): 754~760.
[30]马淑芳，李振春.波动方程叠前深度偏移方法综述.勘探地球物理进展，2007,30 (3):153~160.
[31]马在田，地震成像技术，北京，石油工业出版社，1989.
[32]牟永光，陈小宏，李国发等.地震数据处理方法.北京:石油工业出版社，2007.
[33]潘宏勋，孙开峰，叶勇.地震速度分析技术新进展.勘探地球物理进展，2008,31 (3):173~177.
[34]王华忠，冯波，任浩然.二维Offset平面波有限差分法叠前时间偏移.石油物探，2009,48(1):11~19.
[35]王建立，周纳，土真理等.VTI介质叠前时间偏移及实例分析.石油天然气学报，2010, 32(3): 228~231.
[36]王小卫，姚姚，吕彬等.弯曲射线走时计算方法在Kirchhoff叠前时间偏移中的应用.天然气地球科学，2010, 21(5): 855~858.
[37]渥·伊尔马兹[美].地震资料分析——地震资料处理、反演和解释.北京:石油工业出版社，2006.
[38]肖莉.VTI介质非双曲时差方程叠前时间偏移方法对比.油气地质与采收率，2010, 17(1): 57~61.
[39]熊蠢.地震数据处理方法系统思维.北京:石油工业出版社，1995.
[40]熊蠢编著.复杂地区地震数据处理思路.北京:石油工业出版社，2002.
[41]熊蠢.复杂地区油气地球物理勘探技术.北京:石油工业出版社，1999.
[42]徐琴，李振春鲍伟.波动方程偏移中的成像条件研究进展.勘探地球物理进展.2008,31(4):247~251.
[43]姚姚.地震波场与地震勘探.北京:地质出版社，2006,6.
[44]俞寿朋.高分辨率地震勘探.北京:石油工业出版社，1993.
[45]袁刚，蒋波，曾华会.各向异性叠前时间偏移在塔里木碳酸盐宕资料中的应用.大庆石油学院学报，2010, 34(3): 23~28.
[46]张敏，李振春.偏移速度分析与建模方法综述.勘探地球物理进展，2007, 30 (6 ) 421~427.
[47]张白林，尹成，谢杜生.地震资料数字处理方法.北京:石油工业出版社，1998.
[48]张伟.叠前时间偏移成像影响因素研究[博士学位论文].北京:中国地质大学(北京)，2010.
[49]钟联波.GPU与CPU的比较分析.技术与市场.2009,16(9): 13~14.
[50]周锦明，熊蠢.地震数据精细处理.北京:石油工业出版社，2003.
[51]邹德志.起伏地表叠前时间偏移及并行实现:[硕士学位论文].北京:中国地质大学(北京)，2010.
[52]朱海波张兵.起伏地表叠前时间偏移技术研究.勘探地球物理进展，2007,30 (5):368~372.
[2] Augustin A. Dubrulle. Numerical methods for the migration of constant-offset sections in homogeneous and horizontally layered media. Geophysics, Vol, 48, No. 9 (September 1983); P. 1195~1203.
[3] Bancroft. J C, Geiger H K, Margrave G F. The equivalent offset of prestack time migration. Geophysics, 1998, 63:2 042~2054.
[4] Baysal,E.，Kosloff, D. D., and Sherwood, J.W. C., Reverse time migration, Geophysics, 1983, 48, 1514~1524.
[5] Berryhill, J. R., Wave equation datuming before stack(short note), Geophysics, 1984, 49, 2064~2067.
[6] Chuan He, Chuanwen Sun, et al.，Prestack Kirchhoff Time Migration on High Performance Reconfigurable Computing Platform, 2005, 75th Ann. Internat. Mtg., Soc. Expl.Geophys. Expanded Abstracts: 1903~1905.
[7] Ehinger, A., Lailly, P., and Marfurt, s K. J., Green's function implementation of common-offset wave-equation migration, Geophysics, 1996, 61, 1813~1821.
[8] Ekren, B. O., and Ursin, B., True-amplitude frequency-wavenumber constant-offset migration, Geophysics, 1999, 64,915~924.
[9] Fei, T., Dellinger, J. A., Murphy, G. E., Hensley, J·L., and Gray, S. H·，Anisotropic true-amplitude migration, 68t" Annual Internat. Mtg., Soc, Expl, Geophysics, Expanded Abstracts, 1998, 1677~1679.
[10] Gazdag, J., Wave-equation migration with the phase-shift method, Geophysics, 1978, v.43, p. 1324~1351.[11』Hale, D., Dip-moveout by Fourier transform, Geophysics, 1984, 49,741~757.
[12] John D. Owens, Mike Houston, David Luebke et al. GPU Computing. Proceedings of the IEEE, 2008, 96: 5 8795~ 899.
[13] Judson, D. R., Schultz, P. S., and Sherwood, J.W. C., Equalizing the stacking velocities of dipping events via Devilish, Presented at the 48t" Ann. Internat. Mtg., Soc. Expl. Geophys. 1978. (Brochure published by Digicon Geophysical Corp.)
[14] Mayne, W. H.，1962, Common reflection point horizontal data stacking techniques, Geophysics, 1962, 27, 927~938.
[15] M .Harris, Mappiing computational concepts to GPUs. GPU Gems, 2005, 2: 493~508.
[16] Operto, M. S., Xu, S., and Lambare, G., Can we quantitatively image complex structures with rays. Geophysics, 2000, 65, 1223~1238.
[17] Ruben Martinez, Sugarland，TX(US) ChuanWen Sun, Katy,TX(US) 3D PRESTACK TIME MIGRATlON METHOD. United States Patent, Nov.30, 2004.
[18] Samuel H.Gray, John Etgen, Joe Dellinger et al.Seismic migration problems and solutions. Geophysics, Vol 66, No. 5(September-October 2001); P. 1622~1640.
[19] Stolt, R. H.，Migration by Fourier transform, Geophysics, 1978, v. 43, p. 23~48.
[20] Whitmore, N.D., Iterative depth migration by backward time propagation, 53rd Ann. Internat. Mtg., Soc. Expl.Geophys, Expanded Abstracts, 1983, 382~385.
[21]李肯立，彭俊杰，周仕勇.基于CUDA的Kirchhoff叠前时问偏移算法设计与实现.计算机应用研究.2009, 26(12): 4474~4477.
[22]李庆忠.走向精确勘探的道路——高分辩滤地震勘探系统工程剖析.北京:石油工业出版社，1993.
[23]刘财等编著.勘探地震资料处理新方法及新技术.北京:科学出版社，2006.
[24]刘奇琳，黄跃，唐建明等.波动方程叠前深度偏移的GPU技术.物探化探计算技术，2010, 32(4):386~391.
[25]李振春，张军华.地震数据处理方法.东营:石油大学出版社，2004.8
[26]李振春，朱绪峰，韩文功.真振幅偏移方法综述.勘探地球物理进展，2008,31 (1):11~15.
[27]雪冬，贾光华，全兆岐.Kirchhoff叠前时问偏移处理技术及应用.物探化探计算技术.2009,31(2):126~129.
[28]罗银河，刘江平，董桥梁.Kirchhoff弯曲射线叠前时问偏移及应用.天然气工业，2005, 25(8): 35~37.
[29]麻三怀，杨长春，孙福利等.克希霍夫叠前时间偏移技术在复杂构造带地震资料处理中的应用.地球物理学进展，2008, 23(3): 754~760.
[30]马淑芳，李振春.波动方程叠前深度偏移方法综述.勘探地球物理进展，2007,30 (3):153~160.
[31]马在田，地震成像技术，北京，石油工业出版社，1989.
[32]牟永光，陈小宏，李国发等.地震数据处理方法.北京:石油工业出版社，2007.
[33]潘宏勋，孙开峰，叶勇.地震速度分析技术新进展.勘探地球物理进展，2008,31 (3):173~177.
[34]王华忠，冯波，任浩然.二维Offset平面波有限差分法叠前时间偏移.石油物探，2009,48(1):11~19.
[35]王建立，周纳，土真理等.VTI介质叠前时间偏移及实例分析.石油天然气学报，2010, 32(3): 228~231.
[36]王小卫，姚姚，吕彬等.弯曲射线走时计算方法在Kirchhoff叠前时间偏移中的应用.天然气地球科学，2010, 21(5): 855~858.
[37]渥·伊尔马兹[美].地震资料分析——地震资料处理、反演和解释.北京:石油工业出版社，2006.
[38]肖莉.VTI介质非双曲时差方程叠前时间偏移方法对比.油气地质与采收率，2010, 17(1): 57~61.
[39]熊蠢.地震数据处理方法系统思维.北京:石油工业出版社，1995.
[40]熊蠢编著.复杂地区地震数据处理思路.北京:石油工业出版社，2002.
[41]熊蠢.复杂地区油气地球物理勘探技术.北京:石油工业出版社，1999.
[42]徐琴，李振春鲍伟.波动方程偏移中的成像条件研究进展.勘探地球物理进展.2008,31(4):247~251.
[43]姚姚.地震波场与地震勘探.北京:地质出版社，2006,6.
[44]俞寿朋.高分辨率地震勘探.北京:石油工业出版社，1993.
[45]袁刚，蒋波，曾华会.各向异性叠前时间偏移在塔里木碳酸盐宕资料中的应用.大庆石油学院学报，2010, 34(3): 23~28.
[46]张敏，李振春.偏移速度分析与建模方法综述.勘探地球物理进展，2007, 30 (6 ) 421~427.
[47]张白林，尹成，谢杜生.地震资料数字处理方法.北京:石油工业出版社，1998.
[48]张伟.叠前时间偏移成像影响因素研究[博士学位论文].北京:中国地质大学(北京)，2010.
[49]钟联波.GPU与CPU的比较分析.技术与市场.2009,16(9): 13~14.
[50]周锦明，熊蠢.地震数据精细处理.北京:石油工业出版社，2003.
[51]邹德志.起伏地表叠前时间偏移及并行实现:[硕士学位论文].北京:中国地质大学(北京)，2010.
[52]朱海波张兵.起伏地表叠前时间偏移技术研究.勘探地球物理进展，2007,30 (5):368~372.