Fast least-squares reverse-time migration with adaptive moment estimation
WU Dan1,2, WU Haili1,2, LI Qun1,2, ZHANG Xiangyang1,2, LIU Shuren1,2
1. Northwest Branch, Research Institute of Petroleum Exploration & Development, PetroChina, Lanzhou, Gansu 730030, China; 2. Key Laboratory of Internet of Things, CNPC, Lanzhou, Gansu 730030, China
Abstract:Least-squares reverse-time migration (LSRTM) is a seismic imaging method with high resolution and favorable amplitude fidelity. However, its computational burden is heavy since it often needs to run iterations nearly ten times and takes approximately the computational cost of two full-dataset RTMs for each iteration. Here, we introduce the adaptive moment estimation (Adam) method from the field of deep learning to improve the computational efficiency of LSRTM: At each iteration, only part of the common shot gathers are required to calculate the gradient and the resulting gradient is corrected by the momentum (AdaGrad) method; considering the nonstationary property of the gradient, the root mean square propagation (RMSProp) algorithm is used to eliminate the influence of inadequate illumination. The Adam method, combining the advantages of the AdaGrad method and the RMSProp method, not only reduces the computational burden of each iteration but also improves the convergence speed of the iterations. In addition, this method is straightforward to implement and computationally efficient with a low memory requirement, and thus it is a fast and effective gradient preconditioning method. The A-dam method not only can be applied to LSRTM directly but also is applicable to shot encoding LSRTM. Numerical tests on the SEG/EAGE salt model show that this method can provide a high-precision and high-resolution image at merely the same cost as that of two RTMs. The substantial increase in computational efficiency paves the way for the application of LSRTM in practical seismic data processing.
CLAERBOUT J F.Toward a unified theory of reflector mapping[J].Geophysics,1971,36(3):467-481.
[2]
LAILLY P.The seismic inverse problem as a sequence of before stack migrations[C].Conference on Inverse Scattering,Theory and Applications,Society for Industrial and Applied Mathematics,1983,206-220.
[3]
NEMETH T,WU C J,SCHUSTER G T.Least-squares migration of incomplete reflection data[J].Geophy-sics,1999,64(1):208-221.
[4]
KVHL H,SACCHI M D.Least-squares wave-equation migration for AVP/AVA inversion[J].Geophy-sics,2003,68(1):262-273.
[5]
CLAPP M L,CLAPP R G,BIONDI B L.Regularized least-squares inversion for 3-D subsalt imaging[C].SEG Technical Program Expanded Abstracts,2005,24:1814-1817.
[6]
VALENCIANO A A,BIONDI B,GUITTON A.Target-oriented wave-equation inversion[J].Geophysics,2006,71(4):A35-A38.
[7]
王华忠,盛燊.走向精确地震勘探的道路[J].石油物探,2021,60(5):693-708,720.WANG Huazhong,SHENG Shen.Pathway toward accurate seismic exploration[J].Geophysical Prospecting for Petroleum,2021,60(5):693-708,720.
[8]
段心标,王华忠,邓光校.应用反射理论的最小二乘逆时偏移成像[J].石油地球物理勘探,2020,55(6):1305-1311.DUAN Xinbiao,WANG Huazhong,DENG Guangxiao.Least-squares reverse-time migration based on reflection theory[J].Oil Geophysical Prospecting,2020,55(6):1305-1311.
[9]
陈汉明,周辉,田玉昆.分数阶拉普拉斯算子黏滞声波方程的最小二乘逆时偏移[J].石油地球物理勘探,2020,55(3):616-626.CHEN Hanming,ZHOU Hui,TIAN Yukun.Least-squares reverse-time migration based on a fractional Laplacian viscoacoustic wave equation[J].Oil Geophysical Prospecting,2020,55(3):616-626.
[10]
MORTON S A,OBER C C.Faster shot-record depth migrations using phase encoding[C].SEG Technical Program Expanded Abstracts,1998,17:13-17.
[11]
ROMERO L A,GHIGLIA D C,OBER C C,et al.Phase encoding of shot records in prestack migration[J].Geophysics,2000,65(2):426-436.
[12]
ZHANG Y,SUN J,NOTFORS C,et al.Delayed-shot 3D depth migration[J].Geophysics,2005,70(5):E21-E28.
[13]
LIU F Q,HANSON D W,WHITMORE N D,et al.Toward a unified analysis for source plane-wave migration[J].Geophysics,2006,71(4):S129-S139.
[14]
TANG Y X.Target-oriented wave-equation least-squares migration/inversion with phase-encoded Hessian[J].Geophysics,2009,74(6):WCA95-WCA107.
[15]
GODWIN J,SAVA P.Blended source imaging by amplitude encoding[C].SEG Technical Program Expanded Abstracts,2010,29:3125-3129.
[16]
GAO F C,ATLE A,Williamson P.Full waveform inversion using deterministic source encoding[C].SEG Technical Program Expanded Abstracts,2010,29:1013-1017.
[17]
SCHUSTER G T,WANG X,HUANG Y,et al.Theo-ry of multisource crosstalk reduction by phase-encoded statics[J].Geophysical Journal International,2011,184(3):1289-1303.
[18]
HERRMANN F,LI X.Efficient least-squares imaging with sparsity promotion and compressive sensing[J].Geophysical Prospecting,2012,60(4):696-712.
[19]
LIU Y J,SYMES W W,LI Z C.Multisource least-squares extended reverse time migration with preconditioning guided gradient method[C].SEG Technical Program Expanded Abstracts,2013,32:3709-3715.
[20]
刘玉金,李振春.扩展成像条件下的最小二乘逆时偏移[J].地球物理学报,2015,58(10):3771-3782.LIU Yujin,LI Zhenchun.Least-squares reverse-time migration with extended imaging condition[J].Chinese Journal of Geophysics,2015,58(10):3771-3782.
TANG Y X,LEE S.Preconditioning full waveform inversion with phase-encoded Hessian[C].SEG Technical Program Expanded Abstracts,2010,29:1034-1038.
[23]
PRATT R G.Seismic waveform inversion in the frequency domain,part 1:theory and verification in a physical scale model[J].Geophysics,1999,64(3):888-901.
[24]
SHIN C,JANG S,MIN D J.Improved amplitude preservation for prestack depth migration by inverse scattering theory[J].Geophysical Prospecting,2001,49(5):592-606.
[25]
AOKI N,SCHUSTER G T.Fast least-squares migration with a deblurring filter[J].Geophysics,2009,74(6):WCA83-WCA93.
[26]
姚振岸,孙成禹,喻志超,等.融合点扩散函数的预条件黏声最小二乘逆时偏移[J].石油地球物理勘探,2019,54(1):73-83.YAO Zhen'an,SUN Chengyu,YU Zhichao,et al.Preconditioned visco-acoustic least-squares reverse time migration integrated with point spread function[J].Oil Geophysical Prospecting,2019,54(1):73-83.
[27]
HOU J,SYMES W W.Accelerating extended least-squares migration with weighted conjugate gradient iteration[J].Geophysics,2016,81(4):S165-S179.
[28]
MA Y,HALE D,MENG Z B,et al.Full waveform inversion with image-guided gradient[C].SEG Technical Program Expanded Abstracts,2010,29:1003-1007.
[29]
刘梦丽,黄建平,李闯,等.基于角度滤波成像的最小二乘逆时偏移[J].石油地球物理勘探,2018,53(3):469-477.LIU Mengli,HUANG Jianping,LI Chuang,et al.Least-squares reverse time migration based on the angular filtering imaging condition[J].Oil Geophysical Prospecting,2018,53(3):469-477.
[30]
KINGMA D P,BA J.Adam:a method for stochastic optimization[DB/OL].[2014-12-22],https://arxiv.org/abs/1412.6980.
[31]
DUCHI J,HAZAN E,SINGER Y.Adaptive subgradient methods for online learning and stochastic optimization[J].The Journal of Machine Learning Research,2011,12(7):2121-2159.
[32]
TIELEMAN T,HINTON G.Rmsprop:divide the gradient by running average of its recent magnitude//Courser A:Neural Networks for Machine Learning[EB/OL].https://open.163.com/newview/movie/free?pid=UEV4DEHJP&mid=VEV4DEIUQ.