Frequency Domain Least-Squares Reverse Time Migration with a Modified Scattering-Integral Approach

Jizhong Yang,Yuzhu Liu,Liangguo Dong

DOI: https://doi.org/10.1190/segam2015-5909496.1

2015-01-01

Abstract:PreviousNext No AccessSEG Technical Program Expanded Abstracts 2015Frequency domain least-squares reverse time migration with a modified scattering-integral approachAuthors: Jizhong Yang*Yuzhu LiuLiangguo DongJizhong Yang*State Key Laboratory of Marine Geology, Tongji UniversitySearch for more papers by this author, Yuzhu LiuState Key Laboratory of Marine Geology, Tongji UniversitySearch for more papers by this author, and Liangguo DongState Key Laboratory of Marine Geology, Tongji UniversitySearch for more papers by this authorhttps://doi.org/10.1190/segam2015-5909496.1 SectionsAboutPDF/ePub ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InRedditEmail Abstract Least-squares reverse time migration (LSRTM) is an iterative inversion-based imaging process, which can suppress migration artefacts, balance reflector amplitudes, and enhance spatial resolution compared with standard reverse time migration (RTM). The widely accepted approach for gradient calculation in LSRTM is the adjoint-state method, where the forward propagated source wavefields are zero-lag cross-correlated with the back-propagated receiver wavefields. In this study, we propose a practical alternative for implementing frequency domain RTM (FRTM) and LSRTM (FLSRTM) with a modified scattering-integral approach. The gradient of the misfit function with respect to model parameters is formulated as matrix-vector products, which are calculated with vector operations through the accumulation of decomposed vector-scalar products without explicitly building the kernel matrix. With such a matrix-based scattering-integral approach, we can guarantee that the migration operator is the exact adjoint (conjugate transpose) of the forward Born modeling operator while with reduced memory requirement. At the meantime, the Gauss-Newton normal equation is iteratively solved with a matrix-free conjugate gradient algorithm. The Hessian-vector products are calculated based on a demigration/migration procedure, which are implemented with two nested kernel-vector products. Thus, no additional forward simulations are required any more. The numerical example proves the feasibility of our method. Keywords: frequency-domain, least squares, reverse time migration, wave equationPermalink: https://doi.org/10.1190/segam2015-5909496.1FiguresReferencesRelatedDetailsCited byDensity reconstruction in multiparameter elastic full-waveform inversion8 November 2017 | Journal of Geophysics and Engineering, Vol. 14, No. 6Simultaneous estimation of velocity and density in acoustic multiparameter full-waveform inversion using an improved scattering-integral approachJizhong Yang, Yuzhu Liu, and Liangguo Dong26 September 2016 | GEOPHYSICS, Vol. 81, No. 6Least-squares reverse time migration in the presence of density variationsJizhong Yang, Yuzhu Liu, and Liangguo Dong12 October 2016 | GEOPHYSICS, Vol. 81, No. 6 SEG Technical Program Expanded Abstracts 2015ISSN (print):1052-3812 ISSN (online):1949-4645Copyright: 2015 Pages: 5634 publication data© 2015 Published in electronic format with permission by the Society of Exploration GeophysicistsPublisher:Society of Exploration Geophysicists HistoryPublished Online: 19 Aug 2015 CITATION INFORMATION Jizhong Yang*, Yuzhu Liu, and Liangguo Dong, (2015), "Frequency domain least-squares reverse time migration with a modified scattering-integral approach," SEG Technical Program Expanded Abstracts : 4164-4169. https://doi.org/10.1190/segam2015-5909496.1 Plain-Language Summary Keywordsfrequency-domainleast squaresreverse time migrationwave equationPDF DownloadLoading ...