Fast Inversion of Gravity Data Using Parallel Preconditioned Method
Taihan Wang,Danian Huang,Guoqing Ma,Dailei Zhang,Ye Li,Wenyue Zhou
DOI: https://doi.org/10.1190/igc2017-062
2017-01-01
Abstract:PreviousNext No AccessInternational Geophysical Conference, Qingdao, China, 17-20 April 2017Fast inversion of gravity data using parallel preconditioned methodAuthors: Taihan Wang*Danian HuangGuoqing MaDailei ZhangYe LiWenyue ZhouTaihan Wang*Jilin University, ChangchunSearch for more papers by this author, Danian HuangJilin University, ChangchunSearch for more papers by this author, Guoqing MaJilin University, ChangchunSearch for more papers by this author, Dailei ZhangJilin University, ChangchunSearch for more papers by this author, Ye LiJilin University, ChangchunSearch for more papers by this author, and Wenyue ZhouJilin University, ChangchunSearch for more papers by this authorhttps://doi.org/10.1190/IGC2017-062 SectionsAboutPDF/ePub ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InRedditEmail Abstract With dramatic development of geophysical equipment and exploration technology, the precision and amount of geophysical data is increasing. Consequently, the time-consuming problem in three dimensional (3D) density inversion of large-scale gravity data needs to be considered. Preconditioned conjugate gradient (PCG) algorithm is an efficient way in fast solving linear inversion equations, which can reduce the number of inverse iterations. Considering the additional time of preparing preconditioner per iteration, we propose a parallel implementation of preconditioned conjugate gradient algorithm on CPU and GPU to overcome the calculation efficiency challenge. We adopt and test two different preconditioned strategies with noise-contaminated synthetic data to show the better performance of symmetric successive over-relaxation (SSOR) preconditioner. Higher efficiency and feasibility of proposed parallel method over conventional conjugate gradient algorithm can be seen in 3D gravity inversion results of both synthetic and field data, which lead to up to approximately 19x speed up compared to the serial code on 2.0GHz CPU. Keywords: inversion, parallel, 3D, algorithmPermalink: https://doi.org/10.1190/IGC2017-062FiguresReferencesRelatedDetails International Geophysical Conference, Qingdao, China, 17-20 April 2017ISSN (online):2159-6832Copyright: 2017 Pages: 1525 publication data© 2017 Published in electronic format with permission by the Society of Exploration Geophysicists and Chinese Geophysical SocietyPublisher:Society of Exploration Geophysicists HistoryPublished Online: 31 May 2017 CITATION INFORMATION Taihan Wang*, Danian Huang, Guoqing Ma, Dailei Zhang, Ye Li, and Wenyue Zhou, (2017), "Fast inversion of gravity data using parallel preconditioned method," SEG Global Meeting Abstracts : 234-237. https://doi.org/10.1190/IGC2017-062 Plain-Language Summary Keywordsinversionparallel3DalgorithmPDF DownloadLoading ...