3D elastic locating of microseismic sources: a comparison of GPU and CPU implementation

详细信息    查看全文 | 下载全文 | 推荐本文 |

• 作者：Qingfeng XueYibo WangXu Chang
• 会议时间：2014-10-19
• 关键词：Microseismiclocating ; reverse modeling ; GPUaccelerate
• 作者单位：Qingfeng Xue(Institute of Geology and Geophysics,Chinese Academy of Sciences,Beijing 100029,China;University of Chinese Academy of Sciences,Beijing 100049,China)Yibo Wang,Xu Chang(Institute of Geology and Geophysics,Chinese Academy of Sciences,Beijing 100029,China)
• 母体文献：2014年中国地球科学联合学术年会论文集
• 会议名称：2014年中国地球科学联合学术年会
• 会议地点：北京
• 主办单位：中国地球物理学会
• 语种：chi

摘要

Microseismic events locatingis of utmost importance in seismic exploration,especially in the exploitation process of the unconventional oil and gas resources.Traditional seismic locatingmethods often require clear vertical and horizontal wave travel time information.In this paper,we adopt an approach which use the observed wave field to reversely propagate in timeand then considered the wave fieldas the boundary value for the reverse modeling.Assuming velocity model is accurate,the reversely modelled wave field focuses on the hypocenterof the microseismicevent.In addition,we focus on using the GPU to accelerate the computation of key kernels.Our 3D GPU-based elastic locatingmodeling tests achieved a 20x speedup relative to an OpenMP CPU implementation run on a four-core machine.These GPU-based speedup improvementsallows us to model 3D elastic locating of microseismic eventsat lower hardware cost and with fewer totalcompute resources than heretofore possible.