"Upgrading computational facilities for URI Seismology and Marine Geophysics"

点击翻译按钮获取中文摘要

ABSTRACTOCE-0727919Intellectual Merit. The project will support the computational research of an expanding group of University of Rhode Island young investigators in marine seismology and geophysics. The facility will comprise a 100-processor-core, 25 node Linux Beowulf cluster with associated 30 TB RAID disk array. The provision of this infrastructure will initially facilitate significant advances in finite-frequency seismic tomography, an effort that will enable the group to achieve more accurate images of the Earth's interior structure. A new theory of propagation of real, finite-frequency waves will be applied, and finite-frequency sensitivity kernels will be calculated to determine the degree to which seismic traveltime and seismic amplitude anomalies in seismic data sets are sensitive to changes in 3D Earth structure. The improved computational resources and associated implementations of algorithms using the aforementioned theoretical developments will enable to group to solve nonlinear tomographic problems without present-day high frequency or structural limitations. In addition to the improved 3D tomographic imaging capability for the current URI group of investigators, faculty new to URI (Savage) depend on high performance computational clusters to calculate 3D synthetic seismograms, and to apply these methods to practical problems such as the creation of a model of seismic velocity, attenuation and anisotropy of the Gulf of California (supported through the NSF Margins Program). Savage further will use this cluster to continue the development and improvement of a widely-used multi-processor program (Specfem3D). Zhao has recently joined the URI group, where he will use the cluster to continue work on the scattering-integral method for calculating finite-frequency sensitivity kernels. Zhang will generate 3D models on the cluster with irregular surfaces, seafloor topography, anisotropy and seismic heterogeneity. These newly-joined URI group members have all been carrying out their calculations using machines at other institutions/countries equivalent to that to be established through this project. Thus the establishment of this class of resource at URI will provide the team with the necessary resources to carry out their core research.Broader impacts. The computational cluster will enhance URI's research infrastructure by providing a modern, high-performance supercomputer with advanced 3D graphics capabilities. The facility will continue to be used in research and undergraduate and graduate education. Results from simulations of wave propagation and mantle convection will continue to be integrated into geophysical courses to help students to gain physical intuition. A graduate course on computational geophysics will use the proposed facility. The facility is essential for graduate students to participate in research projects.