This award provides funding for a new Linux computer cluster for seismological research at UCLA for an early career PI. The cluster will be incorporated into the campus-wide cluster so that unused cycles can be shared. Research will involve both forward and inverse modeling. Forward modeling will focus on seismic velocity and anisotropy, and quantitative model uncertainties. Solving the non-linear inverse problems will over-challenge current computing resources. Earthscope USArray Transportable Array data will be processed using the new cluster. Future work constraining seismic anisotropy in the mantle and core will involve full spectral coupling, which is computationally demanding. Wave propagation code will also be used to test models. Three-dimensional core structure models will assist in defining an inner inner core, phase and/or compositional changes within the inner core, and defining density changes at the core boundary, ultimately furthering our understanding of the geotherm. Other study areas for the new cluster include mantle deformation geometry, plate motion-sub-lithospheric deformation coupling, asthenospheric deformation versus fossil lithospheric deformation, mid-ocean ridge spreading, surface tectonic signatures of upper mantle processes, crust, upper mantle lithosphere and the asthenosphere interactions, and thermal versus compositional nature of the mantle at various depths. The new cluster will benefit researchers from many different fields, in Earth Sciences, Astrophysics, Physics,Biology, and Computer Sciences as it will be incorporated into the existing campus-wide cluster. The cluster will provide data on basic Earth system functioning. New modeling techniques and scientific results will be published, and models will be freely disseminated over the internet. The cluster will be used for student training both in Geosciences and within other disciplines - producing a new generation of scientists with the ability to understand forward modeling and parallel computer programming. The cluster will be incorporated into the campus-wide shared cluster. UCLA Academic Technology Services (ATS) then provides maintenance and installation at no cost. ATS also provides system administration including OS maintenance and upgrade, hardware resolution and scheduling. ATS also provides training.
***
Seismic tomography, or the mapping of Earth’s three-dimensional structure in terms of seismic velocities and their directional dependence (termed seismic anisotropy), is important for our understanding of Earth’s interior and how it relates to plate tectonics and the generation of earthquakes. Fundamental progress has been made in the Earth science community during the past 20 years regarding Earth’s internal composition, its thermal state, dynamics, and evolution. However, the advancement of knowledge often brings new challenges and new questions for the next generation of scientists. For instance, the composition, rheology, and deformation regime of the mantle are not well known, and neither is the effect of deep mantle deformation on surface tectonic features. In addition, fundamental questions remain concerning the compositional and physical state of the core, and its formation history. The PI will be able to shed new light on these problems by obtaining reliable and accurate three- dimensional models of seismic wave velocities and anisotropy of the Earth’s interior. Modern seismological research has become more and more dependent on heavy computational resources. It often requires dealing with large datasets that have to be collected, archived, processed, and modeled (Beghein, Snoke et al. 2009; Stubailo, Beghein et al, under review). A good example is the massive amount of data generated by the NSF funded Earthscope project. In addition, the PI’s fundamentally new research efforts emphasize the exploration of a wide range of physical parameters that can explain seismic data, and the statistical analysis of the models obtained. This is done with a forward modeling technique (Sambridge 1999), which is best run in parallel as the CPU time increases rapidly with the size of the model space. Previous research done by the PI with this method has proven promising and successful (Beghein, Resovsky et al. 2002; Beghein and Trampert 2003; Beghein and Trampert 2004a; 2004b; Beghein, Trampert et al. 2006; Beghein, Resovsky et al. 2008; Yao, Beghein et al. 2008), but running these codes in parallel on a large number of processors is essential for the PI’s future work. The computer cluster the PI acquired with the funds provided by NSF will provide adequate resources to efficiently run these codes, which will push forward our understanding of Earth’s interior. We purchased sixteen nodes, each node having a dual-six core Intel Xeon 2.66 GHz processor with 48GB of memory and QDR Infiniband. These were selected in consultation with the UCLA Institute for Digital Research and Education, which helped obtain quotes from vendors, and which is hosting the cluster and providing administrative support and maintenance. In exchange for their support, the cluster has become part of a larger, campus-wide cluster, where each contributing group shares unused cycles with other researchers. This benefits a large number of researchers from many different fields, in Earth Sciences, Astrophysics, Physics, Biology, Computer Sciences, etc. So far, the PI has used the new cluster to obtain a model shear-wave velocities and azimuthal anisotropy in the upper mantle under the Nevada Great Basin, a region that has undergone substantial extensional deformation (manuscript under revision). In addition to enabling the PI to solve various scientific problems of interest for the scientific community, the purchased cluster will benefit her current and future students, and future postdocs. They are being trained and acquire knowledge in the fields of seismic tomography, anisotropy, inverse and forward modeling techniques, and will also be exposed to state-of-the-art geophysical software and analysis tools, and various mathematical and visualization softwares. It will give them valuable skills and experience with Linux Operating Systems, Fortran and C programming, and parallel computing with MPI. In addition, the UCLA Institute for Digital Research and Education regularly offers free high performance computing workshops on Parallel Computing.
