SCEC Award Number 19014 View PDF
Proposal Category Individual Proposal (Data Gathering and Products)
Proposal Title Assimilating SSIP data into a Full 3D Tomography (F3DT) model of the Salton Trough
Investigator(s)
Name Organization
Patricia Persaud Louisiana State University
Other Participants Rasheed Ajala (PhD student at Louisiana State University)
Alan Juarez (No-cost collaborator; PhD student at University of Southern California)
Dr. Gboyega Ayeni (No-cost collaborator; Supervisor, Geophysical Processing and FWI Applications at ExxonMobil)
SCEC Priorities 4a, 3a, 2b SCEC Groups Seismology, CXM, GM
Report Due Date 03/15/2020 Date Report Submitted 11/14/2019
Project Abstract
 To improve earthquake hazard assessments in the Salton Trough, we must accurately forecast strong ground motions through realistic 3D earth models. Recently, we have utilized active-source data from reflection and refraction seismic experiments in the Salton Trough such as the 2011 Salton Seismic Imaging Project to produce 3D travel time velocity models for Coachella and Imperial valleys (Ajala et al., 2019; Persaud et al., 2016). Comparisons of our models with two popular community velocity models reveal significant differences in basin geometry and crustal heterogeneity. To investigate the accuracy of the models in ground motion prediction, we registered our velocity models into the SCEC Unified Community Velocity Model software framework. This facilitates the construction of hybrid velocity models and mesh generation.
Using the SPECFEM3D Cartesian package , we have simulated ground motions for a M5.2 earthquake in four velocity models with geotechnical layers in the top: (1) our travel time velocity models embedded in CVM-H15.1, (2) our travel time velocity models embedded in CVM-S4.26, (3) CVM-H15.1 only, and (4) CVM-S4.26 only. Preliminary results show significant improvement in waveform misfit, especially in the sedimentary basins when our travel time velocity models are embedded into CVM-H15.1, compared to CVM-H15.1 alone. This highlights the importance of active-source data in developing accurate crustal models. Ongoing work involves reducing waveform misfit due to inaccuracy from the model and mesh construction. The best model will subsequently be improved by incorporating existing active-source data in a full-waveform inversion to develop an improved crustal model for the Salton Trough.
Intellectual Merit Our model has practical significance for improving the accuracy of earthquake hazard studies by providing a more accurate seismic velocity model for the region surrounding the southern San Andreas fault system. The accuracy of ground motion estimates strongly depends on the seismic velocity structure, especially the basin structure which is key in determining shaking intensity (Lee et al., 2014). Fault geometry and earth models with realistic material properties are key ingredients of dynamic rupture simulations of the earthquake process (Barall and Harris, 2015).
Data from the 2011 Salton Seismic Imaging Project offers a new opportunity to improve the existing CVMs with active source data that lack the standard source location and origin time uncertainties of earthquake-only data sets and densely sample the upper crust where earthquakes are sparse. For example, using SSIP data, Persaud et al. (2016) identify concealed faults in Imperial Valley that are not associated with mapped surface faults and currently do not exist in the CFM but are aligned with well-defined seismicity lineaments. Similarly, Ajala et al. (2019) show an irregular basement structure in Coachella Valley that will result in different ground shaking estimates than for a symmetric basin and regular basement structure in current regional community velocity models used in seismic hazard analysis for Southern California. Incorporating SSIP data into the existing large-scale CVMs is thus a crucial next step in creating the next generation of CVMs that can then be improved through full waveform inversion.
Broader Impacts This award has supported a PhD student and the research program of an early-career tenure-track faculty. A manuscript was published with the 3-D velocity model for the Coachella Valley (Ajala et al., 2019). Preliminary project results showing the hybrid model and wave simulations were presented at the 2019 SCEC Annual Meeting.
We have made our wavefield simulations publicly available at our LSU research webpage (https://www.geol.lsu.edu/persaud/Data.html). We have provided our 3-D velocity model as well as the derived basement and estimated Z2.5 surfaces important for seismic hazard assessment to the organizers of the SCEC CVM for incorporation into the Community Modeling Environment. These files are also publicly available for download at our LSU research webpage. All final models will be made available to SCEC community in UCVM.
Exemplary Figure Figure 2: Preliminary unstructured hexahedral mesh of the Salton Trough region showing P wave velocities constructed from a combination of our travel time velocity models and CVM-S4.26. The mesh is developed using the internal meshing program of the SPECFEM3D Cartesian package and includes topography and Moho interfaces, and two refinement layers at 10 km below sea level and the Moho. The mesh contains a total of 1,958,400 elements, with the smallest and largest element size being 532 m and 5300 m, respectively. Each element is sampled with five grid points along each dimension, i.e., 125 grid points in each element. Red lines are surface traces of mapped faults in the region.