SCEC Award Number 18094 View PDF
Proposal Category Collaborative Proposal (Integration and Theory)
Proposal Title Tests of a New Automated Method for Remeshing the CFM for Use by Earthquake Simulators
Name Organization
Terry Tullis Brown University Michael Barall Invisible Software, Inc. Scott Marshall Appalachian State University John Shaw Harvard University Keith Richards-Dinger University of California, Riverside Christine Goulet University of Southern California
Other Participants Andreas Plesch, Harvard University
Ossian O'Reilly, USC
Ahmed Elbanna, U. Illinois, Urbana-Champaign
Alexander Breuer, UC San Diego
SCEC Priorities 5a, 3a, 2e SCEC Groups EFP, CME, CISM
Report Due Date 03/15/2019 Date Report Submitted 03/27/2019
Project Abstract
This project developed and tested a semi-automated method to convert any raw triangulated version of the CFM, into a remeshed form that is appropriate for use by earthquake simulators. We have fit the faults with B-splines by mapping the boundary of each to a rectangle, which is subsequently warped to match the outline of the non-rectangular fault. After fitting, we mesh the spline surface at any desired resolution using Cubit. We have selected nine faults with progressively degrees of complexity to test our method. We have been able to remesh the simplest of these faults, but we have yet to demonstrate the method’s ability to deal with the more difficult ones to our satisfaction. At the stage of using Cubit to fit meshes to the spline surfaces we have found that poor meshes are often created due to two types of problems: 1) severe distortion of the inherently rectangular spline surface is needed to fit a non-rectangular fault, and 2) gaps and overlaps result when two intersecting faults in the CFM do not intersect perfectly. We have an idea for a revised spline-fitting approach that is a likely solution to both these problems. It involves fitting the fault with the best fitting rectangular plane using splines and then cutting out its shape rather than warping the rectangle. We are presently unable to say whether this modified approach can provide the basis for a sufficiently automated process to envision using it to remesh the entire CFM for use by earthquake simulators.
Intellectual Merit The project explores the possibility of converting the SCEC CFM into a format that would allow the CFM faults to be used by earthquake simulators. It has developed the creative and original concept of representing the CFM faults using B-splines which allows them to be used in ways not possible with the original CFM native representations. This includes describing them with a self-consistent set of triangular meshes at any desired level of detail.
Broader Impacts One of the potential uses of SCEC’s CFM is to model the occurrence of earthquakes on this complex set of faults. This is not possible without remeshing the CFM as this proposal proposes to learn how to do. UCERF is used by CEA to help set insurance rates in California. It is likely that future versions of UCERF will increasingly be based on simulated histories of earthquake occurrence that come from earthquake simulators. The best way to be sure that earthquake simulators create realistic earthquake sequences is to compare them with real earthquake sequences. This is only possible to do for earthquakes in the magnitude 3 to 6 range because of their relatively frequent occurrence. Using the CFM as the basis for modeling small as well as large earthquakes is the best way to test and use earthquake simulators for the benefit of society.
Exemplary Figure Figure 3. Three meshes of the blind Northridge thrust fault. a) Native CFM5.2 mesh. The heart shaped part of the fault with the finer mesh was defined using relocated hypocenters and is of higher-resolution than the rest of the fault. b) Our automated mesh produced by the spline-based method and Cubit, using part a as input. c) A mesh made manually by Scott Marshall starting from a.