SCEC Award Number 24200 View PDF
Proposal Category Individual Research Project (Single Investigator / Institution)
Proposal Title Benchmark Simulations for Developing, Testing, and Comparing Earthquake Location Algorithms
Investigator(s)
Name Organization
Gregory Beroza Stanford University
SCEC Milestones A1-2, A2-3 SCEC Groups Seismology, CEM, PBS
Report Due Date 03/15/2025 Date Report Submitted 03/24/2025
Project Abstract
Under this project we evaluated eight earthquake location methods (GrowClust, HypoDD, Hypoinverse, HypoSVI, NonLinLoc, NonLinLoc_SSST, VELEST, and XCORLOC) using a synthetic computational experiment on 1000 clustered earthquakes based on the setting of the 2019 Ridgecrest, California, earthquake sequence. We construct a travel-time dataset using the fast-marching method and a 3D velocity model extracted from the CVM with a a von Karman perturbation to represent small-scale heterogeneity. We introduce realistic picking errors, phase availability, and outliers to mimic difficulties encountered in seismic network monitoring. We compare the location results from eight programs applied to the same travel-time data-set and 1D velocity structure against the ground-truth locations. We find superior accuracy and precision of differential time-based location methods compared to single-event location methods. This validates the effectiveness of compensating for deviations from assumed 1D velocity structure either by path or site correction modeling or by cancellation of unmodeled structure using differential arrival times. We also find that each of these location methods underestimate true uncertainty
SCEC Community Models Used Community Velocity Model (CVM)
Usage Description We used the SCEC CVM to represent large scale-length velocity heterogeneity in the Ridgecrest, California area.
Intellectual Merit Earthquake location is foundational to much of what SCEC does. It bears on seismological efforts, on earthquake forecasting, and in the development of the CVM and CFM. It is the key technology for getting a clearer view of fault activity at depth. It is also an area of active research, so creating ground-truth benchmarks for algorithm developers is an important contribution.
Broader Impacts The project supported graduate student Yifan Yu. The project strengthened essential research infrastructure. Three of the program developers identified bugs or other shortcomings in their programs as a result of this exercise. The paper that resulted can serve as a sort of users guide for selecting and effectively running a selected location algorithm. It was also used to benchmark during development of new location algorithms developed by Weiqiang Zhu (using RANSAC)and Ian McBrearty (Graph-DD).
Project Participants Greg Beroza, Bill Ellsworth, Yifan Yu (Stanford) carried out the project. During the project we consulted with the developers of the earthquake location programs, including: Anthony Lomax (ALomax Scientific), Dan Trugman (U Nevada, Reno), Guoqing Lin (University of Miami), Zach Ross (Caltech), and Felix Waldhauser (LDEO).
Exemplary Figure Figure 1. (a–f) Depth profiles for Ridgecrest events for ground truth vs. each single-event location program’ s output. (g–k) Histograms of depth error distributions with red bars for accuracy errors and blue bar for precision errors. The x-axis represents error magnitude, and the y-axis indicates event occurrence. (after Yu et al., 2025).
Linked Publications

Add missing publication or edit citation shown. Enter the SCEC project ID to link publication.