Exciting news! We're transitioning to the Statewide California Earthquake Center. Our new website is under construction, but we'll continue using this website for SCEC business in the meantime. We're also archiving the Southern Center site to preserve its rich history. A new and improved platform is coming soon!
Home  /  SCEC Workshops  /  SCEC Dynamic Rupture TAG – 2020 Ingredients Workshop – Rock Properties

SCEC Dynamic Rupture TAG – 2020 Ingredients Workshop – Rock Properties

Conveners: Ruth Harris and Michael Barall
Date/Time:  October 27, 2020 (09:00-13:15 Pacific Time)
SCEC Award and Report: 20188

Figure 1 (modified from Harris et al., 2018). The "ingredients" necessary for a dynamic (spontaneous) rupture simulation include assumptions of initial stresses on the fault (and off the fault if medium is not elastic), fault geometry, off-fault materials, and a failure criterion which describes how fault friction works. These physics-based computer simulations can produce results such as patterns of fault slip, ground and sub-surface shaking, and heat generation (see Harris, 2004).

The SCEC-USGS Dynamic Rupture TAG convened "The 2020 Ingredients Workshop – Rock Properties” on October 27, 2020 via Zoom. A total of 55 people participated, including scientists from the U.S., Australia, China, England, Germany, Japan, Saudi Arabia, and Switzerland. Twenty were early career scientists (14 students and 6 postdocs). 

Four ingredients are required for dynamic earthquake rupture simulations: fault geometry, fault friction, rock properties, and initial stress conditions (Figure 1). A series of SCEC5 workshops were designed to evaluate the relative importance of each. The November 2018 SCEC workshop focused fault geometry (ingredient #1) and the January 2020 SCEC workshop focused on fault friction (ingredient #2). In October 2020, we focused on rock properties (ingredient #3). This third workshop in the series included presentations about how our choices of velocity structures and off-fault yielding affect computationally simulated earthquake rupture behavior and ground motion. One presentation introduced a new 3D rupture dynamics benchmark that simulates thermal pressurization, and showed the results from the five codes that simulated this benchmark. The workshop concluded with a group discussion of the relative importance of each ingredient evaluated thus far.

Workshop Presentations

Goals and introductions. Ruth Harris (U.S. Geological Survey) welcomed participants and briefly summarized how dynamic rupture simulations work. She noted that the SCEC-USGS Dynamic Rupture Code Verification Group has done a good job checking that code results are reproducible for a variety of assumptions about the initial stresses, fault geometry, rock properties, and fault friction (e.g., as discussed in Harris et al., SRL, 2018). Thus what is needed is a basis for choosing among the wide range of possibilities for these ingredients. Participants were asked to consider the following questions throughout the workshop: Do we need to include rock property heterogeneity at all scales? Is it sufficient to assume elastic behavior, or is plastic behavior required? In terms of the ingredients, how do the rock property effects compare in relative importance to the effects of the fault geometry, the fault friction, and the stress?

Thermal pressurization 3D benchmark and results.  Alice Gabriel (Ludwig Maximilians University, Munich) presented the new code verification exercise designed to test dynamic rupture codes’ abilities to simulate the process of thermal pressurization (thermal heating of fluids, that allows for a rapid drop in fault strength during dynamic earthquake rupture). Gabriel noted selecting parameters for the 3D thermal pressurization exercise was not simple, because most values lead to runaway rupture that does not stop (an infinite earthquake)—as discovered when the related 2D exercise was constructed by Eric Dunham and Michael Barall a number of years ago. For the 3D exercise, the goal, which was achieved, was to find values that would create a finite rupture area, while also producing simulated temperatures that are not too high (Gabriel et al., 2020). Gabriel showed the results from 5 groups of dynamic rupture modelers who used their codes to simulate the 3D thermal pressurization benchmark. Their results matched well qualitatively, as well as quantitatively using metrics developed by Barall and Harris (2015). For further information about the 3D thermal pressurization benchmark, including the detailed benchmark description, please see our group’s webpage.

Kim Olsen, Yihe Huang and David Oglesby similarly asked about the main effect of incorporating thermal pressurization versus not including it. Gabriel answered that thermal pressurization creates strong weakening and a more energetic rupture, making it hard to find a balance that allows for a well-behaved rupture which will eventually stop. Baoning Wu asked what the length scale is for the nucleation length in the simulation. Gabriel answered that it is not easy to determine, because a complex nucleation method (designed by Dunham) was used.

SCEC CVM-H. Returning to the theme of rock properties, Andreas Plesch (Harvard) presented the Harvard version of the SCEC Community Velocity Model, CVM-H (Shaw et al., 2015). CMV-H provides a 3D model of Vp, Vs, and density. Plesch noted that 3D velocity structures in the Earth, particularly in tectonically active regions, are complex and there is heterogeneity at many scales. An additional challenge is that our observations of these 3D structures vary in method, abundance, and frequency. The CVM-H model incorporates information from many observations, including surface geology, 100,000 km of 2D industry seismic data, 3D seismic surveys, and 10,000 wells (mostly available because in California companies are required to send data to the state). Plesch then described major components of CVM-H, with a particular focus on basin structures. He showed how the shallowest layers are modeled, including the geotechnical layer, and how the deepest layers are modeled, including the basement.

Ruth Harris asked if it is better to use CVM-H than another SCEC community velocity model, CVM-S, where ‘S’ stands for SCEC (https://strike.scec.org/scecpedia/CVM-S4.26). Plesch replied it depends on the application, CVM-S is an inversion to optimize travel times using many earthquakes. CVM-H also uses many other observations, so CVM-S may be better for waveform travel times, but CVM-H is a more detailed model. Phil Maechling asked what the next developments are for CVM-H, and Plesch answers included expansion to central California and also moving to a bigger scale model.

Effects of rock property structures on dynamic ruptures and ground shaking. The next three talks discussed the effects of rock property structures on dynamic ruptures and on ground shaking. Kim Olsen (San Diego State University) presented a comprehensive overview of the effects of 3D velocity structures on dynamic rupture simulations and on ground motions. He started with how velocity structures affect dynamic rupture propagation and included studies of heterogeneous large-scale structures, small-scale perturbations, and damage zones. He showed an example of a large-scale velocity variation that was included in work by Susana Custodio, Shuo Ma, and Ralph Archuleta for their simulations of the 2004 M6 Parkfield earthquake. The velocity variation had a minor effect on the dynamic rupture process, but the major effect was that vertical velocity gradient amplified slip. He also noted only minor effects from having a different velocity structure on each side of the fault (a bimaterial case). Next, Olsen presented results from studies of small-scale velocity structures by Sam Bydlon and Eric Dunham who examined 2D simulations of dynamic rupture on rough faults, using von Karman fields. While small scale velocity perturbations can arrest rupture on a planar fault, inclusion of geometrical fault roughness overwhelms the effect of the velocity structure. Olsen then presented results with damage zones (also see Yihe Huang’s talk below), and mentioned that the presence of a low-velocity zone affects rupture speeds, slip pulse shape, and rise time (for example, as shown by Ruth Harris and Steve Day, and Yihe Huang and coauthors).

Olsen next reviewed the effects of rock property structure on ground motions. He noted that the choice between implementing a 3D or a 1D velocity structure can significantly change the modeled ground motion amplitude and duration. Including stochastic small scale heterogeneity in the velocity structure, for example as modeled by Bill Savran and Olsen, helps models better fit observed ground motions at distances farther from a rupture and at higher frequencies. Olsen next discussed the effects of including anelastic attenuation Q(f), citing work by Kyle Withers, Olsen, and Steve Day. Q(f) is most important for longer wave propagation paths and has more of an effect at higher frequencies (e.g., 3 Hz or higher) than at lower frequencies. When examining inclusion of low-velocity zones and implementation of plasticity, from work by Daniel Roten, Olsen, Day, and Yifeng Cui, Olsen concluded there is a tradeoff between the two; including a low-velocity zone might increase the simulated ground motions, but including plasticity would tamp them back down to a reasonable value.

After Olsen’s presentation, Ralph Archuleta noted that plasticity is less likely to come into effect if the shear modulus is also lower in the low-velocity zone. Christine Goulet asked about the relative impact on ground motion simulations at close and medium distances, of including roughness versus including small scale rock property heterogeneity. Olsen answered that fault roughness is more important.

Next, Yongfei Wang (University of Southern California) presented his work about pulses in ground motion, their effects, potential causes, and what might prevent them from occurring. He noted that fault geometry is responsible for pulse-like ground motion in duration and amplitude, and that forward directivity can create a pulse, although plasticity can act to prevent a pulse from occurring. Wang showed that inclusion of plasticity in simulations can produce a flower structure damage pattern (e.g., the work of Shuo Ma) and can also produce lower slip near the Earth’s surface than at depth. He showed how inelastic yielding weakens the fault-normal pulse relative to its appearance in an elastic medium, and that inelastic yielding can induce near-fault saturation of Peak Ground Velocity (PGV) for large stress drops. Wang concluded that pulse-like ground motion poses a higher risk to structures, that fault geometry and directivity can lead to pulse-like motions, and that plasticity can modify this pulse-like behavior.

Following Wang’s talk, Nadia Lapusta asked if it would be better to instead use damage models, which allow for reduction in shear modulus. Wang replied that it depends on the time scale. Bounded stress can reproduce the data, but damage might need to be directly included in more complex situations. Marlon Ramos asked if there is a way to infer source processes from observations. Wang answered that one can try to diagnose crack-like versus pulse-like rupture. On a simple planar fault it might be possible to tell the difference, but in the real world, roughness, heterogeneity, and other factors make it too complicated to directly determine the difference. In some cases, (for example, the 2015 Nepal earthquake) the rupture is inferred from ground motion data to be pulse-like. In most cases it is more challenging.

In the final formal talk of the workshop, Yihe Huang (University of Michigan) presented on how near-fault low-velocity structures affect dynamic rupture and ground motion. In California, damage zones are inferred to be 100-400 meters wide, with a 25%-60% reduction in wave speed. She showed examples of how low-velocity zones change rupture dynamics, how attenuation and plasticity in fault zones change rupture dynamics, and how low-velocity sedimentary basins change ground motions. Huang concluded that low-velocity fault zones can generate slip pulses and supershear ruptures, and that the velocity structure (smooth or layered) affects the frequency dependence of ground motion. She also noted that inclusion of off-fault plastic deformation makes it feasible to compare simulation results with observations.

Group Discussion. The closing discussion revisited the questions posed at the beginning of the workshop, namely what the relative significance is among the ingredients, how important it is to include detailed velocity structure for dynamic rupture or ground motion simulations, and if it is important to include plasticity. In terms of plasticity and fault damage zones, participants thought that they might be useful to include, and important to include near-fault damage zones in simulations. One suggestion was that the decision to include plasticity might depend on the assumed fault geometry. In terms of velocity structure, all respondents noted that velocity structure details are very important for ground motions simulations. When just considering dynamic rupture simulations, some noted that it may be sufficient to implement depth-dependent structure, and perhaps ignore some of the horizontally variable features.

The goal of each of the workshops in this series is to determine the relative importance of each ingredient evaluated. As with the previous fault geometry and fault friction workshops (where those ingredients, respectively, were thought to be important), rock properties were also deemed important by participants. However, the rock properties ingredient was ranked slightly lower than the other two ingredients discussed to date, geometry and friction. This conclusion is for dynamic rupture simulations, that is studies of the earthquake source. As previously mentioned, when considering the resulting ground motions, rock properties play a more significant role.

Presentation videos may be viewed by clicking the links below. PLEASE NOTE: Files are the author’s property. They may contain unpublished or preliminary information and should only be used while viewing the talk. Only the presentations for which SCEC has received permission to post publicly are included below.

09:00 - 09:15 Introduction to the workshop (PDF) Ruth Harris
09:15 - 09:35 Self-Introductions by All Participants All
09:35 - 10:00 Thermal pressurization 3D benchmark and results (VIDEO) Alice Gabriel
10:00 - 10:30 SCEC Community Velocity Model (CVM-H) Andreas Plesch
10:30 - 11:00 Break  
11:00 - 11:30 Effects of velocity and attenuation structure on dynamic rupture and ground motions (VIDEO) Kim Olsen
11:30 - 11:50 Effects of off‐fault inelasticity on near‐fault directivity pulses (VIDEO) Yongfei Wang
11:50 - 12:05 Break  
12:05 - 12:30 How do near-fault low-velocity structures affect dynamic rupture and ground motion? (VIDEO) Yihe Huang
12:30 - 13:15 Discussion and Wrap-up All
13:15 Workshop Adjourns  

Workshop Participants

The Southern California Earthquake Center is committed to providing a safe, productive, and welcoming environment for all participants. We take pride in fostering a diverse and inclusive SCEC community, and therefore expect all participants to abide by the SCEC Activities Code of Conduct SCEC Activities Code of Conduct.

References

  • Barall, M., and R.A. Harris (2015), Metrics for comparing dynamic earthquake rupture simulations, Seism. Res. Lett., 86(1), 223-235, doi:10.1785/0220140122, SCEC Contribution 1951.
  • Gabriel, A., J.C. Vyas, T. Ulrich, J. Ampuero, and M.P. Mai (2020, 08), 3D dynamic rupture modeling with thermal pressurization. Poster Presentation at 2020 SCEC Annual Meeting, SCEC Contribution 10698.
  • Harris, R.A. (2004), Numerical simulations of large earthquakes: dynamic rupture propagation on heterogeneous faults, Pure and Applied Geophysics, 161(11/12) 2171-2181, doi:10.1007/s00024-004-2556-8, SCEC Contribution 830.
  • Harris, R.A., M. Barall, B. Aagaard, S. Ma, D. Roten, K. Olsen, B. Duan, B. Luo, D. Liu, K. Bai, J.-P. Ampuero, Y. Kaneko, A.-A. Gabriel, K. Duru, T. Ulrich, S. Wollherr, Z. Shi, E. Dunham, S. Bydlon, Z. Zhang, X. Chen, S.N. Somala, C. Pelties, J. Tago, V.M. Cruz-Atienza, J. Kozdon, E. Daub, K. Aslam, Y. Kase, K. Withers, and L. Dalguer (2018), A suite of exercises for verifying dynamic earthquake rupture codes, Seism. Res. Lett., 89(3), 1146-1162, doi:10.1785/0220170222, SCEC Contribution 7960.
  • Shaw, J.H., A. Plesch, C. Tape, M. Suess, T.H. Jordan, G. Ely, E. Hauksson, J. Tromp, T. Tanimoto, R. Graves, K. Olsen, C. Nicholson, P.J. Maechling, C. Rivero, P. Lovely, C.M. Brankman, & J. Munster (2015). Unified Structural Representation of the southern California crust and upper mantle. Earth and Planetary Science Letters, 415, 1-15, doi:10.1016/j.epsl.2015.01.016, SCEC Contribution 2004.