Computer simulations help researchers understand the range of possible outcomes of a rupture when key details of the San Andreas Fault are still unknown.
By: Shi En Kim (@goes_by_kim)
Citation: Kim, S. E., 2020, Fault geometry plays a role in San Andreas ruptures, Temblor, http://doi.org/10.32858/temblor.107
The San Andreas Fault runs more than 800 miles (1,200 kilometers) north to south through the state of California. The fault is responsible for some of California’s most notorious earthquakes such as the 1906 San Francisco earthquake. While the central and northern segments of the fault have seen recurring earthquakes as frequently as every 22 years, the southern segment of the San Andreas Fault in Coachella Valley has not had a major shakeup in the last three centuries — far past its recurrence period of 180-200 years. The segment’s proximity to the Los Angeles Basin worries residents that an inevitable rupture will be deadly and costly, affecting millions of lives.
Studies have shown that the southern segment has accumulated sufficient stress to trigger a magnitude-7 earthquake. One team of researchers, led by seismologist Roby Douilly of the University of California, Riverside, is studying various factors, such as fault geometry, to model the evolution of a not-so-hypothetical future earthquake on the San Andreas’s southernmost segment. As reported in Geosphere, the researchers altered the intersection angles and directions of five rupture-prone strands branching from the main San Andreas Fault line to determine the most likely rupture pathway.
The fault in our data
Though the San Andreas is one of the most well-mapped faults in the world, it is still not yet fully understood. The exact configuration of the intertwining fault branches, the pent-up stress levels and the friction laws between the tectonic plates are still open questions. Parts of the San Andreas Fault extend 10 miles (16 kilometers) underground, deeper than our current mapping and seismic methods can plumb. With an incomplete picture of the San Andreas Fault, scientists have to rely on computer simulations to estimate the range of possible rupture outcomes, such as which fault branches will slide and how the seismic energy will be released. Enter Douilly.
Divide and conquer
Douilly and his colleagues used computerized simulations called finite element analyses to simulate how a rupture would propagate from the Coachella Valley segment of the San Andreas, should that segment slip. The method involves constructing a complex 3D geometric model from smaller “discretized” units called elements, similar to how a digital photo is made up of pixels.
“[Our model] has about 47 million elements,” says Douilly. “For a 40-second [rupture propagation] simulation, it takes about four days to run on 12 CPUs.” He estimates that he has run almost 150 simulations over the span of two years for this project.
The researchers started with three previously reported geometric models of the southern San Andreas segment, assumed a simplified version of other parameters (namely the friction and initial stress conditions of the ground before the rupture), then set off earthquakes. The researchers watched the dynamic rupture process play out while paying close attention to the evolution of the displacement, speed and stress distribution.
A slight change in the parameters of the fault could produce drastically different outcomes, the researchers found, with propagations running along different types of faults, including the more dormant faults. Tweaking the effects of the model’s various geometric parameters one-by-one, then scrutinizing each result, the researchers reported a catalog of scenarios. Over the hundreds of scenarios run, it became clear that while the outcome can span a wide range of possibilities, fault geometry plays an important role. But interestingly, in many of the scenarios, the rupture crept northward into the Mission Creek fault strand from the San Andreas, the team reported.
One aspect of the complexity
Tempting as it may be to try, this research is not meant to foretell the exact outcome of a future rupture in the San Andreas Fault, Douilly says. This is because the simulations are still not detailed enough to be representative of the real world. Currently, the model relies on simplified descriptions of many different parameters other than the fault geometry, which still have yet to be confirmed by fieldwork studies. Instead, Douilly says, this work serves as an incremental step toward understanding the roles of specific factors that influence the outcome of an earthquake.
“Earth is so complicated. It’s really difficult to capture all the complexities that actually exist in a single computational model,” says Oliver Stephenson, a graduate student in geophysics at Caltech who was not involved in the study. For example, there are even more complex descriptions of the friction between the tectonic plates, as well as the stress distribution in the ground than what Douilly’s model had used.
“This work is a useful indicator of the impact of one area of complexity: the geometry of the fault,” says Stephenson. “One day we want to be able to combine all of the complexities within a single model.”
This is exactly what Douilly is aiming for in the future. For starters, Douilly wants to account for past earthquakes in his model. Previous earthquakes may redistribute the stress or reset the recurrence clock by relaxing the stresses in the ground, like the spent tantrum of a child. Douilly plans to examine the significant regional earthquakes of the past to fine tune the “initial stress condition” input in his model and hence the accuracy of its simulations.
A major earthquake along the San Andreas Fault looms, as portended by the 2019 Ridgecrest quakes in the Southern California desert. Seismologists such as Douilly are rushing to parse together different pieces of the puzzle that is the next big rupture in the San Andreas Fault.
Concerned about your earthquake risk? Check it at Temblor.
Further reading
Douilly, R., Oglesby, D. D., Cooke, M. L., Hatch, J. L. (2020). Dynamic models of earthquake rupture along branch faults of the eastern San Gorgonio Pass region in California using complex fault structure. Geosphere, 16 (2): 474–489. doi: https://doi.org/10.1130/GES02192.1
Fialko, Y. (2006). Interseismic strain accumulation and the earthquake potential on the southern San Andreas fault system. Nature 441, 968–971. https://doi.org/10.1038/nature04797