Simulating Millions of Years of Earthquakes in California using HPC

Varduhi Kababjyan, Shril P. Panchigar, Anthony G. Lopez, Tomoe Mizutani, Anthony A. Guerra, Scott Callaghan, Jacquelyn J. Gilchrist, Jozi K. Pearson, Gabriela R. Noriega, & Thomas H. Jordan

Submitted August 1, 2018, SCEC Contribution #8250, 2018 SCEC Annual Meeting Poster #312

As part of the 2018 Undergraduate Studies in Earthquake Information Technology (USEIT) internship program, interns worked on a Grand Challenge to evaluate how well the Third Uniform California Earthquake Rupture Forecast (UCERF3) can predict long-term rates of M ≥ 7 ruptures on the Southern San Andreas Fault. The High Performance Computing (HPC) team was tasked with simulating long catalogs of earthquakes using the Rate-State Earthquake Simulator (RSQSim), a physics-based model that simulates long catalogs of seismicity from complex fault models. The HPC team worked on the Blue Waters supercomputer at NCSA using an education allocation of 25,000 node hours.

To configure an RSQSim simulation, one must specify a series of physical parameters, including: the rate- and state-friction coefficients, the initial value of normal stress (MPa), and the earthquake slip rate (m/s). Starting with a base catalog with default parameters, each member of the HPC team generated a short test catalog (50,000-75,000 simulated years) varying different input parameters to study their effects. Using the R programming language, the team conducted analysis on the simulated data with a focus on the recurrence interval and the magnitude frequency distribution of earthquakes on several sections on the San Andreas Fault. In cooperation with the Probabilities team, the HPC team concluded that increasing slip speed by 33% and the normal stress by 25% produced the best fit with the UCERF3 model for most fault sections chosen for the study. The team then used these parameters to simulate two million years of earthquakes, the longest catalog ever simulated using RSQSim. The simulated catalogs were used by the Probabilities team to forecast earthquakes and by the Hazards and Risks Visualization team to analyze the casualties and risks of M ≥ 7 events.

Key Words
HPC, RSQSim, UCERF3

Citation
Kababjyan, V., Panchigar, S. P., Lopez, A. G., Mizutani, T., Guerra, A. A., Callaghan, S., Gilchrist, J. J., Pearson, J. K., Noriega, G. R., & Jordan, T. H. (2018, 08). Simulating Millions of Years of Earthquakes in California using HPC. Poster Presentation at 2018 SCEC Annual Meeting.


Related Projects & Working Groups
Communication, Education, and Outreach (CEO)