Sampling Parametric Rupture Variability using Broadband Ground Motion Simulations

Robert W. Graves

Submitted August 15, 2018, SCEC Contribution #8660, 2018 SCEC Annual Meeting Poster #013

Observations from past earthquakes show the rupture process can be complex, exhibiting spatial variations in rupture speed, slip and slip rate, as well as geometric roughness of the faulting surface. Incorporating these features within kinematic ground motion simulations is challenging due to uncertainty in the expected ranges and inter-correlations of the required parameters. Here, I calculate suites of broadband ground motion simulations for seven large (Mw 6.0 – 7.3) California earthquakes to help constrain appropriate ranges for key rupture parameters. Simulations are computed using the Graves and Pitarka (2010) method and are compared with recorded motions using the goodness-of-fit criteria of Goulet et al (2015). Typically, stations out to a distance of one fault length were included in the analysis with the fewest being 23 sites for North Palm Springs and the most 63 for Northridge. A total of 256 rupture scenarios are examined for each earthquake, with individual scenarios consisting of random realizations generated using the approach of Graves and Pitarka (2016). In addition to spatial variability in slip and fault roughness, each realization randomly samples from uniform distributions of average rupture speed (expressed as fraction of local shear wave velocity, Vr/Vs), down-dip fault width (Fwid) and seismic moment (Mo). For each earthquake, median values of Mo and Fwid are taken from published results, and these are allowed to vary by 25% and 15%, respectively, across the random realizations. The range of Vr/Vs is set at 0.725 to 0.875. For most events, the smallest misfits occur for a limited subset of parameter combinations. In general, the results are most sensitive to average rupture speed, followed by seismic moment and then fault width. The strike-slip and oblique events are modeled best with relatively low average rupture speed (~0.75) whereas the reverse events are modeled better with a relatively high average rupture speed (~0.85). No clear trends were seen for seismic moment and fault width when looking across all events. There were also cases for individual events where different combinations of parameters across the specified ranges produced equally good fits to the observations, highlighting the non-uniqueness of this approach. This reinforces the importance of adequately sampling ranges of rupture parameters when performing validations with past earthquakes, as well as when simulating ground motions for future events.

Key Words
broadband, strong ground motion, validation

Citation
Graves, R. W. (2018, 08). Sampling Parametric Rupture Variability using Broadband Ground Motion Simulations. Poster Presentation at 2018 SCEC Annual Meeting.


Related Projects & Working Groups
Ground Motions