We use the 3D finite‐element method to conduct dynamic models of rupture and resulting ground motion on the Claremont–Casa Loma stepover of the northern San Jacinto fault. We incorporate complex fault geometry (from the U.S. Geological Survey [USGS] Quaternary Faults Database; see Data and Resources), a realistic velocity structure (the Southern California Earthquake Center Community Velocity Model‐S), a realistic regional stress field with an orientation taken from seismicity relocation literature, and several stochastic self‐similar shear stress distributions. As we incorporate more types of complexity, the specific effects of any individual factor become less apparent within the overall rupture behavior. We also find that the distribution of high and low shear stress that arises from combining regional and stochastic stress fields has the strongest control over where the rupture terminates. Using a regional stress field alone, as well as with the combined regional and stochastic stress realization, we find that the stepover presents a significant barrier to rupture, regardless of our choice of initial nucleation point and that it is difficult for rupture to propagate the full length of either fault segment. Greater heterogeneity of stresses tends to produce shorter ruptures. Within this result, we find that the Claremont strand is more favorable for long ruptures than the Casa Loma–Clark strand. Low‐frequency ground‐motion intensity and distribution are controlled largely by the velocity structure rather than by stress heterogeneity. The strongest motions produced in these models are in the San Bernardino basin. Although directivity effects do contribute to the low‐frequency ground‐motion distribution, particularly in the near field, they are secondary to the effects of the velocity structure.
Online Material: Figures of ground motions from models used to calibrate the stress conditions for dynamic rupture propagation.