Annual Meeting - SCEC.org

Report | SCEC Research Accomplishments
Figure 66. Comparison of observed and simulated three-component
ground velocity waveforms for selected sites in the Los Angeles basin
region for one rupture model (same scenario as in FIGURE_GMP5). All
waveforms have been low-pass filtered at f < 0.5 Hz. For each site
waveforms are scaled to the absolute maximum amplitude observed at
that site; the peak amplitude (in cm/s) for each component (simulated or
observed) is indicated at the top right of the observed trace, and the
waveform correlation coefficient computed for the simulated and
observed motion is at left edge of the synthetic traces.
Figure 67. Los Angeles basin excitation sensitivity for slip
perturbations to the ShakeOut rupture scenario that are
parameterized by location (distance along SAF from NW to SE
along the SAF) and rupture velocity. Curves are solid for rupture
from SE to NW, dashed for rupture from NW to SE.
84 | Southern California Earthquake Center
improvement. Simulations of earthquake rupture on the
southern San Andreas fault (SAF) reveal large
amplifications in the San Gabriel and Los Angeles basins
(SGB and LAB) apparently associated with long-range
path effects. Day et al. have developed a method for rapid
calculation of the sensitivity of such predicted wave-field
features to perturbations of the source kinematics, using a
time-reversed (adjoint) wave-field simulation. The
calculations are analogous to those done in adjoint
tomography, and the same time-reversed calculation also
yields path-sensitivity kernels that give further insight
into the excitation mechanism. For rupture on the
southernmost 300 km of SAF, LAB excitation is greatest
for slip concentrated between the northern Coachella
Valley and the Transverse Ranges, propagating to the NE,
and with rupture velocities between 3250 and 3500 m/s
along that fault segment (Figure 67 and Figure 68). That is
within or slightly above the velocity range (between
Rayleigh and S velocities) that is energetically precluded
in the limit of a sharp rupture front, highlighting the
potential value of imposing physical constraints (such as
from spontaneous rupture models) on source
parameterizations. LAB excitation is weak for rupture to
the SW and for ruptures in either direction located north
of the Transverse Ranges, while the Ventura Basin is
preferentially excited by NE ruptures situated north of
the Transverse Ranges. Olsen et al. have continued to
analyze the results from the 0-2 Hz simulation of a Mw 8
San Andreas rupture (see also SPECIAL PROJECTS,
Community Modeling Environment). The 3-D
simulations include strong directivity, rupture
complexity, and basin effects resulting in a much more
complex spatial variation in shaking compared with
empirical ground-motion prediction equations that
estimate median levels of shaking across an ensemble of
events (Figure 69). At rock sites (Vs30 > 1.0 km/s) the
synthetic ground motions exhibit a similar decay in peak
ground velocity with distance from the rupture as the
Boore and Atkinson (2008) and Campbell and Bozorgnia
(2008) ground-motion prediction equations (Vs30 = 760
m/s). In another study Olsen et al. have started
examining the effects of including small-scale stochastic
variations in elastic properties in seismic velocity models.
Preliminary results indicate that weak fractal spatial variations can generate spatial variations in peak velocity (both
amplification and de-amplification) by up to a factor of six. Figure 70 shows snapshots for a purely elastic (infinite Q), 0-2 Hz
simulation of a horizontally incident SH plane wave propagating through a velocity structure containing fractal
inhomogeneities having a Hurst exponent of 0 and a standard deviation of 10%. This structure is imposed on a background 1D
velocity profile taken from the SCEC CVM-S4 for a typical Los Angeles basin site. Ongoing work will examine the tradeoff
between scattering associated with this stochastic variation and intrinsic attenuation associated with anelastic attenuation.