Refinement of Near-Surface P and S Velocities in the SCEC 3D Velocity Model Using 3D Waveform Modeling
Tracy Pattelena
UC Santa Cruz
tracyrxs@cats.ucsc.edu
Mentor: Kim Olsen, UCSB
Active-source industry data were
processed using the tomographic velocity inversion method of Hole
(1992) to create three P wave velocity models at 50 m grid spacing
for the upper 500 m of crust in the Northridge epicentral region
of Southern California's San Fernando Valley (SFV) (Pattelena
et al., 1998). These profiles are named SFV-11, SFV-08, and SFV-12
(Fig. 1). Unusually slow P wave velocities were found along all
three profiles ranging from 900 to 2600 m/s (Fig's. 2 and 3).
Additionally, one of the three profiles, SFV-12, had S wave arrivals
with a resolution sufficient for an S wave velocity model at 50
m grid spacing for the upper 300 m of crust to be generated. Unusually
slow S wave velocities were found ranging from 300 to 900 m/s
(Fig. 3). From this profile, we were able to calculate Vp/Vs,
finding a variable Poisson's ratio of 0.2 near the free surface
to greater than 0.4 throughout most of the model (Fig. 3). These
high-resolution 2D models could provide a valuable constraint
on the SCEC 3D Velocity Model in the SFV where control on the
near-surface S wave velocity, a critical parameter for accurate
prediction of strong ground motion, is mostly indirect and in
many areas not well constrained.
We define a Southern California Earthquake
Center (SCEC) summer internship project that will compare the
near-surface velocities in the tomographic profiles to those in
the SFV portion of the SCEC 3D Velocity Model, Version 2, (SCEC
3D), as well as the ground motion response using the two different
models. The primary goal of these comparisons is to outline any
differences, and thereby potentially improve ground motion estimates
in the SFV. For the method of analysis we use both 2D and 3D fourth-order
staggered-grid visco-elastic finite-difference modeling to generate
synthetic wave propagation. We then compare the accuracy of the
seismic response in terms of amplitude of the SCEC 3D and the
tomographic models against data for Northridge aftershock events.
In addition, we constrain the anelastic attenuation in the near-surface
material by trial and
error of different Q values in the 3D model.
Table of Contents
Model Location and Selection of Northridge Aftershock Events and Stations Used for the Modeling
Isosurface for V(s) = 0.5 km/s
Isosurface for V(s) = 1.0 km/s
S wave Velocites from Tomographic Model SFV-12
P wave Velocites from Tomographic Model SFV-12
Effect of Model Features and Approximations
Shallow vs. Deep 3D SCEC Subset Model
3D Wave Propagation Simulating the M(l)=5.1 Event
Seismogram Plots of 3D Synthetics vs. Data
3D SCEC Subset with Q vs. Data
Combination Model No Q vs Data