Determining the shear wave vel
ocity of
the
Los Angeles
Basin
Matthew Wong and Yuping Wang
Mark Keppel High School
Alhambra, California, 91801
March 31, 2024
Abstract
We use Rayleigh waves to determine the shear wave velocity of
a portion of the Los Angeles
region. Correlations were constructed for two sets of six stations aligned relatively in the north
-
south and east
-
west direction, respectively. The arrival time of surface waves was estimated
between the stations which was the
n used to determine the group velocity for different sets of
frequencies, which we plotted on a dispersion curve to visualize the velocities the wave travels at
different frequencies. Lastly, the dispersion curve was converted into an estimate of the shear
wave velocity as a function of depth.
Introduction
The Los Angeles Basin has millions of residents with a significant earthquake risk. It is
deeper than the other basins in the region such as the San Gabriel basin and the San Fernando
Basin, and has faults within the basin area and is adjacent to the San A
ndreas fault, which is
capable of an 8+ magnitude earthquake [1]. The basin itself is composed of several sedimentary
layers that enhance the level of strong
-
motion shaking during an earthquake [1]. The shape
(mainly depth) of the basins as well as th
e shear strength of the sedimentary material are the
important factors that control the level of amplification of ground motions from earthquakes [1].
As a proxy for the strength of the material, we use the shear
-
wave velocity.
To measure the shear velocity, we use Rayleigh surface waves. These waves travel along the
surface of the earth between the earthquake and the receiver [2]. In the study we present here, we
use Rayleigh waves generated by the cross
-
correlation of the
seismic recordings at two stations.
This method uses the ambient noise generated primarily by the ocean waves breaking on
California’s coast [1]. Fortunately in Los Angeles, this is a strong signal. This method obviates
the need for expensive and environme
ntally disturbing active sources such as large vibrator
trucks or explosions [3].
By cross
-
correlating different stations, waveforms can be detected and their travel times can
be measured, which can be used to determine their velocities. Understanding the velocities of
these waves is crucial to determining the level to which strong
ground motions are amplified.
Specifically, shear wave velocities are important because when the shear wave velocity is higher,
the material is stiffer, and less ground shaking will occur. Velocity is dependent on the medium
in which it travels through, a
nd shear wave velocity is lower in soft soil than in hard soil or rock.
Method and Analysis
Open
-
source code was modified and used to retrieve seismic data from the cloud portion of
the Southern California Earthquake Data Center (SCEDC), rather than using the webservices of a
Jupyter Notebook because the latter had significant data dropouts [5].
Cross
-
correlation utilizes
two seismic stations that are running synchronously, and enhances the seismic signals picked up
by both stations over the same period. It filters unnecessary noise that will not be used, allowing
us to have a clearer image w
hen picking points on a seismogram. All seismic stations chosen for
cross
-
correlation in this study are from the Southern California Seismic Network (SCSN). We
use the vertical component (BHZ channel) for the correlations because this channel is
particularly sensitive to Rayleigh waves.
We chose 6 stations in the north
-
south geographical orientation and 6 stations in the east
-
west
geographical orientation. The stations in the north
-
south geographical orientation are especially
important because ocean waves hit southern Los Angeles fr
om the south [], and thus these
stations will pick up signals from the ocean the best. One station in the middle of all the stations
aligned in a linear direction is chosen to be correlated with the other stations and act as the
virtual vertical source tha
t radiates waves to the other stations (See Figure 1 for a map showing
the locations of our stations chosen). In our study, station USC was chosen as the station for both
the north
-
south and east
-
west measurements. The correlation data was then filtered fr
om 0.05
-
0.15 Hz, 0.15
-
0.25 Hz, 0.35
-
0.45 Hz, etc. increasing in intervals of 0.1 Hz up to 1.45
-
1.55 Hz to
target the data for a specific range of frequencies, i.e.: the interval 0.05
-
0.15 Hz was used to
visualize the frequency of 0.10 Hz. A seismogram is t
hen generated to show the cross
-
correlation
results between two stations at that frequency, and we plot those seismograms near each other on
a distance vs time graph, with the source station being set at a distance of 0 meters. An example
is shown in Figur
e 2.
Figure 1
-
Map of stations (East
-
West and North
-
South)
These are maps indicating the locations of the
seismometers we used for the cross
-
correlations. The first image details the seismometers used for the east
-
west
correlations, while the second image details the seismometers used for the north
-
south correlati
ons.
Figure 2
-
Correlations across stations (0.35
-
0.45 Hz)
This is the plot of all the correlations between each of the
east
-
west and north
-
south stations from 0.35
-
0.45 Hz. Here, we have drawn points on each of the peaks of the
perturbations to visualize how we chose them. The arrival time is measured from the x
-
axis of the peak, and the
distance between the stations is measured directly from their coordinates.
Figure 3
-
Correlations across stations (0.75
-
0.85 Hz)
This is the plot of all the correlations between each of the
east
-
west and north
-
south stations from 0.65
-
0.75 Hz, where the peaks of the arrivals are difficult to discern to
because there are multiple waves at some stations. From the top to the bottom of
the graph, the east
-
west stations
were aligned from east to west geographically, while the north
-
south stations were aligned from north to south
geographically.
This allows us to see the velocities by calculating it through the time the station sees the
perturbation and the distance the station is from the source station. In frequency intervals where
the perturbations were difficult to discern, the data was n
ot used, assuming it was an outlier
caused by a scattered wave. Figure 3 is below showing an example when perturbations were
difficult to discern. While there is a general trend we can see, it was not enough for us to
accurately pick the peaks.
Figure 4.1
-
Dispersion Curve (East
-
West)
This is the
dispersion curve with the error bars plotted with the
east
-
west stations. Each point represents a different set
of correlations.
The velocities that each station had when the perturbations were received were then averaged for
that specific frequency; when all the frequency sets with viable data had velocities averaged, we
plotted the velocities on a velocity versus frequency graph,
known as a dispersion curve, which
allowed us to visualize how velocity changes as the frequency changes. The standard deviation
was calculated to plot the error bars, and the dispersion curves demonstrated a downward trend,
with decreasing velocities as t
he frequency increased. This is explained by the fact that the
further down you go on Earth, the more compact the medium becomes, which decreases velocity
[]. Based on the equation v=
f • λ, because velocity is only affected by the medium in which it
propagates through, wavelength decreases as you go deeper into earth, which then causes the
frequency of the waves to increase.
Figure 4.2
-
Dispersion Curve (North
-
South)
This is
the dispersion curve with the error bars plotted with the
north
-
south stations, with each point representing a
different set of correlations.
Figure 5
-
Phase Velocity
These are the phase velocities shown for the east
-
west and north
-
south waveforms. It is
shown that as the depth increases, the velocity increases as well. The blue plus
-
signs represents the observed trend
from the dispersion curve (based on the points we p
icked in Figure 4) and the red plus
-
signs represents the
modeled trend from Geogiga’s software. The thicker blue line is the velocity determined from points we picked,
while the red line shows velocities based on the modeled trend. The gray area shows the
range of velocities that
the velocity could have been or the standard deviation.
Discussions and Conclusion
Finally, the dispersion curve was transformed into a shear wave velocity versus depth graph,
which shows the velocity of the wave at different depths, which was done using the "Surface"
software from the Geogiga Technology Corporation [6].
To conclude, our data shows that the shear wave velocities in the Los Angeles basin travel up
to around 1500 m/s in both the east
-
west and north
-
south direction when the depth exceeds about
1500 m. And as the depth increased, the velocity of the waves
increased as well. In the east
-
west
graph, the reference model’s velocity was slower than the observed trend up until about 970 m in
depth. Afterwards, the reference model’s velocity is greater than the observed trend. For the
north
-
south graph, the refer
ence model’s velocity was slower than the observed trend until about
1560 m in depth. Afterwards, the reference model’s velocity is greater than the observed trend.
We believe that these differences are simply results of the algorithm rather than an actual
trend in
the subsurface.
Knowing these shear wave velocities are crucial because shear wave velocities are important
for site amplification of strong motion waves. We expect that when the San Andreas fault rupture
occurs or when other earthquakes in southern Los Angeles happe
n, the shear wave velocity will
be the main determining factor on the level of shaking. These results are helpful when
earthquake scientists prepare for upcoming earthquakes in southern California like the long
-
expected San Andreas fault rupture.
Acknowledgements
The authors thank Robert Clayton of the Caltech Seismological Laboratory who provided his
expertise and support with this research. The authors thanks Yan Yang and James Atterholt, both
PhD students at the Caltech Seismological Laboratory, for their a
ssistance with the code used for
cross
-
correlating seismic stations. The authors thank Michael Gurnis of the Caltech
Seismological Laboratory for introducing this research opportunity.
References
1. Schulz, S. S., & Wallace, R. E. (n.d.). The San Andreas Fault.
https://pubs.usgs.gov/gip/earthq3/safaultgip.html
2. Surface Waves
. Michigan Technological University. (n.d.).
https://www.mtu.edu/geo/community/seismology/learn/seismology
-
study/surface
-
wave/
3.
(
2013, April 24).
Development of a low Cost Method to Estimate the Seismic Signature of a
Geothermal Field
[Presentation].
https://www.energy.gov/eere/geothermal/articles/develpment
-
low
-
cost
-
method
-
estimate
-
seismic
-
signiture
-
geothemal
-
field
4.
Building Resonance: Structural stability during earthquakes
. Seismological Facility for the
Advancement of Geoscience. (n.d.).
https://www.iris.edu/hq/inclass/animation/building_resonance_the_resonant_frequency_of_differ
ent_seismic_waves
5. Hadziioannou, C., & Rijal, A. (n.d.). Seismo
-
Live
-
GitHub Pages. https://seismo
-
live.github.io/html/Ambient%20Seismic%20Noise/NoiseCorrelation_wrapper.html
6.
Surface
. Geogiga Technology Corp. (n.d.). https://www.geogiga.com/products/surface/
Supplementary Information
Figure S1
-
Cross
-
correlations across stations (0.25
-
0.35 Hz, 0.35
-
0.45 Hz, 0.45
-
0.55 Hz,...
1.45
-
1.55 Hz)
Below are the plots of all the correlations between each of the east
-
west stations
from 0.25
-
0.35, increasing in intervals of 0.10 up to 1.45
-
1.55 Hz. Here, the peaks of the
perturbations are difficult to discern to draw the group velocity, as there were m
ultiple
perturbations for some stations. The blue graph represents the station SMF2, orange represents
LCG, green represents LGB, red represents RHC2,
and purple represents WLT.
Figure S2
-
Notice that for the north
-
south line, the peaks were picked going from the top right
to the bottom left because the waveforms come from the south, and thus it takes less time for the
waves to reach stations closer to the south than the north.
Figure S3
-
Peaks picked for the north
-
south direction and the east
-
west direction
correlations
The chart below shows the distance and the time for each peak that we picked that
forms a diagonal line based on figure S2. It also shows the average velocity and standard
deviation that we used to generate the phase velocity versus depth graph in figure 5
.