University of Southern California Dissertations and Theses

Elements of seismic structures near major faults from the surface to the Moho

(USC Thesis Other)

Elements of seismic structures near major faults from the surface to the Moho

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content 1

ELEMENTS OF SEISMIC STRUCTURES NEAR MAJOR FAULTS FROM THE
SURFACE TO THE MOHO

by

Yaman Ozakin

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(GEOLOGICAL SCIENCES)

December 2015

Copyright 2015 Yaman Ozakin

2

In memory of Mehmet Ali Özel.

3
Acknowledgments
I am grateful to my Mom, Dad, Mustafa Aktar, Yehuda Ben-Zion, Barbara Tomaszewicz, Christiann
Boutwell, Jessica Donovan, Melanie Gérault, Arkadaş Özakın and my committee members Frank
Vernon, Charlie Sammis, Sergey Lototsky; and many others who helped me in this journey.

4
Contents
Acknowledgments ..................................................................................................................................... 3

Contents ..................................................................................................................................................... 4

Abstract ...................................................................................................................................................... 6

1
Introduction ......................................................................................................................................... 7

2
Velocity contrast across the 1944 rupture zone of the North Anatolian fault east of Ismetpasa
from analysis of teleseismic arrivals ........................................................................................................ 9

2.1
Abstract .......................................................................................................................................... 9

2.2
Introduction .................................................................................................................................... 9

2.3
Seismic experiment and data ....................................................................................................... 10

2.4
Using teleseismic arrivals to image velocity contrast across a fault ............................................ 11

2.5
Results .......................................................................................................................................... 14

2.6
Discussion .................................................................................................................................... 15

2.7
Acknowledgments ........................................................................................................................ 16

Appendix A: Figures and Tables .......................................................................................................... 17

3
Systematic receiver function analysis of the Moho geometry in the southern California plate-
boundary region ...................................................................................................................................... 34

3.1
Abstract ........................................................................................................................................ 34

3.2
Introduction .................................................................................................................................. 34

3.3
Data and Pre-Processing .............................................................................................................. 36

3.4
Receiver Function Techniques ..................................................................................................... 37

3.5
Results .......................................................................................................................................... 41

3.5.1
Profiles of the Stacks ............................................................................................................ 43

3.5.2
NW of Cajon Pass ................................................................................................................. 48

3.5.3
Anza Region .......................................................................................................................... 49

3.6
Discussion .................................................................................................................................... 50

3.7
Acknowledgements ...................................................................................................................... 52

Appendix B: Supplementary Figures .................................................................................................... 53

5
4
Estimating Attenuation Coefficients and P-Wave Velocities of the Shallow San Jacinto Fault
Zone from Betsy Gunshots Data Recorded by a Spatially Dense Array with 1108 Sensors ........... 55

4.1
Abstract ........................................................................................................................................ 55

4.2
Introduction .................................................................................................................................. 55

4.3
Data .............................................................................................................................................. 55

4.4
Method ......................................................................................................................................... 57

4.4.1
Calculating The Body Wave Velocities ................................................................................ 59

4.4.2
Calculating the Body Wave Energies ................................................................................... 61

4.4.3
Calculating the Dominant Frequencies ................................................................................. 61

4.4.4
Stacking the results ............................................................................................................... 62

4.5
Results .......................................................................................................................................... 62

4.6
Discussion .................................................................................................................................... 63

4.7
Acknowledgements ...................................................................................................................... 64

5
Conclusion ......................................................................................................................................... 65

References ................................................................................................................................................ 66

6

Abstract
The crust along major strike-slip fault zones goes under tremendous changes during their
existence. On the large scale, different parts of crust with different lithologies and/or geometries can
come into contact as a result of large cumulative slips, interaction with other fault zones can create
complicated Moho topographies. On smaller scales, recurring unilateral rupture can create asymmetrical
damage, which affects the distribution of attenuation coefficients of body waves. Because of intrinsic
ambiguity in techniques for estimating crustal structure, it is difficult to resolve the ambiguity of seismic
velocity vs. crustal thickness, since both can have the same effect in observations. Here, we investigate
the existence of Moho offsets in Southern California, seismic velocity contrasts across North Anatolian
Fault and distribution of attenuation around San Jacinto Fault Zone using special techniques, developed
to resolve such ambiguities. Moho offsets are studied using receiver function techniques and incorporate
a detailed 3D velocity model obtained from a double difference tomography study, which enables us to
overcome the ambiguity mentioned above. The results show evidence that supports major offsets across
3 major faults, ranging between ~8 to ~20 km. On North Anatolian fault, using data from a small
seismic network, we compare observed teleseismic arrival times with expected ones. The results show a
prominent difference in arrival times between stations that are on different sides of the fault. The ratio of
average seismic velocities for the crust calculated using these differences is 8.3% if the contrast is
confined to the upper half of the crust. To study the shallow crustal properties, we use data from an
active source experiment from a dense array on San Jacinto Fault Zone. We calculate Q values for body
waves by linear regression analysis to estimate the decay rate of the energy in 33 shots. We also estimate
an average velocity structure using the automatic picks from the shot data. The Q values range from 5-
30, which is much lower than Q values associated with greater depths. The body wave velocities range
from 500 m/s to 1250 m/s and both show strong local variations.
7

1 Introduction
Unlike astronomers and astrophysicists, the ability of seismologists to study their subject is
severely crippled by the inability to simply look at it. Instead we have to use indirect methods, i.e. we
have to study how the elastic waves change while propagating through the earth and figure out what
kind of structures along the path of travel those changes correspond to. We try to find out how and
where the velocity of the waves changes, how and where they lose energy to internal friction, how and
where they change their polarization.
One of the most common methods of studying these properties of the earth is using inversion
techniques, for which we create a model and a source signal and run a simulation. Then we check
whether different aspects of the calculated signal matches to what we observe. If it doesn ’t, we change
the model and try again until an error function of our choosing yields a value below an acceptable
threshold. There are two problems with these techniques.
First, the sources that generate signals with largest amounts of energy are usually earthquakes and
we usually don ’t know what the actual source signal is like since even the shallowest earthquakes
happen 5 km beneath the earth ’s surface. Even though there are methods to go around this limitation,
none can be done without major assumptions.
Second, due to limitations in the observations, there may be more than one model that predicts
identical observations and usually don ’t have enough reason to choose one over the other. This non-
uniqueness is embedded in all the fields of seismology using inversion techniques. It is, therefore,
important to develop methods that don ’t depend on inversion techniques to both overcome the non-
uniqueness and also enable us to gain intuitive understanding of the relationship between data and
model.
The non-uniqueness plays especially an important role in the study of the crust near major faults.
Major faults are known to split the whole crust from the surface to the lower boundary known as Moho
discontinuity. It is possible that the motion along these faults to bring structurally and/or geometrically
different parts of the crust together. It is an important objective to find out if and where this happens as
both have implications in the mechanisms that drive earthquakes. Ben-Zion [1989] suggested that if two
sides of the fault consist of materials that have different seismic velocities, the earthquake on that fault
could rupture only in one direction instead of two, which would result in a more destructive earthquake
in the direction it ruptures towards.
In the three chapters presented in this dissertation, the properties of shallow and deep crust are
studied at different scales and we try to overcome the non-uniqueness problem by inventing linear
methods, by incorporating as many results as we can from several other studies and in the case of
chapter 4, by generating our own source signals using explosions.
In chapter 2, we use the arrival times of distant earthquakes to infer whether two sides of the crust
around the North Anatolian Fault in Turkey have significantly different seismic velocities. We compare
the differences between the expected and observed arrival times that propagate through the whole crust.
We check whether or not there is a statistically significant difference between the arrival times observed
in seismic waves traveling through different sides of the fault.
In chapter 3, we concentrate on the major faults in the southern California region by again using
teleseismic earthquakes with estimating “receiver functions ”. We use the fact that the fastest traveling P-
waves convert into S waves when they hit the crust. We use the time series of the converted phases as
8
the response signal and P-waves as the original signal and use deconvolution to estimate the impulse
response of the crust. We then incorporate the 3D velocity structure from a tomography study to convert
the impulse response into the geometry of the crustal structures.
In chapter 4, we estimate attenuation coefficients of a small region around San Jacinto Fault Zone
using ordinary linear regression analysis. The very dense array that we deployed around the fault
consists of 1108 geophones. We performed 33 Betsy-gun shots and we use the energy decay of the
seismic waves due to these shots when estimating the Q values. The results indicate that the Q values in
this region are within 5-30 range.

9
2 Velocity contrast across the 1944 rupture zone of the North Anatolian
fault east of Ismetpasa from analysis of teleseismic arrivals
This chapter was published in Geophysical Research Letters in 2012, according to the following author
list: Y. Ozakin, Y. Ben-Zion, M. Aktar, H. Karabulut and Z. Peng.

2.1 Abstract
We use differences between arrival times of teleseismic events at sets of stations crossing the North
Anatolian fault east of Ismetpasa, where shallow creep has been observed, to detect and quantify a contrast
of seismic velocities across the fault. Waveform cross correlations are utilized to calculate phase delays of P
waves with respect to expected teleseismic arrivals with incident angles corresponding to the generating
events. Compiled delay times associated with 121 teleseismic events indicate about 4.3% average P wave
velocity contrast across the fault over the top 36 km, with faster velocity on the north side. The estimated
contrast is about 8.3% if the velocity contrast is limited to the top 18 km. The sense of velocity contrast is
consistent with the overall tectonic setting and inference made for the examined fault section based on
theoretical expectations for bimaterial ruptures and observed asymmetry of rock damage across the fault.
Our data indicate lack of significant microseismicity near the fault, suggesting that creep in the area is
limited to the depth section above the seismogenic zone.
2.2 Introduction
Active faulting over geological time brings into contact materials that were originally separated and are
thus likely to have different elastic properties. Geological studies show that the principal slip zone in large
fault structures is often localized along bimaterial interfaces that separate rock units with considerably
different properties [e.g. Sengor et al., 2005; Dor et al., 2008; Mitchell et al., 2011]. Bimaterial interfaces
can have a number of important roles in earthquake and fault zone seismology. A lithology contrast in
earthquake source regions produces an ambiguity in inferred seismic moments, associated with the multiple
available choices for the assumed rigidity [e.g., Ben-Zion, 1989; Heaton and Heaton, 1989]. Earthquake
ruptures on a bimaterial interface may propagate as a pulse with a preferred propagation direction [e.g.,
Weertman, 1980; Ben-Zion and Andrews, 1998; Zaliapin and Ben-Zion, 2011; Lengline and Got 2011].
Ignoring a velocity contrast across the fault can produce biases and errors in derived earthquake locations,
fault plane solutions, and other source and structure properties [e.g., McNally and McEvilly, 1977;
Oppenheimer et al., 1988; Ben-Zion and Malin, 1991; Schulte-Pelkum and Ben-Zion, 2012].
The most diagnostic information on bimaterial interfaces can be obtained from fault zone head waves
that propagate along, and hence owe their existence to, velocity contrast interfaces [e.g. Ben-Zion, 1989,
1990]. Using head waves is best done with record sections of seismograms generated by numerous
earthquakes on the fault of interest and observed at near-fault stations [e.g. McGuire and Ben-Zion, 2005;
Zhao et al., 2010]. Since dense near-fault seismic data are not always available, it is desirable to develop
simple techniques that could be used to image fault bimaterial interfaces with more commonly available
data. In the present paper we develop and use such a method based on teleseismic arrivals at stations on the
opposite sides of a fault. The study is done in the context of the 1944 M7.3 earthquake rupture zone on the
North Anatolian fault (NAF).
10

Figure 2.1: (a) Map of the seismic stations (blue triangles) across the North Anatolian
fault zone (black line) east of Ismetpasa used in this study. One station (white triangle) did
not record sufficient high quality data (see text) and was not used. The bottom insets show
the back-azimuth and incidence angle distributions of the used events. The top inset has a
large-scale map. The background colors indicate topography with white being high and
red/brown/green being low. (b) Summary of data coverage at different stations.

Ruptures on a bimaterial interface having a high angle to the maximum compressive background stress,
which is representative for large continental strike-slip faults, is expected to produce [e.g. Ben-Zion and Shi,
2005] signficantly more shallow off-fault damage on the stiffer side of the fault. Dor et al. [2008]
performed detailed mapping of rock damage along several sections of the NAF including the 1944 rupture
zone. From clear asymmetry of rock damage, they suggested that the north side of the 1944 rupture zone
has faster seismic velocity at seismogenic depth than the south side. In this paper we test this inference
using seismic data recorded by a small seismic array across the surface trace of the 1944 rupture east of
Ismetpasa. In the next section we describe briefly the experiment and basic properties of the data set. In
section 2.3 we outline the employed imaging method. The results are presented in section 2.4 and discussed
in section 2.5.

2.3 Seismic experiment and data
We conducted a small seismic experiment in the region of the 1944 rupture of the NAF east of
Ismetpasa with a line of 6-12 seismometers that cross the fault (Figure 1a). The network operated for about
2 years with different periods of recording at different stations (Figure 1b). The instruments were a mixture
of Guralp 40T seismometers with Reftek 130 recorders and Guralp 6TD integrated seismometer plus
digitizers. All stations were deployed in bedrocks with comparable (best possible) site conditions. The
position and elevation of the stations are listed in Table 1 of the supplementary material. The location of the
experiment was chosen because it is within the area where Dor et al. [2008] found strong asymmetry of
rock damage that may reflect repeating ruptures with preferred propagation direction on a bimaterial
BLKS
CMDR
HTPL
BLKN
CRDK
CMLK
TKMK
KRNZ
AHLR
BLKV
BYDR
PASA
10
20
3

# of # of Events ts
10
20
30
N
B
L
K
S
C
M
D
R
H
T
P
L
B
L
K
N
C
R
D
K
C
M
L
K
K
R
N
Z
A
H
L
R
B
L
K
V
B
Y
D
R
T
K
M
K
P
A
S
A
DATA NO DATA
(a) (b) 32˚36' 32˚48'
40˚48' 41˚00'
0
1 0
20
30
40
20 30 40 50
Number of Events
Incidence Angle
11
interface. The location also coincides with a section of the NAF that is partially creeping at least at shallow
depth. The creep rate decayed from a maximum of 4-6 cm/yr following the 1944 earthquake to a present
value of 0.7 cm/yr [Cakir et al., 2005].
Earthquake detection in the recorded data was done by a manual inspection of automatic identification
of candidate events. The events were located using the code hypo and the velocity model of Cambaz and
Karabulut [2010]. To date we were able to detect only ~235 events in the magnitude range −1 to 2.5 within
a radius of 45 km from the center of the network. Only 15-20 events are located within 3 km of the fault and
they all have M
L
<1.0. Using these as templates for detecting more events was not successful so far as the
signals produced by these events are close to the background noise. The small number of events close to the
fault is in marked contrast to the numerous microearthquakes [e.g. Hill et al., 1990] along the creeping
section of the San Andreas fault (SAF), suggesting that the NAF east of Ismetpasa does not creep at
seismogenic depth. Given the small number of local earthquakes, we use teleseismic data to detect and
quantify the possible existence of different rock bodies across the fault.

2.4 Using teleseismic arrivals to image velocity contrast across a fault
The basic data processing in this study is done with the ObsPy toolkit [Beyreuther et al., 2010].
Teleseismic waves which arrive with near-vertical incidence angles sample the crustal structures below the
stations. We calculate the expected arrival times of such teleseismic waves in the absence of a velocity
contrast across the fault with the TauP toolkit [Crotwell et al., 1999]. The calculations account for
differences in arrival times related to the geometry of the stations and incoming planar teleseismic waves
(Figure 2). If the crustal structures across the fault are different, this should be manifested by systematic
differences between the actual arrivals and expectations at stations on the opposite sides of the fault. We
assume that there is no velocity contrast in the mantle.
For each pair of stations, we compile a data set consisting of the differences (Figure 2a) between the
observed and estimated arrivals at the two stations,
2 1 2 1
t t t Δ − Δ = Δ
−
, using all available teleseismic arrivals
at the stations. The differences of teleseismic arrival times are measured using waveform cross correlations.
Figure 3a shows example results of
2 1−
Δt values associated with stations AHLR and CMLK. If the number
of observations for a given data set (pair of stations) is less than 5, or the standard deviation of the measured
arrival time differences is greater than 0.15 sec, the data set is deemed not having sufficient quality and is
ignored. The adopted thresholds for data quality are somewhat arbitrary, but the obtained results are not
very sensitive to the used values. This procedure provides 42 data sets of arrival time differences (Figure
A1, supplementary material) that are used for further analysis.
12

Figure 2.2: (a) Example waveforms observed at stations BLKN and BLKV located north
and south of the fault, respectively. (b) Illustration of differences between arrival times of
teleseismic waves with incidence angle θ and reference expectations at stations across the
fault. Systematic earlier arrivals on a given side of the fault indicate faster seismic
velocities. See text for additional details.
∆t
1
v
1
h
v
2
∆t
2
Expected arrival
Station 1
(N of the fault)
Station 2
(S of the fault)
Fault Plane
Observed arrival
Moho
Surface
Incoming teleseismic wavefront
θ θ
Incidence angle
Incidence angle
(a)
(b)
∆t
1
∆t
2
Estimated
Estimated
Observed
Observed
∆t
1
- ∆t
2
= ∆t
1-2
Time (sec)
Station 1 (N)
Station 2 (S)
∆t
1
< ∆t
2
=> v
1
> v
2
∆t
1
> ∆t
2
=> v
1
< v
2
M Moho
Observed arrival
13

Figure 2.3: Illustration of the method used to compile data on contrast of seismic velocity
across the fault. (a) (a) Arrival time differences (dots) between observations at stations
AHLR and CMLK for all teleseismic events recorded at both stations. The horizontal line
and shaded band denote, respectively, the mean with one standard-deviation on each side.
The events are sorted by back-azimuth shown by the black line and scale on the right. (b)
Corresponding results for velocity ratios based on the arrival time differences in (a) and
Equation (2.2). (c) A 4-qudrant diagram with stations sorted on both axes from north to
south. A location in the chart indicates the relevant pairs of stations and the color at that
point indicates the velocity ratio. (Inset) Clusters of ratios greater or smaller than one in
two diagonal quadrants are indicative of a velocity contrast across the fault.

To estimate the velocity contrast between a given pair of stations, we use a reference constant crustal
thickness h and average crustal P wave velocities beneath the two stations v
1
and v
2
. The incidence angles
of waves are generally station-dependent, but because teleseismic data are used we assume that the angles at
the different stations are essentially the same. From the basic geometry illustrated in Figure 2b, the travel
time difference of a planar teleseismic wave with incidence angle θ at the two stations is

(2.1)

The corresponding ratio of the average crustal velocities beneath the two stations is

(2.2)
Equation (2.2) is used to convert the time differences in each data set (Figures 3a and A1) to velocity
ratios (Figure 3b). The mean of the calculated ratios is assigned to the associated pair of stations. Example
set of estimated velocity ratios is shown in Figure 3b, assuming that h = 36 km [Ozakin and Aktar, 2012]
and v
2
= 6.8 km/s [Cambaz and Karabulut, 2010]. Analogous results obtained for all pairs of stations are
Mean=0.965 (b)
v
1
/v
2
Event Number
BAZ (deg)
0.8
0.85
0.9
0.95
1
1.05
1.1
1.15
1.2
0 10 20 30 40 50
0
50
100
150
200
250
300
350
Station 1
(S of the fault)
Station 2
(N of the fault)
Fault
Fault
N
N
S
S
v
1
/v
2
Faster S Faster N
N S
S
N
N S
S
N
N S
S
N
No Contrast
>1
>1 <1
<1
∆t
1
-∆t
2
(sec)
Mean=0.21 sec
St. Dev. = 0.083 sec
AHLR - CMLK
(a)
BAZ (deg)
-1
-0.5
0
0.5
1
0 10 20 30 40 50
0
50
100
150
200
250
300
350
(c)
) cos( ) cos(
2 1
2 1
θ θ v
h
v
h
t − = Δ
−
) cos(
2 2 1 2
1
θ v t h
h
v
v
−
Δ +
=
14
used to construct a matrix of velocity ratios at different positions (Figure 3c). The matrix is indexed using
stations that are sorted by distances from the fault where stations on the north side have positive
coordinates. A value less than 1.0 in the matrix indicates that the velocity beneath the station of the relevant
column is smaller than the one beneath the station in the corresponding row. If the velocity ratios between
stations on the north and south sides are systematically greater than 1.0, the crustal section to the north
should be on the average faster and vice versa (assuming the crustal thickness does not vary too much
across the fault). If such relations hold when comparing all station pairs that are on the opposite sides of the
fault, we should get clusters of positive and negative values in corresponding quadrants of the matrix.
Based on a receiver function study [Ozakin and Aktar, 2012], the average crustal thickness in our study
area is 36 km and there is an average increase of the crustal thickness of about 1.5-2.0 km over a distance of
100 km from north to south (assuming smooth thickness variation). The NS extent of the network is about
20 km, giving about 0.3-0.4 km of potential crustal thickness variations. With an average P wave velocity
of 6.8 km/s [Cambaz and Karabulut, 2010], this corresponds to less than 0.1 sec difference in arrival times.
These values are sufficiently small not to affect the results obtained in the next section.

2.5 Results
We use 121 teleseismic events between 2007 and 2009 taken from the USGS web site
http://earthquake.usgs.gov/earthquakes/eqarchives/epic/epic_global.php (Table A2 of the online
supplementary material). The great circle distances (Δ in degrees) of the teleseismic events are converted to
incidence angles using the relation

(2.3)
This expression is based on least-squares fit to travel time calculations in a layered earth model [Lay and
Wallace, 1995, Table 8.1]. Most incidence angles of the employed data are
!
25 ≤ θ (Figure 1a bottom left),
consistent with our assumption of near vertical propagation. Eliminating the data associated with
!
25 > θ
does not change appreciably the estimated velocity contrast values.
Figure 2.4 shows a color map of velocity contrast ratios obtained by interpolating the ratios derived for
all pairs of stations as discussed before (Figure 2.3) using h = 36 km and v
2
= 6.8 km/s. The results have
clear clusters of ratios larger and smaller than 1.0 indicative of a velocity contrast across the fault.
Specifically, the upper left quadrant indicates velocity ratios between stations on the south (along the
horizontal axis) and stations on the north (along the vertical axis). The general existence of ratios less than
1.0 in this quadrant implies that the average crustal P velocity in the south is smaller than in the north.
Similar information is given in a slightly different geometrical form at the bottom right quadrant. With h =
36 km and v
2
= 6.84 km/s, the average velocity contrast across the fault is 4.3%. The differences in station
elevations (Figure 2.4 bottom) are unlikely to affect significantly the obtained results, as the elevation
changes across the fault are smaller than the internal variations on the north side. Figure A2 (online
supplementary material) shows that there is no clear correlation between the obtained velocity contrast and
station elevations.
33 . 49 72 . 0 00383 . 0
2
+ Δ − Δ = θ
15

Figure 2.4: (Top) Velocity ratios (colors) based on time differences between teleseismic
arrivals and reference expectations at pairs of stations as illustrated in Figure 2.3. The circles
with white outline denote velocity ratios for all pairs of stations and the smooth color map is
a Delaunay mesh of block-mean of data points. The observed coherent colors in the top-left
and bottom-right quadrants indicate that the north side of the fault has faster P wave velocity.
(Bottom) Elevations of stations used in the study. See text for additional details.

2.6 Discussion
We developed and implemented a simple method to estimate the average contrast of crustal seismic
velocities across a fault from teleseismic arrivals at several stations on the opposite sides of the fault. As
mentioned in the introduction, a lithology contrast across a fault can influence the directivity of earthquake
ruptures (and hence expected shaking hazard in different along-strike directions), and may be relevant for
various other aspects of fault mechanics and seismology [e.g. Ben-Zion and Andrews, 1998; Oppenheimer
et al. 1988]. In many places dense seismic networks do not exist and it is important to develop simple tools
that may be used to detect and provide first-order estimates of velocity contrasts across faults. Our method
1400
Elevation (m)
900
-8000
-6000
-4000
-2000
0
2000
4000
6000
8000
Distance From the Fault (m)
-8000 -6000 -4000 -2000 0 2000 4000 6000 8000
Distance From the Fault (m)
CMDR
HTPL
BLKN
CRDK
CMLK
TKMK
KRNZ
AHLR
BLKV
BYDR
PASA
CMDR
HTPL
BLKN
CRDK
CMLK
TKMK
KRNZ
AHLR
BLKV
BYDR
PASA
NAF
NAF
N
S
N S
0.95
1.00
1.05
v
1
/v
2
16
provides such a tool, which may be used with a small number of stations across faults that are in seismically
quiet periods.
Some background information is needed to reduce the non-uniqueness associated with trade-offs
between changes of crustal thickness and velocity contrast across a fault, as done in the present study based
on the results of Ozakin and Aktar [2012]. We note that standard receiver function studies, ignoring the
possible existence of lithology variation across the fault, tend to overestimate changes of crustal thickness
across the fault [Schulte-Pelkum and Ben-Zion, 2012]. Therefore, our estimate of a possible bias assuming
constant crustal thickness is likely an upper bound. A support for our interpretation in terms of velocity
contrast is provided by the fact that the estimated contrast values generally increase, within fluctuations, as
one uses pairs of stations that are progressively closer to the fault. Also, a change in the Moho depth near
the fault is expected to broaden the surface deformation [Lyakhovsky and Ben-Zion, 2009], whereas the
observed deformation in the study area is highly localized [Cakir et al., 2005].
Our results indicate that the average velocity contrast in the top 36 km across the NAF east of Ismetpasa
using all stations is 4.3%. Using only the 2 stations just north and south of the fault with similar elevation
(BLKN and AHLR), yields an estimated contrast value of about 5.2%. The velocity contrast is expected to
decrease with depth [e.g., Ben-Zion et al., 1992; Lewis et al., 2007; Roux et al., 2011]. Assuming for
example that the velocity contrast is limited to the upper 18 km of the crust, and using for that depth section
v
2
= 6.3 km/s [Cambaz and Karabulut, 2010] gives an average contrast value of 8.3%. This is similar to
values found for several sections of the SAF [e.g. McGuire and Ben-Zion, 2005; Zhao et al., 2010]. The
obtained sense of velocity contrast in our study area is consistent with the inference made by Dor et al.
[2008] in the context of bimaterial ruptures based on observed asymmetry of rock damage, as well as the
large scale geological framework associated with the NAF [Sengor et al., 2005]. Detailed analysis of shear
wave anisotropy close to the NAF [Peng and Ben-Zion, 2004, 2005] does not indicate strong systematic
differences between the blocks on the opposite sides of the fault. The average difference of P anisotropy
across the fault (if such exists) is likely to be even smaller, so anisotropy is unlikely to influence our results
significantly.
The best diagnostic signals for imaging fault bimaterial interfaces are head waves generated by
seismicity that is localized on a fault [e.g. Ben-Zion and Malin, 1991]. Based on our detection efforts to
date, the seismic records during the ~2 yr of our experiment (Figure 2.1) contain very few events near the
NAF down to M
L
<1.0. This is in marked contrast to the creeping section of the SAF [e.g. Hill et al., 1990],
and the Hayward fault in CA [e.g. Schmidt et al., 2005] which is partially creeping as the NAF fault section
examined in this study. Bulut et al. [2012] used recently changes in polarizations of waves in early P
waveforms to detect fault zone head waves and measure the arrival times of the P head and regular body
waves. This method can be used (as the one used in our study) with a small number of stations, but it
requires again seismicity on the fault. Noise-based techniques can be used to provide structural images in
places without seismicity [e.g. Shapiro et al., 2005; Roux et al., 2011]. Future studies using a combination
of these and other techniques can improve on the results of this paper.

2.7 Acknowledgments
The study was funded by the Bogazici University Research Foundation (grant 09M103) and the
National Science Foundation (grant EAR-0823695). YBZ acknowledges support from the Alexander von
Humboldt foundation. ZP acknowledges support from the Georgia Tech Research Foundation. The
manuscript benefitted from highly useful comments by two anonymous referees.

17
Appendix A: Figures and Tables

Figure A1. Time differences (dots) and back-azimuth distribution (black line) associated
with all pairs of employed stations. The horizontal red line and shaded band denote,
respectively, the mean with one standard-deviation on each side.

18
Figure A1 continued.

19
Figure A1 continued.

20
Figure A1 continued.

21
Figure A1 continued.

22
Figure A1 continued.

23
Figure A1 continued.

24
Figure A1 continued.

25
Figure A1 continued.

26
Figure A1 continued.

27

Figure A1 continued.

28

Figure A2: Elevation differences and derived velocity contrasts associated with
different pairs of stations. The data sets do not appear to be correlated.

29
Name Side Distance from NAF
(m)
Elevation (m) Latitude Longitude
CMLK N 7235 1361 40.965 32.79367
CMDR N 5440 1216 40.94017 32.741
TKMK N 3789 1280 40.92983 32.773
BLKN N 2687 1180 40.90933 32.72283
KRNZ N 1379 983 40.90933 32.78467
HTPL N 1094 1111 40.888 32.70167
PASA N 443 990 40.86933 32.62417
AHLR S 925 1238 40.88667 32.7735
BLKV S 3138 1274 40.8615 32.75167
CRDK S 3527 1195 40.84833 32.72417
BLKS S 5304 1250 40.82367 32.67333
BYDR S 7754 1207 40.82367 32.774

Table A1: Station information.

30

Year Month Day Hour Min Latitude Longitude Mag Distance
(Deg)
Back-Azimuth
2007 11 25 16 24 -8.29 118.37 6.5 92.0508 99.1053
2007 11 25 18 0 -2.24 100.41 6 74.7555 106.706
2007 11 25 20 15 -8.22 118.47 6.5 92.0802 98.9879
2007 11 29 19 27 14.94 -61.27 7.4 83.3748 283.904
2007 12 19 9 56 51.36 -179.51 7.2 83.9101 19.6807
2007 12 21 7 5 51.37 -178.98 6.3 84.0352 19.371
2007 12 22 12 23 2.09 96.81 6.1 69.1862 106.01
2007 12 25 14 31 38.5 142.03 6.1 77.9922 49.2277
2007 12 26 22 31 52.56 -168.22 6.4 85.1649 12.6797
2008 1 4 7 1 -2.78 101.03 6 75.5741 106.669
2008 1 5 11 53 51.25 -130.75 6.6 87.1196 349.719
2008 1 5 12 0 51.16 -130.54 6.4 87.179 349.57
2008 1 6 5 10 37.22 22.69 6.2 8.65586 248.228
2008 1 9 8 44 32.29 85.17 6.4 42.3967 84.375
2008 1 9 15 0 51.65 -131.18 6.1 86.7854 350.068
2008 1 10 2 0 43.78 -127.26 6.3 93.818 345.648
2008 1 20 20 9 2.35 126.82 6.1 91.5414 85.5906
2008 1 22 17 11 1.01 97.44 6.2 70.38 106.383
2008 2 8 9 31 10.67 -41.9 6.9 71.5227 268.191
2008 2 9 18 8 -0.24 125.08 6 91.904 88.6764
2008 2 14 10 27 36.5 21.67 6.9 9.72035 246.79
2008 2 14 12 59 36.35 21.86 6.5 9.66498 245.561
2008 2 20 8 29 2.77 95.96 7.4 68.1048 106.106
2008 2 20 18 14 36.29 21.77 6.2 9.75819 245.482
2008 2 21 3 0 77.08 18.57 6.1 36.8013 354.712
2008 2 21 14 18 41.15 -114.87 6 93.223 336.121
2008 2 24 15 0 -2.4 99.93 6.5 74.5167 107.167
2008 2 25 8 12 -2.49 99.97 7.2 74.6072 107.206
2008 2 25 18 41 -2.33 99.89 6.6 74.4398 107.144
2008 2 25 21 55 -2.24 99.81 6.7 74.3203 107.134
2008 2 27 7 4 26.82 142.44 6.2 86.2959 57.4877
2008 3 3 2 4 -2.18 99.82 6.2 74.2865 107.082
2008 3 3 9 5 46.41 153.18 6.5 78.1559 37.5701
2008 3 3 14 32 19.91 121.33 6 76.1897 75.6179
2008 3 3 14 43 13.35 125.63 6.9 83.5217 78.0592
2008 3 13 13 22 -45.49 35.01 6 86.0173 178.423
2008 3 14 22 54 26.99 142.6 6 86.2797 57.2663
2008 3 15 15 0 2.71 94.6 6 67.1577 107.161
2008 3 20 14 33 6.19 126.93 6.1 89.1288 82.6336
2008 3 20 22 22 35.49 81.47 7.2 38.2563 81.9236
2008 3 22 21 26 52.18 -178.72 6.2 83.3563 18.8956
2008 3 29 17 47 2.86 95.3 6.3 67.5633 106.523
Table A2: Event Information. Earthquakes with magnitude ≥ 6 were used in the study.
See main text for the employed catalog.

31
Year Month Day Hour Min Latitude Longitude Mag Distance
(Deg)
Back-Azimuth
2007 11 25 16 24 -8.29 118.37 6.5 92.0508 99.1053
2007 11 25 18 0 -2.24 100.41 6 74.7555 106.706
2007 11 25 20 15 -8.22 118.47 6.5 92.0802 98.9879
2007 11 29 19 27 14.94 -61.27 7.4 83.3748 283.904
2007 12 19 9 56 51.36 -179.51 7.2 83.9101 19.6807
2007 12 21 7 5 51.37 -178.98 6.3 84.0352 19.371
2007 12 22 12 23 2.09 96.81 6.1 69.1862 106.01
2007 12 25 14 31 38.5 142.03 6.1 77.9922 49.2277
2007 12 26 22 31 52.56 -168.22 6.4 85.1649 12.6797
2008 4 15 23 22 51.86 -179.36 6.4 83.4906 19.3899
2008 4 16 6 50 51.88 -179.16 6.6 83.5225 19.2676
2008 4 19 3 51 -7.82 125.69 6.1 97.2592 94.0238
2008 4 19 10 47 -7.88 125.72 6 97.3204 94.0503
2008 4 23 18 24 22.88 121.62 6 74.5646 73.0515
2008 4 24 12 29 -1.18 -23.47 6.5 65.9292 245.561
2008 5 2 1 9 51.86 -177.53 6.6 83.9392 18.3412
2008 5 7 16 10 36.18 141.54 6.2 79.2866 51.2266
2008 5 7 16 44 36.16 141.76 6.1 79.4301 51.1176
2008 5 7 17 0 36.16 141.53 6.9 79.2943 51.247
2008 5 9 22 54 12.52 143.18 6.8 96.7597 67.1684
2008 5 12 6 46 31 103.32 7.9 56.5646 75.9855
2008 5 12 11 50 31.21 103.62 6.1 56.6772 75.6118
2008 5 19 14 6 1.64 99.15 6 71.2019 104.663
2008 5 20 14 4 51.16 178.76 6.3 83.6388 20.7585
2008 5 29 16 0 64 -21.01 6.3 38.488 325.154
2008 5 31 5 0 -41.2 80.48 6.4 92.4568 146.034
2008 6 1 2 11 20.12 121.35 6.3 76.0739 75.4378
2008 6 3 22 43 -8.1 120.23 6 93.3225 97.7589
2008 6 8 12 29 37.96 21.52 6.4 9.18548 255.087
2008 6 14 0 0 39.03 140.88 6.9 76.9723 49.4526
2008 6 23 0 35 67.7 141.28 6.1 59.2794 24.8967
2008 6 27 12 0 11.01 91.82 6.6 59.5622 102.419
2008 6 28 13 51 10.85 91.71 6.1 59.5853 102.64
2008 6 29 21 51 45.16 137.45 6 71.0404 46.3594
2008 7 5 2 32 53.88 152.89 7.7 72.5576 32.4595
2008 7 8 8 0 27.53 128.33 6 76.4573 65.3823
2008 7 15 3 50 35.8 27.86 6.4 6.33015 219.166
2008 7 19 3 0 37.55 142.21 7 78.7842 49.8265
2008 7 21 11 34 37.19 142.05 6 78.9369 50.1826
2008 7 23 15 3 39.8 141.46 6.8 76.8273 48.5475
2008 7 24 2 0 50.97 157.58 6.2 76.6654 32.245
2008 8 5 10 0 32.76 105.49 6 57.251 73.0578
Table A2 continued

32
Year Month Day Hour Min Latitude Longitude Mag Distance
(Deg)
Back-Azimuth
2007 11 25 16 24 -8.29 118.37 6.5 92.0508 99.1053
2007 11 25 18 0 -2.24 100.41 6 74.7555 106.706
2007 11 25 20 15 -8.22 118.47 6.5 92.0802 98.9879
2007 11 29 19 27 14.94 -61.27 7.4 83.3748 283.904
2007 12 19 9 56 51.36 -179.51 7.2 83.9101 19.6807
2007 12 21 7 5 51.37 -178.98 6.3 84.0352 19.371
2007 12 22 12 23 2.09 96.81 6.1 69.1862 106.01
2007 12 25 14 31 38.5 142.03 6.1 77.9922 49.2277
2007 12 26 22 31 52.56 -168.22 6.4 85.1649 12.6797
2008 8 10 8 37 11.06 91.81 6.2 59.5222 102.383
2008 8 11 23 10 -1.02 -21.84 6 64.694 244.377
2008 8 15 10 43 12.9 124.32 6 82.837 79.2337
2008 8 21 12 9 25.04 97.7 6 55.5344 85.2357
2008 8 22 8 0 -17.77 65.39 6 65.7496 145.714
2008 8 25 13 51 30.9 83.52 6.7 41.7917 87.1806
2008 8 27 1 21 51.61 104.16 6.3 48.6503 51.9259
2008 8 28 15 31 -0.25 -17.36 6.3 61.1079 241.241
2008 8 30 8 50 26.24 101.89 6 58.0304 81.5506
2008 9 10 13 26 8.09 -38.71 6.6 70.757 263.978
2008 9 11 0 3 1.88 127.36 6.6 92.255 85.5908
2008 9 11 0 42 41.89 143.75 6.8 76.6382 45.7718
2008 9 25 2 0 30.84 83.49 6 41.7973 87.278
2008 10 5 16 0 39.53 73.82 6.7 31.1762 78.8264
2008 10 5 23 39 33.89 69.47 6 29.8206 91.6247
2008 10 6 8 24 29.81 90.35 6.3 47.4822 84.4604
2008 10 11 11 0 19.16 -64.83 6.1 83.2292 289.343
2008 10 28 23 5 30.64 67.35 6.4 29.6923 99.0888
2008 10 29 11 52 30.6 67.46 6.4 29.7944 99.0638
2008 11 2 14 16 51.55 -174.37 6.1 84.9479 16.6169
2008 11 10 1 45 37.56 95.83 6.3 48.0246 72.348
2008 11 16 17 25 1.27 122.09 7.4 88.6593 89.4891
2008 11 22 16 52 -4.35 101.26 6.3 76.8136 107.672
2008 11 22 19 0 -1.23 -13.93 6.3 59.6078 237.519
2008 11 24 9 40 54.2 154.32 7.3 72.868 31.5984
2008 12 6 11 12 -7.39 124.75 6.4 96.2721 94.3024
2008 12 20 10 15 36.54 142.43 6.3 79.5636 50.4596
2008 12 25 3 56 5.75 125.38 6.3 88.2472 83.9722
2009 1 3 20 55 36.42 70.74 6.6 29.8613 86.1439
2009 1 13 1 1 -13.15 66.08 6 61.9761 142.703
2009 1 15 18 41 46.86 155.15 7.4 78.7192 36.2021
2009 2 11 17 15 3.89 126.39 7.2 90.2524 84.6807
2009 2 11 18 11 4.03 126.78 6 90.4557 84.3209
Table A2 continued

33
Year Month Day Hour Min Latitude Longitude Mag Distance
(Deg)
Back-Azimuth
2007 11 25 16 24 -8.29 118.37 6.5 92.0508 99.1053
2007 11 25 18 0 -2.24 100.41 6 74.7555 106.706
2007 11 25 20 15 -8.22 118.47 6.5 92.0802 98.9879
2007 11 29 19 27 14.94 -61.27 7.4 83.3748 283.904
2007 12 19 9 56 51.36 -179.51 7.2 83.9101 19.6807
2007 12 21 7 5 51.37 -178.98 6.3 84.0352 19.371
2007 12 22 12 23 2.09 96.81 6.1 69.1862 106.01
2007 12 25 14 31 38.5 142.03 6.1 77.9922 49.2277
2007 12 26 22 31 52.56 -168.22 6.4 85.1649 12.6797
2009 2 12 4 39 3.95 126.41 6 90.2285 84.6225
2009 2 12 8 17 3.97 126.72 6 90.4494 84.4052
2009 2 12 13 55 4.03 126.55 6.3 90.2461 84.5042
2009 2 22 18 0 3.68 126.56 6 90.481 84.7608
2009 3 6 11 0 80.32 -1.85 6.5 41.5621 351.666
2009 3 16 14 55 3.81 126.55 6.3 90.4086 84.648
2009 3 30 7 41 56.55 -152.74 6 82.8187 3.07418
2009 4 4 5 52 5.15 127.2 6.3 90.0263 83.2177
2009 4 7 4 55 46.05 151.55 6.9 77.6653 38.6496
2009 4 15 20 37 -3.12 100.47 6.3 75.4437 107.289
2009 4 18 19 55 46.01 151.43 6.6 77.6051 38.7539
2009 4 19 5 55 4.14 126.68 6.1 90.292 84.3153
2009 4 21 5 55 50.83 155.01 6.2 75.6285 33.6107

Table A2 continued

34
3 Systematic receiver function analysis of the Moho geometry in the
southern California plate-boundary region
This chapter was published in Pure and Applied Geophysics in 2014 according to the following author
list: Y. Ozakin and Y. Ben-Zion.
3.1 Abstract
We investigate the geometry of the Moho interface in the southern California region including the
San Andreas fault (SAF), San Jacinto fault zone (SJFZ), Elsinore fault (EF) and Eastern California
Shear Zone with systematic analysis of receiver functions. The data set consists of 145 teleseismic
events recorded at 188 broadband stations throughout the region. The analysis utilizes a 3D velocity
model associated with detailed double-difference tomographic results for the seismogenic depth section
around the SAF, SJFZ and EF combined with a larger scale community model. A 3D ray tracing
algorithm is used to produce effective 1D velocity models along each source-receiver teleseismic ray.
Common Conversion Point (CCP) stacks are calculated using the set of velocity models extracted for
each ray. The CCP stacks are analyzed with volumetric plots, maps of maximum CCP stack values, and
projections along profiles that cross major faults and other features of interest. The results indicate that
the Moho geometry in the study area is very complex and characterized by large prominent undulations
along the NE-SW direction. A zone of relatively deep Moho (~35-40 km) with overall N-S direction
crosses the SAF, SJFZ and EF. A section of very shallow Moho (~10 km) below and to the SE of the
Salton Trough, likely associated with young oceanic crust, produces large Moho offsets at its margins.
Locations with significant changes of Moho depth appear to be correlated with fault complexity in the
brittle crust. The observations also show vertical Moho offsets of ~8 km across the SAF and the SJFZ
close to Cajon pass, and sections with no clear Moho phase underneath Cajon pass and adjacent to the
SJFZ likely produced by complex local velocity structures in the brittle upper crust. These features are
robust with respect to various parameters of the analysis procedure.
3.2 Introduction
The shear tectonic motion across the plate boundary region in southern California is accommodated
by a system of sub-parallel strike slip faults (e.g. Wallace et al., 1990), with the largest three being the
San Andreas fault (SAF), San Jacinto Fault Zone (SJFZ) and Elsinore fault (EF). The cumulative offsets
across the faults and ongoing deformation processes produce complex velocity structures in the brittle
upper crust. These may include juxtaposition of crustal blocks with different seismic velocities and/or
different thicknesses (e.g. Hauksson 2000; McGuire and Ben-Zion 2005; Özeren and Holt, 2010; Ozakin
et al. 2012), and low velocity damage zones around the faults (e.g. Fialko, 2004; Hong and Menke,
2006; Allam and Ben-Zion 2012; Rempe et al. 2013).
There is a long-standing debate on whether deformation below the brittle crust is dominated by
distributed ductile processes largely decoupled from the seismogenic faults (e.g. England and
McKenzie, 1982; Bourne et al., 1998; Molnar 1992), or by localized processes at the downward
continuation of the major faults (e.g., Peltzer and Tapponnier, 1988; Gilbert et al., 1994; Meyer et al.,
1998). A common seismological method of addressing this issue is to use receiver functions, associated
with converted P-to-S waves at prominent (sub-)horizontal interfaces, to image the possible existence of
vertical offsets of the Moho across large faults (e.g., Burdick and Langston, 1977; Jones and Phinney,
1998; Zhu and Kanamori, 2000; Wittlinger et al., 2004).
35

Figure 3.1: A map of the study area with used broadband seismic stations (inverted triangles) and
surface traces of the San Andreas fault (SAF), San Jacinto fault zone (SJFZ) and Elsinore fault
(EF). The town Hemet and Cajon pass (CP) are indicated by white circles. The gray rectangle
encloses the region covered by the velocity model of Allam and Ben-Zion (2012). The big and
small green boxes indicate areas of focus in sections 3.4.2 and 3.4.3. Results from station BBR
(white triangle) are shown in Figure 3.3. Projections of results along profiles (A-G) indicated by
red lines are shown in Figures 3.7 and 3.8.

To interpret receiver function results in terms of crustal thicknesses, they must be converted from the
time domain into space using an assumed velocity model. This is usually implemented with 1D velocity
models that do not account for systematic variations of lithologies near and across the faults. The Moho
interface is often constructed in receiver function analysis using a common conversion point (CCP)
stack, where receiver function amplitudes are projected along ray paths in an assumed velocity model,
and locations at depth associated with the most coherent large amplitudes are interpreted as associated
with the Moho. Schulte-Pelkum and Ben-Zion (2012) demonstrated that observed velocity contrasts
across faults and asymmetric sedimentary covers or damage zones can lead, if ignored, to erroneous
inferences of vertical Moho offset of 5 km or more. The derived Moho images can be biased by different
velocity structures across faults also when using a 3D velocity model to construct local 1D models
below each station, and using these 1D models in the CCP stacking for the ray paths of incoming
teleseismic signals from different directions.
Zhu (2000) and Miller et al. (2014) inferred from receiver function analyses on vertical Moho steps
of ~8 km across, respectively, the Mojave section of the SAF and Anza section of the SJFZ. In the
-118.00˚ -117.50˚ -117.00˚ -116.50˚ -116.00˚ -115.50˚ -115.00˚
32.50˚
33.00˚
33.50˚
34.00˚
34.50˚
0 50
CP
Hemet
ADO
BBS
BFS
BLA2
BOM
BOR
BTC
CHN
CJM
CLT
CTC
DEV
DGR
DNR
DPP
DRE
ERR
IDO
IPT
JEM
LPC
LUC2
LUG
MGE
MLS
MSC
MSJ
MUR
NSS2
OLI
PDU
PER
PLM
PLS
PMD
PSD
PSR
RSS
RVR
RXH
SAL
SBPX
SLB
SLR
SNO
SNR
SVD
SWS
TOR
VTV
WES
WLT
WWC
BBR
BAR
BC3
BEL
BRE
DAN
EML
GOR
IKP
IRM
JVA
LLS
MCT
OLP
RIO
SBB2
SDD
SDG
SDR
SOF
SRN
STG
TA2
BZN
CPE
CRY
FRD
HWB
KNW
LVA2
MONP2
PFO
RDM
SMER
SND
SOL
TRO
WMC
A
B
C
D
E
F
G
Figure 1
SAF SAF
SAF SAF
SJFZ SJFZ
Salton Trough Salton Trough
EF EF
36
present work we perform a detailed receiver function analysis in the southern California plate boundary
region (Figure 3.1). The study area includes the entire lengths of the SJFZ and EF, and covers the
southern SAF from the Mojave slightly NW of Cajon pass to the SE of the Salton Trough. The analysis
utilizes the detailed 3D double-difference tomography model of Allam and Ben-Zion (2012) for the
seismogenic crust around the SAF, SJFZ and EF, combined with the regional community velocity model
of Southern California Earthquake Center (SCEC) at larger-scales. The combined 3D velocity model is
used to construct separate effective 1D velocity models for each ray path of the teleseismic arrivals. The
results indicate prominent undulations in the Moho topography, which may be controlled to some extent
by the faults, and flat or dipping moho interface at most places across the faults. We also observe a large
vertical Moho offset (> 20 km) across the Salton Trough that may involve a Moho offset across the SAF
in the area, a steeply dipping or vertical Moho offset across the Mojave section of the SAF to the NW of
Cajon pass, and a vertical Moho offset across the SJFZ just SE of Cajon pass.

Figure 3.2: (a) The epicenter distribution of the teleseismic events used in this work. The
concentric circles represent approximately 30 and 100 degrees of great circle distance
from the study area. (b) The back azimuth distribution of the events.

3.3 Data and Pre-Processing
We use three component seismograms recorded by 188 broadband seismic stations across
Southern California (Figure 3.1). We select for analysis 145 teleseismic events occurring between
January 2001 and March 2004, with a magnitude range of 5.5 to 7.4 and epicentral distances from 30 to
100 degrees (Figure 3.2). The data pre-processing consists of the following steps. After removing the
mean and trend, the teleseismic waveforms are cut between 10 s before and 60 s after the calculated P
wave arrival times based on a travel time calculator (TauP), which uses the IASP91 velocity model
(Crotwell et al., 1999). Given the relatively long size of the used time window, imprecision in the
calculated P arrival times do not affect the results. The waveform segments are bandpass filtered with a
4 pole, 2-pass Butterworth filter with corner frequencies of 0.1 and 2 Hz, and a 0.2 s wide taper is
applied to the filtered data. The resulting horizontal component seismograms for each event are then
rotated to be parallel and perpendicular to the great circle path to obtain radial (R) and transverse (T)
seismograms.

10 20 30
Back Azimuth
No of Events
N
Figure 2
37
3.4 Receiver Function Techniques
The receiver functions are impulse responses of the earth near the receiver. To calculate radial and
transverse receiver functions, the vertical component of motion Z(t) is deconvolved from the R(t) or T(t)
seismograms, respectively, to eliminate source and instrument effects. The amplitude of radial receiver
functions correlates with the amount of impedance contrast. According to synthetic tests by Cassidy
(1992), changes in polarities in transverse receiver functions are indicative of a dipping interface.
Among a set of transverse receiver functions sorted in increasing order by their back-azimuths, the
transition from positive to negative polarities at a given time can indicate the direction of down-dip of
the corresponding interface.
The P receiver functions are estimated here using the iterative time domain deconvolution
algorithm of Ligorria and Ammon (1999). This is associated with summing Gaussian pulses with means
that are determined by the time-shift of the peaks in the cross-correlation of Z(t) and R(t). The accuracy
of the results is estimated by comparing synthetic radial seismograms, calculated by convolving the
obtained receiver function with the vertical component data, and the observed R(t). Our analysis
procedure involves adding iteratively 100 Gaussian pulses, each with a width of about 1.0 s. This
produces progressive fit changes at the final iterations below 0.01%, which is similar to the criterion
used by Ligorria and Ammon (1999). Figure 3.3a shows a representative set of receiver functions
calculated for station BBR. We eliminate receiver functions with RMS misfit between the final
estimated and observed radial waveforms larger than 40%, which corresponds to ~1/3 of the all
calculated receiver functions. Corresponding results for station BBR are shown in Figure 3.3b. We
explored using RMS misfits in wide range of values and found that the main conclusions of this work on
Moho topography and offsets (sections 3.4 and 3.5) do not change. There is no manual elimination of
receiver functions to avoid any subjective bias in the results.
38

Figure 3.3: Receiver functions from data of station BBR sorted by their back-azimuths.
The triangles and squares on the right sub-panels indicate, respectively, the back azimuth
and epicentral distance of each receiver function. (a) All (129) receiver functions with
positive amplitudes colored by the RMS misfit. (b) A set of 109 receiver functions with
RMS misfits <40%.

Dueker and Sheehan (1997) used a common midpoint (CMP) technique, which groups and stacks
receiver functions in a 1D velocity model with piercing points in the same geographic region after
move-out. In a further step utilizing on the CMP technique, Zhu (2000) developed the CCP method that
first maps the receiver functions along the ray paths and then stacks them in the spatial domain. In that
method, the time delay between P and converted Ps phase ( ∆T
ps
) is related to the depth (D) of the feature
causing the conversion

ΔT
ps
(p,V
s
(z),R
v
)= [ 1−p
2
V
S
2
(z)
D
0
∫
− R
V
−2
−p
2
V
S
2
(z)]dz (3.1)

where p is the ray parameter and R
V
is the ratio V
P
(z)/V
S
(z) of the P and S velocities assumed to be
constant. The ray parameter p can be estimated from the IASP91 model using the travel time calculator
(TauP) mentioned above. To avoid biases produced by strong lateral heterogeneities across the faults
(Schulte-Pelkum and Ben-Zion, 2012), we use in our analysis a modified CCP stacking procedure
utilizing a different effective 1D velocity model for each incoming teleseismic ray. The sets of effective
20 10 0
Time (sec)
0 100 200 300
Back Azimuth (deg)
50 100
Distance (deg)
20 10 0
Time (sec)
0 100 200 300
Back Azimuth (deg)
50 100
Distance (deg)
RMS Misfit
0% 100%
(a) (b)
Figure 3
39
1D velocity models are constructed from a 3D velocity model for the study region based on a
combination of the detailed 3D tomographic results of Allam and Ben-Zion (2012) for the region around
the San Jacinto fault and the larger scale SCEC Community Velocity Model-H (Magistrale et al., 2000;
S üss and Shaw, 2003; Plesch et al., 2011). The model of Allam and Ben-Zion (2012) is used for the
depth section 5-15 km of the large grey rectangular region in Figure 3.1, where it has reliable results,
and the SCEC Community model is used in the outer regions to that volume. The combined results are
smoothed using 3D Gaussians of 1 km width to avoid artifacts at the boundaries between the two
different models, while maintaining the velocity contrasts across the major faults present in the results of
Allam and Ben-Zion (2012).
To generate sets of effective 1D velocity models representative of the structures sampled by each
teleseismic ray, we use the 3D ray tracing of Sambridge and Kennett (1990) in the combined velocity
model to calculate the path of the incoming rays. Then we extract the velocity structure along each path
as a function of depth V
P
(z), which in turn is converted to V
S
(z) using V
P
/V
S
= 1.732. In principle, more
accurate results may be obtained by incorporating space variations of V
P
/V
S
. However, the results of
Allam and Ben-Zion (2012) for V
P
and V
S
are associated with significantly different amount of data,
leading to (spatially-variable) artifacts in V
P
/V
S
ratios. The same typically holds for other tomographic
images. We therefore perform our calculations using the more reliable V
P
results and fixed V
P
/V
S
ratio
of a Poisson solid. Miller et al. (2014) used a similar procedure with effective 1D velocity models along
each ray path in the CCP stacking algorithm, based on the SCEC community velocity model. The
incorporation here of the more detailed velocity model in the seismogenic depth section of the plate-
boundary region is expected to provide more accurate results.
We bin the obtained CCP stacks in a 3D grid with cubic cells of 1 km dimension. We analyze the
3D discrete scalar field in this grid to produce volumetric plots and various projections onto 2D profiles
by stacking the amplitudes from all bins within certain widths. The 2D profiles have a different grid size
to account for scattering of the incoming waves. We use rectangular cells with dimension of 5 km size
horizontally. This is approximately the radius of the first Fresnel zone
2 2
) 4 / ( D f V D r
S
− + =
(Sheriff, 1977), assuming dominant frequency f of 1 Hz, average V
S
value of 3.5 km/s and average Moho
depth D of 30 km. We first produce images for profiles along lines A-G of Figure 3.1 by projecting
results from cells that are within 40 km of the profiles. This width is chosen because it includes enough
cells to construct a reasonably continuous Moho across the profiles. We then check that our main
conclusions do not depend strongly on the used width by performing similar analyses for width values in
the range 20-60 km.
40

Figure 3.4: A histogram of common conversion point (CCP) stacks in each cells of the 3D
grid. A threshold value (0.15) indicated with a vertical black line.

To map and analyze the large-scale characteristics of the Moho depth variations for given regions,
we isolate the most coherent conversion phase within the depth range typical for Moho in southern
California. (This is done to produce Figures 3.9a and 3.11a in sections 3.4.2 and 3.4.3.) Figure 3.4 shows
the distribution of derived stack amplitudes in the study area. The value 0.15 may be used to distinguish
high amplitudes from low ones. We mark the depth z of the maximum amplitude for each (x, y) grid
point over the section deeper than 15 km that is larger than 0.15. We ignore the top 15 km to avoid near-
surface results associated with the direct P waves. We note that identifying the points of phase
conversion with such an automated procedure eliminates the need for labor intensive manual selection
(associated with ~27,000 receiver functions in our study), and also reduces possible subjective biases.
To eliminate outliers and obtain a more coherent picture for the overall Moho topography in the
entire study area, we use the following ‘horizontal coherency filter’. We calculate an incoherency
coefficient C given by,

4
) (
50
z z C
N
i
− =
∑
(2)

where N
50
indicates the 50 nearest neighbors, z
i
is the depth of the i
th
neighboring point and z is the depth
of the candidate point. The value of C is basically a measure of total difference in heights with the
neighboring points. We select a cutoff value C
0
based on the distribution of C among the candidate
points and eliminate points with C>C
0
. For the whole study area, we chose C
0
to be 5.0x10
5
, which
eliminates ~1/3 of the candidate points. We calculate the block-mean of the remaining points with non-
overlapping 50 km wide volumes for smoothing. The values at the centers of the blocks are then
connected by fitting a continuous curvature surface using the function “surface” in GMT package by
Wessel and Smith (1998) with a tension factor of 0.25. The resulting surface geometry depends on the
assumed tension factor, so such images are not suitable for analyzing details of the Moho such as offsets
across fault zones. However, this processing is useful for examining general regional trends, and we
found by trial and error that a tension factor of 0.25 is suitable for such a purpose. (The above procedure
0
1000
2000
Number
0.1 0.01 0.2 0.3 0.4 0.5
Amplitude
0.15
Figure 4
41
is used only to generate Figure 3.6 in section 3.4.) The results change somewhat for different filter
parameters N and C
0
but the large-scale characteristics remain similar.

Figure 3.5: A volumetric plot of the CCP stacks in the 3D grid. The coordinate system is
UTM in km with an origin at (32.47, -117.974). The X-axis is along the east-west
direction, the Y-axis is along the north-south direction, and the sea level has a zero depth
value. The color and opacity of each cell is determined by the local value of the CCP
stack. The color-opacity scale with respect to a white background is at the bottom. Red
colors correspond to positive values and indicate a sharp decrease in the velocity. The top
face of the box corresponds to the map in Figure 3.1. The viewpoints in both panels are at
a depth of 25 km. (a) The study area as seen from the east looking west. (b) The study
area as seen from the south looking north.

3.5 Results
The volumetric results over the employed 3D grid in the entire study area point to a complex Moho
geometry (Figure 3.5). Several persisting features include undulations along both fault-normal and
parallel directions to the three major faults and dipping surfaces. Figure 3.5a shows the volume from the
east looking to the west, with features exhibiting generally large complexity. To the south (right side of
the figure), the Moho phase becomes shallow and loses its coherency. Figure 3.5b shows the same
volume from the south looking north and includes a prominent Moho undulation around 30 km depth
manifested by two crests and a trough. To the east (right side of the figure), the Moho phase disappears
likely because of artifacts caused by early multiple arrivals associated with a shallow Moho in the SE.
Figure 5
Receiver Function Amplitude
South North
West East
-0.1 0.1
-0.04 0.08 -0.08 0 0.04
(a)
(b)
42
We examine further the regional aspects of the Moho depth by constructing a surface (Figure 3.6) using
the filter associated with equation (2).
We observe various large-scale characteristics that are consistent with previous studies in the area
(Zhu and Kanamori 2000, Lewis et al. 2000, Yan and Clayton 2007). The Moho boundary has a broad,
NS trending, deep band that reaches >40 km below Cajon pass (Figure 3.1), and extends to the Eastern
California Shear Zone in the north. The Moho depth is gradually decreasing to the SE and is ~20 km
near the Salton Trough. The thin crust in this region is believed to be the northern end of the Baja
California rift zone (Zhu and Kanamori, 2000). In the next subsection we discuss additional features that
may be associated with the SAF, SJFZ and EF by examining results at profiles normal to the faults.

Figure 3.6: The large-scale geometry of the inferred Moho interface colored by the
depth. The inferred Moho conversion points (indicated by colored squares) correspond to
depths of the cell with highest CCP stack processed through the ‘vertical coherency filter ’
associated with equation (2). The topography map is obtained by fitting a continuous
curvature surface to these using the ‘surface’ program from the GMT package (Wessel
and Smith, 1991, 1998) with a tension factor of 0.25. The other symbols and lines are as
in Figure 3.1.
-118.00˚ -117.50˚ -117.00˚ -116.50˚ -116.00˚ -115.50˚ -115.00˚
32.50˚
33.00˚
33.50˚
34.00˚
34.50˚
ADO
BBR
BBS
BFS
BLA2
BOM
BOR
BTC
CHN
CJM
CLT
CTC
DEV
DGR
DPP
DRE
ERR
IDO
IPT
JEM
LPC
LUC2
LUG
MGE
MLS
MSC
MSJ
MUR
NSS2
OLI
PDU
PER
PLM
PLS
PMD
PSD
PSR
RSS
RVR
RXH
SAL
SBPX
SLB
SLR
SNO
SNR
SVD
SWS
TOR
VTV
WES
WLT
WWC
BAR
BC3
BEL
BRE
DAN
EML
GOR
IKP
IRM
JVA
LLS
MCT
OLP
RIO
SBB2
SDD
SDG
SDR
SOF
SRN
STG
TA2
CPE
HWB
KNW
LVA2
MONP2
PFO
RDM
SMER
SOL
TRO
20 25 30 35 40
Depth (km)
Figure 6
A
B
C
D
E
F
G
DNR
BZN
CRY
FRD
SND
WMC
43
3.5.1 Profiles of the Stacks
Figure 3.7 displays projections of results from the 3D grid within 40 km of profiles A-G of Figure
3.1. Since the width of the projected volume is relatively large, we get a continuous Moho phase
throughout most of the profiles. However, each plot also shows

Figure 3.7: Amplitudes of CCP stacks projected onto the profiles indicated in Figure 3.1. All
depths are with respect to sea level. Positive amplitudes (red color) indicate a sharp velocity
decrease. Each projection is from a volume 40 km wide on each side of the profile line. Vertical
colored lines indicate locations of the surface traces of faults given in the legend. The gray
contours enclose 1 km x 1 km cells with minimum hit count of 1. The ray count is 0 close to the
surface and the margins. The 3D grid is projected and stacked within 5 km x 1 km cells (see text).
0
10
20
30
40
50
33.7 33.8 33.9 34.0 34.1 34.2 34.3 34.4 34.5 34.6 34.7 34.8 34.9 35.0 35.1
0
10
20
30
40
50
33.5 33.6 33.7 33.8 33.9 34.0 34.1 34.2 34.3 34.4 34.5 34.6 34.7 34.8
0
10
20
30
40
50
33.2 33.3 33.4 33.5 33.6 33.7 33.8 33.9 34.0 34.1 34.2 34.3 34.4 34.5 34.6
0
10
20
30
40
50
33.0 33.1 33.2 33.3 33.4 33.5 33.6 33.7 33.8 33.9 34.0 34.1 34.2 34.3
0
10
20
30
40
50
32.7 32.8 32.9 33.0 33.1 33.2 33.3 33.4 33.5 33.6 33.7 33.8 33.9 34.0 34.1
0
10
20
30
40
50
32.5 32.6 32.7 32.8 32.9 33.0 33.1 33.2 33.3 33.4 33.5 33.6 33.7 33.8
0
10
20
30
40
50
32.2 32.3 32.4 32.5 32.6 32.7 32.8 32.9 33.0 33.1 33.2 33.3 33.4 33.5 33.6
A.
B
C.
D.
E.
F.
G.
Latitude
Depth (km)
San Andreas Fault
San Jacinto Fault
Elsinore Fault
0.1
0.0
-0.1
Amplitude
Figure 7
44
The smooth image is obtained by bilinearly interpolating the center points of neighboring cells in
all profiles. The solid black line is a tentative Moho surface traced manually with cubic splines
over the maxima of the most coherent phase. The vertical dashed lines in profiles B and E are
possible Moho offsets across the SJFZ and at the boundary of a young rift surrounded by
continental crust.

variations of the Moho depth in the direction perpendicular to the profiles. This implies that one must be
careful in making conclusions based on these profiles alone. Our overall interpretations of the Moho
depth in each profile of Figure 3.7 are indicated by solid black lines, produced by manually tracing the
maxima of the most coherent phase deeper than 15 km with cubic Bezier splines. The robustness of
several inferred features of interest in profiles B and D across the SAF and SJFZ are examined by
varying the projection width, and verifying that the inferred lateral variations of the Moho depth are not
sensitive to the projection width.
There is a similar pattern of undulation in Figure 3.7 for the different profiles. The positions of
crests and troughs of the Moho topography generally coincide in profiles A through E where the results
are relatively clear. Since the orientation of the profiles are approximately perpendicular to the SAF,
SJFZ and EF, which strike approximately in the NW-SE direction, this similarity suggests that the
geometrical relation of these faults with the plate motion plays a dominant role in determining the
characteristics of the Moho topography along the NE-SW direction. The lack of a deep coherent phase
in profiles F and G in the SE section of the study area may be interpreted as a very thin crust near and to
the SE to the Salton Trough associated with the young oceanic crust. The shape of the rift, narrower in
the NW end and wider to the SE, results in a very narrow zone of thin oceanic crust that is surrounded
by thicker continental crust. Below we discuss further various features in each of the profiles.
Profile A is normal to the Mojave section of the SAF, just NW of Cajon Pass where the SJFZ
branches from the SAF. The average crustal thickness in this region is ~38 km, consistent with the
findings of Zhu and Kanamori (2000), Lewis et al. (2000) and Yan and Clayton (2007). The crust has a
thickness of ~35 km between latitudes 34.0N and 34.4N. There is a strong change in the slope of Moho
across the SAF (yellow vertical bar), as its thickness increases from ~35 km to ~42 km. This feature
may be associated with a vertical Moho offset rather than dipping surface as suggested by Zhu (2000).
We focus on this area further in section 3.4.2.
Profile B cuts the SAF and SJFZ just to the SE of Cajon Pass. The Moho is about 35 km deep in
the western portion of the profile. There is a gradual increase of Moho depth across the EF (green
vertical bar), and a sharp increase across the SJFZ (black vertical bar). The density of stations along this
profile, especially between and near SJFZ and SAF, is enough to make this sharp transition at SJFZ a
viable candidate for a vertical Moho offset of 10 km. There is no apparent discontinuity of the Moho
across SAF, where the crust is deepest at ~45 km and then gradually rises to a depth of ~31 km at the
eastern end of the profile. We also note the lack of clear Moho phase in the region below Cajon pass
(between the SAN and SJFZ) in this profile.
Profile C runs through a complex region of the SFJZ near the town of Hemet that include several
fault traces. The Moho has a depth of ~30 km in the west, and again it gradually increases from 32 km to
40 near the EF and SFJZ as in profile B. The transition across the EF is a candidate for a Moho offset of
~5 km. However, the two stations nearest to the EF are separated by a considerable (~7 km) along-fault-
distance. It is therefore difficult to determine whether this transition is an actual offset or an artifact
caused by a dipping continuous Moho. The changes across the SJFZ and SAF are best interpreted as
parts of the overall undulations mentioned before.
45
Profile D crosses the SJFZ near the transition between the simpler Anza seismic gap section to the
NW and the more complex trifurcation region to the SE. This profile includes the area studied by Miller
et al. (2014). We observe an overall undulation pattern of the Moho phase, as in the previous profiles,
that appears to be coherent throughout the profile. On the NE side of the SFJZ, the results show a
complex structure that lacks a coherent Moho phase adjacent to the fault, which coincides (Allam and
Ben-Zion, 2012; Allam et al., 2014) with significant low velocity damage zone in the upper crust. Miller
et al. (2014) suggested that there is a Moho offset of ~8 km across the SJFZ in this area, but this is not
seen clearly in our results. There is another candidate for a Moho offset of ~10 km near latitude 33.77,
which does not coincide with any major faults. This feature may be interpreted as part of a strong Moho
undulation within 20 km of the surface trace of the SAF that is at a crest of the undulation. We discuss
results from this region further in section 3.4.3.
Profile E crosses the SFJZ at the SE portion of the trifurcation area and passes through the NW
end of the Salton Trough adjacent to the SAF. Beneath the Salton Trough there is a distinct absence of
deep (~30 km) coherent Moho signal that seems to be replaced by a much shallower coherent signal
(~10 km). This feature is observed using receiver functions from 3 stations (NSS2, SLB and BOM) that
are due to the west of the SAF, and is also observed with profiles associated with projections of different
widths (Figure 3.8). We therefore conclude that the large change of Moho near the Salton Trough is
robust and likely associated with transition from the thick continental crust to a newly forming oceanic
crust. The NE end of this offset may coincide with the SAF at depth, if the SAF is dipping to the NE as
suggested by hypocenter locations and geodetic data in the region (Allam and Ben-Zion 2012; Lindsey
et al., 2013).
Profiles F and G also has signatures of the shallow Moho farther to the SE of the Salton Trough.
In profile F, the Moho phase starts at ~40 km in the west and gradually turns shallower with an ~15 deg
slope until it becomes indistinguishable from other phases that might be the multiples of the Pms phase.
The lack of any deep coherent phase in profiles F and G makes it difficult to interpret the Moho
topography, but given the transition from continental to oceanic crust in the area it is likely that the
results reflect relatively thin depth (~10 km deep).
46

Figure 3.8: Amplitudes of CCP stacks for profile E in Figure 3.7 using different
projection widths given in the upper right corner of each plot. The color scale and
symbols are as in Figure 3.7. The results associated with widths of 30, 50 and 60 km
support the robustness of the Moho offset seen in profile E.
Latitude
Depth (km)
San Andreas Fault
San Jacinto Fault
Elsinore Fault
0.1
0.0
-0.1
Amplitude
Figure 8
0
10
20
30
40
50
32.7 32.8 32.9 33.0 33.1 33.2 33.3 33.4 33.5 33.6 33.7 33.8 33.9 34.0 34.1
0
10
20
30
40
50
0
10
20
30
40
50
0
10
20
30
40
50
20 km
30 km
60 km
50 km
47

Figure 3.9: (a) Distribution of points corresponding to Moho conversion phases across
the Mojave section of SAF near Cajon Pass. The depths and locations of the points
correspond to the strongest phase conversions in the 3D grid with depths greater than 15
km and amplitudes larger than 0.15 (see section 3.3 for details). (b) Amplitudes of CCP
stacks projected from a section 40 km wide on each side onto profile AA ’ in (a). The
scales of horizontal and vertical axes are chosen to have aspect ratio of 1. The vertical
yellow line indicates the location of the surface trace of SAF. The other symbols and
lines are as in Figure 3.7.
ADO BFS CHF CJM CLT IPT LPC LUG MLS PDU RVR SBPX TA2
0
10
20
30
40
50
34.1 34.2 34.3 34.4 34.5
0.1
0.0
-0.1
Amplitude
Depth (km)
-118.00˚ -117.80˚ -117.60˚ -117.40˚ -117.20˚ -117.00˚
34.00˚
34.20˚
34.40˚
34.60˚
0 10
ADO
BFS
CHN
CJM
CLT
IPT
LPC
LUG
MLS
OLI
PDU
PSR
RSS
RVR
SBPX
SVD
VTV
WLT
RIO
SBB2
TA2
20
25
30
35
40
Depth (km)
(a)
(b)
Figure 9
Latitude
A
A’
A A’
48
3.5.2 NW of Cajon Pass
The geometrical properties of the SAF change strongly at Cajon pass, from a single major fault at
the Mojave to a complex set of branching faults SE of Cajon pass that include the SJFZ. As mentioned,
Zhu (2000) inferred from receiver function analysis

Figure 3.10: Amplitudes of CCP stacks for profile AA ’ in Figure 3.9 using different
projection widths given at the top of each plot. The color scale and symbols are as in
Figure 3.9. The results support the robustness of the Moho offset indicated in Figure 3.9b.

a Moho offset of 6-8 km across the Mojave section of the SAF, while our results (Figure 3.7, profile A)
may be interpreted as a strongly dipping Moho. Figures 3.9 and 3.10 give additional results attempting
to clarify the Moho geometry across the SAF to the NW of Cajon pass. Figure 3.9a shows the lateral
locations and depths of the strong conversion phases assumed to be associated with the Moho (section
3.4.3) around profile A of Figure 3.7.
More detailed cross-section views of these points are shown in Figure B1. The CCP stacks
projected onto the profile using data from 40 km on each side of the fault (Figure 3.9b) indicate
undulating Moho depth of 32-35 km to the SW of the SAF, and a less undulating depth of ~40 km to the
NE of the fault. The change in the average depth and undulation character of the most coherent high
amplitude phase suggest a possible Moho offset of ~8 km (dashed vertical line) right below the SAF
20 km 30 km
50 km 60 km
0.1
0.0
-0.1
Amplitude
Depth (km) Depth (km) Figure 10
Latitude Latitude
0
10
20
30
40
50
0
10
20
30
40
50
34.1 34.2 34.3 34.4 34.5 34.1 34.2 34.3 34.4 34.5
49
consistent with the results of Zhu (2000). To the NE of the SAF the inferred Moho depth gradually
decreases along the profile from ~43 km to ~38 km.
The depth map of the points corresponding to the Moho in Figure 3.9a further supports the
interpretation of an offset across the SAF. Between stations TA2 and LPC, there is a cluster of points on
the NE side of SAF with an average depth of 27 km. To the north of station TA2 there is another cluster
of points with average depth of 38 km. Since the along-fault distance between these clusters is very
small, it is likely that they correspond to the Moho on the opposite sides of a vertical offset. The depth
distribution of conversion points suggests further that the Moho offset disappears as one moves SE
along the SAF, as the difference in depths between clusters of points across the fault diminishes. The
Moho depth is also observed to be similar on the opposite sides of the SAF as the fault perpendicular
distance increases. These observations suggest that the difference in crustal thickness is a local feature in
this region. The discussed geometrical properties of the Moho across the SAF remain when using
projections of different widths (Figure 3.10).
3.5.3 Anza Region
Similarly to section 3.4.2, we revisit here the results of profile D in Figure 3.7 associated with the
SJFZ at the transition from the relatively simple Anza section to the complex trifurcation area. Miller et
al. (2014) inferred from receiver function analysis a vertical Moho offset of ~8 km in that area, and
concluded from synthetic calculations that this can not be an artifact caused by the velocity contrast
observed by Allam and Ben-Zion (2012) for this part of the fault. In an effort to clarify the Moho
geometry in that area, we show the lateral locations and depths of the Moho conversion phases (Figure
3.11a) and the CCP stacks projected onto the profile using data from 40 km on each side of the fault
(Figure 3.11b). More detailed cross-section views of these points are provided in Figure B1. We note
that the results are based on 3 additional stations (black triangles in Figure 3.11a) compared to the study
of Miller et al. (2014), one of which in close proximity of the trace of SJFZ, providing additional data on
the Moho structure near the fault.
The coordinates and depth values of the strong conversion points interpreted to be associated with
the Moho (Figure 3.11a) have a trend that suggests a dipping Moho to the NW. To test the existence of a
dipping interface, we inspected the transverse receiver functions from stations in the area. The back
azimuths corresponding to a polarity switch in these stations agrees with the results only in some of the
stations, which suggests that the local structure is more complicated than a flat dipping Moho. In
contrast to the results in section 3.4.2, we do not observe points with significant depth difference across
the fault consistent with a vertical Moho offset. The deep coherent phase of the projected CCP stacks
(Figure 3.11b) shows undulations on both sides of the fault with no clear evidence of a vertical Moho
offset. As noted in section 3.4.1, there is a region of high complexity to the NE of the SJFZ along the
profile in Figure 3.11b, which lacks a Moho phase for a 5 km lateral distance from the fault. This
apparent complexity in the Moho phase coincides with laterally, and may be an artifact of, a significant
low velocity damage zone in the upper crust (Allam and Ben-Zion, 2012; Allam et al., 2014).
50
The discussed features persist in profiles associated with projections of different widths (Figure
3.12). We speculate that a combination of a dipping Moho and the specific station configuration in this
region might produce an apparent Moho offset. On the NE side of the SFJZ most stations sample
relatively shallow Moho, whereas on SW side the more distributed stations sample both shallow and
deep parts of the Moho. A projection of CCP stacks from these stations onto a profile across the fault
will result in a shallow Moho to the NE and a deeper Moho to the SW, which might be interpreted as a
vertical offset.
3.6 Discussion
Receiver functions have been used extensively to image the Moho geometry in various regions
(Burdick and Langston, 1977; Jones and Phinney, 1998; Zhu and Kanamori, 2000; Wittlinger et al.,

Figure 3.11: (a) Same as Figure 3.9a for the Anza region. (b) Amplitudes of CCP stacks
projected from a section 40 km wide on each side onto profile BB ’ in (a). The scales of
horizontal and vertical axes are chosen to have aspect ratio of 1. The vertical black lines
indicate locations of surface traces of the SJFZ. The black triangles indicate additional
stations to those used in Miller et al. (2014).
(a)
(b)
-116.80˚ -116.60˚ -116.40˚
33.40˚
33.60˚
DNR
PLM
PMD
BZN
CRY
FRD
KNW
LVA2
PFO
RDM
SND
TRO
WMC
010
B
B’
Figure 11
20
25
30
35
40
Depth (km)
0.1
0.0
-0.1
Amplitude
0
10
20
30
40
50
33.5 33.6 33.7
BZN
CRY FRD
KNW
PFO
RDM
SND TRO
WMC
DGR
DNR SLR
B B’
Latitude
Depth (km)
51

Figure 3.12: Amplitudes of CCP stacks for profile BB ’ in Figure 3.11 using different
projection widths given at the top of each plot. The color scale and symbols are as in
Figure 3.11. The results support the robustness of the features discussed in the context
of Figures 3.7 and 3.11.

2004; Miller et al., 2014). In the present study we examine systematically variations of the Moho
interface in the southern California plate-boundary, using receiver functions from 145 events in 188
broadband stations, and detailed 3D-based velocity models for each ray path used to construct CCP
stacking. The results show that the Moho in the study area has highly complex topography with
prominent undulations throughout most of the region. The undulations are not correlated with
sedimentary basins and other strong velocity perturbations in the top few km of the crust in the area
(Allam and Ben-Zion, 2012; Zigone et al. 2014). At several places there are localized Moho features
including vertical offsets and relatively steep slopes. The observed large-scale characteristics of the
Moho structure are consistent with previous studies in the area (Zhu and Kanamori 2000, Lewis et al.
2000, Yan and Clayton 2007). We verify that the main discussed features are not sensitive to specific
parameter values (e.g., threshold value for RMS misfit) and details of the processing procedure (e.g.,
widths of zones used for projections along the profiles). Our method for mapping points of strong phase
33.5 33.6 33.7
0
10
20
30
40
50
33.5 33.6 33.7
33.5 33.6 33.7
20 km 30 km
50 km 60 km
0.1
0.0
-0.1
Amplitude
Depth (km) Depth (km)
33.5 33.6 33.7
0
10
20
30
40
50
Figure 12
Latitude Latitude
52
conversions within a 3D volumetric grid of the CCP stacks (Figure 3.5) allows for objective and
efficient generation of a regional Moho map (Figure 3.6) from large data sets. The highly complex
observed Moho topography makes the comparison of the Moho depth map with various cross-sections
of the stacks (Figures 3.7-3.12) essential for analyzing local features of the Moho such as vertical
offsets.
Lyakhovsky and Ben-Zion (2009) demonstrated with numerical simulations of evolving fault zone
structures that significant changes of the Moho depth are expected to suppress fault localization in the
brittle crust and broaden the surface deformation. This is consistent with our observations (Figure 3.6) of
significant changes of Moho depth around the complex region of the SAF near Cajon pass and further to
the SE, at the western boundary of the complex Eastern California Shear Zone, and around the Salton
Trough area.
We observe two prominent Moho offsets of ~20 km near the Salton Trough (profile E in Figure
3.7), at the boundary between the continental crust and a young rift it encloses. The estimated Moho
depth below and to the SE of the Salton Trough is only ~10 km. The eastern boundary of the thin rift
region may coincide with the SAF, depending on the dip angle of the SAF in this region, resulting in a
relatively large offset across a strike slip fault. Lekic et al. (2011) inferred a deeper Moho in the northern
Salton Trough of ~20 km. However, this result was based on H-κ stacking for a given station, which
may have missed some of the lateral variations observed in this paper. If one interprets the largest
positive Sp phase beneath the Salton Trough of Lekic et al. (2011) as the Moho, its depth would be
significantly closer to the ~10 km depth that we infer (K. Fischer, pers. communication, 2014).
We investigate in detail two regions (green boxes in Figure 3.1), where previous receiver function
studies inferred ~8 km vertical offsets across the Mojave section of the SAF (Zhu, 2000) and the Anza
section of the SJFZ (Miller et al. 2014). Our analysis supports the existence of a Moho offset across the
SAF slightly to the NW of Cajon pass, although the Moho there may have a steep dip rather than a sharp
offset (Figures 3.9-3.10). However, the derived results do not indicate a vertical Moho offset across the
SJFZ near Anza (Figures 3.11-3.12). We suggest that an apparent vertical offset in that region may be
produced by the geometry of the station distribution at the surface above a dipping Moho. On the other
hand, the obtained images point to a possible vertical Moho offset across the SJFZ slightly to the SE of
Cajon pass (profile B in Figure 3.7). The localities of the inferred Moho offsets across the SAF and
SJFZ in close proximity to Cajon pass suggests that the branching of the two major faults there may be
related to the Moho offsets.
As discussed by Schulte-Pelkum and Ben-Zion (2012), local velocity structures in the brittle upper
crust near major faults can bias images based on receiver functions. In addition to overall velocity
contrasts, significant local damage zones can obscure receiver function results in sections adjacent to a
fault or near structural complexities (e.g., to the right of the SJFZ and below Cajon Pass in profiles D
and B of Figure 3.7), and may play a part in possible erroneous inferences at such locations. It is worth
noting that anisotropy, which may characterize damage zones near faults and shear zones, can also bias
interpretations of receiver function results (Schulte-Pelkum and Mahan, 2014). This is not accounted for
in our work but may be a subject of continuing study.
3.7 Acknowledgements
This chapter benefited from useful comments of two anonymous reviewers and Karen Fischer. We
thank Iain Bailey for providing a core code for CCP stacks and Meghan Miller and Amir Allam for
discussions. The data used in this work are archived and distributed by the southern California
earthquake data center. The study was supported by the National Science Foundation (grant EAR-
0908903).

53
Appendix B: Supplementary Figures

Figure B1: (a) Moho conversion points near Cajon Pass in Figure 9 separated into 4 groups based on
their perpendicular distances from the profile line AA’. Each group is plotted with a different color
using same horizontal and vertical scales. (b) Projections of the points in (a) onto profile AA’. The
offset in Moho depth across the SAF is clearly visible especially among the blue points close to the
fault.
54

Figure B2: (a) Moho conversion points near the Anza region in Figure 11 separated into 4 groups based
on their perpendicular distances from the profile line BB ’. Each group is plotted with a different color
using same horizontal and vertical scales. (b) Projections of the points in (a) onto profile BB ’. No clear
offset is seen in any of the groups.
55
4 Estimating Attenuation Coefficients and P-Wave Velocities of the
Shallow San Jacinto Fault Zone from Betsy Gunshots Data Recorded
by a Spatially Dense Array with 1108 Sensors
4.1 Abstract
We estimate values of P wave velocity and P attenuation coefficients (𝑄
!
) for the subsurface
material at the Sage Brush Flat site along the Clark branch of the San Jacinto Fault Zone. The data are
generated by 33 Betsy gunshots and recorded by a spatially dense array of 1108 vertical component
geophones deployed in a rectangular grid that is approximately 600 m x 600 m. We automatically pick
the arrival times of the seismic body waves from each explosion arriving at stations within 200 m. These
measurements are used to derive an average velocity map with velocity values ranging from 500 m/s to
1250 m/s. We estimate the energy of the early P waves by squaring the amplitudes in a short window
relative to the automatic picks. These energies are fitted to a decay function representing the geometrical
spreading and intrinsic attenuation. By separating the stations into spatial bins and calculating
attenuation values for each by linear regression, we construct a 𝑄
!
values map. Most of the 𝑄
!
values
are in 5-20 range, which is consistent with other studies of shallow fault zone regions.

4.2 Introduction
The attenuation properties of the shallow crust play an important role in waveform modeling,
tomography, hazard analysis and various other topics. The velocities in these layers is known to be
relatively slow compared to the rest of the crust, which makes them important in ground motion
estimation.
Despite its importance, very little is known about the properties of these top layers, especially in
fault zones, where the material can be damaged, sometimes asymmetrically (Rempe et al., 2013), due to
high dynamic stress changes. The damage asymmetry is indicative of bimaterial interfaces, which can
cause a preferred rupture direction, hence having implications in seismic hazard. Attenuation
coefficients are believed to be correlated with damage and can be studied using various techniques.
The very shallow part of the crust is difficult to study if the region lacks any seismic
instrumentation and/or boreholes, since otherwise none of the ray paths of any of the direct phases
reaching seismic instruments in other locations pass through it. The dense array we deployed inside and
around the damage zone of Clark branch of SJFZ is thus very important in the sense that it gives us a
unique opportunity to study fault zone properties in high resolution, particularly the distribution of
damaged zones.
Most common methods for calculating attenuation involves correlation of background seismic noise
and inversion techniques, which is, despite the possibility of giving high-resolution results, not an
intuitive method and can suffer from non-uniqueness.
In this study, we develop a new, intuitive method based on linear regression analysis made on
energy decay of seismic waves generated in explosions. The results give us insight about the distribution
of attenuation and damage around the fault zone in an intuitively understandable way.
4.3 Data
We installed 1108 vertical-component geophones on the SJFZ. The array was shaped as a
rectangular grid with dimensions of approximately 600 m x 600 m around the Clark Fault, which is the
main branch off of SJFZ south-east of Anza. The instrument spacing was 10 m fault normal direction
56
(20 m in SW end of the array) and 30 m in fault parallel direction (Figure 4.1). Data were recorded
continuously at 500 Hz between May 7th and June 13th in 2014. We performed 33 explosions with the
Betsy gun inside the network. Each explosion was next to an instrument, at approximately ~1 m
distance. The GPS used to locate the explosions was not as accurate as the one used to locate the
instruments. For this reason we used the location of the closest station to the explosion as a proxy for the
location of the explosion. However, our analysis has been designed in a way that doesn’t require the
accuracy of coordinates of the explosions to be smaller than a few meters.

Figure 4.1: The station distribution of the array. The colored circles are stations; the
color indicates the relative median energy calculated for the whole duration of the
experiment in seismometer units (count
2
). White lines are the fault traces. White stars
enclose the stations with explosions done next to them. The regional map is given in the
upper left inset. The red rectangle encloses the area shown in the lower left inset, where
the location of the array is indicated with the red square on the local map.

These explosions generated coherent signals within ~500 m. Figure 4.2 is a section plot from
explosion 27 and is representative of rest of the data from other explosions. The direct wave phase can
be seen to be overtaken by a head-wave, probably from deeper sub-horizontal layers, around 150-200 m
distance from the explosion.

57

Figure 4.2: Section plot for explosion 27. Colors indicate the velocity amplitudes
normalized for each seismogram. Distance is the 3D linear distance from the explosion to
the station. Each horizontal line is a stack of seismograms for one or more stations,
distances of which are within the same 1 m interval.
4.4 Method
Amplitude of particle velocity of a seismic plane wave propagating in an inelastic medium obeys

𝐴 𝑟 =𝐴
!
exp −
𝜋𝜈𝑟
𝑐𝑄

(4.1)
where 𝐴
!
is the maximum amplitude, 𝑟 is the distance from the source, 𝑐 is the velocity in the medium,
𝜈 is the frequency and 𝑄 is the attenuation coefficient (Aki and Richards, 2002). This is correct if 𝑟 is
large enough so that the plane wave approximation is valid. For regions close to the source, we also have
to consider the geometrical spreading effect. Due to conservation of energy, the seismic amplitude from
a point source in an elastic medium decays as

𝐴 𝑟 𝛼
1
𝑟

(4.2)
By combining equations (4.1) and (4.2), we model the amplitude of the seismic wave at a station at a
distance 𝑟 from the source as
𝐴(𝑟)=
𝐴
!
𝑟
exp −
𝜋𝜈𝑟
𝑐𝑄
!
(4.3)
We convert (4.3) to units of energy by taking the square and replacing some of the coefficients,
58
𝐸(𝑟)=
!
!
!
!
exp w
!
𝑟 , (4.4)
where 𝑤
!
is a coefficient related to total energy generated by the explosion, site and instrument
responses and 𝑤
!
is the attenuation coefficient related to 𝑄
!
by
𝑄
!
=−
!!"
!!
!
. (4.5)
We take the logarithm of both sides in equation (4.3),
log[𝐸𝑟
!
]= log𝑤
!
+𝑤
!
𝑟 (4.6)
log[𝐸𝑟
!
]=𝑤
!
!
+𝑤
!
𝑟 (4.7)
We use ordinary least squares to estimate the parameters 𝑤
!
!
and 𝑤
!
that give us the smallest RMS error
in equation (4.7). To get rid of the outliers, we exclude any data points if the following is not valid:
abs (log[𝐸𝑟
!
]) −𝑤
!
!
−𝑤
!
𝑟<𝑇
!
. We chose the threshold value 𝑇
!
to be 2.0 by visual inspection of
data. The results are not sensitive to this number. We then recalculate 𝑤
!
!
and 𝑤
!
with the remaining data
points using linear regression once more.
We ignore stations farther than 200 m, the average distance at which the direct wave starts being
taken over by deeper phases. We also ignored stations closer than 20 m because the seismograms from
stations at these distances are at risk for clipping.
In order to gain spatial resolution, the stations are separated into 4 bins based on their azimuths in
roughly fault parallel and fault normal directions. We fit the equation (4.7) separately to each of these
bins and obtain values for 𝑤
!
and 𝑤
!
for each bin.
In order to convert the coefficient 𝑤
!
to 𝑄
!
for each bin, we use the dominant frequency 𝜈 of the
station that is closest to the explosion in the same angular range, since the dominant frequency at that
station is most representative of the source frequency distribution. We use the average 𝑐 at each bin.
Details of estimating 𝜈 and 𝑐 for individual stations are given in sections 4.3.1 and 4.4.2.
We ignore the whole bin if the 𝑄
!
value is negative. Otherwise, we assign this 𝑄
!
value to all the
stations in the bin. We repeat this process for all the bins at all 33 explosions. We then visually inspect
the section plots and the accuracy of the picks and throw away the results from the explosions whose
picks are off.

59

Figure 4.3: The energy vs distance plots for explosion 27 for each azimuthal bin. The
points are colored based on the bin they are in, locations of which are shown in the insets
on the map. The NE and SE quadrants yield negative Q values and therefore have been
excluded in the final stack.
4.4.1 Calculating The Body Wave Velocities
We pick the arrival times (𝑇
!
!
) at all the stations. We use the automatic picker of Ross and Ben-
Zion (2014) with parameters LTA window length=8.0 s, STA window length=0.5 s, trigger on
threshold=5.0, trigger off threshold=3.0 (Figure 4.4); and then we apply the following to the picks:
(i) We use ordinary linear regression to fit the average velocity of the direct wave 𝑐 and the
origin time 𝑇
!
to the travel time equation

𝑇
!
=
𝑟
𝑐
+𝑇
!
(4.8)
(ii) We exclude any picks that are not within a threshold of 0.1 s of the 𝑇
!
to get rid of the
outliers.
60

Figure 4.4: The normalized traces from explosion 27. The inset shows the area indicated
with the red rectangle. The red and blue elongated (+) symbols indicate the accepted and
rejected automatic picks respectively. The black sloping straight line indicates the best fit
line for all the accepted automatic picks with origin being fixed to the hand-picked origin
time.

We apply (i) and (ii) a second time with a threshold of 0.05 seconds to the remaining picks to
further refine the fit. The estimated 𝑐 is the average velocity for each explosion (Figure 4.4). We also
obtain individual velocities for each station using the equation (4.8), which we use to derive the average
velocity map in Figure 4.5.

𝑐
!
=
!
!
!
!!
!
!
, (4.10)
These velocities (𝑐
!
) associated with each station at each explosion correspond to average
velocities between the explosions and stations. We generate a 5mx5m grid of cells and calculate ray
paths for each station-explosion pair. We then assign corresponding velocities of each ray to the cells
they go through and average the values in each cell from all crossing rays. The results are not aimed to
give a detailed representation of the velocity structure but rather give a general order-of-magnitude idea
of the P wave velocities in the region (Figure 4.5).

61

Figure 4.5: Velocity map obtained by averaging the velocities of the rays from each
explosion to each station within 200 m over a grid of 5m x 5m cells.
4.4.2 Calculating the Body Wave Energies
We calculate the median energy in the time window starting from 0.02 s before 𝑇
!
!
and is 0.2 s long
to avoid the outliers in the amplitudes of the seismograms. The chosen interval covers direct waves,
which is confirmed by visually inspecting the seismograms. The median energy values should robust
and still representative of the energy level at the station.

4.4.3 Calculating the Dominant Frequencies
We calculate the power spectral density (PSD) using the multi-taper method (Thomson 1982) at
each station for the same window we use to calculate the median energy mentioned in chapter 4.4.2. The
frequency with the maximum expected power in the PSD is the dominant frequency of the direct wave
phase at the station.
62
As expected, the dominant frequency decreases with increased distance from the explosion (Figure
4.6).

Figure 4.6: (a) The dominant frequency from all the stations for explosion 27 as function
of distance. The red horizontal line indicates the median of the dominant frequencies at 25
Hz. (b) The PSD from all the stations for the same explosion. Red ‘x’s mark the maximum
of the PSD’s. The red bars are the histograms showing the distribution of dominant
frequencies.

4.4.4 Stacking the results
We stack the 𝑄 values obtained from all the explosions in the vicinity at each station and calculate
the median. If a station has less than 3 Q values associated with it, we ignore the station in the final
results. Otherwise, the median Q value is our final result for that station.
4.5 Results
We calculated Q values for 27 explosions. In the end we obtained Q values for 900 stations. The Q
values lie mostly in the 5-20 range, which is in agreement with the results by Liu et al. (2015). The Q
values are presented in Figure 4.7.
The Q values in a given location seem to agree among different explosions. The errors can be due
to the noise in observations, complex geometry of propagating wavefronts and maybe damage
anisotropy.
The region near the fault trace with relatively high Q values (~10) is expected since the fault zones
tend to consist of damaged material. However, this damage zone extends outside the fault zone on the
NE side of the fault. Considering that the velocities on this side of the fault are also low, this could
indicate damage asymmetry in this region with the NE side consisting of more damaged material.
The SW side of the fault seems to have a more rigid structure with the Q values ranging from 5-10.
The NW corner of the array has the highest seismic velocities with a maximum Vp of ~1250 m/s.
63

Figure 4.7: The Q values based on the stacked values. Stations with less than 3 results
have been ignored. The histogram shows the number of stations with given Q values in the
final stack.

4.6 Discussion
We demonstrated the new method we developed for calculating Q values in a spatially dense
seismic network. Calculating the Q values by linear regression gives us an intuitive understanding of the
relationship between the data and the results.
64
The velocities that we calculated for converting the coefficients in linear regression to Q values
also gives us an average idea about the seismic velocities of body waves which are most likely P waves.
It is possible to perform an inversion to obtain a more accurate velocity map, which is beyond the scope
of this chapter.
One aspect of the method that we have ignored is the site effect. We implicitly made the
assumption that the site effects at each station do not correlate with the distances from the explosion.
This might be true up to a degree but it is reasonable to expect some kind of a smooth spatial distribution
of site effects, which would affect the fit of the energy function 𝐸(𝑟).
It is possible to apply this method to larger scale networks with possibly a few changes, such as the
size and the shape of the bins, the geometry of the rays of different phases and also incorporating site
effects. Study large areas can be done very quickly since the computational requirements of the method
are minimal.
4.7 Acknowledgements
We are grateful to Bud Wellman for all of his help and permission he gave us to install the
instruments on his land. We also thank Geoff Davis, Jon Meyer, Pieter-Ewald Share, Xin Liu, Cooper
Harris, Maxwell Dalquist and Hongrui Qiu for their help with the field work. Rob Clayton and Paul
Davis kindly provided the Betsy gun.
65
5 Conclusion
In the presented chapters, we have explored the crustal structure, mostly near fault zones, in different
scales and depths. The results indicate how complex the crust can be in complicated tectonic settings.
We have studied major fault zones in Southern California and North Anatolia, which have similar
aspects; including vertical dip, tectonic setting and cumulative slip.
The results underline different extremes in terms of fault zone crustal structure. In Chapter 2, a
velocity contrast, which could be as high as 8.3%, across North Anatolian Fault, for which there is no
evidence in receiver function studies that supports the existence of a Moho offset, is observed using a
novel method of comparing teleseismic arrival times. In contrast, the receiver function study in Chapter
3 shows evidence of major offsets in Moho across San Andreas Fault and many complicated structures
and Moho topographies in Southern California.
The conclusion supported by these findings is that major Moho offsets and high velocity contrasts
are two extremes, both or any mix of which could result in sharp changes in characteristics across fault
zones observed in many fault zones. Hence, one must take both into consideration in any type of fault
zone or crustal modeling.
For future work, it is possible to use the methods in the presented studies for different networks and
with different scales. The method in Chapter 2 can be used in any array covering a fault zone. The
simplicity of the methodology makes it a viable tool for detecting velocity contrasts and quantifying it.
There are several arrays installed on different sections of many major faults around the earth. The fact
that the method is automated makes it possible and very easy to use it on data sets from such faults.
The specialized receiver function technique in Chapter 3 can be used in any region for which we
have a detailed 3D velocity model. The existence and accuracy of such a model increases the accuracy
of our crustal structure and enables us to make a better interpretation of the driving forces behind it.
It is also possible to incorporate the 3D velocity structure better in inversion techniques involving
receiver functions. The main limitation in receiver function inversion techniques is the lack of sensitivity
to lateral variations. Inclusion of the effects of the lateral velocity changes using the modified CCP stack
technique in our study can enable us to resolve the structures near fault zones in inversion techniques.
The technique in Chapter 4 for estimation of attenuation coefficient can very easily be applied to
any type of array or network of stations with some modifications. The most obvious choice would be the
US Array for obtaining a continent scale Q-value map. For such large scales, using earthquakes instead
of active sources is much more feasible. The modifications to the method would have to account for the
radiation pattern and the different geometrical spreading factor for such events, which would be more
complex than what is in our study.

66
References
Aki, K. and Richards, P. G. (2002). Quantitative seismology. .
Allam, A. and Ben-Zion, Y. (2012). Seismic velocity structures in the southern California plate-
boundary environment from double-difference tomography. Geophysical Journal International,
190(2), 1181--1196.
Allam, A., Ben-Zion, Y., Kurzon, I., and Vernon, F. L. (2014). Seismic velocity structure in the Hot
Springs and Trifurcation Areas of the San Jacinto Fault Zone. Geophys. J. Int., in press.
Ben-Zion, Y. (1989). The response of two joined quarter spaces to SH line sources located at the
material discontinuity interface. Geophysical Journal International, 98(2), 213--222.
Ben-Zion, Y. (1990). The response of two half spaces to point dislocations at the material interface.
Geophysical Journal International, 101(3), 507--528.
Ben-Zion, Y. and Andrews, D. (1998). Properties and implications of dynamic rupture along a material
interface. Bulletin of the Seismological Society of America, 88(4), 1085--1094.
Ben-Zion, Y. and Shi, Z. (2005). Dynamic rupture on a material interface with spontaneous generation
of plastic strain in the bulk. Earth and Planetary Science Letters, 236(1), 486--496.
Beyreuther, M., Barsch, R., Krischer, L., Megies, T., Behr, Y., and Wassermann, J. (2010). ObsPy: A
Python toolbox for seismology. Seismological Research Letters, 81(3), 530--533.
Bourne, S., England, P., and Parsons, B. (1998). The motion of crustal blocks driven by flow of the
lower lithosphere and implications for slip rates of continental strike-slip faults. Nature,
391(6668), 655--659.
Bulut, F., Ben-Zion, Y., and Bohnhoff, M. (2012). Evidence for a bimaterial interface along the
Mudurnu segment of the North Anatolian Fault Zone from polarization analysis of P waves.
Earth and Planetary Science Letters, 327, 17--22.
Cakir, Z., Akoglu, A. M., Belabbes, S., Ergintav, S., and Meghraoui, M. (2005). Creeping along the
Ismetpasa section of the North Anatolian fault (Western Turkey): Rate and extent from InSAR.
Earth and Planetary Science Letters, 238(1), 225--234.
Cambaz, M. D. and Karabulut, H. (2010). Love-wave group velocity maps of Turkey and surrounding
regions. Geophysical Journal International, 181(1), 502--520.
Cassidy, J. (1992). Numerical experiments in broadband receiver function analysis. Bulletin of the
Seismological Society of America, 82(3), 1453--1474.
Crotwell, H. P., Owens, T. J., and Ritsema, J. (1999). The TauP Toolkit: Flexible seismic travel-time
and ray-path utilities. Seismological Research Letters, 70(2), 154--160.
Dor, O., Yildirim, C., Rockwell, T. K., Ben-Zion, Y., Emre, O., Sisk, M., and Duman, T. Y. (2008).
Geological and geomorphologic asymmetry across the rupture zones of the 1943 and 1944
earthquakes on the North Anatolian Fault: possible signals for preferred earthquake propagation
direction. Geophysical Journal International, 173(2), 483--504.
Dueker, K. G. and Sheehan, A. F. (1997). Mantle discontinuity structure from midpoint stacks of
converted P to S waves across the Yellowstone hotspot track. Journal of Geophysical Research:
Solid Earth (1978--2012), 102(B4), 8313--8327.
England, P. and McKenzie, D. (1982). A thin viscous sheet model for continental deformation.
Geophysical Journal of the Royal Astronomical Society, 70(2), 295--321.
Fialko, Y. (2004). Probing the mechanical properties of seismically active crust with space geodesy:
Study of the coseismic deformation due to the 1992 Mw7. 3 Landers (southern California)
earthquake. Journal of Geophysical Research: Solid Earth (1978--2012), 109(B3).
67
Gilbert, L. E., Scholz, C. H., and Beavan, J. (1994). Strain localization along the San Andreas fault:
consequences for loading mechanisms. Journal of Geophysical Research: Solid Earth (1978--
2012), 99(B12), 23975--23984.
Hauksson, E. (2000). Crustal structure and seismicity distribution adjacent to the Pacific and North
America plate boundary in southern California. Journal of Geophysical Research: Solid Earth
(1978--2012), 105(B6), 13875--13903.
Hong, T.-K. and Menke, W. (2006). Tomographic investigation of the wear along the San Jacinto fault,
southern California. Physics of the earth and Planetary Interiors, 155(3), 236--248.
Jones, C. H. and Phinney, R. A. (1998). Seismic structure of the lithosphere from teleseismic converted
arrivals observed at small arrays in the southern Sierra Nevada and vicinity, California. Journal
of Geophysical Research: Solid Earth (1978--2012), 103(B5), 10065--10090.
Lekic, V., French, S. W., and Fischer, K. M. (2011). Lithospheric thinning beneath rifted regions of
southern California. Science, 334(6057), 783--787.
Lenglin é, O. and Got, J.-L. (2011). Rupture directivity of microearthquake sequences near Parkfield,
California. Geophysical Research Letters, 38(8).
Lewis, J. L., Day, S. M., Magistrale, H., Eakins, J., and Vernon, F. (2000). Regional crustal thickness
variations of the Peninsular Ranges, southern California. Geology, 28(4), 303--306.
Ligorria, J. P. and Ammon, C. J. (1999). Iterative deconvolution and receiver-function estimation.
Bulletin of the Seismological Society of America, 89, 303-306.
Lindsey, E. and Fialko, Y. (2013). Geodetic slip rates in the southern San Andreas Fault system:
Effects of elastic heterogeneity and fault geometry. Journal of Geophysical Research: Solid
Earth, 118(2), 689--697.
Lyakhovsky, V. and Ben-Zion, Y. (2009). Evolving geometrical and material properties of fault zones
in a damage rheology model. Geochemistry, Geophysics, Geosystems, 10(11).
Magistrale, H., Day, S., Clayton, R. W., and Graves, R. (2000). The SCEC Southern California
reference three-dimensional seismic velocity model version 2. Bulletin of the Seismological
Society of America, 90(6B), S65--S76.
McGuire, J. and Ben-Zion, Y. (2005). High-resolution imaging of the Bear Valley section of the San
Andreas Fault at seismogenic depths with fault-zone head waves and relocated seismicity.
Geophysical Journal International, 163(1), 152--164.
McNally, K. C. and McEvilly, T. V. (1977). Velocity contrast across the San Andreas fault in central
California: small-scale variations from P-wave nodal plane distortion. Bulletin of the
Seismological Society of America, 67(6), 1565--1576.
Megies, T., Beyreuther, M., Barsch, R., Krischer, L., and Wassermann, J. (2011). ObsPy--What can it
do for data centers and observatories?. Annals Of Geophysics, 54(1), 47--58.
Meyer, B., Tapponnier, P., Bourjot, L., Metivier, F., Gaudemer, Y., Peltzer, G., Shunmin, G., and Zhitai,
C. (1998). Crustal thickening in Gansu-Qinghai, lithospheric mantle subduction, and oblique,
strike-slip controlled growth of the Tibet plateau. Geophysical Journal International, 135(1), 1--
47.
Miller, M. S., Zhang, P., and Dolan, J. F. (2014). Moho structure across the San Jacinto fault zone:
Insights into strain localization at depth. Lithosphere, 6(1), 43--47.
Mitchell, T., Ben-Zion, Y., and Shimamoto, T. (2011). Pulverized fault rocks and damage asymmetry
along the Arima-Takatsuki Tectonic Line, Japan. Earth and Planetary Science Letters, 308(3),
284--297.
68
Molnar, P. (1992). Brace-Goetze strength profiles, the partitioning of strike-slip and thrust faulting at
zones of oblique convergence, and the stress-heat flow paradox of the San Andreas fault.
International Geophysics, 51, 435--459.
Oppenheimer, D. H., Reasenberg, P. A., and Simpson, R. W. (1988). Fault plane solutions for the 1984
Morgan Hill, California, earthquake sequence: Evidence for the state of stress on the Calaveras
fault. Journal of Geophysical Research: Solid Earth (1978--2012), 93(B8), 9007--9026.
Ozakin, Y., Ben-Zion, Y., Aktar, M., Karabulut, H., and Peng, Z. (2012). Velocity contrast across the
1944 rupture zone of the North Anatolian fault east of Ismetpasa from analysis of teleseismic
arrivals. Geophysical Research Letters, 39(8).
Ozakin, Y. and Aktar, M. (2012). The Moho topography of Turkey from P-receiver functions.
Geophys. J. Int. in review.
Ozakin, Y. and Ben-Zion, Y. (2014). Systematic Receiver Function Analysis of the Moho Geometry in
the Southern California Plate-Boundary Region. Pure and Applied Geophysics, 1--18.
Özeren, M. S. and Holt, W. E. (2010). The dynamics of the eastern Mediterranean and eastern Turkey.
Geophysical Journal International, 183(3), 1165--1184.
Peltzer, G. and Tapponnier, P. (1988). Formation and evolution of strike-slip faults, rifts, and basins
during the India-Asia collision: An experimental approach. Journal of Geophysical Research:
Solid Earth (1978--2012), 93(B12), 15085--15117.
Peng, Z. and Ben-Zion, Y. (2004). Systematic analysis of crustal anisotropy along the Karadere —
D üzce branch of the North Anatolian fault. Geophysical Journal International, 159(1), 253--274.
Peng, Z. and Ben-Zion, Y. (2005). Spatiotemporal variations of crustal anisotropy from similar events
in aftershocks of the 1999 M7. 4 Izmit and M7. 1 D üzce, Turkey, earthquake sequences.
Geophysical Journal International, 160(3), 1027--1043.
Plesch, A., Tape, C., Graves, J., Small, P., Ely, G., and Shaw, J. (2011). Updates for the CVM-H
including new representations of the offshore Santa Maria and San Bernardino basin and a new
Moho surface. In 2011 SCEC Annual Meeting, Proceedings and Abstracts (pp. B128).
Rempe, M., Mitchell, T., Renner, J., Nippress, S., Ben-Zion, Y., and Rockwell, T. (2013). Damage and
seismic velocity structure of pulverized rocks near the San Andreas Fault. Journal of
Geophysical Research: Solid Earth, 118(6), 2813--2831. http://dx.doi.org/10.1002/jgrb.50184
Ross, Z. E. and Ben-Zion, Y. (2014). An earthquake detection algorithm with pseudo-probabilities of
multiple indicators. Geophysical Journal International, 197(1), 458--463.
Roux, P., Wathelet, M., and Roueff, A. (2011). The San Andreas Fault revisited through seismic-noise
and surface-wave tomography. Geophysical Research Letters, 38(13).
Sambridge, M. and Kennett, B. (1990). Boundary value ray tracing in a heterogeneous medium: a
simple and versatile algorithm. Geophysical Journal International, 101(1), 157--168.
Schmidt, D., B ürgmann, R., Nadeau, R., and d'Alessio, M. (2005). Distribution of aseismic slip rate on
the Hayward fault inferred from seismic and geodetic data. Journal of Geophysical Research:
Solid Earth (1978--2012), 110(B8).
Schulte-Pelkum, V. and Mahan, K. H. (in press). Imaging faults and shear zones using receiver
functions. Pure Appl. Geophys.
Schulte-Pelkum, V. and Ben-Zion, Y. (2012). Apparent Vertical Moho Offsets under Continental
Strike-Slip Faults from Lithology Contrasts in the Seismogenic Crust. Bulletin of the
Seismological Society of America, 102(6), 2757--2763.
Sengör, A., T üys üz, O., Imren, C., Sakinc, M., Eyidogan, H., G ör ür, N., Le Pichon, X., and Rangin, C.
(2005). The North Anatolian fault: A new look. Annu. Rev. Earth Planet. Sci., 33, 37--112.
69
Shapiro, N. M., Campillo, M., Stehly, L., and Ritzwoller, M. H. (2005). High-resolution surface-wave
tomography from ambient seismic noise. Science, 307(5715), 1615--1618.
Sheriff, R. E. (1977). Limitations on Resolution of Seismic Reflections and Geologic Detail Derivable
from Them: Section 1. Fundamentals of Stratigraphic Interpretation of Seismic Data. .
S üss, M. P. and Shaw, J. H. (2003). P wave seismic velocity structure derived from sonic logs and
industry reflection data in the Los Angeles basin, California. Journal of Geophysical Research:
Solid Earth (1978--2012), 108(B3).
Thomson, D. J. (1982). Spectrum estimation and harmonic analysis. Proceedings of the IEEE, 70(9),
1055--1096.
Wallace, R. E. (1990). The San Andreas fault system, California. US Government Printing Office
Washington, DC.

Weertman, J. (1980). Unstable slippage across a fault that separates elastic media of different elastic
constants. Journal of Geophysical Research: Solid Earth (1978--2012), 85(B3), 1455--1461.
Wessel, P. and Smith, W. H. (1991). Free software helps map and display data. Eos, Transactions
American Geophysical Union, 72(41), 441--446.
Wessel, P. and Smith, W. H. (1998). New, improved version of Generic Mapping Tools released. Eos,
Transactions American Geophysical Union, 79(47), 579--579.
Wittlinger, G., Vergne, J., Tapponnier, P., Farra, V., Poupinet, G., Jiang, M., Su, H., Herquel, G., and
Paul, A. (2004). Teleseismic imaging of subducting lithosphere and Moho offsets beneath
western Tibet. Earth and Planetary Science Letters, 221(1), 117--130.
Yan, Z. and Clayton, R. (2007). Regional mapping of the crustal structure in southern California from
receiver functions. Journal of Geophysical Research: Solid Earth (1978--2012), 112(B5).
Zaliapin, I. and Ben-Zion, Y. (2011). Asymmetric distribution of aftershocks on large faults in
California. Geophysical Journal International, 185(3), 1288--1304.
Zhao, P., Peng, Z., Shi, Z., Lewis, M. A., and Ben-Zion, Y. (2010). Variations of the velocity contrast
and rupture properties of M6 earthquakes along the Parkfield section of the San Andreas fault.
Geophysical Journal International, 180(2), 765--780.
Zhu, L. and Helmberger, D. V. (1998). Moho offset across the northern margin of the Tibetan Plateau.
Science, 281(5380), 1170--1172.
Zhu, L. (2000). Crustal structure across the San Andreas Fault, southern California from teleseismic
converted waves. Earth and Planetary Science Letters, 179(1), 183--190.
Zhu, L. and Kanamori, H. (2000). Moho depth variation in southern California from teleseismic
receiver functions. Journal of Geophysical Research: Solid Earth (1978--2012), 105(B2), 2969--
2980.
Zigone, D., Ben-Zion, Y., Campillo, M., and Roux, P. (in review). Seismic Tomography of the
Southern California plate boundary region from noise-based Rayleigh and Love Waves. Pure
Appl. Geophys.

Abstract (if available)

Abstract The crust along major strike-slip fault zones goes under tremendous changes during their existence. On the large scale, different parts of crust with different lithologies and/or geometries can come into contact as a result of large cumulative slips, interaction with other fault zones can create complicated Moho topographies. On smaller scales, recurring unilateral rupture can create asymmetrical damage, which affects the distribution of attenuation coefficients of body waves. Because of intrinsic ambiguity in techniques for estimating crustal structure, it is difficult to resolve the ambiguity of seismic velocity vs. crustal thickness, since both can have the same effect in observations. Here, we investigate the existence of Moho offsets in Southern California, seismic velocity contrasts across North Anatolian Fault and distribution of attenuation around San Jacinto Fault Zone using special techniques, developed to resolve such ambiguities. Moho offsets are studied using receiver function techniques and incorporate a detailed 3D velocity model obtained from a double difference tomography study, which enables us to overcome the ambiguity mentioned above. The results show evidence that supports major offsets across 3 major faults, ranging between ~8 to ~20 km. On North Anatolian fault, using data from a small seismic network, we compare observed teleseismic arrival times with expected ones. The results show a prominent difference in arrival times between stations that are on different sides of the fault. The ratio of average seismic velocities for the crust calculated using these differences is 8.3% if the contrast is confined to the upper half of the crust. To study the shallow crustal properties, we use data from an active source experiment from a dense array on San Jacinto Fault Zone. We calculate Q values for body waves by linear regression analysis to estimate the decay rate of the energy in 33 shots. We also estimate an average velocity structure using the automatic picks from the shot data. The Q values range from 5-30, which is much lower than Q values associated with greater depths. The body wave velocities range from 500 m/s to 1250 m/s and both show strong local variations.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

High-resolution imaging and monitoring of fault zones and shallow structures: case studies in southern California and on Mars

PDF

Mapping the Moho in southern California using P receiver functions

PDF

Statistical analyses of ambient seismic noise spectra with applications to detecting imperfectly diffuse wave field and deriving attenuation and phase velocity information

PDF

Structural clustering analysis of CVMS-4.26: a 3D seismic velocity model for southern California

PDF

Multi-scale imaging and monitoring of crustal and fault zone structures in southern California

PDF

Applying automated techniques to large seismic datasets for systematic analyses of phases, source, and structure

PDF

Multi-scale imaging of the fault zone velocity structure: double-difference tomography, inversion of fault zone headwaves, and fault zone sensitivity kernels

PDF

Stress-strain characterization of complex seismic sources

PDF

Seismic velocity structure and spatiotemporal patterns of microseismicity in active fault zones of Southern California

PDF

Multi-scale damage signatures across major strike-slip faults

PDF

Multi-scale imaging of major fault zones in Southern California

PDF

Volumetric interactions between major ruptures and fault zones illuminated by small earthquake properties

PDF

Detailed properties of seismic waveforms and earthquake sources in Southern California

PDF

Heterogeneity of earthquake stress drops, focal mechanisms and active fault zones

PDF

Analysis of waveform and catalog data of aftershocks for properties of earthquakes and faults

PDF

Detection and modeling of slow slip events as creep instabilities beneath major fault zones

PDF

Symmetry properties, pulverized rocks and damage architecture as signatures of earthquake ruptures

PDF

Spatiotemporal variations of stress field in the San Jacinto Fault Zone and South Central Transverse Ranges

PDF

Paleoseismologic and slip rate studies of three major faults in southern California: understanding the complex behavior of plate boundary fault systems over millenial timescales

PDF

Microseismicity, fault structure, & the seismic cycle: insights from laboratory stick-slip experiments

Asset Metadata

Creator Ozakin, Yaman (author)

Core Title Elements of seismic structures near major faults from the surface to the Moho

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Geological Sciences

Publication Date 10/20/2015

Defense Date 03/23/2015

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag attenuation,earth's crust,Moho discontinuity,OAI-PMH Harvest,receiver functions

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Ben-Zion, Yehuda (committee chair), Lototsky, Sergey (committee member), Sammis, Charles (committee member), Vernon, Frank (committee member)

Creator Email dandik@gmail.com,ozakin@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-190923

Unique identifier UC11277923

Identifier etd-OzakinYama-3978.pdf (filename),usctheses-c40-190923 (legacy record id)

Legacy Identifier etd-OzakinYama-3978-1.pdf

Dmrecord 190923

Document Type Dissertation

Format application/pdf (imt)

Rights Ozakin, Yaman

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA