Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Multi-scale damage signatures across major strike-slip faults
(USC Thesis Other)
Multi-scale damage signatures across major strike-slip faults
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
MULTI-SCALE DAMAGE SIGNATURES ACROSS MAJOR STRIKE-SLIP FAULTS
by
Neta Wechsler
________________________________________________________________________
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)
August 2010
Copyright 2010 Neta Wechsler
ii
Dedication
This work is dedicated with love to Ronen, without whom it would not exist.
iii
Acknowledgements
First and foremost, I would like to thank my adviser, Yehuda Ben-Zion, whose
belief in my abilities and tenacity in guiding me towards scientific excellence was
instrumental in bringing this work to fruition. I would also like to thank Tom Rockwell,
my co-adviser and much more, who helped me in any way he could including letting me
into his home and family. I thank Judi Chester for all of her help with the core project and
for the use of the probe and SEM. Gary Girty is thanked for his help with thin sections
and geochemistry of the core samples. Thanks are due to Shari Cristofferson for her work
on the second chapter. Chapter 3 could not have been written without the dedicated work
of Emily Allen. My special thanks to Cindy Waite, the department’s academic advisor,
for her devoted help with all the gnarly administrative happenstances that seemed to
plague me continuously.
Many people contributed to this work in many different ways, and I am very
thankful to all of them. They include Joan Kimbrough, Ramon Arrowsmith, David
Tarboton, Ray Guillemette, Emily Brodsky, Amir Sagi, Ken Hudnut, the SDSU
quaternary lab students, and the USC geophysics group.
I would like to express my gratitude to my family, for bestowing on me a sense of
importance for studying and research, and for their loving support that reached me from
the other side of the planet. Last, but surely not least, I am eternally grateful to my friend
and husband Ronen Gersman for his endless patience and endurance, for spending time
helping me with the research, either in the field or on the computer, and for all of his
unconditional encouragement and love.
iv
Table of Contents
Dedication ii
Acknowledgements iii
List of Tables vi
List of Figures vii
Abstract x
Introduction 1
Chapter 1: Evolving Geometrical Heterogeneities of Fault Trace Data 7
Chapter 1 Abstract 7
Introduction 8
Data and Methods 13
Data 13
Measured Quantities 19
Binning 22
Controlling Variables 24
Results 28
Discussion 41
Chapter 2: Application of High Resolution DEM Data to Detect Rock Damage from
Geomorphic Signals along the Central San Jacinto Fault 47
Chapter 2 Abstract 47
Introduction 48
Regional Setting 53
Data and Methods 56
Data 56
Analysis Methods 60
Chosen Drainages Approach 63
Spatial Approach 64
Results 67
Chosen Drainages Approach 67
SRTM Dataset 67
LiDAR Dataset 71
Spatial Approach 74
SRTM Dataset 74
LiDAR Dataset 77
Discussion 83
Conclusions 87
Chapter 3: Characterization of Pulverized Granitoids in a Shallow Core along the San
Andreas Fault, Little Rock, CA 90
Introduction 90
Core Characterization 94
Geological Settings 94
Lithology and Structure 96
v
Particle Size Distribution 100
Sample Preparation 101
Automated Method Calibration 101
Results 105
Geochemistry 107
Methods 107
Results 109
XRF Elemental Analysis 109
XRD Mineralogy 113
Discussion 115
PSD and D-values 115
Whole Rock Chemistry 120
Surface Weathering 122
Origin of Pulverization 123
Conclusions 125
Chapter 4: Particle Size Analysis of Damage Textures in Fault Zone Rocks 127
Introduction 127
Overview 127
Particle Size Distribution 129
Material and Methods 131
Rock Samples 131
Damage Textures 133
SEM Image Analysis 134
Image Processing 137
Measured Parameters 138
Electron Microprobe 139
Results 143
SEM Image Analysis 143
PSD 143
Shape Analysis 144
Probe 146
Discussion 150
PSD 150
Affects of Mineralogy on PSD and Shape 152
Processes of Particle Break-up 154
Fracture Surface Energy 154
Conclusions 155
Summary 157
Bibliography 162
Appendices
Appendix 1: Little Rock Creek core logs 179
Appendix 2: XRF Data 196
vi
List of Tables
Table 1.1: Fault zones, abbreviation codes, number of segments in each fault zone, cumulative slip
and estimated variation in slip 16
Table 1.2: Statistics of circular data – various parameters 24
Table 1.3: Additional fault zone parameters, related to segment lengths 34
Table 2.1: Geological formations and their description for the LiDAR dataset locations 58
Table 2.2: Comparison analysis of drainages delineated from the SRTM dataset 69
Table 2.3: Comparison analysis of drainages delineated from the LiDAR dataset 72
Table 2.4: Comparisons of drainage density in different geological units 79
Table 2.5: Comparisons of drainage density between offset rock bodies along the fault 81
Table 3.1: point counts of core thin sections 97
Table 4.1: D-values and cut-off values for each region and for each magnification 140
Table 4.2: Probe maps elemental combinations and interpretations 141
Table 4.3: Mineralogical composition results derived from the elemental mapping 148
Table B-1: XRF major and trace elements data for surface and core samples, as well as CIA
values, median grain size in microns, depth in meters and lithology 196
vii
List of Figures
Fig. 1.1: Active fault map of California with the faults used in this study 14
Fig. 1.2: Diagrammatic representation of the analysis of a fault zone 19
Fig. 1.3: Segment length L
i
vs. segment misalignment 25
Fig. 1.4: (a) Same as Fig. 3 for two fault zones. (b) Same as a, but for the equal interval bins 26
Fig. 1.5: Rose diagrams of the segments for each fault zone 29
Fig. 1.6: The relation between the circular standard deviation/circular standard error (a and b,
respectively), the misalignment from plate motion direction and the cumulative slip for
each fault zone 30
Fig. 1.7: The relation between the circular standard deviation/circular standard error (a and b,
respectively), the misalignment from plate motion direction and the recent slip rate for
each fault zone 30
Fig. 1.8: The relation between stdev(Yi) and the cumulative slip (a), and slip rate (b), between Y
and the cumulative slip (c), and slip rate (d), and between the sum of segments lengths
and the cumulative slip (e), and slip rate (f) 32
Fig. 1.9: The relation between stdev(Zi) and the cumulative slip (a), and slip rate (b), and between
Z and the cumulative slip (c), and slip rate (d) 33
Fig. 1.10: Slope (m-value) of the linear fit to the equal number bins using cutoff value of 1000 m
vs. cumulative slip (a) and slip rate (b) 35
Fig. 1.11: Circular standard deviation ν vs. Slope (m-value) of the exponential fit to the equal
interval bins 36
Fig. 1.12: Number of segments larger than a certain length (Nf) as a function of segment length in
meters (L) on a log-log scale for all fault zones 38
Fig. 1.13: Same as Fig. 12, with log Nf as the y-axis 39
Fig. 1.14: Obtained b-values (eqs. 11, 12) vs. cumulative slip (a), slip-rate (b) and misalignment
(c) 40
Fig. 1.15:
% 1
L versus cumulative slip (a) and slip rate (b) 41
Fig. 2.1: Schematic illustration of damage asymmetry 51
Fig. 2.2: Regional setting of the study area 54
Fig. 2.3: Location of the LiDAR datasets relative to the mapped fault traces 59
viii
Fig. 2.4: An example of the results of drainage delineation along the SJF using the LiDAR
dataset 62
Fig. 2.5: Covariance function of down-slope distance vs. length scale 66
Fig. 2.6: Geology and location of chosen drainages with hypsometric integral values derived from
SRTM data 68
Fig. 2.7: Spatial drainage density and slope for the SRTM dataset 75
Fig. 2.8: A map of spatial Dd using an averaging filter with 100 m radius on the LiDAR data 77
Fig. 2.9: Results of the spatial approach using the LiDAR dataset 80
Fig. 2.10: A comparison of offset rock bodies from the two sides of the fault 82
Fig. 3.1: Location map of the drill site 94
Fig. 3.2: A simplified geological map of the study area 95
Fig. 3.3: Photomicrographs of core samples 98
Fig. 3.4: (a) A photomicrograph of grain mount, sample LR41. (b) A comparison of microscope
grain diameter measurements with measurements of fractions in the Horiba analyzer 102
Fig. 3.5: A comparison of the PSD from the classical sieve-pipette method (SP) and the automated
Horiba-Camsizer (HC) measurement 104
Fig. 3.6: Median particle size vs. depth 106
Fig. 3.7: An example of D-value calculation for sample LR18, a granite from 7.4 m and D-value
vs. median particle size 107
Fig. 3.8: Silica variation diagrams of selected major and trace elements 108
Fig. 3.9: Depth variation diagrams and major shears and alteration bands, as well as gouge zones
in the core 110
Fig. 3.10: Comparisons of major and trace elements for core samples 111
Fig. 3.11: Core samples bulk compositions plotted in A-CN-K and A-CNK-FM space 112
Fig. 3.12: XRD diffractograms of three samples 114
Fig. 3.13: PSD curvature, represented by b 116
Fig. 3.14: (a) PSD on a log-log plot for samples with different D-values. (b) Height ratio vs. D-
value for the same samples as in (a). (c) An example of fitting the sum of two Gaussians
to sample’s PSD 119
Fig. 3.15: Comparisons of major and trace elements 121
ix
Fig. 4.1: Location map of the drill site 133
Fig. 4.2: Whole section scans under polarized light of (a) sample 0307 and (b) sample 1168A 135
Fig. 4.3: Schematic of procedure for grain size analysis 137
Fig. 4.4: An example of elemental mapping 142
Fig. 4.5: Combined log-log plot for entire range of magnification for sample 0307 (left), and
individual D-values for each magnification separately (right) 144
Fig. 4.6: Combined log-log plot for entire range of magnification for sample 1168A (left), and
individual D-values for each magnification separately (right) 144
Fig. 4.7: Average shape factors of samples, per magnification 145
Fig. 4.8: Elemental mapping of two zones in the cataclastic region in sample 0307 147
Fig. 4.9: Elemental mapping of two zones in the cataclastic region in sample 1168A 149
Fig. 4.10: Comparisons of PSD derived from SEM measurements and laser particle analyzer 150
Fig. 4.11: SEM images 152
Fig. A-1: The complete log of the first Little Rock core (1.5 – 40.2 meters depth) 179
Fig. A-2: The complete log of the second Little Rock core (35-42.7 meters depth) 192
x
Abstract
This thesis presents a compilation of results from studies of active fault zone
geometry, structural properties, and macro- and micro-scale damage fabrics. Multi-scale
observations using a wide range of techniques were made along the transform plate
boundary of the Pacific and the North-American plates. A new method for quantifying
fault trace heterogeneity using Geographic Information System was outlined and used on
the database of active faults in California. Several parameters were defined for
quantifying fault trace heterogeneity and the range or dispersion in the data. The
cumulative slip and slip rate proved effective measures of fault zone maturity.
High resolution topographic models acquired by remote-sensing techniques were
utilized to demonstrate how damage to the host rock is related to drainage development
about a fault, and how the drainage density can be used as a proxy to study damage zone
geometry. A strong correlation between drainage density and proximity to the fault was
interpreted as an effect of degree of rock damage. Results of damage mapping using
drainage density indicate that the northeast side of the SJF is generally more damaged.
The observed asymmetry could be geological evidence for a preferred rupture
propagation direction, because a preferred propagation direction is predicted to produce
asymmetric damage structure that would be recorded in the volume of rock surrounding a
fault. The fault damage zone, as inferred from drainage properties, is more pronounced
near areas of complexities in the surface trace. Heterogeneities seen in the fault trace can
create stress concentrations and are correlated with observations of higher damage levels.
xi
An extensive exploration of the properties of a damage zone phenomenon –
pulverized granitic rocks, was performed with an objective of characterizing their
chemistry and the changes they undergo due to their proximity to the San Andreas Fault.
X-ray Diffraction and X-ray Fluorescence were used to describe the rocks composition,
thin sections were used to describe mineralogy and damage fabrics, and a laser particle
analyzer was employed to measure the particle size distribution (PSD). SEM images were
used to study the PSD of specific damage textures in pulverized rocks. Additionally, a
microprobe was used to study the mineralogy of the fragments and the matrix.
The mean particle size in pulverized rocks is mostly fine sand and silt in size and
the rock’s constituent minerals react similarly to deformation, with no observed size
preference in the mineral fractions. The rocks contain evidence of shear deformation, and
the observed damage is not homogeneous in its intensity. Fluid-rock interactions and
calcite metasomatism occur in spatial context with secondary faults. Observed clay
minerals are a result of faulting and damage, not surface weathering. Results point to a
mixed-mode deformation of pulverization and shear in the rocks, in conjunction to non-
fractal particle size reduction processes, such as mineral alteration. Taking into account
pulverized rocks, shear zones, and a wide damage zone, the amount of energy required
for creating and maintaining that damage zone is on the order of a few percent of the total
released seismic energy.
1
Introduction
The characterization of the properties of large plate-boundary fault zones is crucial
for understanding earthquakes and deformation mechanisms in the crust. The expression
of faults at the surface is geometrically complex and often consists of multiple segments,
branching structures, step-overs and curves. These and other structural heterogeneities
can strongly affect the nucleation, propagation, radiation and arrest of earthquake
ruptures associated with a given fault zone, as well as the associated seismicity patterns
(e.g., Ben-Zion, 2008, and references therein).
Another aspect of fault zone evolution with wide implications for earthquake
physics is its internal geological structure. The commonly used model of fault zone
structure in the brittle crust is a fault core surrounded by damage zone. The entire fault
zone is defined as the volume of rock about the fault that exhibits higher deformation
intensity compared with the background levels (Chester et al. 2004). The fault core is
relatively narrow, and most of the slip is localized within it (Chester and Chester 1998). It
can also be a wider zone that contains multiple strands of fault cores (Faulkner et al.,
2003; Faulkner et al., 2008). Typical core structures include altered and comminuted
rocks such as gouge, foliated cataclasites and ultracataclasites (Chester et al., 1993;
Cowie and Scholz, 1992; Flinn, 1977; Sibson, 1977; Wibberley and Shimamoto, 2003).
Elements that define the fault core, such as fine grain size and mineral alteration, are
thought to reflect high shear strain, extreme comminution, and enhanced fluid-rock
interactions (Chester et al., 2004).
2
The fault core is surrounded by a much wider damage zone which is characterized
by lower intensity deformation features such as subsidiary faults, fractures and
microfractures, pulverized rocks, localized alteration zones and mineralization. The
damage intensity decreases away from the fault core, reaching background levels in the
surrounding host rock (e.g. Chester et al., 2004; Mitchell and Faulkner, 2009; Wilson et
al., 2003). The damage zone geometry and the damage intensity can vary along the fault,
and different faults may have different damage zone geometry. Although it appears that
the thickness of gouge and ultracataclasite layers and total thickness of fault zones
generally increase with displacement (e.g. Savage and Brodsky, 2010), the distribution of
deformation within a fault zone is highly variable.
The importance of understanding fault zone structure and properties is in its
implications for understanding the processes that govern earthquakes. From the structure
it is possible to infer the fault’s mechanical behavior (Ben-Zion and Shi, 2005; Biegel
and Sammis, 2004; Faulkner et al., 2003), the stress field surrounding it (Zoback et al.,
1987; Faulker et al., 2006), and fluid flow properties (Miller et al., 2004). The properties
and evolution of the damage zone are some of the most poorly constrained aspects of
fault zone structure. The importance of the damage zone dictates the need for quantitative
field data on its physical dimensions, the intensity and distribution of damage.
So far two aspects of fault zone evolution were mentioned – fault-trace
heterogeneity and damage zone development. These two aspects of fault zone evolution
can sometimes seem contradicting. In terms of the surficial expression of earthquakes and
faulting, models of fault zone evolution predict regularization of the structural
3
heterogeneities through smoothing of geometrical asperities (e.g., Candela et al., 2009,
Sagy et al., 2007, Ben-Zion and Sammis, 2003). From fault trace evolution studies it is
predicted that as faults mature and accumulate slip, they coalesce into a localized, narrow
zone of deformation, leaving only a relict damage zone around them (Tchalenko, 1970;
Wesnousky, 1988; 1994). On the other hand, structural and geological studies of active
and exhumed fault zones conclude that concurrently with slip localization there is a
development and widening of the surrounding damage zone as slip accumulates. It
follows that the fault zone as defined above by Chester et al. (2004) is actually getting
wider, not narrower with cumulative slip.
Some of the most intriguing damage phenomena, pulverized rocks, have been
described in the damage zone of several large displacement faults (Dor et al., 2006;
Mitchell et al. 2009; Rockwell et al. 2009; Wechsler et al. 2009). They contain very little
shear and maintain the original igneous texture, yet they are highly damaged, incohesive,
and full of microfractures. Their extent in the damage zone, either in depth or in width is
not well known, but they have been observed in some places at distances of up to 300 m
away from the main fault trace. These rocks are thought to be produced during the
passage of an earthquake (Rockwell et al. 2009), but they have not yet been observed
around passively exhumed faults, leading to the question of whether or not they are only
shallow features. The amount of energy spent in the creation and maintenance of the
damage zone is poorly constrained, in part due to incomplete knowledge about its
geometry.
4
This thesis focuses on active fault zones, and it encompasses observations on their
geometry, structural properties, and macro- and micro-scale damage fabrics. Multi-scale
observations using a wide range of techniques were made along the transform plate
boundary of the Pacific and the North-American plates. The techniques employed cover a
myriad of geological and geophysical methods at scales varying from microns to
kilometers, and include GIS analysis, remote-sensing, structural and geological mapping,
chemical analysis with X-ray diffraction and fluorescence, electron and optical
microscopy, and image analysis.
The first chapter discusses the evolution of faults with time and cumulative slip, by
studying their surface traces (Wechsler et al., 2010). This is a large scale study which
uses fault trace maps of a transform plate boundary. A new method for quantifying fault
trace heterogeneity is outlined and used on the database of active faults in California.
Several quantities are considered as potential controlling variables on structural
complexities of fault traces: 1. the misalignment of the fault zone orientation from the
plate motion direction, 2. the total slip accommodated by the fault zone, and 3. the slip
rate across the fault as measured by geodesy. Through correlation between the parameters
that quantify heterogeneity on a fault and the potential controlling variables, I show that
cumulative slip and slip-rate have significant correlation with the fault trace
heterogeneity, while misalignment does not.
In the second chapter the geometry of the damage zone is studied in a scale of
hundreds of meters (Wechsler et al., 2009). High resolution topographic models acquired
by remote-sensing techniques are used to demonstrate how damage to the host rock is
5
related to drainage development about a fault, and how the drainage density can be used
as a proxy to study damage zone geometry. I use GIS and ALSM/LiDAR (Airborne
Laser Swath Mapping, or Light Detection And Ranging) data to analyze drainage density,
and to study the damage zone geometry about the San Jacinto. The observed damage
zone width using this method is a few hundreds of meters, and an asymmetric damage
distribution is found across the fault.
The third and fourth chapters discuss the results of an extensive exploration of the
properties of pulverized granitic rocks. The scale of observations ranges from outcrop
mapping and hand samples to micron scales. The third chapter describes results from
shallow coring into pulverized granitic rocks in the Mojave section of the San Andreas
fault, with the objective of characterizing the chemistry of the rocks and the changes they
go through due to their proximity to a large fault. The study describes the composition of
pulverized granites using XRD and XRF, their mineralogy and observed damage in thin
sections, and their particle size distributions (PSD) using a laser particle analyzer. The
original goal of this study was to find and eliminate any surface weathering signal that
previous authors (e.g. Rockwell et al., 2009) supposed was affecting their PSD in
pulverized rocks.
The fourth chapter delves into the issue of PSD further, using SEM images to study
the PSD of pulverized rocks and their damage textures. Thin section observations of
shear and cataclastic types of deformation within the pulverized rocks led to the
supposition that the different damage textures would have different PSD. SEM was used
to focus on specific deformation textures and measure their PSD separately, using thin
6
sections from the core. Additionally, a microprobe was used to study the mineralogy of
the fragments and the matrix. The goal was to better understand the processes that created
this unique damage textures in the damage zone.
The observations and analyses discussed in chapter 1 elucidate the relation between
fault evolution and surface trace expression. Chapter 2 demonstrates a new remote-
sensing technique for mapping damage zone extent and gives an example from the San-
Jacinto fault. Chapters 3 and 4 contain a thorough description of an important and, until
recently, looked-over rock damage phenomenon – pulverized rocks, and the possible
implications that those rocks have on the study of earthquakes and faults.
By combining field observations and new data analysis techniques, this work
contributes to the development of a stronger connection between observations and theory
in the earthquake research field.
7
Chapter 1: Evolving Geometrical Heterogeneities of Fault Trace Data
Published in Geophysical Journal International, June 1, 2010.
Co-Authors: Yehuda Ben-Zion
1
and Shari Christofferson
1
1. Department of Earth Sciences, University of Southern California Los Angeles, CA 90089
Chapter 1 Abstract
We perform a systematic comparative analysis of geometrical fault zone
heterogeneities using derived measures from digitized fault maps that are not very
sensitive to mapping resolution. We employ the digital GIS map of California faults
(version 2.0) and analyze the surface traces of active strike-slip fault zones with evidence
of Quaternary and historic movements. Each fault zone is broken into segments that are
defined as a continuous length of fault bounded by changes of angle larger than 1°.
Measurements of the orientations and lengths of fault zone segments are used to calculate
the mean direction and misalignment of each fault zone from the local plate motion
direction, and to define several quantities that represent the fault zone disorder. These
include circular standard deviation and circular standard error of segments, orientation of
long and short segments with respect to the mean direction, and normal separation
distances of fault segments. We examine the correlations between various calculated
parameters of fault zone disorder and the following three potential controlling variables:
cumulative slip, slip rate, and fault zone misalignment from the plate motion direction.
The analysis indicates that the circular standard deviation and circular standard error of
8
segments decrease overall with increasing cumulative slip and increasing slip rate of the
fault zones. The results imply that the circular standard deviation and error, quantifying
the range or dispersion in the data, provide effective measures of the fault zone disorder,
and that the cumulative slip and slip rate (or more generally slip rate normalized by
healing rate) represent the fault zone maturity. The fault zone misalignment from plate
motion direction does not seem to play a major role in controlling the fault trace
heterogeneities. The frequency-size statistics of fault segment lengths can be fitted well
by an exponential function over the entire range of observations.
Introduction
The observation of structural heterogeneities in earthquake fault zones is
ubiquitous. In a homogeneous solid, dynamic branching during propagation of
earthquake ruptures provides a generic mechanism for generation of geometrical fault
zone heterogeneities (e.g., Kame and Yamashita, 1999, Poliakov et al., 2002, Sharon et
al., 1995). On the other hand, repeating ruptures within the same region are expected to
regularize the structural heterogeneities through smoothing of geometrical asperities (e.g.,
Candela et al., 2009, Sagy et al., 2007) and creation of damage zones and bimaterial
interfaces that can suppress branching (e.g., Ben-Zion and Andrews, 1998, Brietzke and
Ben-Zion, 2006). These expectations are generally compatible with a wide variety of
laboratory and field data, which indicate that faults are formed with high initial
geometrical heterogeneity, and that increasing offset leads to the emergence of simpler
through-going structures on which subsequent large-scale motion is localized (Ben-Zion
9
and Sammis, 2003, and references therein). However, owing to the limited resolution and
complexity of the available data, the quantification of evolving fault zone heterogeneities
as a function of key governing variables remains a challenging task.
On time scales of several large earthquake cycles, the existing structural
heterogeneities can strongly affect the nucleation, propagation, radiation and arrest of
earthquake ruptures (e.g., Aochi et al., 2000, Papageorgiou and Aki, 1983, Rice et al.,
1994), as well the frequency-size distribution, temporal statistics and other properties of
earthquake population on the fault (Ben-Zion and Rice, 1993, Hillers et al., 2007, Zoller
et al., 2006). It is thus important to obtain robust quantitative measures of structural
heterogeneity that can characterize the evolutionary stage and geometrical properties of
active faults. Wesnousky (1994), Stirling et al. (1996) and de Joussineau and Aydin
(2009) characterized structural heterogeneity as the density of steps per unit distance
along the surface traces of faults, and concluded that the stepover density decreases with
increasing total slip on the fault. As discussed below, while this conclusion appears to be
correct, the stepover density does not provide a full characterization of the structural
complexity in relation to fault mechanics, and the evolution of fault heterogeneities may
depend on additional variables.
Fault zones with regular stepover offsets at similar intervals may have high or low
offset-density, but both cases are geometrically and mechanically simpler than faults with
a diversity of different stepover distances at irregular intervals (e.g., Ben-Zion, 2008, and
references therein). Since the equations of elasticity governing stress transfer are scale
invariant, uniform changes of all length scales, including the density of fault stepovers,
10
only shifts (to first order) the earthquake response to different scales without changing
the response type (e.g., the form of earthquake statistics). On the other hand, changes in
the degree of regularity or diversity of fault stepovers can both shift to different scales
and change the form of the resulting earthquake properties. For these reasons, Ben-Zion
and Rice (1993, 1995) and Ben-Zion (1996) suggested that the range of size scales
characterizing fault heterogeneities is a better parameter than the density of fault
stepovers. Ben-Zion et al. (2003), Zöller et al. (2005), Hillers et al. (2006, 2007) and
Bailey and Ben-Zion (2009) demonstrated with detailed model calculations that the range
of size scales of stress and strength heterogeneities indeed governs the forms and
properties of earthquake statistics on the fault.
In general, structural complexities are expected to depend on the following three
variables. 1. The misalignment of the fault zone orientation from the plate motion
direction, with larger misalignment associated with greater complexity. As demonstrated
by Tchalenko (1970) with clay experiments and later studies, the development of
deformation structures occurs in stages, with fault zone segments first created as Riedel
shears at high angles to the imposed shear direction. Increasing deformation leads to
rotation of the Riedel shears towards the imposed shear direction and development of
new fractures (P and Y shear zones) oriented parallel or sub parallel to the shear
direction. The overall rotation of fault segments towards alignment with the loading
direction is part of the maturation process of a fault zone, while the orientation of fault
segments relative to each another is associated with various types (bends, steps, jogs) of
segmentation. 2. The total slip accommodated by the fault zone, with large cumulative
11
slip leading to reduced complexity. This is expected from the results of Tchalenko
(1970), Wesnousky (1994), Stirling et al. (1996) and additional geological and
geophysical observations summarized by Ben-Zion and Sammis (2003). 3. The ratio Q of
the loading rate over damage healing rate. Relatively high Q values are associated with
persistence of weak zones throughout large earthquake cycles, leading to the
development of geometrically regular structures, while relatively low Q values are
associated with fast strength recovery and long term persistence of disordered fault zones
(Ben-Zion et al., 1999, Lyakhovsky et al., 2001). If the damage healing rate is
approximately constant, the slip rate of a fault can be used as a proxy for Q.
In the present paper we develop and apply a procedure for quantifying the
complexity of fault zone structures with measures that are related to the mechanics of
faulting and can be determined objectively from maps of fault zone traces. We analyze a
digital map of strike-slip fault zones in California (Jennings, 1994), define fault zones
and fault segments in a method which takes into account segment length and curvature,
and examine derived statistical quantities representing the diversity of fault zone
heterogeneities versus several controlling variable. A primary measurable property in this
work is the angle
i
θ
ˆ
of a segment i in a fault zone with respect to a given reference
system (here the overall fault zone direction). This is easy to measure from maps of fault
traces and is far less sensitive to mapping resolution and user decisions than the more
typically measured fault steps and jogs.
The range of angles of fault segments in a fault zone provides a reasonably robust
measure of the structural complexity of the fault zone. In addition, we measure the
12
lengths
i
L of fault segments i and calculate from the
i
θ
ˆ
and
i
L
values the normal
projections
i
Y of fault segments with respect to the average fault zone direction, and the
normal distances
i
Z
of segments from a line representing the average fault zone. While
the determination of L, Z and Y is less robust than that of
i
θ
ˆ
, these length scales play
important roles in the propagation of earthquakes. Ruptures propagating on longer
segments (having greater L values) become increasingly more difficult to stop because of
the increasing stress concentration at the rupture tip. See e.g. Fig. 13 of Ben-Zion (2008)
and related text. The Y values represent the misaligned portion of a segment, so wide
variations of Y values correspond to a complex fault zone with many misaligned
segments. A misaligned segment will tend to promote rupture arrest rather than
propagation. The Z values represent a quantity similar to the size of fault stepovers,
which was used in previous quantification attempts, but is less sensitive to interpretation
and resolution issues because it is measured for all segments in a fault zone relative to a
reference frame. A smooth mature fault zone is expected to have small variations of Z
values, as a result of segments coalescence over long periods of activity.
In the next section we describe the data used in this study, employed methods and
measured quantities. In section 3 we present correlations between statistical parameters of
the measured quantities and several potential controlling variables (cumulative slip, slip
rate and misalignment of the fault zone from the local plate motion direction) that might
govern the evolution of fault zone complexities. As faults grow and become more mature,
the diversity of measured quantities is expected to decrease, the distribution of L values is
13
expected to be shifted to higher values and the distributions of Y and Z values are
expected to be shifted to lower values. These expectations are supported overall by the
obtained correlations between various parameters derived from the fault trace data,
cumulative slip and slip rate. We also find that the frequency-size statistics of fault
segments follow a power law distribution only over a very narrow range, and are better
described overall by an exponential distribution. The results provide important
constraints for studies concerned with geometrical fault zone heterogeneities.
Data and Methods
Data
Our goal is to systematically analyze geometrical fault trace complexity using clear
rules that can be specified as computer algorithms and applied to digital data sets. We use
the ‘Digital Database of Quaternary and Younger Faults from the Fault Activity Map of
California, Version 2.0’ (Jennings, 1994), published in 2005 (Fig. 1.1). This map is at a
scale of 1:250,000, with resolution of approximately 250 meters. The map compilers state
that “…Most significant late Quaternary and younger faults are now better-portrayed
digitally in version 2.0, both with respect to location and the depiction of surface trace
complexity…” (Excerpt from Map metadata), which implies a better resolution for the
new version. The metadata also mentions that the maps were digitized with about 0.01
inch resolution, which gives ~70 m error for a map of the aforementioned scale.
14
Fig. 1.1: Active fault map of California with the faults used in this study. The fault zones are
marked by their defining rectangles. See Table 1.1 for names and fault zone related parameters.
The fault zones chosen for this study are active strike-slip faults with evidence of
Quaternary movements including historical ruptures. We first define fault zones for the
study using the association of various segments with a given fault zone and grouping
those together. We then use polygons that are at most 20 km wide to select fault zones
15
from the maps. This is done to avoid overemphasizing the historical naming of faults, to
separate large fault zones into continuous sections, and to consolidate distributed
segments into zones. The 20 km value is slightly larger than the average seismogenic
thickness in California. The length of each 20 km wide box defining a fault zone is first
oriented to preserve the overall fault zone direction. Then the maximum number of fault
segments in each zone is selected as described below, while excluding segments that are
in the box but belong to another fault zone (Fig. 1.1). We determine grouping or
separation of segments by initially fitting a 20 km wide box along each fault zone. If a
near-by fault fits into the same box, it is added to the fault zone. If one box cannot
accommodate the entire fault length, the fault zone is separated into sections. Different
faults can be grouped together even when they are first defined as separate fault zone, if
there is a possibility of through-going rupture as defined by their geometry. Additionally,
if two faults were known to have ruptured together historically, we group them into one
fault zone. The San Andreas fault zone is separated into 4 sections according to changes
in the fault overall direction, so that each section fits into a rectangular box. The grouped
or separated fault names are summarized in Table 1.1.
16
Table 1.1: Fault zones, abbreviation codes, number of segments in each fault zone, cumulative
slip and estimated variation in slip.
Fault Zone Code
# of
segments
Slip
(km)
±
(km)
Minimum
slip rate
(mm/yr)
Maximum
slip rate
(mm/yr)
Calaveras
calaveras 772 50 10 2 18
Central San Andreas
csaf 772 300 15 29 39
Eastern California Shear Zone 1
(Helendale, south Lockhart)
ecsz1 289 3 0.2 0.1 1.5
Eastern California Shear Zone 2
(Lenwood – Lockheart, Johnson
valley)
ecsz2 400 2.25 0.75 0.1 1.5
Eastern California Shear Zone 3
(Camprock – Emerson, Homestead
Valley, Landers)
ecsz3 755 3 1 0.1 2
Eastern California Shear Zone 4
(Calico – Pisgah, Mesquite,
Blackwater)
ecsz4 621 9.6 4.5 1 10
Hayward Maacama
hwrdmac 832 75 25 7 11
Newport Inglewood - Rose canyon
nirc 489 5.1 2.1 1 2
Northern San Andreas
nsaf 573 300 15 20 28
Rinconada
rinconada 648 43 4 1 2
Round Valley - Bartlet springs
rvbs 385 17.5 2.5 3 9
San Jacinto
sj 2551 24 1 6 18
Southern San Andreas - Mojave
mojave 528 300 15 21 41
Southern San Andreas - Salton
salton 845 200 40 10 36
Whittier Elsinore
wtels 712 12.5 2.5 2 7
Notes: When more than one value for cumulative slip appears in the literature, or in a fault zone that
combines 2 or more faults with different slip, the slip given is the average of those values. Minimum and
maximum slip rates were taken for the most recent movements – either from geodesy or, when not
available, from Holocene slip rates. (Sources: Dokka and Travis, 1990, Garfunkel, 1974, Graymer et al.,
2002, McCaffrey, 2005, Petersen and Wesnousky, 1994, Sedlock and Hamilton, 1991, Stirling et al., 1996,
Powell, 1993, Matti and Morton, 1993)
17
We note that more detailed data may exist for given fault sections, but selecting and
compiling data from different sources may produce biases. The same holds for applying
“filters” based on age and other attributes, which may increase the resolution at places
but can create biases and decrease the amount of available data. For a systematic
comparative analysis of geometrical attributes of different fault zones, it is important to
use data from a single source, complied by others, as the employed digital map. We
define an individual fault segment within the mapped trace of a fault zone as a continuous
fault length bound by any change in orientation larger than 1º, including kinks, bends,
steps, and curves. The faults in the digitized map are represented as polylines, namely
lines that are made of multiple segments, connected at their vertices. Those polylines
usually contain many small segments which are a result of the digitization process, so
that a curved segment is actually composed of many short, straight segments. The change
in orientation is defined in GIS as the angle between two lines that share one vertex,
meaning they belong to the same polyline. If this angle is more than 1º, then those lines
will be separated into 2 segments. If the angle is less than 1º, the connecting vertex will
be eliminated, leaving one straight line. A long but curved segment, or a fault trace with
kinks or bends, will be broken down into several straight line parts.
In order to extract straight segments we apply a simplification scheme on the data.
The polylines are simplified by the point removal algorithm of Douglas and Peucker
(1973) which removes vertices and consolidates small segments into longer ones, as long
as a simplified polyline does not deviate from the original shape by more than a threshold
value. We choose the value to be 50 meters, which is below the stated best case map
18
resolution. The simplified polylines are then split at their vertices into straight line
segments (Fig. 1.2). We are aware that this algorithm does not directly convert two lines
with an angle less than 1º into one line, and may still result in neighboring segments that
have an angle of less than 1º between them. However, the Douglas-Peucker algorithm is a
simple, efficient and easily reproducible procedure, which was shown to preserve the
Euclidean geometric properties of data. After applying the algorithm on our data, less
than 0.1% of the segments still had an angle smaller than 1º between two neighbors, and
those were simplified manually by removing the vertex between two segments and
joining them into one straight line. When breaking the faults into straight segments we
may produce occasionally segments that are smaller than the stated resolution of the map.
We retain those segments if they are a part of a longer segment that is broken because of
a kink or a bend.
19
Fig. 1.2: Diagrammatic representation of the analysis of a fault zone. Fault zones consist of
hundred of segments, marked in red. The average fault zone direction θ is marked in dashed
grey line, and serves as the reference frame. A close-up on the area marked by a black rectangle
(after simplification) shows a diagrammatic representation of the geometrical quantities
mentioned in the measured quantities section. Each individual segment is bounded by black
vertical lines. The quantities Li, Yi, Zi and
i
θ
ˆ
are demonstrated.
Measured Quantities
As mentioned, we measure the azimuth
i
θ and length
i
L
of each fault-segment i
and calculate from these data
i
Y and
i
Z (Fig. 1.2). We work primarily with orientations
20
since they are less sensitive to mapping resolution than other proposed measures of fault
zone complexity such as lengths of fault steps and jogs (e.g., Wesnousky, 1988).
We convert the
i
θ and
i
L values into vectors and compute the average fault zone
direction θ for each fault zone from the vector sum of the segments (see equation (1) of
Jones (2006)). We treat the segments as axial data (i.e. a segment with ° = 90
i
θ is the
same as a segment with ° = 270
i
θ ) and use the approach suggested by Fisher (1993,
section 2.4) to deal with axial data when calculating statistical quantities. The segment
lengths are used as weights, so that longer segments have more weight and therefore
more influence on the calculated average fault zone directions. We compute the
misalignment angle of each fault zone (boxes in Fig. 1.1) from the local plate motion
direction φ as
(1.1)
φ θ γ − =
The local plate motion direction φ
is computed using the relative directions of the
Pacific and North American plates at the midpoint of each fault zone using NUVEL-1A
(DeMets et al., 1994). We compute for each segment the minimum angle from the
average fault direction
(1.2)
° ≤ − < °
° ≤ − < °
° ≤ − < °
° ≤ − < °
− − °
° −
− − °
−
=
360 270
270 180
180 90
90 0
if
if
if
if
360
180 -
180
ˆ
i
i
i
i
i
i
i
i
i
θ θ
θ θ
θ θ
θ θ
θ θ
θ θ
θ θ
θ θ
θ
21
We calculate the normal projections
i
Y with respect to the average fault zone
direction for each fault zone (Fig. 1.2) using
(1.3)
i i i
L Y θ
ˆ
sin =
We define a reference line with azimuth θ that passes through the fault zone center
(geometrical center of the confining box), and calculate the normal distance Z
i
of each
segments’ center from that reference line (Fig. 1.2). For both Y
i
and Z
i
calculations we
use only segments that are longer than 1000 m and examine the standard deviations of the
calculated Y
i
and Z
i
values. Because the standard deviation tends to be larger for larger
populations, we normalize the standard deviations stdev(Y
i
) and stdev(Z
i
) using the sums
of the segments lengths, which represent the total length L
tot
of the fault zones.
Specifically, we calculate
(1.4) ,for L
i
>1000 m
∑
=
i
i tot
L L
and define
(1.5) Y = stdev(Y
i
) / L
tot
(1.6) Z = stdev(Z
i
) / L
tot
To characterize the ranges of values or diversity of the measured parameters within
each fault zone, we calculate the circular standard deviation ν and circular standard error
σ (Fisher, 1993, Jones, 2006) of the vectors describing the segments. The circular
standard deviation is an indication of data dispersion about the mean and is defined
(Fisher, 1993) as
22
(1.7) log 2 R − = ν
where R
is the mean resultant length of the vector summation of the
i
θ and
i
L
values
given by
(1.8)
∑ ∑ ∑
+ =
i
i
i
i i
i
i i
L L L R
2 2
) cos ( ) sin ( θ θ
The circular standard error σ is defined as
(1.9) / n δ σ
=
where δ
is the circular dispersion and n is the number of samples. The circular
dispersion is defined as
(1.10) ) 2 )( 1 (
2
2
R ρ δ − =
where
2
ρ is the second trigonometric moment of the vector population (Fisher, 1993).
The computed quantities ν , σ , Y and Z provide different representative parameters for
the diversity of geometrical heterogeneities in a given fault zone section.
Binning
We analyze the length of segments L
i
versus their relative misalignment to the
overall fault zone direction θ expressed as
i
θ
ˆ
sin . We observe that longer segments tend
to be more aligned with the mean fault zone direction (Fig. 1.3). However, trying to
quantify this observation is no simple matter, as small segments are abundant in almost
every direction, while longer segments are less frequent so their statistical effect
diminishes in comparison to the smaller segments. The segments are sorted by their
23
relative misalignment from the fault zone direction θ and we bin the segment data
according to two different schemes. In the first, referred to below as “equal number
binning”, each bin contains an equal number of segments, so that for N data points and k
bins, each bin will contain N/k segments. In the second scheme, referred to below as
“equal interval binning”, each of the k bins spans an equal interval of misalignment
k / 1 = ∆ θ . For each segment i and bin n, if θ θ θ ∆ + < < ∆ ) 1 (
ˆ
sin n n
i
, then segment i is in
bin n.
The bins misalignment values are the average values of all segments they contain.
The bins values are fitted with a linear function of the form y = mx + b. The m-value is a
measure of how many favorably oriented segments there are compared to less favorably
oriented ones, so higher m-value indicates less fault zone heterogeneity. In some of the
analysis we treat segments shorter than some cut-off length as background data. Such
segments may be constantly created and merged during the fault life time, and hence may
represent properties of a steady-state regime. Further, the inherent resolution of the data
implies that small segments may be data artifacts and should not be used. Fig. 1.4 shows,
as examples, the relation between the length and misalignment of fault segments within
the Newport Inglewood - Rose Canyon and the Northern San Andreas fault zones (as
defined in Fig. 1.1), using the equal number binning method with cutoff lengths of 0, 500
and 1000 m. The results generally vary between fault zones – for faults with more slip the
slopes (m-values) mostly increase with increased cutoff length, while those of faults with
less slip decrease. These and other correlations are examined more systematically in the
results section.
24
Controlling Variables
As mentioned, we assume that fault zone heterogeneities depend on the following
three controlling variables: 1) the misalignment of the fault zone orientation from the
overall plate motion direction, 2) the total slip accommodated by the fault zone, and 3)
the ratio Q of the time scale for material healing over the time scale of fault loading. The
first controlling variable is intuitive, the second was used in several previous studies (e.g.,
Wesnousky, 1988; Stirling et al., 1996) and the third emerges from model calculations of
evolving fault zone heterogeneities (e.g., Lyakhovsky et al., 2001; Ben-Zion, 2008).
Table 1.2: Statistics of circular data – various parameters (Fisher, 1993).
Fault zone
φ
θ
γ
ν Skewness Kurtosis δ
σ 95% conf. interval
calaveras 326 332.5 6.47 0.598 0.184 -38.12 0.334 0.021 2.338
csaf 324 315.1 8.92 0.492 -2.007 -98.39 0.215 0.017 1.873
ecsz1 322 315.0 6.99 0.861 -0.702 -6.80 0.603 0.046 5.138
ecsz2 322 322.3 0.35 0.878 0.459 -4.57 0.807 0.045 5.053
ecsz3 322 322.2 0.16 0.794 0.791 -9.21 0.572 0.028 3.094
ecsz4 322 323.9 1.87 0.662 0.030 -25.08 0.346 0.024 2.651
hwrdmac 327 327.6 0.55 0.535 -0.260 -67.33 0.245 0.017 1.928
nirc 320 334.1 14.09 0.940 0.422 -2.77 0.989 0.045 5.057
nsaf 326 322.4 3.62 0.401 -1.386 -246.88 0.144 0.016 1.781
rinconada 325 324.4 0.57 0.542 -0.391 -61.18 0.277 0.021 2.323
rvbs 328 327.8 0.19 0.629 0.224 -29.54 0.372 0.031 3.492
sj 320 313.9 6.07 0.959 -0.525 2.23 0.675 0.016 1.828
wtels 320 309.5 10.51 1.060 -0.308 1.85 0.849 0.035 3.881
mojave 320 291.7 28.34 0.491 1.560 3.58 0.214 0.020 2.260
Salton 320 301.1 18.89 0.846 0.093 0.70 0.714 0.029 3.268
Notes: φ - Local plate motion direction. θ - Average fault zone direction. γ - Misalignment angle from φ.
ν - Circular standard deviation. δ
- Circular dispersion. σ - Circular standard error.
25
Fig. 1.3: Segment length Li vs. segment misalignment, represented as
i
θ
ˆ
sin , for all fault zones.
Each dot represents one fault segment.
26
Fig. 1.4: (a) Same as Fig. 1.3 for two fault zones. Green dots are the original data. Blue circles
represent the averaged equal number bins (each bin has the same number of segments). Red line
is the least squares fit to the averaged bins, with the indicated equation. The cutoff value for each
plot is specified. (b) Same as (a), but for the equal interval bins. Blue circles represent the
summed bins (corresponding log-scale axis in blue). Red line is the least squares fit to the
averaged bins, with the indicated equation. The cutoff value for each plot is specified.
27
Fig. 1.4, Continued.
The time scale for material healing depends primarily on the temperature, pressure
and fluid content (e.g., Beeler and Hickman, 2004, Dieterich and Kilgore, 1996, Hickman
and Evans, 1992, Nakatani and Scholz, 2004). Given the limited available data, we
assume the healing time scale is similar on the average for the different examined fault
28
zones, and represent Q by the measured slip rate across a fault zone. The cumulative slip
and slip rate for each of the fault zones are summarized in Table 1.1. The slip values are
taken from various sources, and when a range of values is available or a fault zone
consists of more than one named fault, we use the average quantity of the reported
values. The error for each fault zone represents the range of values in the literature. The
slip rate values are taken from geodetic measurements when available, and otherwise the
Holocene slip rates are used. The misalignments of fault zone orientations from the plate
motion direction are listed in Table 1.2.
Results
Using the above definitions, we analyze 15 right-lateral strike-slip fault zones in
California. Fig. 1.5 displays rose-diagrams of the fault segments in each fault zone
section and Table 1.2 summarizes circular statistics results. Figs 1.6a and 1.6b show the
relations between
ν
,
σ
,
misalignment γ (see measured quantities section for
definitions), and the cumulative slip for each of the fault zone sections. Figs 1.7a and
1.7b show similar relations between the circular statistics parameters, the fault zone
misalignment and the slip rate.
29
Fig. 1.5: Rose diagrams of the segments for each fault zone. The arrow marks the average fault
zone direction θ . Diagrams were generated using GEorient (Rod Holcombe).
30
Fig. 1.6: (a) The relation between the circular standard deviation, the misalignment from plate
motion direction and the cumulative slip for each fault zone. Black line is the least squares fit for
the data using a log-linear relation. The grey line is the same, but dropping the Salton data point.
The corresponding equations and R
2
values are next to each fit. (b) same as (a), but for the
circular standard error.
Fig. 1.7: (a) The relation between the circular standard deviation, the misalignment from plate
motion direction and the recent slip rate for each fault zone. Black line is the least squares fit for
the data using a log-linear relation. The grey line is the same, but dropping the Rinconada data
point. The corresponding equations and R
2
values are next to each fit. (b) same as (a), but for the
circular standard error.
31
The results indicate that fault zones with higher cumulative slip have generally
lower values of ν
and σ . We also find that ν
and σ
values calculated from the
measurements generally decrease with increasing slip rates. We do not find clear
relations between the misalignment of a fault zone (color scale in Figs 1.6 and 1.7) and its
cumulative slip or slip rate, but fault zones with lower γ values fall on dipping lines and
appear generally below fault zones with higher γ values. The faults in all sections of the
Eastern California Shear Zone (ECSZ) have generally higher values of ν than the SAF,
despite having average γ values that are generally smaller. The Mojave section of the
SAF, which has a very straight and relatively simple surface trace, is one of the most
misaligned segments, while having the smallest standard deviation value in the study
area. The correlation between the statistical parameters and the cumulative slip improves
with the removal of the Salton section of the SAF, which has a high cumulative slip, and
high values of
ν
and σ (Fig. 1.6). The removal of the Rinconada fault, which has a very
low slip rate compared to its high cumulative slip, improves the correlation of the
statistical parameters and the slip rate (Fig. 1.7).
Table 1.3 summarizes the results of the length scale analysis and the binning
approaches. Fig. 1.8 demonstrates the relations between stdev(Y
i
), the related normalized
quantity Y (Eq. 5), and the cumulative slip and slip rate for the different fault sections
(Figs 1.8a-d), as well as the relation between the sum of the segments length and the
cumulative slip and slip rate (Figs 1.8e and 1.8f). The relations between stdev(Z
i
), the
related normalized quantity Z (Eq. 6), and the cumulative slip and slip rate for the
different fault sections are shown in Fig. 1.9.
32
Fig. 1.8: Relations between stdev(Yi) and the cumulative slip (a), and slip rate (b), between Y and
the cumulative slip (c), and slip rate (d), and between the sum of segments lengths and the
cumulative slip (e), and slip rate (f). Correlation coefficients are displayed on each chart.
33
Fig. 1.9: Relations between stdev(Zi) and the cumulative slip (a), and slip rate (b), and between Z
and the cumulative slip (c), and slip rate (d). Correlation coefficients are displayed on each chart.
There are no clear correlations between the standard deviations of Yi and Zi with
either cumulative slip or slip rate, but when normalized by the sum of the segments
length the correlation improves markedly for both. The sum of the segments length is
itself correlated with cumulative slip and slip rate (Fig. 1.8e and 1.8f), but to a lesser
34
extent than the normalized Y or Z values. The misalignment does not correlate with any
of these quantities.
Table 1.3: Additional fault zone parameters, related to segment lengths (see eqs. 1.3-1.6).
Fault zone Stdev(Y
i
) Y Stdev(Z
i
) Z tot
L
m-value
equal
number
m-value
equal
interval
% 1
L
calaveras 291.6 9.0 3042.92 93.8 324389.6 446 5.009 4698
csaf 331.4 4.9 3389.84 49.8 680732.8 1285.5 6.192 6415
ecsz1 393.1 20.9 2086.87 110.7 188443.5 810 2.626 4816
ecsz2 520.3 17.1 4838.34 159.3 303770.3 -61 2.727 5641
ecsz3 386.8 14.4 3732.83 138.9 268787.6 63.5 2.98 4321
ecsz4 324.3 6.6 5479.00 110.8 494562.0 401.7 3.6 7147
hwrdmac 305.1 4.3 3924.58 55.7 705069.4 1097.4 4.994 5929
nirc 561.6 10.3 7345.86 134.2 547556.5 217.4 2.261 6128
nsaf 275.7 4.0 2944.12 42.5 692301.8 3133.27 7.326 9517
rinconada 288.8 9.6 2653.58 88.7 299330.4 1358.2 5.851 4445
rvbs 381.2 8.6 2692.38 60.4 445742.6 686.7 5.011 5483
sj 431.9 8.3 3718.64 71.3 521315.5 957.3 2.337 4877
wtels 664.9 12.6 4106.55 78.0 526791.6 870.1 1.96 7445
mojave 303.2 7.8 2401.60 61.4 390902.2 2004.4 6.39 8950
salton 553.3 7.9 3057.02 43.9 696023.1 405.4 2.759 6090
Fig 1.10 summarizes results on the correlations between the m-values, cumulative
slip and slip rate, using the equal number binning method with a cutoff length of 1000 m
(see binning section). This cutoff value provides the overall best correlations. As
mentioned before, using several cutoff lengths had different effects on different faults.
However, in the comparative statistical analysis done here, we use fixed cutoff lengths
for all data.
35
Fig. 1.10: Slope (m-value) of the linear fit to the equal number bins using cutoff value of 1000 m
vs. cumulative slip (a) and slip rate (b). The black line is the least squares fit for the data using a
log-linear relation. The equations and R
2
value of the fits are displayed.
It is possible that each fault zone has a different cutoff scale, where segments below
it are generated during individual failure events and then eroded away, only to be
generated at the next event. Knowing this scale for each fault may add valuable
information about general fault behavior. However, determining such length scales
require additional higher resolution data and is beyond the scope of this paper.
Conversely, this could be a result of the original mapping resolution of each fault zone. If
a fault was mapped in very high resolution then the small segments are truly a part of the
fault and represent transient deformation, but the small segments belonging to a fault
mapped in lower resolution may be the result of broken down polylines and thus
represent a part of an originally long segment that was broken up by our method.
36
In the equal interval binning case, the resulted m-values also correlate with
cumulative slip and slip rate. However, plotting the m-value versus the circular standard
deviation ν demonstrates that they are highly correlated and therefore provide two ways
of representing the same basic data property of dispersion (Fig. 1.11).
Fig. 1.11: Circular standard deviation ν vs. Slope (m-value) of the exponential fit to the equal
interval bins, using a cutoff value of 0 m.
Following many previous studies (e.g., Barton, 1995, Turcotte, 1986), we examine
the frequency-size statistics of the lengths of individual segments in each fault zone. We
plot the cumulative number N
f
of segments with length greater than L (Fig. 1.12). We
attempt to fit the data using both a power law for the straight part in the log-log plot,
(1.11)
1
1
b
f
L a N
−
=
as well as with a log-linear (or exponential) function for the entire data,
(1.12)
) log(
2 2
L b a N
f
− =
37
The power law range of the dataset extends only over one order of magnitude (Fig.
1.12) and it is clear that the log-linear (exponential) relation fits in all cases the data much
better (Fig. 1.13). The R
2
values associated with fitting the data using the log-linear
function are all higher than 0.97. We consider the slope, or b-value, of either of the fitting
functions (Eqs. 1.11 and 1.12) as a possible quantity that may represent the fault zone
heterogeneity (Fig. 1.14). We find a weak correlation between the b
2
values and slip rate
(Fig. 1.14b), and essentially no correlation between the b
2
values and cumulative slip
(Fig. 1.14a) or misalignment (Fig. 1.14c).
We finally calculate the average length of the longest 1% of the segments
% 1
L for
each fault zone to smooth outliers. Figs 1.15a and 1.15b show
% 1
L versus cumulative
slip and slip rate, respectively. The results indicate that while the existence of longer
straight segments correlates with the fault zone maturity as defined by its cumulative slip
and slip rate, the correlations are not strong.
38
Fig. 1.12: Number of segments larger than a certain length (Nf) as a function of segment length in
meters (L) on a log-log scale for all fault zones (see eq. 1.11).
39
Fig. 1.13: Same as Fig. 1.12, with log Nf as the y-axis. The red line is the least squares fit for the
data, using eq. 1.12. The b-values and R
2
for each fault zone are summarized in the table.
40
Fig. 1.14: Obtained b-values (equations 1.11, 1.12) vs. cumulative slip (a), slip-rate (b) and
misalignment (c) for the two fits described in the text.
41
Fig. 1.15:
% 1
L versus cumulative slip (a) and slip rate (b).
Discussion
We performed quantitative analyses of geometrical heterogeneities of active strike-
slip fault zones in California, using a GIS-based fault map and computer algorithm based
on simple rules, with the goal of clarifying the evolution of fault zone disorder as a
function of several possible controlling variables. The basic measured quantities are the
lengths and angles of fault segments with respect to the local plate motion direction.
These observables are not very sensitive to precise mapping at the ends of the segments
and are therefore more robust than measurements of distances (stepovers, jogs, etc.)
associated with the end regions of segments. Using the angles and the lengths, we
calculate two measures (Y
i
and Z
i
) that are related to separation distances between the
fault segments. The analysis focuses on the range or dispersion (i.e. circular standard
deviation and related parameters) of the various quantities as a function of cumulative
slip, slip rate, and overall fault zone misalignment. The emphasis on the range of
42
quantities is motivated by theoretical studies which indicate that the range of geometrical
heterogeneities characterizing a fault zone controls the dynamics (see Fig. 24 of Ben-
Zion 2008, and related text and references) of the earthquakes sustained by the fault zone.
We attempted to reduce from the scale problem that is inherent in such studies in
several ways. First, the smoothing of fault traces using the Douglas-Peuker algorithm
removes kinks or bends that are smaller than the specified resolution. This method of
smoothing affects all derived quantities and provides a basis for our comparative
statistical study. Second, our focus on orientations and the use of segment lengths as
scaling (weight) parameters in the statistics of circular data reduce the sensitivity to
small-scale mapping resolution issues. In addition, small segments are ignored using a
cut-off value in the binning method and for calculating the Y and Z values. The approach
can be used to analyze data covering various scales above the cutoff value, although this
leads to some loss of data since the comparison must be made using the lowest common
resolution. Some of the data scattering may result from mapping inaccuracies and other
inherent errors that the employ data inevitably contain.
The results indicate that the measures of data dispersion associated with circular
statistics (circular standard deviation and error) provide the most useful parameters for
quantifying the degree of fault zone disorder. These parameters decrease generally with
increasing cumulative slip and increasing slip rate (Figs 1.6 and 1.7), indicating that the
latter quantities represent the maturity of the fault zone. The observed reduction of
structural complexity with cumulative slip supports similar inferences made in several
previous studies (e.g. de Joussineau and Aydin, 2009, Wesnousky, 1988). The reduction
43
of fault zone disorder with increasing slip rate is consistent (assuming similar healing
rates for the different fault zones) with theoretical results on evolving fault zone
structures in a damage rheology model (e.g. Ben-Zion et al., 1999, Lyakhovsky et al.,
2001). A binning approach that defines a quantity representing the tendency of long
segments to align with the mean fault direction is also a useful measure of fault zone
heterogeneity. The misalignment from plate motion direction does not seem to play a
major role in controlling the analyzed fault trace heterogeneities, for the range of values
in the examined data, but it might have a secondary effect on the heterogeneity and
explain some of the outliers. Scholz et al. (2009) suggested based on data compilation by
Ando et al (2009) that a critical level of fault misalignment leads to the formation of
major splays.
The Rinconada fault zone has a very low value of recent slip rate (1 mm/yr or less)
compared with its cumulative slip, probably because the slip migrated east to the current
location of the San Andreas fault. On the other hand, it is expected to have a relatively
ordered mature fault trace due to its high cumulative slip. It is therefore no surprise that
the heterogeneity parameters for the Rinconada fault zone have fairly small values, even
though its slip rate is low.
The Salton fault zone, as defined in our study, stands out compared to other parts of
the San Andreas system. The heterogeneity based on the employed parameters is always
quite higher for this fault zone than other fault zones with similar slip. When removing
the Salton data point, the correlation between some heterogeneity parameters and the
cumulative slip improves markedly (i.e. Figs 1.6 and 1.10). This could be because our
44
definition of the Salton fault zone includes the San-Gorgonio pass area, where the San
Andreas steps left in a restraining bend, resulting in complex fault geometry (irregular
and discontinuous right-lateral, reverse, thrust, and oblique-normal faults) and relatively
lower slip rates (Dair and Cooke, 2009, and references therein). Another reason may be
that the southern part of the San Andreas fault is on the east margin of the Salton trough,
and was repeatedly flooded by the Pleistocene Lake Cahuilla (Sieh and Williams, 1990).
The fast sedimentation rate across the fault trace is resetting the trace, and in essence
healing the surface, which may contribute to the complexity of the surface trace in that
area despite the comparatively high slip rate for this part of the San Andreas fault.
The results in Figs 1.12 and 1.13 indicate that the frequency-size statistics of fault
segment lengths can be described well by an exponential function over the entire range of
observations. The statistics cannot be fitted by a power law over more than one order of
magnitude. This is in contrast to many inferences that the fault lengths statistics follow a
power law distribution (e.g. King 1983; Turcotte 1997, and references therein). We note
that the power law range of the data in most such studies does not extend (as in Fig. 1.12)
much beyond one order of magnitude. It is also possible that our data are depleted for the
smaller segment lengths, due to the limited resolution, and that a more complete data
would have a broader power law range. In any case, the slope of the power law range
does not seem to correlate with any of our controlling variables and therefore is not
useful as a proxy for evolving heterogeneity.
The weak correlation observed for
% 1
L versus the controlling parameters is of
limited use only. One possibility concerning the limited information in relatively long
45
straight segments is that while small- and medium-length segments can be ephemeral
features, long segments are probably self preserving. A long straight segment that was
once created in the past – even as a proto-fault feature, may be preserved during the
recent fault activity. It is possible that a more detailed study that distinguishes between
long segments that are related to recent activity and ones that are inherited structures will
show stronger correlations. This is left for future work.
Numerous theoretical and observational studies highlighted various important
possible connections between geometrical disorder of fault zones and dynamical aspects
of the associated earthquake populations (e.g. Bailey et al., 2010, Elliott et al., 2009,
Ben-Zion, 2008, Bhat et al., 2007, Hillers et al., 2007, Stirling et al., 1996, Harris and
Day, 1993, Sibson, 1986). We conclude that the standard deviation and standard error of
circular statistics, which combine information on the orientations and lengths of fault
segments, provide effective characterizations of fault zone disorder. The clear
correlations of these parameters with the cumulative slip and slip rate suggest that they
are also correlated with the overall seismic potential and related quantities (e.g.
frequency-size and temporal statisitics) of the earthquake populations on the faults. Our
results demonstrate that geometrical properties of fault zones cannot be described by a
steady-state process associated with universal scale-invariant functions, but rather exhibit
clear evolutionary trends. The progressive regularization of geometrical heterogeneities is
expected to lead to a more efficient mechanical process, associated with larger dynamic
weakening during failure, higher slip- and rupture-velocities and higher seismic radiation
46
(e.g., Ben-Zion, 2008). Future studies should test these expectations by combining
analysis of the type done here with analysis of various earthquake quantities on the faults.
We finally note that there has been considerable interest in quantifying the surface
trace complexity of single rupture events (e.g. King and Nabelek, 1985, Sharon et al.,
1995, Pucci et al., 2007, Wesnousky, 2008). Although our study focused on the long term
surface expression of fault traces and overall fault zone evolution, the method and
quantities used in this work can be implemented on datasets of single ruptures.
47
Chapter 2: Application of High Resolution DEM Data to Detect Rock
Damage from Geomorphic Signals along the Central San Jacinto
Fault
Published in Geomorphology, Vol. 113, pp. 82-96, 2009.
Co-Authors: Tom Rockwell
2
and Yehuda Ben-Zion
1
1. Department of Earth Sciences, University of Southern California Los Angeles, CA
2. Department of Geological Sciences, San Diego State University, San Diego, CA
Chapter 2 Abstract
We analyze geomorphic properties extracted from LiDAR and SRTM (Shuttle
Radar Topography Mission) data to test whether the damage zone along the central San
Jacinto Fault (SJF) zone can be resolved with remotely-sensed data in a quantitative
fashion. The SJF is one of the most active faults in southern California, with well
expressed geomorphology and a fast slip rate, as seen in the geology and by GPS. We use
ArcMap and the TauDEM toolbox to compare several morphometric parameters,
including drainage density (Dd), on both sides of the fault, using a 1 km and a 5 km
buffer for the LiDAR and SRTM data, respectively. We also analyze the spatial patterns
of Dd near the fault, using two different definitions of spatial Dd. The high resolution of
the LiDAR data allows us to focus on a single fault, eliminating the effects of parallel
nearby faults. From the LiDAR data we find that the highest Dd values occur in areas
between two fault strands, followed generally by rocks on the northeast side of the fault,
with the lowest Dd values occurring on the southwest side of the fault. The SRTM data
shows a band of high Dd values centered on the main fault trace with ~1 km width. Our
48
results indicate that there is a strong correlation between drainage density and proximity
to the fault, with zones of structural complexity along the fault displaying the highest Dd.
We interpret this to largely be an effect of degree of rock damage, as these are areas that
are expected to be more damaged, and field observations support this contention. If we
are correct, then it appears that the northeast side of the SJF is generally more damaged.
South of the trifurcation area there is evidence that the signal is reversed on the larger
scale, with more damage on the southwest side of the fault inferred from the SRTM data,
possibly caused by extension between the Coyote Creek and Clark faults. The
implications of the observed asymmetry could be geological evidence for rupture
propagation direction, because a preferred propagation direction is predicted to produce
asymmetric damage structure that would be recorded in the volume of rock surrounding a
fault.
Introduction
Understanding the structure of active fault zones is important for many branches of
earth sciences, including earthquake and fault mechanics and crustal hydrology. Because
faults grow and evolve as a result of crustal stresses generated by multiple earthquakes,
there are many connections between the processes that govern earthquake ruptures and
the structural properties of fault zones. In a typical fault structure, the principal slip zone
is surrounded by gouge and embedded within a tabular or wedge shaped damage zone
which can extend to several kilometers depth and several hundred meters width (Ben-
Zion and Sammis, 2003, and references therein). However, most of the movement across
49
the fault is accommodated within the narrow localized principal slip zone (Rockwell and
Ben-Zion, 2007). The broad damage zones are observed in gravity and geodetic surveys
around active faults (e.g., Stierman, 1984; Hamiel and Fialko, 2007), and are also seen in
numerical modeling of evolving fault zone structures (Finzi et al., 2009), but they appear
to accommodate only minor adjustments in the long term motion of faults (e.g., Chester
and Chester, 1998; Rockwell and Ben-Zion, 2007). Clarifying the in-situ properties of
fault zone structures can provide important information on the mechanisms that generate
each structural component, as well as the stress fields that are operative during and
between earthquakes.
In this paper we examine the effect of seismic-induced rock damage on the
development of drainages near an active fault zone. In the past, studies that looked at the
interaction between active faults and geomorphology (e.g. Whipple, 2004; Densmore et
al., 2007) used the tectonic factor only as the source of relief, which caused increased
erosion as a result of the increased elevation difference. It was recently suggested by
Molnar et al. (2007) that tectonics play an important role in causing rapid erosion of
hillslopes by fragmenting the upper crust down to the scale of boulders or smaller. They
suggested that fracture density is an important factor affecting rock erodibility. It was
shown that both micro- and macrofracture density increase in proximity to faults (Chester
and Chester, 1998; Wilson et al., 2003), and that along active faults there is a low-
velocity zone that is seen in fault zone trapped waves and is associated with intense
damage (Li et al., 1999). Another recent observation regarding fault zones was the
existence of highly damaged or pulverized bodies of rock along active strike-slip faults,
50
at distances of up to hundreds of meters away from the fault (Dor et al., 2006a) and with
increasing intensity of pulverization closer to the fault (Rockwell et al., 2009). The
pulverized rocks undergo intense fracturing in the microscale which reduces their grain
size significantly (Rockwell et al., 2009). This reduction in grain size can have an effect
of decreasing the rock permeability by increasing moisture retention in the near surface,
thereby decreasing infiltration capacity, which in turn increases runoff and promotes the
initiation of channels. Along with the general decrease in the strength of the “rock”, the
end result may be to cause higher drainage density where damage is more intense, i.e.
close to the fault.
Theoretical results indicate that on a fault that separates different elastic solids,
ruptures tend to propagate in a wrinkle-like pulse predominately in the direction of slip
on the compliant side of the fault (e.g. Weertman, 1980; Ampuero and Ben-Zion, 2008).
Such ruptures produce dynamic dilation at the tip that propagates in the direction of
motion on the more compliant side of the fault, and dynamic compression at the tip
propagating in the other direction. Due to these opposite rupture tip changes in normal
stresses, ruptures tend to propagate in the direction of motion of the block with slower
seismic velocities at depth, which is referred to as the preferred direction. On bimaterial
faults that produce a preferred propagation direction for earthquake ruptures, most of the
rock damage is expected to accumulate on the stiffer side, which persistently experiences
a tensile stress field during earthquake ruptures (Ben-Zion and Shi, 2005), as it is easier
to damage rocks under tension than under compression (Fig. 2.1). If there is no preferred
rupture direction, such as in a homogenous solid, superposition of the damage generated
51
by many earthquakes is expected to be approximately symmetric across the fault. The
existence of a preferred propagation direction for ruptures on large faults can have
fundamental consequences for many aspects of earthquake physics and estimates of
seismic shaking hazard in major metropolitan areas near large faults (e.g. Ben-Zion,
2001).
Fig. 2.1: Plastic strain (black to white scale) generated by repeating ruptures propagating in the
direction of the arrow on a strike-slip fault (black line). More damage is expected at the upper
right part, where the primary tensional quadrant of the dynamic rupture tip is located. Schematic
illustration modified from Ben-Zion and Shi (2005).
Asymmetric patterns of rock damage across large strike-slip faults have recently
been documented at several localities over several scales. In southern California, Dor et
al. (2006b) observed significant asymmetric distribution of damage elements on one side
of the principle slip surfaces of the San Jacinto, Punchbowl and San Andreas faults at a
scale of a few meters. Lewis et al. (2005, 2007) found clear asymmetry of low-velocity
damaged fault zone layers from analysis of seismic fault zone trapped and head waves
along sections of the San Jacinto and San Andreas faults. The fault zone layers imaged in
these studies are about 100 m wide and extend through the top few km of the crust. Dor
52
et al. (2008) observed damage asymmetry at scales ranging from sub-meter to over a km
across parts of the North Anatolian Fault in Turkey, consistent with the (opposite) rupture
directions of the 1943 and 1944 earthquakes on the fault, which are thought to represent
long term preferred propagation directions on the corresponding sections of the North
Anatolian Fault.
The asymmetry of rock damage across faults may be expressed by differences in
the surface hydrology of the drainage systems on the opposite sides of the fault. The
increase in damage to the rocks is expected to correlate with more erosion on the one
hand, and higher drainage density on the other hand. If indeed the damage is significantly
higher on one side compared to the other, then we expect to see the influence of the
damage asymmetry in the drainages on both sides of the fault. In general, the erosion
intensity and drainage density are affected by many other intrinsic and extrinsic
parameters, which include climate, rock unit, soil type, slope, aspect, relief, land use,
basin development stage, etc. However, in cases where those variables are similar across
terrains, the different levels of rock damage across the fault may be the influential factor
on erosion and the development of drainage patterns.
In this paper, we study two neighboring terrains, which lay on the two sides of a
major fault - the San Jacinto Fault. Quantitative comparison of geomorphic parameters
related to erosion and drainage patterns is used to study the underlying distribution of
rock damage, and to examine the symmetry properties of damage across the fault.
Because the chosen terrains are approximately similar in their generic parameters that
may affect erosion, having similar climate, geology and geomorphic history, earthquake-
53
induced damage may be invoked to explain any observed differences in the erosion
intensity and drainage patterns that are indicated by the analyses done in this work.
Regional Setting
The San Jacinto Fault (SJF) is one of the major branches of the San Andreas Fault
(SAF) system in southern California, and extends from the Transverse Ranges
southeastward into the Salton Trough (Fig. 2.2). It is presently the most seismically active
fault in southern California, with a geologic slip rate of 12–14 mm/yr (Rockwell et al.,
1990) and post-Cretaceous cumulative offset of ~24 km (Sharp, 1967). The SJF is a
young fault, but its age is poorly constrained to 1–2.5 Ma (Matti and Morton, 1993;
Dorsey and Roering, 2006). Becker et al. (2005) found a strong GPS strain signal of 15
mm/yr across the fault. Fialko (2006) used InSAR to study the interseismic deformation
across the southern SAF system and showed that there is nearly equal interseismic strain
accumulation on the SAF and the SJF, with ~19 mm/yr inferred slip rate for the SJF on
the Borrego Mountain section.
54
Fig. 2.2: Regional setting of the study area. Inset: location map of southern California. Rectangle
on inset shows location of main map. Main map: shaded relief derived from the SRTM data
showing the topography. LiDAR data coverage is marked with hatched fill. Active faults are
plotted from the Quaternary faults database (US Geological Survey, 2006) and are marked with
black lines. SJF - San Jacinto Fault. HSF - Hot Springs Fault. TMF - Thomas Mountain Fault.
BRF - Buck Ridge Fault. CCF - Coyote Creek Fault. CR - Coyote Ridge.
The Anza section of the San Jacinto Fault zone has been termed the “Anza
SeismicGap” due to its dearth of microseismicity, and is one of the segments of the fault
zone that may not have ruptured in historical times (Sanders and Kanamori, 1984),
although paleoseismic observations suggest it may have ruptured in the November 22,
1800 earthquake (Rockwell et al., 2006; Rockwell and Seitz, 2008). The San Jacinto
Fault is well expressed geomorphically near Anza, and there is apparently only one major
55
active strand - the Clark Fault (Rockwell et al., 1990). Other faults in the area to the south
include the Coyote Creek and Buck Ridge faults, both of which are sub-parallel to the
main strand and both terminate their north ends in the Anza area - the trifurcation zone.
The slip distribution is ~22 km of cumulative slip on the Clark strand northwest of Anza,
with a few kilometers maximum displacement on the Hot Springs Fault to the northeast.
South of the trifurcation zone, the cumulative slip on the Clark Fault drops to about 14.5
km, there is about 5 km of cumulative slip on the Coyote Creek Fault and only minor slip
on the Buck Ridge fault (Le et al., in review). Two left jogs in the SJF main strand create
areas of compression, northwest and southeast of Anza. The subsurface structure of the
fault is dipping steeply to the NE throughout the area (Sanders and Kanamori, 1984;
Lewis et al., 2005).
Several studies have focused on fault related damage in the area. Dorsey and
Roering (2006) studied basin evolution northeast of the Clark fault, along Horse canyon
and Buck ridge, and showed that as basins move through the restraining bend in Horse
canyon they change their profiles from convex to concave. North of the trifurcation area,
Dor et al. (2006b) studied three outcrops of the fault core and described its asymmetry, as
well as asymmetry in the damage of the adjacent Pleistocene sediments, and concluded
that the northeast side of the fault is more damaged northwest of Anza. Lewis et al.
(2005) used fault zone trapped waves to image the internal structure of the different
branches of the SJF south of the trifurcation area. They interpreted the existence of ~100
meter wide trapping structure that extends to a depth of 3–5 km and is offset to the
northeast from the surface trace of each fault branch. These results suggest a broad
56
damage zone along the northeast side of the Clark segment south of the trifurcation at
depth. However, Stillings (2007) mapped the Horse Canyon area southeast of Anza and
observed more damage in the form of pulverization on the southwest side of the fault's
primary northern strand, extending out several tens of meters but with greatest damage
proximal to the active fault core. This relationship argues that the pulverization is related
to slip on the Clark fault. The northeast side of the fault in the Horse Canyon area only
expresses intense damage outwards for a few meters fromthe fault core at the surface, in
apparent contrast to the observations of Lewis et al. (2005) at depth, although the country
rock is highly fractured for a greater distance.
The SJF trifurcation area is ideal for a hydrological analysis, being a semi-arid
region where the climate is generally similar, and the vegetation cover is relatively
sparse, consisting mostly of succulents at low elevations with some juniper trees at higher
elevations (above 1 km). The relatively minor total fault offset ensures that until the
initiation of slip on the fault, the rocks on both sides probably experienced similar
geological and climate histories. An observed difference between corresponding rock
bodies would therefore be related to the more recent fault activity, which is believed to
have initiated in the early Quaternary (Kirby et al., 2007).
Data and Methods
Data
We use two datasets with different resolutions to examine the geomorphic
parameters of the study area. The first is a 30 m pixel DEM derived from SRTM (Shuttle
57
Radar Topography Mission) data, obtained from the USGS seamless server
(http://seamless.usgs.gov). This will be referred to as the SRTM dataset. The second is a
1 m pixel DEM derived from the point cloud data of the B4 LiDAR (Light Distance And
Ranging, a.k.a. ALSM (Airborne Laser Swath Mapping)) project, obtained from the
GEON portal (http://www.geongrid.org). This will be referred to as the LiDAR dataset.
The B4 LiDAR data covers the southern part of the SAF and the SJF. The data were
acquired in 2005 and cover the main fault traces with ~1 km combined swath width. The
swaths usually overlap, so that the ground is sampled at multiple points per square meter.
LiDAR has the advantage of the ability to penetrate through vegetation, so that some of
the recorded returns are from vegetation and others are from the ground.
The point cloud data were converted into grid using a local binning algorithm with
a radius of 1 m and choosing the local minimum (Arrowsmith and Crosby, 2006; Kim et
al., 2006). The local minimum approach is smoothing some of the more convoluted
portions of the topography, thus helping the later determination of flow path. By using a
local minima binning scheme, we managed to eliminate most of the vegetation returns,
but not all. The vegetation in the study area consists mainly of desert shrubs and
succulents, whose canopy is locally dense at higher elevations, and therefore harder to
penetrate using LiDAR. However, at lower elevations, it is quite sparse and therefore less
disrupting to geomorphic analysis. Missing elevation values, where the point cloud
density was smaller than the search radius, were filled by using a larger (2 m) search
radius to estimate the elevation. Thus, holes larger than 1 pixel were left without values,
but most of the study area was indeed covered.
58
With the SRTM dataset, we use the geological map of California, obtained in GIS
format from the California Dept. of Conservation, Division of Mines and Geology, and
based on Jennings (1977, 1985, 1994). The map scale is stated to be 1:750,000 so that the
margin of error for a location of a line is about 250 m, according to the metadata. We
focus on igneous and metamorphic rocks, assuming that most of the surface damagewas
caused by the activity of the San Jacinto Fault. Sharp (1967) notes that where the fault
traverses crystalline rocks, it generally occupies a central position in a crushed zone of
several tens of meters in width, along which a rift valley has been eroded. This rift valley
is visible in the shaded relief topography of both the SRTM and the LiDAR datasets (Figs
2.2 and 2.3). There are two relevant crystalline rock units within the study area: gr-m -
defined as undifferentiated granitic and metamorphic rocks of pre- Cenozoic age, and
grMz - defined as Mesozoic granitic rocks. The balance of the study area is covered by
Quaternary alluvial deposits, which we disregard.
Table 2.1: Geological formations and their description for the LiDAR dataset locations, after
Sharp (1967).
Name Map name(s) Description Age
Ka Ka, Kga Adamellitic (quartz monzonite) rocks mid-Cretaceous
Kt Kt, Ktg Tonalitic rocks mid-Cretaceous
pKm pKm Metamorphic rocks of mainly gneiss composition, with
small intrusive bodies similar to the mid-Cretaceous
rocks composition
pre-mid-Cretaceous
Qb Qb Bautista beds – poorly consolidated continental
sediments
Pleistocene
59
Fig. 2.3: Location of the LiDAR datasets relative to the mapped fault traces. The area is divided
into three parts numbered 1–3 for presentation purposes. Shaded relief is derived from the B4
LiDAR data. Active faults from the Quaternary faults database are marked with black lines (US
Geological Survey, 2006). The geological map is after Sharp (1967). See Table 2.1 for details
about the rock units. The dashed rectangle shows the locations of Figs. 2.4 and 2.8.
60
For the LiDAR dataset we use the geological map of the SJF by Sharp (1967),
digitized by the authors in the vicinity of the main fault strand. The original map scale is
1:25,000. Table 2.1 summarizes the geological formations of interest. Several formations
are present only on one side of the fault and are therefore not as useful for our purpose.
We include the Pleistocene Bautista Formation because the only fault activity that could
have affected it most likely originated from the San Jacinto Fault. However, that unit is
composed of medium to poorly consolidated sedimentary strata, and its erosion is
expected to be more affected by other factors such as slope and aspect, or varying
sediment properties within the unit.
The database of Quaternary faults from the US Geological Survey (2006) is used to
delineate the SJF strands in the study area. This is a vector layer of polylines, each
representing a fault or a fault segment that has been active in the last 1.6 million years.
The layer scale is stated to be 1:250,000. However, the dataset was compiled from many
sources of varying resolution and the stated scale is an overestimation. In our study area,
the source of the mapped faults is Sharp (1967), so the actual scale of the data is
1:25,000. The Quaternary faults dataset is used to create a buffer in the analysis of both
the LiDAR and SRTM datasets, as will be described below.
Analysis Methods
Both the SRTM and the LiDAR datasets were analyzed using ESRI ArcGIS 9.2
with the TauDEM toolbox (Tarboton, 1997). The overall drainage delineation scheme is
summarized in Stepinski and Collier (2004). Stream networks were derived using the
Strahler classification method (Strahler,1952) with a threshold value of stream order=3
61
for the SRTM data, and verified using air photos. For the LiDAR data, the Slope–Area
method was used with a threshold value of (A/b) S2=200 m, where A is drainage area, b
is unit contour length (in this case 1 m) and S is slope (Montgomery and Dietrich, 1992).
Threshold values were chosen using the drop-analysis function of TauDEM, which
selects the smallest threshold for which the absolute value of the t-statistic is less than 2.
This selects the highest resolution network consistent with the “constant drop law”
(Tarboton et al., 1991, 1992). The choice of different delineation methods for the stream
networks for the two datasets (stream order for the SRTM dataset, slope–area for the
LiDAR dataset) was a result of unsatisfying delineation results when one method was
applied for both datasets. Fig. 2.4 is an example of the result of drainage delineation
using the LiDAR dataset.
For the SRTM dataset, we use a 2.5 km buffer around the main fault, which means
a 5 km wide zone. The LiDAR dataset is already effectively buffered to a ~1 km wide
zone surrounding the main fault. We divide the study area into two parts - NW of the
trifurcation and SE of the trifurcation, which will be called NW of Anza and SE of Anza.
Each side of the fault is studied separately and compared to the other side (NE, SW) in
both parts, as well as for the entire study area.
62
Fig. 2.4: An example of the results of drainage delineation along the SJF using the LiDAR
dataset. The south side exhibits a denser network of drainages, and the SJF in this locale steps to
the left and therefore this is an area of compression. (A) Hill shade image of the fault zone. (B)
The drainage network produced using TauDEM, with geological units (see Table 1.1 for unit
names).
63
We approach the problem of measuring damage with drainage networks in two
ways. The first is to look at various drainage parameters of chosen drainages and to
compare them across the fault. A similar approach was used by Dor et al. (2008) for a
study along the North Anatolian Fault. The second approach is to look for spatial patterns
of variations in the drainage network. This approach was used by Tucker et al. (2001) but
for a different purpose. Both methods are not infallible, although we think they are
complimentary. Fig. 2.4 demonstrates how the drainage networks can differ in character
and density on both sides of the fault.
Chosen Drainages Approach
For each type of rock unit and for each side of the fault, we chose several drainages
that are completely or almost completely contained within the fault buffer zone, on one
side of the fault (do not cross it). Areas with a high degree of human activity (mainly near
Anza) are ignored. For all the chosen drainages, several geomorphic and geographic
parameters are calculated and a comparison of drainages at different locations is made.
We calculate the following for each of the drainages: Average slope, average
aspect, area, perimeter, and the following hydrological parameters: Hypsometric Integral
(Hi) and Drainage Density (Dd). Hi is an integral of the dimensionless area–altitude
distribution curve, which relates the horizontal cross-sectional area of a drainage basin to
the relative elevation above the basin mouth (Strahler, 1952). Hi can range between 0 and
1, usually ranging 0.25–0.75, and lower values are associated with higher levels of
erosion, since more material is removed from the basin. Drainages with similar area–
slope–elevation relations can still have different hypsometric curves, if their drainage
64
networks are different (Willgoose and Hancock, 1998). Dd is the ratio between the total
stream lengths (L) and the drainage basin area (A) (Horton, 1945). Dd is associated with
rates of erosion (higher where erosion is more intense) and so can potentially be used to
infer tectonic and geomorphic history (Tucker et al., 2001). In contrast to Dor et al.
(2008), we do not use the Horton–Strahler parameters (Horton, 1945; Strahler, 1964) of
bifurcation ration, slope ratio, etc. because it has been shown that Strahler's network
statistics do not necessarily reflect differences in drainage properties (Kirchner, 1993;
Hancock, 2005).
All of the above parameters are expected to change as a function of slope, aspect,
rock unit, and climate. If there is indeed more damage on one side of the fault, this side
will presumably be more erodible, promoting the transport of material downstream
(lower hypsometric integral) and encouraging the formation of badland topography (high
drainage density). By choosing drainage systems on different sides of the fault and
controlling for rock unit, slope, aspect and climate, it may be possible to isolate the
differences in drainage parameters that occur simply due to the damage asymmetry. If the
damage asymmetry indeed exists, we expect that the Dd values will be higher on the
damaged side, and the Hi values are expected to be lower.
Spatial Approach
We examine the drainage density as a spatially varying function and compare it
across similar terrains that are located on both sides of the fault. First we compute spatial
Dd simply as the total length of streams in a certain area, following the definition of
Horton (1945). For this we use a raster of the area where pixels with streams are of value
65
1 and pixels without streams are of value 0 (one of the products of TauDEM). A
summation of pixels within an area is then performed, and the result is divided by the
area to give a number that represents the drainage density. This action is similar to
counting the length of the stream per area.
Tucker et al. (2001) claimed that Dd is not a continuous terrain property and
therefore does not reflect the large spatial variability of hillslope lengths, but when
considered as the hillslope flow path, its spatial variation can tell something about various
geomorphic processes. They suggested using the down-slope distance to the nearest
channel as a spatial quantity related to drainage density. Following Tucker et al. (2001),
we look at the variation of the drainage density in space, using the down-slope distance to
the nearest channel. A spatial average can be performed either for a certain area, or for
the entire dataset using a moving averaging filter with a radius determined by the spatial
covariance method, as described by Tucker et al. (2001). For the SRTM data, the filter
radius is found to be 1000 m (Fig. 2.5a), similar to the value found by Tucker et al.
(2001) for a similar dataset. For the LiDAR data, the value is 100 m (Fig. 2.5b). In this
approach, shorter distances to the nearest channel (a lower value of down-slope distance)
correspond to higher drainage density.
66
Fig. 2.5: Covariance function of down-slope distance vs. length scale (meters). The covariance
represents the degree of average correlation between values of down-slope distance at points
separated by a given length scale. Where the covariance levels out, the correlation breaks down,
and we choose the corresponding length scale as our averaging filter size. (A) For the SRTM data
the chosen filter size is 1000 m, similar to Tucker et al. (2001). (B) For the LiDAR data the
chosen filter size is 100 m.
Use of the averaging filter can give smoothed Dd images in both methods, but it
does not differentiate between different rock types. We therefore perform another
averaging, this time taking the rock unit polygons and computing the mean values for
each rock type at each location. Both methods enable us to look at the spatial variations
in the drainage density near the fault, without the limitation of singling out drainages.
67
Results
Chosen Drainages Approach
SRTM Dataset
We select 19 drainages NE of the fault and 15 drainages SW of the fault for the
comparison of parameters. The chosen drainages are all contained within the buffer area
and the igneous–metamorphic rock units. Fig. 2.6 illustrates the drainage locations, the
rock units and the Hi values. The comparison results are summarized in Table 2.2. Two
parameters are compared across the fault for each rock unit - the hypsometric integral
(Hi) and Drainage Density (Dd). Looking at the study area as a whole, the Hi values are a
bit higher on the SW side of the fault for the grMz rock unit, but the opposite is true for
the gr-m rock unit, where Hi values are higher on the northeast side. We use the
Kolmogorov–Smirnov (K–S) method (Chakravarti et al., 1967) to test if there is a
statistically significant difference between the populations of Hi values in different rock
units and different sides of the fault. The results indicate that the Hi values of drainages
in both rock units are from different populations, with more than 95% confidence. From
this we can conclude that Hi values are significantly affected by rock type. However, the
hypothesis that the Hi values are different across the fault is not rejected by the K–S test,
which means that the differences in Hi values across the fault are not significant enough
to imply different erosion rates. From this we conclude that it is possible that the small
differences in Hi values do not reflect differences in damage, and are likely due to
averaging values from drainages of various slopes. Another possibility is that by
68
averaging values for each side of the fault we are obscuring any spatial variability in Hi
values that may be caused by variability of damage along the fault trace.
Fig. 2.6: Geology and location of chosen drainages with hypsometric integral values derived from
SRTM data. Lower Hi values, which we interpret as higher rock damage, are in lighter shades.
The corresponding data is presented in Table 2.2.
69
Table 2.2: Comparison analysis of drainages delineated from the SRTM dataset. For geology see
text. Hi average - the average hypsometric integral. Dd average - the average drainage density.
Average area - the average area of the drainages. Lower Hi value/higher Dd value correlates with
more damage. For the entire study area, the Hi values are similar within each rock type, while the
Dd values indicate a possible higher damage on the SW. When the area is divided to NW and SE
of the trifurcation, the Hi values for the SE part are lower in the SW side of the fault for both rock
types, indicating a possible damage asymmetry.
Location
relative to
trifurcation
Geology Location
relative to
fault
Hi
average
Average
slope
(degrees)
Dd average
( ⋅10
-2
)
Average area
( ⋅10
6
m
2
)
(undivided) grMz NE 0.600 19.0 1.15 1.23
SW 0.606 14.2 1.20 1.35
gr-m NE 0.543 23.1 1.21 1.62
SW 0.533 16.7 1.22 1.23
NW grMz NE 0.547 21.2 1.12 1.38
gr-m SW 0.540 19.7 1.16 1.06
SE grMz NE 0.705 14.6 1.20 0.93
SW 0.606 14.2 1.20 1.35
gr-m NE 0.543 23.1 1.21 1.62
SW 0.530 14.9 1.25 1.33
To study the possibility of damage varying along the fault, we separate the study
area into two parts, northwest and southeast of Anza. The K–S method indicates there's a
statistically significant difference between the populations of Hi values for drainages in
the two areas.
Southeast of Anza, the Hi values are consistently lower on the southwest side of the
fault, which may point to more damage on this side. In the grMz rock unit, the average
slope is almost equal for both sides of the fault, but the average Hi value is significantly
lower on the southwest side, meaning that the difference in Hi values cannot be attributed
to slope differences. In the gr-m rock unit, the average Hi value is lower on the southwest
70
side, while the average slope is also lower on the SW side. Higher slope is usually
considered to promote erosion, therefore it is expected that the differences in slope would
result in lower Hi values where the slope is steeper. This relation does not hold for the gr-
m drainages, where even though the average slope is much higher on the northeast side,
the southwest side still has the lower average Hi value. This discrepancy can be explained
by our damage asymmetry hypothesis - the southwest side suffered more damage which
made it more susceptible to erosion, causing the Hi values to be overall higher.
To the northwest of Anza, a direct comparison is not possible due to differences in
rock units. However, taking into account that values of Hi in the grMz unit are usually
higher, and that the average slope is similar for both sides, the small difference between
the average Hi values on the two sides of the fault may indicate more damage on the
northeast side, consistent with the observations of Dor et al. (2006a). The above results
suggest an asymmetric distribution of damage across the fault, so that northwest of Anza
the northeast side of the fault is more damaged, while southeast of Anza the southwest
side of the fault is more damaged.
Comparisons of drainage density values show consistently higher Dd values on the
southwest side (Table 2.2). It is important to note that drainage density is correlated to the
drainage area, so that higher Dd values occur in drainages with smaller areas. This
correlation is evident from the Dd definition and the fact that it has a dimension. Taking
that into account sheds doubt on the higher values of Dd on the southwest side, especially
since the differences are so small. One exception may be the average Dd value of
drainages in the grMz rock unit to the southeast of Anza, which is equal for both sides of
71
the fault despite the larger average drainage area on the southwest side. This may indicate
higher damage on the southwest side, which causes a higher Dd value than expected for
the larger area. This result is compatible with the Hi comparison result that southeast of
Anza the southwest side displays more damage.
LiDAR Dataset
We select 18 drainages SW of the fault, 42 drainages NE of the fault and 14
drainages that are in between two fault strands using the LiDAR dataset. The
comparisons of results from the analysis are summarized in Table 2.3. Two parameters
are compared across the fault for each rock unit - the hypsometric integral (Hi) and
Drainage Density (Dd). Due to the distribution of rock types in the study area, we are
able to compare only drainages from three different lithologic groups - pKm, Kt and Qb
(see Table 2.1). The Qb unit is the Bautista Formation sandstone and conglomerate and is
not directly comparable to the other two rock units, as it is largely derived from them.
The higher resolution of the LiDAR dataset enabled us to choose drainages that fall in
between the two main strands of the fault in an area where a thrust component is present,
so the locations of some drainages is defined as C for “center”.
72
Table 2.3: Comparison analysis of drainages delineated from the LiDAR dataset. For geology see
Table 2.1. Hi average - the average hypsometric integral. Dd average - the average drainage
density. Average area - the average area of the drainages. Lower Hi value/higher Dd value
correlates with more damage. Location - relative to fault. Indicates if the rock unit is northeast
(NE) of the fault, southwest (SW) of the fault, or in the center between two fault strands (C).
Location in parenthesis - relative to trifurcation (when applicable). The drainages in the area that
falls between fault strands (center) have lower Hi and higher Dd values, consistent with it being
highly damaged due to its location. For most rock units, the northeast side displays higher Dd and
lower Hi values, which is interpreted as higher degree of rock damage.
Geology Location Hi average
Average slope
(degrees) Dd average ( ⋅10
-2
) Average area ( ⋅10
4
m
2
)
pKm C 0.566 30.7 6.41 5.12
NE 0.583 28.3 11.7 5.06
SW 0.662 24.3 6.32 7.48
Qb NE 0.508 29.7 6.81 5.10
SW 0.537 31.9 6.78 8.42
NE (NW) 0.493 36.4 7.79 4.31
SW (NW) 0.502 30.7 6.25 13.7
NE (SE) 0.524 23.0 5.83 5.89
SW (SE) 0.572 33.0 7.30 3.15
Kt NE 0.557 30.7 8.54 4.55
SW 0.538 24.5 4.99 4.69
NE (SE) 0.571 26.2 6.62 3.42
SW (SE) 0.538 24.5 4.99 4.69
The average Hi values for drainages in the pKm rock unit clearly show a
progressive change, where the center has the lowest value, followed by the northeast side
of the fault, and finally the southwest side has the highest Hi values. However, the
average slope values follow the same progression, and may be invoked to explain the
change in Hi values. The average Dd values are higher in the northeast, even though the
average area is similar. Assuming that higher Dd and lower Hi values correspond to
73
higher degree of damage, the results imply that the center and NE side are more damaged
compared with the SW side of the fault. The locations of the pKm drainages are
distributed such that most of the northeastern drainages fall southeast of Anza, and most
of the southwest drainages fall northwest of Anza. From this we conclude that the
northeast side has more damage southeast of Anza than the southwest side northwest of
Anza.
Drainages in the Kt rock unit have similar Hi values across the fault, and when
considering the average slope, no significant difference is apparent. However, the
drainage density is higher on the northeast side, while the average area is similar. A
comparison focusing on the area southeast of Anza hints to a slight difference, with lower
Hi values on the southwest, but higher Dd values on the northeast. The Dd value is
probably affected by smaller drainage areas in the northeast, while the Hi value may be
affected by steeper slope in the northeast, making the comparison less useful for inferring
damage signal.
In the Qb rock unit, the average Hi value is slightly lower on the northeast side of
the fault northwest of Anza, and lower on the southwest side southeast of Anza. The
average slope southeast of Anza is higher on the SW side, so that the difference in Hi
cannot be explained by it. It can be explained by higher damage on the northeast side of
the fault, consistent with the results from the pKm unit. The Dd values are less useful due
to different drainage area values.
The results of comparing drainages using the LiDAR dataset are compatible with
the results from the SRTM data to the northwest of the trifurcation, implying more
74
damage on the northeast side. Southeast of Anza, the LiDAR results imply more damage
on the northeast side, which is the opposite of what the SRTM data imply.
Spatial Approach
The spatial approach is driven by the observed correlation between the drainage
density and the drainage area for individual drainages. We compute spatial drainage
density that is not related to specific drainages using the two methods. The first uses the
down-slope distance to the nearest channel (Tucker et al., 2001), and the second uses the
sum of the length of streams divided by area, but is not limited to specific drainages. The
two methods are referred to as the down-slope distance and the streams-sum methods,
respectively. We analyze the spatial patterns by using a circular averaging filter with size
chosen according to the data autocorrelation (see Section 3.2.2). We also compare
drainage densities for each rock unit separately, using the geological map shapefiles. In
the first method, small downslope distance corresponds to high drainage density.
SRTM Dataset
Fig. 2.7 shows the results of a 1000 m radius averaging filter applied to the down-
slope distance and streams-sum rasters, as well as to the slope raster. To determine how
much the pattern is influenced by the slope, we calculate the spatial correlation between
the three rasters. The correlation between the slope and the down-slope distance rasters is
−0.05, and the correlation between the slope and the streams-sum rasters is −0.2 (a value
of 0 indicate no correlation, while a value of 1 indicates direct correlation and a value of
−1 indicates reverse correlation). Therefore, the slope does not seem to have much effect
on the pattern. We mask the areas that do not correspond to bedrock (either grMz or
75
gr-m), but we expect that the fringes of the bedrock area would be affected because of the
filter radius.
Fig. 2.7: Spatial drainage density and slope for the SRTM dataset, filtered with 1 km averaging
radius and masked so that only data for igneous or metamorphic units' location are visible. The
drainage density is higher near the main fault, and we interpret this observation as a manifestation
of the damage zone that surrounds the fault. (A) Drainage density using the streams-sum method.
Darker shades denote higher drainage density. (B) Drainage density using the down-slope
distance method. Darker shades denote smaller down-slope distance to nearest channel, which
corresponds to higher drainage density. (C) Slope (in degrees).
Looking at the spatial patterns, we see that down-slope and stream-sum rasters are
similar but not identical. Northwest of Anza, there is a higher drainage density zone near
76
the fault, but it is slightly off-center to the northeast. The width of the high Dd band
seems to be about 1 km. In the trifurcation area, there seems to be higher Dd between the
Clark and the Buck Ridge strands, and between the Clark and the Thomas Mountain
strands. Southeast of the trifurcation, the higher Dd concentrates in the areas where the
fault bends or splays. The streams-sum data show a band of high Dd centered on the
Clark strand and north of the normal faults that cut Coyote Ridge (Fig. 2.7a). The higher
drainage density spot on the southeast is related to the Jackass flats, a large area of late
Quaternary alluvium which is masked but still has an effect on the Dd because of the
filter size. The minor dependency on slope can be seen by comparing parts A and B of
Fig. 2.7 to part C.
It is important to note that this method does not distinguish between rock units, so
it is only useful for looking at general spatial patterns. The averaging filter does not
discriminate between bedrock and alluvium, and “smears” the drainage density between
rock units. Even masking the data so that only Dd data for bedrock is available does not
entirely remove the affects of large areas of alluvial cover that can have high Dd values
due to the multiple streams and braiding that occur in such flat areas. The resolution of
the data enables us to see general patterns of drainage density using the two different
methods. If indeed the higher Dd band near the fault is an expression of damage to the
rocks, it appears that the best scale to search for fault related damage is a kilometer wide
band around the fault, which is what the LiDAR data provides us.
77
Fig. 2.8: A map of spatial Dd (streams-sum method) using an averaging filter with 100 m radius
on the LiDAR data. The area in the map is the same as in Fig. 2.4. The filtering action smoothes
out the data and makes it easier to see regional patterns. In this map, the central and southern
parts have visibly higher Dd.
LiDAR Dataset
Fig. 2.8 shows the results of a 100 m radius averaging filter applied to the streams-
sum raster for part of the LiDAR dataset. The SJF in this area bends into a thrust and
splays into several segments. The south side is the hanging block, where more damage is
expected in the rocks. The higher drainage density is evident on the south side of the
fault, as well as between the strands, consistent with the thrust component on the fault at
this location. However, there are different rock types across the fault which prevents us
from a direct comparison (see Fig. 2.4). In order to catalog the data according to different
rock units and locations, we divide each rock type into groups according to location (NE
or SW of the fault, NW or SE of Anza) and compute the spatial average of drainage
78
density obtained by the two methods for each rock unit in each group. The results are
summarized in Fig. 2.9 and Table 2.4. When looking at the overall averages, without
partition with relation to the trifurcation zone, for most rock units, the northeast side of
the fault has higher Dd values, except for the pKm rock type, where the Dd value derived
from the down-slope distance is similar for both sides (the center has the highest Dd
values), and the Dd value derived from the stream-sum is a bit higher on the southwest
side. When focusing only on outcrops that are southeast of Anza, the northeast side again
displays higher Dd, except for the Kt rock unit, where the Dd value derived from down-
slope distance is slightly lower on the southwest side. There are not enough outcrops of
similar rock types to compare to northwest of Anza.
It is worth noting that the Qb and Ka outcrops display the highest difference in
down-slope distance values. While outcrops of pKm rock unit that fall in the center have
the highest damage, this is not true for the Qb outcrops, which may be due to their
unconsolidated nature. It is also likely that some areas of highly damaged rock are more
deeply eroded, and therefore are overlain by young alluvium and could not be used in this
analysis.
79
Table 2.4: Comparisons of drainage density in different geological units, calculated by the two
methods - the down-slope distance and the streams-sum per area. Table corresponds to Fig. 2.9.
Location relative to fault indicates if the rock unit is northeast (NE) of the fault, southwest (SW)
of the fault, or in the center between two fault strands (C). The data is presented for the entire
area (all) and for the area southeast of the trifurcation only (SE). For each rock unit at a certain
location (northeast of the fault, southeast of the trifurcation, etc) the average Dd is calculated
using either the down-slope distance or the streams length sum per area method. Values in bold
indicate higher drainage density, therefore more damage. Note that the lower values in the down-
slope distance mean higher drainage density, because the distance to the nearest channel is
smaller. The rock units that have the higher Dd are usually those located NE of the fault or in the
center between strands.
Rock
unit
Location
relative to fault
Down-slope
distance (all)
Stream length
sum/area ⋅10
-2
(all)
Down-slope
distance (SE)
Stream length
sum/area ⋅10
-2
(SE)
Qb NE 31.67 3.41 34.90 3.20
SW 40.86 2.88 50.67 2.84
C 36.47 2.85 36.47 2.85
pKm NE 29.37 3.78 29.37 3.78
SW 29.56 4.18 30.56 2.53
C 23.95 3.37 23.95 3.37
Ka NE 34.09 4.29 33.18 4.63
SW 69.80 3.26 113.02 1.35
Kt NE 34.33 4.71 39.51 4.37
SW 35.81 3.01 35.81 3.01
Using the LiDAR data to focus on the area close to the fault, we see a similar sense
of damage asymmetry, as reflected in the drainage density southeast of the trifurcation, to
what the individual drainage approach shows. The results of the LiDAR data point to the
northeast side as having higher Dd values and therefore being more damaged across the
study area, which is contrary to some field observations. However, when averaging for
rock units, we disregard difference in distance from the fault, which is known to affect
the degree of damage. Some units occur closer to the fault on one side, which might skew
80
the results. Another concern is the lithological difference within each rock unit, which
may affect its strength and the amount of damage it sustained.
Fig. 2.9: Results of the spatial approach using the LiDAR dataset and comparing drainage density
in four different geological units, calculated by two methods — the down-slope distance and the
streams-sum methods. The corresponding data is presented in Table 2.4. On the left are the results
for the entire study area, whereas on the right are the results for the area southeast of Anza.
Distance — the down-slope distance method. Lower values correspond to higher Dd. The values
are consistently lower on the NE side, which we interpret as more damage to the rocks. # of
streams — the streams-sum per area method. Higher values correspond to higher Dd. The values
are consistently higher on the NE side, which we interpret as more damage to the rocks. Values
are divided by the maximum value to scale them between 0 and 1.
Another approach is therefore taken, where similar areas in corresponding offset
bodies of rock are compared. Using the known cumulative offset on the San Jacinto
Fault, we locate two outcrops of the same rock body and compare the Dd of two similarly
81
shaped areas at similar distance from the fault trace. Thus we eliminate possible
lithological differences that can occur within each rock unit, and ensure that we are
indeed comparing similar rocks with similar histories. The results are summarized in
Table 2.5 and Fig. 2.10.
Table 2.5: Comparisons of drainage density between offset rock bodies along the fault, calculated
by the two methods - the down-slope distance and the streams-sum per area. Table corresponds to
Fig. 2.10. Each pair has one outcrop NE of the fault and one outcrop SW of the fault. The pair's
location relative to the trifurcation (or Anza) is indicated. Split means that one outcrop is NW of
Anza and the other is SE of Anza. Id is the outcrop identifying number. The outcrops were
chosen such that their area and distance from the fault would be as similar as possible.
Rock unit and
location relative
to Anza
Id Location
relative to
fault
Area
( ⋅10
5
m
2
)
Average
slope
Down-slope
distance
Stream length
sum/area( ⋅10
-2
)
Qb SE 1 NE 1.37 18.52 51.83 2.53
8 SW 1.35 32.33 56.76 2.10
Qb SE 6 NE 1.78 28.39 32.98 4.60
13 SW 1.77 23.11 26.36 3.79
Qb Split 14 NE 2.87 28.40 15.40 3.50
19 SW 2.94 25.81 20.24 3.22
Ka SE 2 NE 2.94 35.00 20.03 8.60
9 SW 3.07 40.25 13.89 5.22
Ka SE 11 NE 3.12 27.47 25.66 4.62
15 SW 3.16 24.41 29.98 2.55
pKm SE 7 NE 2.62 26.39 22.71 3.01
16 SW 2.57 18.83 29.32 2.83
pKm Split 17 NE 2.71 20.57 30.68 1.94
24 SW 2.72 24.18 22.46 4.07
pKm Split 18 NE 3.00 25.83 22.32 3.64
21 SW 2.92 25.60 20.26 3.39
pKm NW 22 SW 4.06 31.91 20.37 6.97
25 NE 4.02 25.43 23.12 3.25
82
Fig. 2.10: A comparison of offset rock bodies from the two sides of the fault using the LiDAR
dataset and two methods of drainage density calculations. On the left side there are examples of
two pairs of comparable offset bodies chosen for this analysis. The extent of each chosen polygon
is marked with a dashed line. The polygon id (corresponding to Id in Table 2.5) is given for each
map, and in the location map at the bottom, near the corresponding location rectangle. Active
faults from the Quaternary faults database are marked with black lines in the upper maps, and in
grey lines in the lower map. Geology is the same as in Fig. 2.3. On the right side are histograms
that illustrate the results presented in Table 2.5. (A) Drainage density using the down-slope
distance method (values are divided by the maximum value to scale them between 0 and 1).
Lower values correspond to higher Dd. The results do not indicate one side as having a
consistently higher Dd. (B) Drainage density using the streams-sum per area method. Higher
values correspond to higher Dd. The values are consistently higher on the NE side, which we
interpret as more damage to the rocks. In the histogram, split means that one outcrop is NW of
Anza and the other is SE of Anza. Otherwise the location with relation to Anza is indicated.
83
The large offset on the Clark Fault makes it difficult to find a pair of outcrops to
compare northwest of Anza, so most of the pairs either come from southeast of the
trifurcation, or they are “split” - the outcrop on the SW side is northwest of the
trifurcation, and the outcrop on the NE side is southeast of the trifurcation. The results of
the down-slope distance method do not yield conclusive results as there is no consistent
side that displays lower values. The results of the streams-sum method consistently show
higher values of Dd on the northeast side of the fault, regardless of slope, with one
exception in the pKm rock unit. Since the down-slope distance method is supposed to be
comparable to the “classical” definition of drainage density, it is somewhat surprising
that there is not much correlation between the results from the two methods. It is possible
that the down-slope distance method is being affected by streams outside the chosen
outcrop polygon, because a pixel close to the edge can have a lower down-slope distance
value if it is closest to a channel that is outside the polygon, thus reducing the average
value and increasing the measure of Dd.
Discussion
We have presented an analysis of the geomorphology near the San Jacinto Fault
using several different methods to explore drainage properties. We show that drainage
density increases in close proximity to the fault, and demonstrate that some geomorphic
differences between the drainages on the two sides of the fault cannot be explained by the
usual factors that control drainage morphology. This suggests that fault related damage is
a likely factor that can produce those differences. Here we discuss some concerns and
84
limitations we encountered while conducting the analysis and how they may have
affected our results. We also examine the interpretation of some drainage properties as
damage proxies and how the results correspond to field observations.
When choosing drainages for comparison, we were limited by several factors. The
drainages should be inside the zone of interest, they should mainly lie within one rock
unit, and the sizes of the drainages should be similar. It was therefore difficult to choose a
large enough population of drainages for the results to be statistically significant, if
damage is the principal factor to be compared and the rest of the affecting parameters
discarded. We conclude that the hypsometric integral seems to be the best quantity for
comparing individual drainages across the fault, being non-dimensional and least
correlated to the drainage size. Drainage density is less suited for this type of comparison,
when used on individual drainages, unless there is high similarity in drainage areas. We
tried to choose drainages that have similar aspects, and succeeded to do so in the SRTM
case, where most of the drainages face southwest (Fig. 2.6). This was not possible in the
LiDAR dataset, because the active fault is controlling the geomorphology and creating a
valley into which the streams are draining from both sides. Another limiting factor in the
LiDAR data was its limited extent, determined by the fault locality and not by
geomorphology, which caused some drainage basins to be only partially mapped and
therefore unsuited for analysis.
When choosing the spatial approach, we conclude that the meaningful scale to look
for fault related damage is at most a 1 km wide zone around the fault. The spatial analysis
of drainage density using the SRTM data demonstrates higher Dd near the fault, with
85
width of ~1 km (Fig. 2.8). It seems that the scale of 5 km width across the fault, used
with the SRTM data, is too wide and is only partly capturing fault related damage. A
damage zonewidth of ~1 km is best reflected in the hydrology, and is supported by field
observations of badland morphology around the fault (Stillings, 2007). Therefore, the use
of high resolution LiDAR data is more suited for looking at fault related damage, because
it is a representation of the surface at a more appropriate scale for our purpose. The best
validation of the spatial approach comes from the result that outcrops of rock occurring
between two strands exhibit the highest drainage density, which correlates to the highest
damaged zone (Table 2.4). The spatial approach, while having the advantage of
smoothing and eliminating the need to choose drainages, is not without its problems.
There is still a need to differentiate between rock units, and even though the area
dependency is easily overcome by choosing regions with equal area, the slope and aspect
contributions are harder to control due to the area's geomorphology. We find that using
the classical definition of drainage density (sum of streams length per area) produces
better results than using the down-slope distance proxy of Tucker et al. (2001), and that
focusing on offset correlative bodies of rock yields the best results in that case.
The chosen drainages approach gave better results southeast of Anza, possibly
because the smaller total offset of rock units allows for an easier comparison between the
two sides of the fault in that region. There is some disagreement between the SRTM and
the LiDAR data results, where the SRTM data indicate more damage on the southwest
side, southeast of Anza, while the LiDAR data indicate more damage on the northeast
side throughout the fault zone. This damage asymmetry is also indicated by the spatial
86
approach results. It may be that the SRTM data, which is using a larger buffer, is
recording the effects of the activity on the Coyote Creek fault and the extension in the
Coyote Ridge area southeast of the trifurcation. The rocks in the Coyote Ridge area are
possibly more fractured as it is easier to damage rocks in an extensional regime. The
narrower focus of the LiDAR dataset allowed us to eliminate most of the influence of the
extensional step-over between the Coyote Creek and Clark faults which may have
influenced the observations from the SRTM data. However, we note that the inference on
higher damage on the southwest side of the fault does not match the results of the LiDAR
analysis, nor those of Lewis et al. (2005). This may reflect, in part, the nature and scale of
the damage field as it is reflected in the drainage geomorphology.
Field observations show that highly pulverized granites are mostly located in areas
that are between two fault strands, as well as on the hanging wall of the fault where it
bends, resulting in areas where thrusting is the main slip component. This is in agreement
with the observed high values of Dd in the spatial SRTM data near fault kinks, double
strands, and fault junctions (Fig. 2.7). Stillings (2007) observed that in some locations,
the most intensely damaged rock occurs outward at a distance of tens of meters on the
southwest side of the fault, with pulverization localized between the two fault strands.
Our analysis methods focused on larger scales than tens of meters, and we were limited
by the data resolution, so we were not able to detect this type of observed damage. We
are probably observing a mix of signals from an evolving damage structure. The more
widely-spaced fracturing northeast of the fault probably controls the signal, with a
smaller influence from the pulverization zones in areas of fault trace complexity.
87
It is possible that the zones with the highest damage are also the ones that are the
most eroded, and are now covered with alluvium, which are areas that we eliminated
from the dataset. One such area is Jackass flat, where the fault crosses a wide basin with
Quaternary alluvial fill, and is not exposed in bedrock. In the Jackass flat area the
bedrock on the northeast side is farther away from the active fault trace than on the
southwest side (and out of the LiDAR data extent), which could indicate more erosion
and therefore more damage on the northeast. However, the situation is reversed in Horse
canyon, where bedrock abuts the fault on the northeast, and on the southwest there is a
wide alluvial flat underlain by the damage zone and a secondary fault.
Conclusions
The rock damage signal along the Clark strand of the San Jacinto Fault was
inferred from observations derived from two elevation datasets and a suit of hydrological
and GIS tools, and there appears to be a correlation between the amount of damage across
and within the fault zone and the drainage properties in those areas. The results from the
lower resolution SRTM data point to a ~1 km wide damage zone centered on the main
strand of the San Jacinto Fault. The use of high resolution topography from LiDAR data
enabled us to narrow our focus to the fault of interest and remove influences from nearby
faults. The LiDAR data proved more useful for observations at the scales of the fault
process zone than the low-resolution SRTM data.
The fault damage zone, as inferred from drainage properties, is more pronounced
near areas of complexities in the surface trace, such as kinks, double strands and fault
88
junctions. The results from the LiDAR spatial analysis indicate that the highest damage
occurs in between fault strands, followed by higher overall damage to the northeast. The
damage that we infer is probably in the form of widely spaced fracturing at the 5–20 cm
scale, as described by Molnar et al. (2007), as opposed to pulverization which occurs at
the tens of microns scale and is thought to be a marker for preferred propagation direction
(Dor et al., 2006a; Lewis et al., 2007). If indeed the signal that is registered in the LiDAR
data is related to the macro-fracturing of rock, then we are left with the problem of
resolving which damage signal is more representative of the overall preferred rupture
direction, as they seem to give opposite results. From field observation, pulverization
along the SJF occurs mostly in areas of compression or between two fault strands, which
may indicate that pulverization is more related to structural complexities rather than
rupture direction, and that the overall higher drainage density in those areas reflects that
relationship.
The higher damage on the northeast compared to the southwest side of the fault
throughout the study area may point to a consistent propagation direction along the entire
studied length of the fault, from southeast to northwest, consistent with the observations
of Lewis et al. (2005) of a higher level of damage on the northeast side of each branch of
the SJF. A southeast to northwest rupture propagation on the Clark segment of the SJF,
means more energy will propagate towards the more populated areas of San Bernardino
and Riverside. This is also important for hazard assessment for the Los Angeles basin
area, because it has been shown by simulations that a northward propagating earthquake
will cause much longer shaking in the basin (Olsen et al., 2008).
89
While trying to quantify observed asymmetry in the drainage morphology across
the fault, we encountered many difficulties, such as removing the affects of slope and
aspect, finding appropriate areas to compare, and taking into account other possible
sources of damage besides the preferred propagation direction hypothesis. Our results are
by no means conclusive as to the sense of asymmetry on the fault, and further work needs
to be done in order to resolve those issues. If indeed the two sides of the fault have a
significant difference in rock damage, it should be reflected in the slope–area relation of
different drainages, which we have yet to explore. Even higher resolution datasets can be
derived from the LiDAR point cloud, which would allow one to focus on a narrower zone
and perhaps detect different damage signatures, such as rock pulverization and seismic
trapping structures which typically occur only within 100 m from the fault. Looking at
several different faults at various stages of development, and conducting the same type of
analysis for comparison, should help determine if damage and damage intensity are
indeed affecting drainage patterns. Faults of the Eastern California Shear Zone could be
ideal for this purpose, being in an arid location and having accessible LiDAR data.
90
Chapter 3: Characterization of Pulverized Granitoids in a Shallow Core
along the San Andreas Fault, Little Rock, CA
Co-Authors: Emily E. Allen
2
, Thomas K. Rockwell
2
, Gary Girty
2
, Judith S.
Chester
3
and Yehuda Ben-Zion
1
1. Department of Earth Sciences, University of Southern California Los Angeles, CA 90089
2. Department of Geological Sciences, San Diego State University, San Diego, CA 92182
3. Department of Geology and Geophysics, Texas A&M University, College Station, TX
77843
Introduction
Highly fractured rock (referred to as rock flour, rock powder, breccia, gouge, and
pulverized rock) has long been recognized along surface traces of the San Andreas and
other strike-slip faults (e.g., Flinn, 1977; Anderson et al., 1980, 1983). Primarily, the
fractured rock has been described as having a powdery texture in outcrop, reflecting
deformation at the microscopic scale dominated by Mode I (opening) fractures that
display little or no shear displacement (e.g. Dor et al., 2006). Recently, this fault-
associated rock has received considerable attention, as it is recognized as a fundamental
characteristic of the damage zone along the San Andreas Fault (SAF), particularly for
granite bodies (Wilson et al., 2005; Dor et al., 2006). Pulverized granite (PG) also has
been documented recently along portions of the Garlock Fault (Rockwell et al., 2009), the
San Jacinto Fault (Stillings, 2007), and the Arima-Takatsuki Fault zone in Japan
(Mitchell et al., 2009). Some of these studies have analyzed the details of PG chemistry
and its physical properties.
91
Pulverization is believed to be associated with dynamic reduction of normal stress
during earthquake ruptures (Brune et al., 1993; Ben-Zion and Shi, 2005; Wilson et al.,
2005), which is expected to be enhanced for ruptures on a bimaterial interface (e.g., Ben-
Zion and Andrews 1998; Brietzke et al. 2009). Recent laboratory experiments indicate
that pulverization requires high strain rates (Doan and Gary, 2009), suggesting that
pulverization provides evidence for supershear rupture. As shown analytically by Ben-
Zion (2001) and demonstrated in various numerical simulations (e.g. Ben-Zion and
Huang, 2002; Dalguer and Day, 2009), rupture on a bimaterial interface can produce a
very high slip rate (e.g., tens of m/s) which is expected to produce a very high co-seismic
strain rate.
The nature of pulverized rocks along the Mojave section of the San Andreas Fault
(SAF) has attracted attention since Brune (2001) argued that highly damaged crystalline
rocks in several locations along the fault lack evidence of fault parallel shear. Dor et al.
(2006) defined rocks from the damage zone of the SAF near Tejon pass and along the
Mojave segment as “pulverized” to indicate lack of shear deformation. To determine the
distribution of similarly pulverized rock along the Mojave section of the fault, Dor et al.
(2006) systematically mapped the distribution and intensity of pulverized crystalline
rock. They found that almost all the crystalline rocks within 50 to 200 meters from the
SAF are pulverized to varying degrees, occupying an approximately 100 meter wide sub-
vertical tabular zone parallel to the fault (Dor et al., 2006). Rockwell et al. (2009) and
Stillings (2007) studied pulverized granite (PG) along portions of the Garlock and San
Jacinto faults, respectively, and demonstrated that the pulverization is spatially related to
92
the normal distance from the fault. Those previous studies all focused on surface and
near-surface exposures (up to 2 m depth). Consequently, there is debate as to the role of
surface weathering and other surficial effects in the development of PG along faults, and
the depth range of pulverization.
Wilson et al. (2005) found little evidence of weathering in the pulverized granites
from Tejon Pass, and concluded that the pulverization reflects a mechanical process. The
mean grain size reported, based on a laser particle analyses, is in the sub-micron size
range (Wilson et al., 2005). Furthermore, they suggested that the observed extreme
mechanical comminution implies that approximately 50% of the earthquake energy
budget is spent in creating new fracture surfaces in the fault zone. This result contradicts
energy budget calculations by Chester et al. (2005), who show that fracture surface
energy only accounts for a small fraction of the total earthquake energy budget.
Subsequent analysis of the Tejon Lookout granites by Rockwell et al. (2009), taken
from the same locality studied by Wilson et al. (2005), demonstrates significant
weathering of the <4 micron grain-size fraction. This group also found that the mean
grain size (26 to 208 microns) at Tejon Lookout and at Tejon Ranch along the Garlock
Fault, is substantially larger than that reported by Wilson et al. (2005), in general
agreement with a small fraction of the earthquake energy budget going towards fracturing
as suggested by Chester et al. (2005). Rockwell et al. (2009) concluded that the different
estimates of particle size reflect the technique and equipment used. In a laser particle size
analyzer, pulverized samples tend to settle during long runs at low circulation speeds
producing an apparent final particle size distribution that is much finer.
93
Questions stemming from these previous studies indicate that there is a critical
need to systematically characterize the composition, particle size, and deformation
character of pulverized rock in the damage zones of major continental faults. In this study
we focus on (1) characterizing the properties of pulverized rock and (2) determining the
role of surficial weathering and the extent and depth to which weathering affects
pulverization in the subsurface. In order to study pulverized rocks in detail and to
eliminate signals from surface weathering processes, we obtained a nearly continuous
(~95% recovery), 42 meter-deep, 6.35 cm diameter, oriented core of pulverized fault
zone rock adjacent to the San Andreas Fault near Little Rock, southeast of Palmdale (Fig.
3.1). The core is composed mainly of felsic igneous rocks, and crosses several secondary
fault zones that contain gouge zones that are up to several centimeters thick. Our goal is
to characterize the compositional variation, and the distribution and type of damage as a
function of depth for the cored zone adjacent to the active trace of the SAF at Little Rock.
Below we present a general overview of the location of the drill site, a description
of the rock structure and composition, chemical characterization of the cored interval
using XRF and XRD methods, and the particle size distribution (PSD) measured using
both pipette rack and laser particle analyzer. We then discuss the implications of our
findings in the context of the earthquake energy budget and fault zone related processes.
94
Fig. 3.1: Location map of the drill site. (a) General location and major faults in southern
California. (b) An air-photo of the location of the drill site (star). gru – Mesozoic granitoids. Ta –
Neogene Anaverde Fm. Notice the offset creek (dashed blue line).
Core Characterization
Geological Settings
The drill site is located approximately 80 meters south of the primary active trace
of the SAF near Little Rock Creek, southeast of Palmdale and north of the San Gabriel
Mountains (Fig. 3.1). At this locality, the active trace of the SAF is relatively straight and
active slip is localized. The geology in the study area was mapped by Barrows et al.
(1985). Surface exposures of Mesozoic granitoids, which are typically overlain by
95
Quaternary alluvial deposits, are exposed mostly on hill slopes, and extend from the main
strand of the SAF toward the south (Fig. 3.2). About 500 m south of the SAF, the granites
are overlain by the Neogene Juniper Hills Formation (JHF), an alluvial-fluvial
sedimentary unit that is offset ~20 km along the SAF. Bedding planes within the JHF are
steeply tilted. Barrows et al. (1895) interpreted the JHF to be in fault contact with the
granites. Approximately 1 km south of the SAF, the Nadeau Fault, a thrust fault that
places quartz diorite over the JHF, is mapped in very close proximity to the inferred trace
of the now inactive Punchbowl Fault (PF). The San Gabriel Mountains rise to the south
of the Nadeau Fault, exposing extensive areas of granodiorite (Lowe pluton).
Fig. 3.2: A simplified geological map of the study area modified from Barrows et al. (1985).
Major units and faults are marked. SAF – San Andreas Fault. LRF – Little Rock fault. NNF –
Northern Nadaeu Fault. PF – Punchbowl Fault. Drill site location is marked with a star.
96
The granitic rock body drilled during this study could be a sliver of igneous rock
sandwiched between the SAF and PF. Its surface exposure extends from about halfway
between the PF and the present day active strand of the SAF, and the PF dies out in this
area. The drilling site location was chosen because it is one of the few places along the
Mojave segment of the SAF where there is a large surface exposure of granitic rocks
right up to the fault (Dor et al., 2006).
We employed a standard soil drilling rig (split-spoon auger) and recovered a
continuous, 6.35 cm diameter core to a depth of about 35 m during the first day of
drilling. Unfortunately, the drill seized up the following day, forcing us to abandon some
of the drill stem. We then stepped over a few meters and obtained a core sample from 35-
42 meters in depth. The rocks encountered are severely pulverized, and therefore only
about 40% of the recovered core remained intact when placed in storage boxes.
Additional representative surface samples were collected from outcrops adjacent to the
drill site to allow characterization of the uppermost two meters of rock.
Lithology and Structure
The orientation of the core is known to within ±10 degrees. The core captured three
primary rock types, and crossed several secondary shear zones and localized faults,
consisting of narrow zones of dark, clayey, dense fault gouge. In general, the core is
composed of about 40% granite, 45% granodiorite and 15% quartz diorite, the latter rock
type appearing only in the deepest part of the core. A detailed log of the core is given in
Appendix 1 and the mineralogy of the three primary rock types is presented in Table 3.1.
97
Table 3.1: point counts of core thin sections from various depths, which are representative of the
three key rock types in the core. Upper value is number of counts, lower value is in percent.
Sample
name
LR023 0308S 0468S 1168D 1219S 1225S 1251S 1312S
Depth (m)
9.1 9.3 14.2 35.4 36.9 37.1 37.9 39.7
quartz
%
75
25.0
49
16.3
41
13.7
87
29.0
25
8.3
47
15.7
9
3.0
25
8.3
plagioclase
%
138
46.0
74
24.7
124
41.3
133
44.3
27
9.0
60
20.0
45
15.0
111
37.0
k-spar
%
35
11.7
118
39.3
29
9.7
23
7.7
77
25.7
116
38.7
95
31.7
15
5.0
clays†
%
28
9.3
55
18.3
56
18.7
24
8.0
130
43.3
68
22.7
113
37.7
36
12.0
opaques
%
-
1
0.3
9
3.0
7
2.3
2
0.7
-
3
1.0
-
chlorite
%
10
3.3
-
13
4.3
11
3.7
19
6.3
1
0.3
21
7.0
32
10.7
calcite
%
1
0.3
-
5
1.7
3
1.0
3
1.0
8
2.7
-
11
3.7
biotite
%
-
-
18
6.0
12
4.0
2
0.7
-
-
36
12.0
epidote
%
-
1
0.3
-
-
15
5.0
-
14
4.7
3
1.0
white mica
%
12
4.0
2
0.7
1
0.3
-
-
-
-
-
garnet
%
1
0.3
-
3
1.0
-
-
-
-
-
amphibole
%
-
-
-
-
-
-
-
30
10.0
unknown
%
-
-
1
0.3
-
-
-
-
1
0.3
perthite 13 38 2 - - - 1 -6
IUGS def grano-
diorite
granite grano-
diorite
grano-
diorite
quartz
syenite*
granite quartz
syenite*
quartz
monzo-
diorite
matching
core
sample#
LR023 LR024 LR035 LR072 none LR074 LR075 LR083
† - including clay size
* - more than 35% clays, rock definition probably incorrect
98
Fig. 3.3: Photomicrographs of core samples. (a) Pulverization of various minerals. Plagioclase
breaks along cleavage. Muscovite seems intact (sample depth – 9.3 m). (b) Authigenic calcite in
cracked quartz. Two phases of calcite growth can be seen (sample depth – 14.1 m). (c)
Authigenic calcite growth and twinning in cataclasite (sample depth – 36.9 m). (d) Biotite altered
into Chlorite (sample depth – 14.1 m). (e) Laumentitization of plagioclase and cataclastic shears
(sample depth – 38 m). (f) Clay-filled shear zone surrounded by cataclasis (sample depth – 14.1
m). Q-quartz, K-pottasium feldspar, P-plagioclase, G-garnet, M-muscovite, C-calcite, B-biotite,
Cl- Chlorite.
99
Granite samples are white to pink in color, generally lack mafic minerals, and are
very friable in character. They are composed mostly of quartz, K-feldspar, and
plagioclase (mostly albite), with muscovite and little or no biotite. The K-spars exhibit
perthitic textures. Some of the quartz grains display undulatory extinction and
deformation band development. Calcite is rarely present in veins. The muscovite is
characterized by chlorite and opaque iron oxide along seams parallel to {001}. In thin
section garnet, zircon and apatite are apparent. All grains regardless of mineral type are
fractured to various degrees. Muscovite and feldspar grains tend to break along cleavage
planes, but not exclusively so (Fig. 3.3). Plagioclase grains are often serriticised.
Granodiorite samples are composed of quartz, plagioclase (albite-oligoclase), K-
feldspar, biotite and minor amounts of chlorite, epidote, titanite, calcite and iron oxides.
The granodiorite samples containing chlorite are usually more intact than the granites.
The calcite and iron oxide appear as cement in cataclastic zones. Calcite also grows in
cracks and voids and is twinned and fractured (Fig. 3.3). Biotite is oftentimes replaced by
chlorite, and some plagioclase grains are often serriticised or altered to laumontite (Fig.
3.3). Some of the quartz grains display undulatory extinction and deformation band
development. All grains regardless of mineral type are fractured to various degrees.
Quartz diorites appear only in the bottom of the cored interval. They are mostly
greenish-brown, and are more altered and fractured into clay size fragments. The diorites
are composed of mostly chlorite, quartz, plagioclase, biotite, iron oxides, titanite and
minor amounts of calcite. Various stages of chlorite replacement of biotite are evident in
100
thin section. In addition, clear translucent pore-filling of authigenic chlorite is present in
some specimens.
Numerous sections of the core are cut by centimeter to several centimeter thick
shear bands. The thinnest shear bands observed in hand samples are about 1 mm thick,
and are filled with grayish or greenish clay size grains (Fig. 3.3). The orientations of the
shear bands and secondary faults were recorded where possible (Appendix 1), but no
systematic orientation distribution was established. The suggested random fabric may
partly reflect the uncertainty in core orientation. The overall damaged nature of the cored
interval made it difficult to determine unequivocally, the origin of the mesoscale open
fractures.
Towards the bottom of the core there are several gouge zones of varying thickness,
composed of dark brown to black, highly cohesive clay. Some of these zones contain
small (up to 2-3 mm in diameter) fragments of the surrounding rock. The widest gouge
zone occurs at 36.5 meters depth and is about 30 centimeters thick.
Particle Size Distribution
Multiple splits of samples from the core and two analysis methods were used to
determine particle size distribution (PSD). A standard pipette and dry sieve method
(Rockwell, 2000; Rockwell et al., 2009) was employed to calibrate automated analyses
performed with a laser particle size analyzer (Horiba LA-930 laser diffraction particle
size analyzer) combined with a camsizer (Retsch particle size analyzer) to ensure
accurate and reproducible results. Sample preparation followed that described by
101
Rockwell et al. (2009), with one exception: for the automated method, samples were wet
split at 125 microns instead of 63 microns to produce more consistent results.
Sample Preparation
Samples were gently disaggregated by hand before being run through a mechanical
splitter. One split (about 40 gm) was used for each PSD measurement. Sample splits were
dried, weighed, and shaken in a horizontal box shaker for 24 hours with a dispersant
(0.05% solution of sodium hexa-metaphosphate). Subsequently, samples were wet-sieved
at either 63 or 125 microns (pipette-sieve or automated method, respectively). The fine
fraction was used for the standard pipette method and the Horiba laser particle analyses,
and the coarse fraction was used for in standard sieving and the Retch camsizer analyses.
The coarse fractions were weighed, and then either dry-sieved using phi-interval sieves in
order to combine the data with the pipette results, or run through the Retch camsizer and
combined with the Horiba analyzer results (by weight).
Automated Method Calibration
As noted by Rockwell et al. (2009), laser particle size analyzers do not necessarily
give the same results as the pipette-sieve method, either because of sedimentation in the
machine, or because of the internal algorithms that convert the diffraction data to particle
size distribution. In order to calibrate the results from the Horiba analyzer, we used the
dry-sieved fractions of three previously measured (using pipette-sieve) samples (31-63
microns, 63-125 microns, 125-250 microns). Each fraction was run in the Horiba
analyzer separately using several circulation speeds to produce a distribution.
Additionally, grain mounts were made of each fraction and examined under light
102
microscopy (Fig. 3.4a). Using a microscope mounted camera we were able to measure
the maximum and minimum Feret diameters (caliper widths) of 400 grains in each
fraction. These data were used to calculate the PSD of each fraction. By comparing the
microscopy and analyzer results for a given fraction (Fig. 3.4b) it was determined that the
most consistent results are obtained using the measured minimum Feret diameter to
calculate percent by volume for the mounted samples. It is important to note, however,
that the Horiba analyzer consistently underestimated the particle size for the 125-250
microns fraction, shifting the PSD toward smaller values even when using various
circulation speeds and analyzer parameters. Therefore the upper size-cutoff for the fine
fraction (to be measured in the analyzer) was determined to be 125 microns, even though
the possible analyzer measurement range stated by the manufacturer is up to 2 mm.
Fig. 3.4: (A) A photomicrograph of grain mount, sample LR41. (B) A comparison of microscope
grain diameter measurements (minimum Feret diameter) with measurements of fractions in the
Horiba analyzer, for 3 fractions.
Using the calibration data, it is possible to make several observations regarding the
performance of the Horiba analyzer. First, the “tail” in the coarse fraction observed by
103
Rockwell et al. (2009) is real, and is the result of the ellipsoid shape of some of the
particles. A long and narrow fragment can pass through the sieve mesh in its narrow
dimension during the shaking, yet the analyzer might detect its larger dimension. For
example, when observed under a microscope, the 125-250 micron fraction had grains
with a maximum diameter of ~400 microns and a minimum diameter of ~90 microns.
The result is “tails” of particles that are larger or smaller than the sieve size. Our results
indicate that the volumetric distribution calculated using the minimum Feret diameter
was the most similar to the PSD produced by the Horiba analyzer (Fig. 3.4b).
We used our fraction calibration results to run 18 whole samples through the
Horiba analyzer and compare the results to the standard pipette-sieve method (Fig. 3.5).
Each sample was run and measured at three circulation speeds (3, 4 and 5 in the Horiba).
Thus we determined which instrument parameters are the best for reproducing similar
distributions. Parameters include circulation speed, percent of obscuration, index of
refraction (Sperazza et al. 2004) and smoothness of distribution.
The Horiba analyzer has three index of refraction settings for grain shapes. The
index of refraction that best reproduced the calibration measurements is that of non-
circular, jagged quartz grains. The optimal percent of obscuration, a measure of the
amount of sample in the analyzer’s chamber, was between 15 and 25 percent, similar to
results by Sperazza et al. (2004).
104
Fig. 3.5: A comparison of the PSD from the classical sieve-pipette method (SP) and the
automated Horiba-Camsizer (HC) measurement (rebinned to the same bins used in the classical
method) for two samples - LR14, a granodiorite from 5.7 m, and LR18, a granite from 7.4 m. On
the left is the classical method PSD compared to the rebinned PSD from the automated method.
On the right is the original PSD from the automated method, which is the weighted sum of the
measurements from the Horiba and the Camsizer.
Using the standard samples to calibrate the Horiba analyzer, entire splits were
measured using the methods described above. Following the result of the grain-mounting
calibration, the minimum Feret diameter setting was used for the camsizer. The Horiba
analyzer and camsizer results are then combined by weight percent (Fig. 3.5). Comparing
the classical vs. automated method, it was found that in most samples, the classical
105
method seemed to slightly underestimate the amount of fine material compared with the
automated method (Fig. 3.5), which we attribute to the much finer lower detection limit
of the Horiba analyzer (0.2 microns) compared with the standard pipette method (1-2
microns).
Results
The PSD of PG is mostly fine sand and silt in size. The mean particle size for the
majority of samples falls between 50 and 470 microns (Fig. 3.6), much coarser then
originally proposed by Wilson et al. (2005). There is a slight fining of PSD with depth,
but no apparent correlation with rock composition. As expected, the smallest values of
PSD were obtained from samples taken from shear and gouge zones.
Various fractions from the sieving process (between 31 and 500 microns) were
examined under the microscope to determine if a compositional difference exists between
size fractions (i.e. is there a dominating mineral in certain fractions that is lacking in
others) but no significant difference was found.
We convert the volumetric PSD into linear density and plot the log of the number
of particles vs. the log of their diameter to obtain the D-value of each sample
(Blenkinsop, 1991; Sammis et al., 1987; Rockwell et al., 2009). For consistency, we use
the same range for all samples (0.5-500 microns) and calculate their D-values (Fig. 3.7).
The D-values correlate with the mean particle size over the range of 20-500 microns on a
log-linear plot (Fig. 3.7), and span the range of 2.5-3.1, with no apparent correlation to
rock type.
106
Fig. 3.6: Median particle size vs. depth.
Using the calculated D-values and the smallest grains observed in SEM as a lower
bound (0.02 microns), we can estimate the amount of new surface created by the
pulverization of the rocks, using the same approach as Chester et al. (2005). We find that,
even for the finest samples, the fracture surface area is at most 2.2 x 105 m2 per unit area,
which is an order of magnitude smaller than that inferred for the damage zone of the
Punchbowl Fault based on analysis of microfracture surfaces by Chester et al. (2005), or
the results determined using SEM particle analysis by Keulen et al. (2007).
107
Fig. 3.7: Left – an example of D-value calculation for sample LR18, a granite from 7.4 m. Right -
D-value vs. median particle size.
Geochemistry
Methods
A Phillips MajiX Pro spectrometer and accompanying software were used to
determine major and trace element concentrations for all samples following the method
described by Girty et al. (2006; 2008) and used by Rockwell et al. (2009). The samples
were first powdered in a Spex Certiprep tungsten carbide shatter box. Fused disks were
used for major elemental concentrations, and pressed pellets were used for trace element
concentrations. Loss on Ignition (LOI), the sum of volatile components, was also
determined (Appendix 2).
108
Fig. 3.8: Silica variation diagrams of selected major and trace elements.
Using the aliquot of the finest (<2 microns) material from our PSD samples, an
XRD analysis was performed using a Phillips X’pert multipurpose diffractometer with
copper Kα radiation at 1.5405 Å, and 45 KV and 40 mA settings. Each sample was
109
measured 4 times: untreated, glycolated, heated to 350˚C and heated to 550˚C. Typical
scans were from 2 ˚ to 55˚ 2 θ for untreated samples, and 2 ˚ to 20˚ 2 θ for glycolated and
heat treated specimens.
Results
XRF Elemental Analysis
The three main rock types sampled have distinct bulk rock chemistries (Fig. 3.8,
Appendix 2). Depth variation in composition is minor, except at specific depths that
correspond with the location of secondary faults (Fig. 3.9).
In order to determine the influence of secondary fault zones on the samples
chemistry, several host rock samples taken about 500 meters from the fault zone were
analyzed for comparison. The protolith samples showed only minor changes in bulk rock
composition compared with the fault zone samples, particularly the granodiorite samples.
Compared with the granite protolith, granitic core samples show a slight increase in Fe,
K, Ti, Ba, Y and Nb, and a strong depletion of Ca. The composition of samples taken
from several fault gouge zones falls between that of the granodiorite and the diorite in
bulk rock composition (Fig. 3.10). The only deviation is in sample gouge4, where there is
a relative increase in Rb and U.
We convert our major element data to molecular proportions, and calculate the
chemical index of alteration (CIA) following Nesbitt and Young (1982), as follows:
(3.1) A = Al
2
O
3
/ Al
2
O
3
+CaO*+N
2
O+K
2
O
(3.2) CN = CaO*+N
2
O / Al
2
O
3
+CaO*+N
2
O+K
2
O
(3.3) K = K
2
O / Al
2
O
3
+CaO*+N
2
O+K
2
O
110
Fig. 3.9: (a) Depth variation diagrams of Al
2
O
3
, CaO, Zr and CIA. Grey zones mark zones with outlier samples. (b) Major
shears and alteration bands, as well as gouge zones in the core.
111
Fig. 3.10: Upper – comparisons of major and trace elements for core samples (averages with
standard deviation error bars) and protolith samples. Lower - comparisons of major and trace
elements for gouge samples and averaged granodiorite and diorite core samples.
CaO* refers to CaO associated with the silicate fraction only. In A-CN-K space, the
proportion of molecular Al
2
O
3
(A) is also the chemical index of alteration (CIA). CIA is
only calculated for granite and granodiorite samples. The results are plotted on a ternary
diagram in order to determine the degree of surface related weathering (Nesbit and
Young, 1982). An increase in CIA is expected for weathered samples, while for a typical
un-weathered granitic rock, CIA should be about 0.5. The majority of the samples plot on
the join between plagioclase and K-feldspar (Fig. 3.11), which is consistent with little to
no alteration of feldspars to secondary clay, a characteristic reaction during the
112
weathering process. In other words, the samples do not follow the theoretical
compositional changes seen during progressive weathering for granites (Nesbitt et al.,
1996). The only exceptions are samples taken from nearby secondary fault zones (Fig.
3.9b), where there is enrichment in Ca in the granodiorite, which is the result of calcite
metasomatism. This enrichment is manifested as calcite precipitation in voids, which
means that CaO* does not represent the original igneous composition and this in turn
affects the CIA calculations.
A similar analysis was done using the Fe and Mg oxides (FM), this time lumping
molecular CaO, N
2
O and K
2
O into one category (CNK). On an A-CNK-FM diagram, the
core samples plot about or below the feldspar-biotite join, again indicating little or no
weathering (Fig. 3.11).
Fig. 3.11: Core samples bulk compositions plotted in A-CN-K and A-CNK-FM space. The arrow
depicts the predicted weathering trend for granite. Ka = kaolinite; Chl = chlorite; Plag =
plagioclase; Ksp = K-feldspar; Feld = feldspar; Ill = Illite; Bio = biotite; Cal = calcite.
113
XRD Mineralogy
The diffractograms from all clay smears (<2 micron fraction) derived from the
samples, regardless of depth or type, contain illite (10.1 Å and 5 Å) and chlorite (14.25
Å, 7 Å and 4.8 Å). The typical illite-smectite (I-S) mixed layer diffractogram (an
indication of clay weathering) is recognized by the collapse of the smectite peak (15-20
Å) and its shift into the illite space (10.1 Å) after glycolation and heating. The I-S pattern
is found in most of the surface samples taken from the adjacent outcrop, similar to
previous findings of weathering products in surface samples (Stillings, 2007; Rockwell et
al., 2009). The diffractograms for granite and diorite sampless, however, do not have the
I-S pattern, while the granodiorites have the pattern only down to ~28 m (Fig. 3.12). The
gouge samples contain illite-smectite and possibly kaolinite. Overall, evidence of surface
weathering (as indicated by the presence of I-S pattern) is minor to absent, and what little
there is in the granodiorites disappears with depth in the core. The only exceptions are the
gouge samples, which contain clay weathering products at all depths.
114
Fig. 3.12: XRD diffractograms of three samples. DO13 is granite from an outcrop next to the core
location. LR04 is granite from 2.3 m depth, LR45 is granodiorite from 19.1 m depth. For each
sample 2 measurements are shown – one is the untreated sample and the other is the sample
measured after glycolation and heating to 550 degrees. I – Illite, S – Smectite, Chl – Chlorite.
115
Discussion
PSD and D-values
The obtained mean particle size for the core samples is in agreement with Rockwell
et al. (2009). The pulverized granitoid PSD is mostly fine sand and silt in size, and not as
fine as described by Wilson et al. (2005). All of the rock’s constituent minerals react
similarly to deformation and break into various size grains, with no size preference for
the mineral fractions that we could observe.
Rockwell et al. (2009) discussed the curvature of their PSD on a log-log plot and
interpreted it as a non-fractal population, with changes in curvature as a function of
distance from the fault, correlated to the degree of damage in the rocks. Samples close to
the fault displayed much more curvature than those farther from the fault, and the Tejon
Pass sample at 20 m from the fault could be fit equally well by a linear or a curved
distribution. They used the following “power-log” function to fit their data:
(3.4) 𝑦𝑦 = 𝑎𝑎 ∙ 𝑥𝑥 ( 𝑏𝑏 log 𝑥𝑥 + 𝑐𝑐 )
where y represents the log of particle volumetric density, x is the log of the particle
diameter, and a, b, and c are constants. The value of b represents the curvature of the
distribution, with b=0 as a straight (power-law) line. Rockwell et al. did not use a laser
particle analyzer for the Tejon pass samples, and so a direct comparison to the core
samples is not possible, however by re-binning the core PSD data into phi-scale it is
possible to compare our results to theirs.
Our samples, collected from about 80 m from the principal slip zone, follow a
power-law distribution between 0.5 and 500 microns, with slope (D-value) of 2.5-3.1,
116
when using the analyzer data. By re-binning and fitting a power-log function (equation
3.4) between 0.5-4096 microns (-1 – 12 on phi scale) we examine whether our samples
follow the same trend observed by Rockwell et al., namely that samples with a larger
degree of damage are more curved. We cannot use the distance from the fault as a
comparison parameter, and therefore we examine the changes in b as a function of mean
particle size and D-value (Fig. 3.13). The core samples mostly follow a similar trend to
that observed by Rockwell et al., with decreasing curvature (b closer to zero) correlated
to smaller particles and higher D-values, except for the finest samples. Samples with D-
values larger than 2.85 follow a reversed trend of decreasing curvature with increasing D-
value, and their correlation with mean particle size breaks. This reversed trend can
introduce ambiguity of representation when using the power-log function and we
therefore offer an alternative representation below.
Fig. 3.13: PSD curvature, represented by b (see eq. 3.4), vs. median particle size and D-value.
Core samples are represented by diamonds; open diamonds represent the finest samples that do
not follow the trend of decreasing curvature with decreasing particle size. Samples from the
Tejon outcrop are represented by grey rectangles. Data for Tejon samples was taken from
Rockwell et al. (2009).
117
It is possible to reasonably fit any of our PSD curves with a curve of the following
form (a sum of 2 Gaussian distributions) on a log-log plot:
(3.5)
2
2
2
2
2
1
2
1
2
) (
2
2
2
) (
1
1
2
1
2
1
) (
σ
µ
σ
µ
σ π σ π
−
−
−
−
⋅ + ⋅ =
D D
e a e a D N
Where N(D) is log(volumetric % of particles), D is log(diameter), a
1
and a
2
are
coefficients,
i
µ is the mean and
i
σ is the variance of each Gaussian distribution. In
essence, each PSD would be represented as two populations with normal distribution, one
coarser with a mean of 2.5-3 microns and one finer with a mean of 0.4-1 microns. If the
coarser population is more dominant, the PSD will appear more curved on a log-log plot
and the calculated D-value will be lower. If the finer population is more dominant, the
PSD will be straight and the calculated D-value higher. We demonstrate this idea for a
subset of samples in Fig. 3.14a.
We can also look at the relative weight of the coarse and the fine fractions, as
represented by the two Gaussians. The height of each Gaussian distribution can be
represented as:
(3.6)
i i i
a h σ / =
and the ratio between the coarse and fine Gaussian distributions needed to
reconstruct the PSD would be:
(3.7)
fine
coarse
h
h
This height ratio is large when the coarse population is dominant and small when
the fine population is dominant, and it is correlated to the D-value (Fig. 3.14b). This
118
correlation does not reverse in the finest samples, in contrast to Rockwell et al. power-log
function.
It was speculated by various authors (Ann and Sammis, 1994; Blenkinsop, 1991;
Chester et al., 2004; Sammis and King, 2007; Keulen et al., 2007) that the D-values
represent different deformation mechanisms, where values around 2.6 are characteristic
of constrained comminution and pulverization, while values of 3 or more are indicative of
gouge or foliated cataclasite and zones of high shear displacement. Our samples do not
cluster around either of those values, but rather span the range of 2.5-3.1. We think that
this may be indicative of a mixed mode of deformation, meaning that our samples may
look as if they lack shearing displacement, but do in fact contain zones where there is
additional reduction in particle size due to such shear displacement. The mixture of
several grain populations creates the observed D-values.
119
Fig. 3.14: (a) PSD on a log-log plot for samples with different D-values. (b) Height ratio (eq. 3.7)
vs. D-value for the same samples as in a. (c) An example of fitting the sum of two Gaussians to
sample’s PSD (eq. 3.6).
Combining PSD results, observations of shear bands and secondary faults within
the core and thin section observations of cataclastic zones, we must abandon the idea that
the zone of pulverized rocks contain no evidence for shearing whatsoever (Dor et al.,
2006), and open the possibility that shearing is at least partially involved in the
comminution of grains observed in pulverized rocks. On the surface outcrops it was quite
impossible to see some of the smaller secondary faults due to surface weathering, but we
did observe secondary faulting features at various distances from the main fault trace.
120
In terms of earthquake energy budget, our PSD results indicate that the damage is
not homogeneous in its intensity, even in a single location (the core, in this case), and the
distribution of particle sizes vary considerably within short distances of meters or less.
Even taking into account the finest particles observed, however, our calculated fracture
surface area is an order of magnitude smaller than the estimates of Chester at al. (2005)
for microfractures. This is without taking into account healed/sealed fractures, which are
seen in high density in feldspars in some of the thin sections (for example in Fig. 3.3a). If
we assume that the width of the damage zone around the SAF in this locality is an order
of magnitude larger than that of the Punchbowl and that the damage decreases
logarithmically with distance from the fault, we get values similar to those calculated by
Chester et al. (2005) for overall microfracture abundance.
Whole Rock Chemistry
The changes in bulk rock composition in the granites are compatible with the
observations of authigenic calcite filling veins, as seen in thin sections. The reason for the
mobility of Ca in the rock is possibly the albitization of plagioclase, which frees Ca ions
from the lattice. The free Ca is then moved around by fluids, possibly of meteoric origins,
and is precipitated in cracks and voids. In Fig. 3.9 it is seen that Ca enrichment in the
granodiorite occurs in proximity to secondary faults. The whole rock major element
compositions of granitic samples with the lowest CIA values (<45%) occur in samples
collected adjacent to secondary faults. Notably the CIA values in those samples are
different than other granitic core samples because CIA calculations are affected by calcite
metasomatism. Their major elements composition is more similar to the granitic sample
121
collected away from the fault. On the other hand, the minor element compositions in
these samples are similar to that of the other granitic core samples (Fig. 3.15). The core
samples are different in composition than the outcrop sample probably because of fault
related alterations that changed the rock chemistry. The difference in composition
between the granites within the core may be due to fluid movement that enriches the
granitic samples in proximity to secondary faults with Ca, thus lowering their CIA and
raising their Ca content with very little or no additional compositional changes. This is
supported by the similar LOI values between the two core sample groups, compared with
the outcrop sample.
Fig. 3.15: Comparisons of major and trace elements for the average granitic core samples, the
protolith sample (Granite outcrop), and the low (>45%) CIA granitic core samples next to
secondary faults (Granite Ca rich).
Faults can be either conduits or barriers for fluids, which in turn would interact
with the rock and change its composition. Fluid rock interaction at the Punchbowl and
San Gabriel faults has been observed by various authors (Evans and Chester 1995;
Chester et al. 1993), while others found very little evidence for it (Anderson et al. 1983).
122
In thin sections, some plagioclase is altered to laumontite, and there is also XRD
evidence for its presence in some samples. laumontite can be formed by the albitization
of plagioclase, and so could be related to Ca enrichment. The rest of the changes in major
element compositions are smaller than data variability and can be disregarded. The CIA
(chemical index of alteration) presents a similar picture, with changes from unaltered
values (0.5) next to areas of secondary faults (gouge zones).
The composition of the gouge samples is in between that of the granodiorite and
the diorite, which points to their origin being due to mechanical grinding of a mixture of
the two rock types, which they are next to.
Surface Weathering
In the Tejon Lookout granites, although weathering was minor, Rockwell et al.
(2009) did document the presence of pedogenic clays. Smectite and illite dominated the
finest fraction of their samples, with the scattered occurrence of kaolinite. They
concluded that these additional weathering products added to the cumulative weight of
the finest materials measured, and therefore affected their PSD results. Rockwell et al.
(2009) noted pedogenic clays were also observed in thin sections. In the Little Rock
locality, we found similar evidence of pedogenic clays at the surface – illite-smectite and
possibly kaolinite, but in small quantities. The kaolinite peak, if present, is masked by the
chlorite peak. The presence of clay weathering products in the core samples is a function
of their composition – granitic samples have no smectite and only small amounts of illite,
while granodioritic samples have minor amounts of smectite present at depths of up to 28
m. It is possible that different rocks react differently to weathering, but because the
123
surface granitic samples did contain smectite yet the core samples didn’t (Fig. 3.12), this
indicates that the presence of smectite at depth is not necessarily a product of surface
weathering. Rather, the assemblage illite-smectite-chlorite is thought to be the result of
fluid-rock interaction and brittle deformation under low temperature conditions, as
previously observed along the SAF (Evans and Chester, 1995). Schleicher et al. (2009)
looked at the clay minerals in the SAFOD core and concluded that anomalously high I-S
content correlated with the zones of active deformation and fault weakening. Our gouge
samples from the core may be exhibiting the same behavior, albeit with different rock
types (mudstone vs. granite).
Origin of Pulverization
All of the above observations raise the question of timing – when did each and
every stage of deformation or alteration occur? There are no direct data as to the amount
or exhumation in the Little Rock site, but data do exist for the San Gabriel Mountains
south and west of Little Rock and the SAF. In the Mt. Baldy area, rocks are inferred to
have been exhumed ~3-5 km in the last 13 Myr, whereas in the western San Gabriel
Mountains, the amount of exhumation is considerably less (Blythe et al., 2000). These
estimates are for regions of high topography, so substantial past uplift is expected. In the
vicinity of Little Rock, topography is generally low along the SAF so the amount of
young exhumation is likely small. East of the fault, the Mojave block has sustained little
or no exhumation since the Miocene, as there are Miocene sedimentary strata preserved
(cf. Rainbow Basin) that have neither been greatly buried nor eroded. Considering that
the Neogene strata on each side of the fault (Juniper Hills formation on the west and
124
Anaverde formation on the east) are similarly preserved, one can certainly make the
argument that there has been no substantial exhumation since their deposition in the past
few million years. Bearing in mind that the currently active strand of the SAF, next to
which our study area lies, has been active since the Neogene (Barrows et al., 1985), some
of the healing and alterations observed in the rocks that are characteristic of higher
temperatures (e.g. deeper) may be older residuals from deformation at depth, and not
necessarily related to the latest activity on the SAF. Wilson et al. (2003) inferred the
relative timing of various deformation stages for the Punchbowl fault, and determined
that microfracturing and healing occurred throughout the fault’s activity, while other
alterations were more confined in their timing. In thin sections, we see evidence of
multiple cataclastic phases, where an earlier cemented cataclasite is broken again. There
is also authigenic calcite deformation of twinning and cataclasis. Blenkinsop (1991)
observed laumintitization along cracks and inferred chemically-assisted fracturing.
Nevertheless, we do not see this in our samples, and the quartz is just as fractured as the
altered plagioclase, without any obvious chemical process assistance.
There is a large amount of shear in the rocks, which is seen both in hand samples
and in thin section, but the relative timing of shear and pulverization is unknown. It is
possible that the shears preceded the microfracturing and the two damage textures
represent two different mechanisms for rock damage. It is also possible that they occured
concurrently, but direct data to resolve this was not observed.
125
Conclusions
This paper describes pulverized fault zone rocks recovered from a shallow core
along the San Andreas fault near Little Rock, with the focus being on their mineralogy,
particle size distribution, chemical composition, and damage fabrics. Specifically, a
primary goal was to distinguish the surface weathering signal from damage due to fault
zone processes, as most preceding studies were conducted from surface outcrops.
An important observation is that pulverized rocks at Little Rock do exhibit
evidence of shear, although it cannot be determined whether the shear is in part
responsible for the pulverization. Shear has also been observed in pulverized granitic
rocks in Japan (Mitchell et al., 2009), but again, the relationship between the
pulverization, shearing and their proximity to secondary faults makes the mechanisms of
damage unclear.
The D-values, which range between 2.5 and 3.1, indicate a mixed population of
particles in each sample. From analysis of thin sections, it appears that the smaller
population was comminuted by shear, whereas the larger population was fractured by the
pulverization process. The PSD results support the conclusion that the part of the total
earthquake energy budget expended for breaking or shattering rocks is small.
The bulk rock composition and its changes with depth point to the mobility of some
elements as a result of fluid-rock interaction along secondary faults. XRD results indicate
the production of minor amounts of clay and weathering along secondary faults, and this
clay is apparently not related to surface weathering processes. Alteration and mineral
growth are seen in thin sections, as well as healed microfractures.
126
The location of the study area is well within the fault damage zone, but this study’s
focus is only one point in space. More work is needed in order to study the three-
dimensionality of the damage zone in detail, and determine its effective width with
respect to fluid-rock interactions and the spatial distribution of damage.
127
Chapter 4: Particle Size Analysis of Damage Textures in Fault Zone Rocks
Co-Authors: Judith S. Chester
1
, Thomas K. Rockwell
2
and Yehuda Ben-Zion
3
1. Department of Geological Sciences, San Diego State University, San Diego, CA 92182
2. Department of Geology and Geophysics, Texas A&M University, College Station, TX
3. Department of Earth Sciences, University of Southern California Los Angeles, CA 90089
Introduction
Overview
The rocks surrounding faults in the shallow crust accumulate brittle deformation,
which represents the passage of many earthquakes and the release of energy built-up by
the relative motions of the plates. The rocks break into smaller and smaller particles as
damage accumulates, and that changes their mechanical properties and can influence the
faulting mechanism in the rocks (Mair et al., 2002). Increased deformation causes
particles to break into smaller particles, either by constrained comminution (Sammis and
King, 2007), or by abrasion and spalling (Keulen et al., 2007). It is therefore important to
study changes in particle size and shape, as increasing roundness and decreasing particle
size can be the result of increasing deformation or displacement.
The structure of a typical fault has been described for large displacement strike-slip
faults, combining observations from exhumed and active faults (Chester et al., 2004).
This structure includes the fault core, which is a zone of extreme slip localization
(Rockwell and Ben-Zion, 2007), and the surrounding damage zone, in which the host
rock is damaged to a certain degree and the damage is decreasing away from the fault
128
(Wilson et al., 2003; Mitchell and Faulkner, 2009). To date, most studies of fault zone
material properties have focused on the fault core, where most of the movement occurs.
Characterization of the particle size distribution of fault zone rocks has been done by
various authors, for naturally and experimentally derived material (See Table in Keulen
et al., 2007 for a summary of studies). In recent years, some of the focus has shifted to the
surrounding host rock and the damage it sustains, its importance having been recognized
for the study of earthquake processes. If the creation of such a damage zone necessitates
spending energy in breaking up the rocks, then this energy needs to be accounted for
when estimating total energy release in an earthquake. The energy partition during failure
is still an open question, with important implications for earthquake physics (Ben-Zion,
2008). Several attempts were made to estimate the fracture surface energy per rupture
event (Chester et al., 2005; Wilson et al., 2005). The unknown spatial extent and intensity
of the damage zone can make such calculations widely inaccurate, however. The damage
zone can be an important energy sink, which will influence the propagation or arrest of
earthquake ruptures. Characterizing fault zone structure with a focus on the damage zone
and how it changes with depth is necessary for understanding rupture behavior and
earthquake energy budget.
Some of the most intriguing damage phenomena, pulverized granitic rocks, have
been described in the vicinity of several large displacement faults including the San
Andreas fault (Dor et al., 2006; Rockwell et al., 2009; Wechsler et al., 2010), the Garlock
fault (Rockwell et al., 2009), the San Jacinto fault (Stillings, 2007), and the Arima-
Takatsuki fault zone in Japan (Mitchell et al., 2009). These damage zone rocks of
129
plutonic origin are unique because they are broken down to micron scale particles with
very little shear and without cohesion, and so are different than the classical definition of
either gouge or cataclasite as defined by Sibson (1977). In places along the San Andreas
fault (SAF) the pulverized zone was mapped to several hundred meters in width, where
rocks that are 200 m away from the fault core still lacked cohesion (Dor et al., 2006).
Pulverization is believed to be associated with dynamic reduction of normal stress during
earthquake ruptures (Brune et al., 1993; Ben-Zion and Shi, 2005; Wilson et al., 2005),
which is expected to be enhanced for ruptures on a bimaterial interface (e.g., Ben-Zion
and Andrews, 1998; Brietzke et al., 2009). Recent laboratory experiments indicate that
dynamic pulverization requires high strain rates (Doan and Gary, 2009). This was
interpreted by Doan and Gary (2009) to suggest that pulverization provides evidence for
supershear ruptures.
Particle Size Distribution
Particle size distribution has become a common tool for characterizing fault zone
rocks. Measured in 3-D either by classical sieve-pipette methods (e.g. Anderson et al.,
1980, 1983; An and Sammis, 1994) or by the use of laser dispersion particle analyzer
(Wilson et al., 2005; Rockwell et al., 2009; Wechsler et al., 2010), and in 2-D by SEM
image analysis (e.g. Heilbronner and Keulen, 2006; Keulen et al., 2007; Bjork et al.,
2009), it is usually presented on a log(number)-log(diameter) graph. In such a
representation, the PSD is fitted by a linear function, and its slope is referred to as the
fractal dimension (Sammis et al., 1987) or D (Heilbronner and Keulen, 2006). The value
of D is thought to represent the fragmentation process, and to increase with the number of
130
fracturing events, energy input, strain, and confining pressure (Blenkinsop, 1991). The
reduction of particle size is thought to occur by several processes that represent
progressive shear, displacement, or released energy. Self-similar cataclasis or constrained
comminution process suggested by Sammis et al., (1986) can explain the increase in D-
value up to the dimension of measurement (D=2 in 2 dimensional and D=3 in 3
dimensional methods), but it cannot account for D-values larger than the measurement
dimension, which have been observed in numerous works (Keulen et al., 2007 and
references therein). Additional processes such as wear and attrition, spalling, selective
fracturing and alteration have been invoked as possible mechanisms for further reduction
in particle size (Blenkinsop, 1991; Keulen et al., 2007).
Recent studies have looked at the shapes of particles, in addition to their size
distribution (Storti et al., 2003; Keulen et al., 2007; Bjork et al., 2009). Particle shape can
be characterized in numerous ways which are described in the literature and readily
available in image analysis software packages (Heilbronner and Keulen, 2006), and can
be used in conjunction with particle size to describe fault rocks. The importance of
measuring particle shape is in its influence on the frictional behavior of the material
(Mair et al., 2002), as well as on its porosity and permeability. It was suggested that
particles shape and size change (i.e. smaller and rounder) reflects a change in
deformation mechanism, from intergranular fracturing to abrasion (Bjork et al., 2009). In
the process of continuing deformation, originally angular particles will be rounded by
abrasion, becoming smaller and rounder, and this process could explain D-values larger
than the measurement dimension.
131
Both PSD and particle shape are thought to depend on composition (Heilbronner
and Keulen, 2006; Keulen et al., 2006). For example, plagioclase will tend to break along
cleavage planes and produce irregular shapes, quartz does not have cleavage and will
therefore break less evenly and less intensely, and mica will deform plastically and kink
or shear along its weak cleavage, almost without breaking.
This paper focuses on the damage sustained in pulverized granitic rocks adjacent to
the SAF, which were obtained from a shallow drill-hole near Palmdale, CA. Those rocks
have been studied in detail (Wechsler et al., 2010) and their composition and particle size
distribution (PSD) described. This paper’s goal is to describe PSD and particle shape for
different damage textures using SEM images, and particle compositions using
microprobe and mineral mapping methods. Unlike previous studies, we focus on the
damage zone and not on the fault itself. By studying the different damage textures, we
aim to understand the processes that break up the rocks and create the pulverized damage
texture.
Material and Methods
Rock Samples
In order to study pulverized rocks in detail a nearly continuous (~95% recovery),
42 meter-deep, 6.35 cm diameter, oriented core of pulverized fault zone rock was
obtained from a location adjacent to the San Andreas Fault near Little Rock, southeast of
Palmdale (Fig. 4.1). The drill site is located approximately 80 meters south of the main
active trace of the SAF. The core contains three primary rock types, and crosses several
132
secondary faults, characterized by narrow zones of dark, clayey, dense fault gouge. In
general, the core is composed of about 40% granite, 45% granodiorite and 15% quartz
diorite, the latter rock type appearing only in the deepest part of the core. Detailed logs,
composition and PSD of the core were described by Wechsler et al., (2010). In contrast to
previous studies, they observed numerous shear related damage features in the pulverized
rocks, such as cataclastic shear zones and secondary faults both in hand samples and in
thin sections. Measurements of PSD on core samples using a laser particle analyzer gave
D-values which span the range of 2.5-3.1 with no apparent correlation to rock type.
Wechsler et al., hypothesize that the mixture of several grain populations in one sample,
each with its characteristic D-value, creates the observed intermediate D-values. The
reason for this mixing is that PSD samples were taken as pieces from the core, and it was
not possible to separate sheared and pulverized zones prior to running them through the
laser particle analyzer.
Our goal is to focus on measuring the properties of different damage textures
within a single sample. Two granite samples taken from the core are used, one from a
depth of 9 m (sample 0307) and one from a depth of 35 m (sample 1168A). Sample 0307
modal composition is ~16% quartz, ~25% plagioclase, ~39% K-feldspar and ~18% clay-
size particles (unidentified or clay minerals). Sample 1168A modal composition is ~30%
quartz, ~11% plagioclase, ~27% K-feldspar, ~9% biotite and ~16% clay-size particles.
Both samples were vacuum-impregnated with low viscosity epoxy, and double polished
thin section prepared for light and electron microscopy.
133
Fig. 4.1: Location map of the drill site. (a) General location and major faults in southern
California. (b) An air-photo of the location of the drill site (star). gru – Mesozoic granitoids.
Notice the offset creek (dashed blue line).
Damage Textures
The terminology used in this paper is illustrated in Fig. 4.2, showing areas of
“pulverized” and “cataclastic” regions within the samples. Zones of pulverized texture
are characterized by large host-rock crystals that are fractured primarily by opening-mode
fractures to produce angular particles often ranging from 10-100 microns in diameter, the
fractured parts displaying optical continuity and preserve the original igneous texture of
the rock. The cataclastic zones are characterized by loss of the original igneous texture,
134
smaller (0.5-10 microns) and more rounded grains, a greater clay content, and sometimes
show repeated stages of calcite cementation and shear. This terminology is a geometrical
description of the observed textures in thin section and does not attempt to describe the
texture origin and genesis.
Based on previous works, it is expected that the PSD of the pulverized and the
cataclastic regions will be different, and have different D-values, with a lower value
(around 1.6) for the pulverized regions and a higher value (around 2.0) for the cataclastic
regions.
SEM Image Analysis
In order to characterize the PSD of each texture, sets of back scattered (BSE)
images were obtained using the FEI Quanta FE-SEM instrument in the Texas A&M
microscopy and imaging center. In each sample, two areas of interest are selected, one
representing pulverized texture, the other representing cataclastic texture (Fig. 4.2). Four
levels of magnification are used: x59, x250, x1000 and x4000. The first level was chosen
as the lowest magnification level that still produced a good quality scan, and each higher
level is approximately 4 times magnified. At each magnification level, a mosaic of four
BSE images is recorded (Fig. 4.3a). Using magnification higher than x4000 did not
produce useful images. Those sets are used to measure the PSD of each area of interest.
135
Fig. 4.2: Whole section scans under polarized light of (a) sample 0307 and (b) sample 1168A. In
each scans the pulverized and cataclastic regions that were imaged using SEM are marked with a
yellow rectangle. Insets are examples of SEM-BSE images, one from a cataclastic region in
sample 0307, and one from a pulverized region in sample 1168A. K – K-feldspar, P-plagioclase
(albite-oligoclase), Q-quartz, M-muscovite.
136
Fig. 4.2, Continued
137
Fig. 4.3: Schematic of procedure for grain size analysis. (a) Starting image: SAF core sample;
scanning electron micrograph backscatter contrast. (b) Binary image. (c) Log-log plot of
frequency versus equivalent diameter (10 bins per order of magnitude). The slope of the line fit
yields the D-value. (d) Combined log-log plot for entire range of magnification
Image Processing
Adobe Photoshop and ImageJ (http://rsbweb.nih.gov/ij/index.html) software are
used for image processing and analysis. Each BSE image is adjusted in Photoshop to
enhance contrast between the minerals and the matrix (Fig. 4.3a), and then a threshold is
applied using ImageJ to create a binary image of the particles. The threshold value is
constant for each set of images, but can vary between sets due to differences in the SEM
138
settings. The next step is cleaning the image and correcting for joined particles, using
morphological commands such as erode and dilate in ImageJ. Finally, the resulting binary
file is corrected manually by overlying it on the original BSE image, and a mosaic of the
binary images is created for the analysis (Fig. 4.3b). Before analysis the images are
scaled according to their magnification level, so that the results are in microns instead of
pixels.
Measured Parameters
ImageJ contains a particle analysis package which is used in this paper. Particles
that are on the image edge and particles that are smaller than 20 pixel
2
are excluded from
the analysis. For each remaining particle, the following parameters are calculated using
ImageJ:
Area (A) – the area of the particle (in micron
2
).
Perimeter (P) – the perimeter of the particle (in micron).
Circularity – 4π*A/P
2
. A shape descriptor that is related to the particle angularity.
A value of 1.0 indicates a perfect circle. As the value approaches 0.0, it indicates an
increasingly jagged shape, with a long perimeter compared with its area.
Feret's Diameter - The longest (max Feret) and shortest (min Feret) distances
between any two points along the selection boundary, also known as caliper (in micron).
From those results the following parameters for size and shape description are
calculated:
Equivalent diameter (EqD) – the diameter of a circle with the same area as the
particle (in micron). EqD = 2* √A/π
139
Feret ratio (L/S) – the ratio of the maximum Feret diameter to the minimum Feret
diameter. L/S is a shape descriptor, where a value of 1.0 represents a round particle, and
larger values represent elongated shapes.
The equivalent diameter is plotted vs. the number of particles using evenly spaced
diameter bins, 10 bins per one order of magnitude, on a log-log scale (Fig. 4.3c). We
determine a lower threshold for each magnification set according to where a break in
slope occurs on the log-log plot, and discard the smallest values. The threshold values for
each set are given in Table 4.1. The smallest values are discarded in all sets because the
image resolution causes the measurement results for small particles to be inaccurate,
some particles are not detected and some “fake” particles are created by the image
processing stage. The largest values in all sets except for the lowest magnification set are
also discarded, because when zooming-in we intentionally avoid the largest particles and
therefore the largest diameters are underrepresented. D-value is calculated for each
magnification set using the remaining bins. The results from all magnifications are then
combined into a single plot using a correction factor that depends on the area covered by
the sets (Fig. 4.3d). The average frequency of overlapping bins is used in the combined
PSD plot.
Electron Microprobe
In order to characterize the rock mineralogy and alteration reactions and their
relation to brittle fracturing the Cameca SX50 instrument in the Microprobe facility in
Texas A&M is used to create elemental maps of regions with different damage textures.
140
Table 4.1 – D-values and cut-off values for each region and for each magnification.
Sample
Area
D-value
combined
Range for
D-value
(log(EqD))
Magnification D-value per
Magnification
Lower cut-off
bin size
(log(EqD))
Upper cut-off
bin size
(log(EqD))
0307
pulverized
2.34
1.28
1.8 - 2.7
-0.1 - 1.8
x59
x250
x1000
x4000
1.98
1.19
1.43
1.15
1.5
0.7
0.3
-0.2
2.8*
2
1.4
1.1
cataclastic
2.12 0.3 - 2.6 x59
x250
x1000
x4000
2.14
2.07
1.95
1.59
1.5
0.7
0.4
-0.5
2.8*
2
1.7
1.2
1168A
pulverized
1.36 -0.2 – 2.8 x59
x250
x1000
x4000
1.62
1.22
1.10
1.33
1.5
0.7
0.3
-0.5
2.9*
2
1.6
1
cataclastic
1.94 0.7 - 2.6 x59
x250
x1000
x4000
1.85
1.84
1.79
1.06
1.3
0.6
0.4
-0.7
2.6*
2.2
1.6
0.7
* - Cut-off value is limit of data.
Elemental maps of regions of interest are created by progressively rasterizing the
electron beam point by point over an area of interest, to obtain the spatial distribution of 8
major elements - Si, Al, Na, Ca, K, Fe, Mg, and Ti. The element maps are then combined
to an index-color image, where every number represents a distinct elemental composition
(Fig. 4.4). The process of combination is as follows: Each element map is converted to a
binary image, and each element is assigned a power of 2, from 2
0
to 2
7
. Each binary
image is then multiplied by the assigned value. The 8 images are summed to one image,
in which each cell holds a value between 0-255 which represents a unique combination of
a subset of the 8 major elements. The pixels with values corresponding to a certain
141
mineral are given a color which represents this mineral in the resulting image. For
example, if Si is given the value 2
7
=128, then every pixel that has this value in the final
image represents quartz. If Al is given the value 2
6
=64, and Na the value 2
5
=32, then
every pixel with the value 128+64+32=224 represents a mineral that contain only Si, Al
and Na. In granitic rocks this mineral will most likely be albite. Table 4.2 summarizes the
combinations of elements and our mineralogical interpretation of them. Interpretations
are based on the rock composition and on thin section observations.
Table 4.2 – Probe maps elemental combinations and interpretations.
Number Elements Mineral
4 Fe iron oxides
8 Ca calcite
128 Si quartz
137 Si Ca Ti titanite
192 Si Al kaolinite
196 Si Al Fe garnet/clays
198 Si Al Fe Mg chlorite
200 Si Al Ca laumontite
201 Si Al Ca Ti titanite
203 Si Al Ca Mg Ti titanite
206 Si Al Ca Fe Mg chlorite
208 Si Al K k-feldspar
214 Si Al K Fe Mg biotite†
216 Si Al K Ca k-spar (+ musc/illite)
222 Si Al K Ca Fe Mg biotite†
224 Si Al Na albite
226 Si Al Na Mg albite
232 Si Al Na Ca oligoclase
240 Si Al Na K k-spar (+ musc/illite)
244 Si Al Na K Fe biotite†
246 Si Al Na K Fe Mg biotite†
248 Si Al Na K Ca k-spar (+ musc/illite)
† - Matrix/shear zone fill in sample 0307 which does not contain biotite
142
Fig. 4.4: An example of elemental mapping. The center image is the result of combining the eight
surrounding images; each of those is a density map of an element count. The scale on the bottom
left shows the color intensity scale for the element maps.
143
Results
SEM Image Analysis
PSD
PSD analysis results of the combined sets of SEM images show the D-value of the
pulverized and the cataclastic zones for each sample (Figs 4.5, 4.6). In sample 0307 the
D-value of the cataclastic region is 2.12, and there is a break in slope at EqD = 2 micron
(Fig. 4.5). This is similar to previous observations by Keulen et al., (2007). The slope
break cannot be a result of our choice of magnitude range, because the highest magnitude
used is x4000, and in that magnitude a particle with EqD = 2 micron is equivalent to an
area of ~9400 pixel
2
, which is two orders of magnitude above our low cutoff value of 20
pixel
2
. In Sample 1168A the D-value of the cataclastic region is 1.94, and the break in
slope is at EqD = 5 micron.
The pulverized region in sample 0307 has two different D-values. For the large
particles with EqD = 63-500 microns, the slope is 2.34, while at 63 micron there is a
break and the slope changes to 1.28. This is also evident when D-values of each
magnification are plotted separately (Fig. 4.5b), where the smallest magnification (x59)
D-value is approximately 2, markedly higher than the rest. The pulverized region in
sample 1168A has a constant D-value of 1.36 throughout. The individual D-values show
a similar trend to sample 0307, where the smallest magnification had the largest D-value
= 1.62.
144
Fig. 4.5: Combined log-log plot for entire range of magnification for sample 0307 (left), and
individual D-values for each magnification separately (right). The bars represent 2 sigma for the
linear regression used for calculating the D-value.
Fig. 4.6: Combined log-log plot for entire range of magnification for sample 1168A (left), and
individual D-values for each magnification separately (right). The bars represent 2 standard
deviation for the linear regression used for calculating the D-value.
145
Shape Analysis
Two shape descriptors are used – circularity and L/S (see methods section for
definitions). The average shape descriptors are calculated only for particles larger than
the smallest cut-off value for each magnification (see Table 4.1). Fig. 4.7 shows the
average circularity and L/S values for each magnification level in each region, for both
samples. The results indicate that the particles are more elongated (larger average L/S) in
the pulverized samples in general. The circularity, which is a measure of angularity,
indicates that pulverized regions have more angular particles (smaller circularity). There
is one exception in sample 0307 in the x250 magnification level, where the cataclastic
region has a slightly smaller average circularity, compared with its pulverized
counterpart.
Fig. 4.7: Average shape factors of samples, per magnification.
The particles in the cataclastic regions, although in general more circular, as
expressed by their L/S values that are close to 1, can have a low circularity if they have
uneven perimeter, which is the case in some samples. The reason for this unevenness can
146
be due to chemical processes of alteration that act on particles perimeter. Manually
correcting the image helps in overcoming this problem in part, but not completely. As a
consequence the circularity is considered to be highly affected by the image analysis
procedure and therefore a less robust shape descriptor.
Probe
Several maps of the cataclastic zones were obtained in both samples, each map
covering an area of 1x1 mm. In sample 0307 (Fig. 4.8), the main minerals are K-feldspar,
quartz and plagioclase. The K-feldspar can be purely potassium-aluminum-silicate, or it
can contain some calcium and sodium, in which case the mineral may be illite, a clay
alteration product which was detected in core samples by Wechsler et al., (2010). The
potassium feldspar can have perthitic texture with albite (e.g. top of image 1 in Fig. 4.8).
Albite can also appear by itself or as part of an oligoclase particle. Quartz has a more
angular appearance compared with the other particles. The matrix seems mostly
composed of illite or clay minerals of similar composition. There is a small amount of
chlorite, and a crack or a shear zone containing Fe/Mg rich aluminosilicates. No biotite
was observed in the thin section.
In sample 1168A (Fig. 4.9) the main minerals are albite, oligoclase, quartz, K-
feldspar, biotite, chlorite and laumontite. Additional minerals include titanite, calcite, iron
oxides, and possibly kaolinite. The albite is altered to laumontite and illite, and
laumontite appears also in the matrix. The biotite is altered to chlorite, and both also
appear in the matrix. Titanite, calcite and iron oxides appear as isolated particles, and are
easily recognized in the BSE image by their brightness. Quartz particles have kaolinite
147
rims, filled cracks and poke-marks. The albite/oligoclase alteration to laumontite is seen
in the BSE image by the poke-marked or “scarred” appearance of the particle surface.
This scarring is not observed in quartz or feldspar. The zones of laumontite alteration in
the particles do not necessarily follow cracks or particle edges, and completely altered
laumontite particles can be seen in the matrix. From this it is possible to infer that the
alteration and the mechanical breaking of the particles are not related and do not affect
each other. One exception is a cracked quartz particle (right side of image 1, Fig. 4.9)
which has laumontite filling the crack. Laumontite is not an alteration product of quartz,
most likely it is either matrix or calcite filling in a crack that has kaolinite at the edges.
The matrix composition is a combination of small particles of the original minerals and
clay minerals.
Fig. 4.8: Elemental mapping of two zones in the cataclastic region in sample 0307. The maps are
1x1 mm or 512x512 pixels in size.
148
Table 4.3: Mineralogical composition results derived from the elemental mapping. The pixels
representing each mineral in Table 4.2 are summed and divided by the total number of pixel in
the image, minus the pixels representing epoxy. See Figs 4.8 and 4.9 for maps, Table 4.2 for
mineral mapping definitions.
Mineral 0307 zone 1 0307 zone 2 1168A zone 1 1168A zone 2
Quartz 11.87 12.88 10.07 7.44
Albite 12.81 12.11 24.95 22.26
Oligoclase 8.35 5.64 16.92 11.33
Kspar (clean) 26.21 30.35 3.00 4.71
Kspar (+ Musc/Illite) 34.37 29.39 10.60 10.75
Biotite† 0.64 1.82 2.20 4.82
Chlorite 0.02 0.03 1.55 3.91
Titanite - - 0.33 0.92
Laumontite 0.07 0.08 7.49 7.35
Kaolinite? 3.07 3.47 5.69 4.80
Calcite - - 0.03 0.07
Garnet/clays 0.05 0.04 0.23 0.31
Iron Oxides - - 0.02 0.14
Other (mainly fines) 2.54 4.17 16.91 21.16
† - Matrix/shear zone fill in sample 0307 which does not contain biotite
The noisy appearance of the image is due to beam width and scattering, which can
sometimes create artificial border effects where two minerals are touching. For example,
the contact between K-feldspar and albite in Fig. 4.8 will be mapped as sodium bearing
K-feldspar, even if the contact is actually sharp. A calcite filled crack in quartz with
kaolinite rims (Fig. 4.9) can be mapped as laumontite because of beam width.
Even though the two samples are of similar composition, the probe generated maps
of their cataclastic zones are quite different, and the amount and spread of alteration in
sample 1168A is much larger. It also contains more Fe, Mg and Ti bearing minerals.
Table 4.3 summarizes the results of the mapping according to the detected minerals. The
149
cataclastic region in sample 1168A is next to a clear zone of shear (Fig. 4.2) with more
clays and clay size minerals. Therefore it contains a larger amount of clay minerals and
fines.
Fig. 4.9: Elemental mapping of two zones in the cataclastic region in sample 1168A. The maps
are 1x1 mm or 512x512 pixels in size.
150
Discussion
PSD
Wechsler et al. (2010) measured the PSD of adjacent samples from the same core
section using laser particle analyzer, and calculated their D-values. For samples 0307 and
1168A they obtained D = 2.74 and D = 2.79, respectively. The D-value was calculated
over the range of 0.5-500 micron. Considering the 3-D measurement method, it is
necessary to subtract one dimension for a comparison with this paper 2-D estimates,
resulting in D = 1.74 and D = 1.79. Fig. 4.10 compares Wechsler et al.’s results to this
paper. The PSD curves were converted to 2-D by diving each bin’s value by its diameter.
The intermediate value of D for a whole rock sample, not separated by damage texture, is
indicative of a mixture of the various particle size populations.
Fig. 4.10: Comparisons of PSD derived from SEM measurements (this study) and laser particle
analyzer (Wechsler et al., 2010). PSD values are offset on the y-axis for clarity. The line
represents Wechsler et al. D-value for the indicated sample.
151
We find that a break in slope occurs at an equivalent diameter of 2-5 micron in the
cataclastic zones PSD, which slightly larger than reported by Keulen et al. (2007), who
found the break at an equivalent radius 0.9-2 micron. The break in slope is observed in
this study only in cataclastic samples, while Keulen et al. found a break in slope both in
their “cracked” (similar to pulverized) and “gouge” (similar to cataclastic) samples.
Wechsler et al. did not find a break in slope at around 2-5 micron, which can be
explained by a varying sensitivity of the laser particle analyzer at various scales. When
comparing the results from the laser particle analyzer to the pipette method, however,
Wechsler et al. found that the laser particle analyzer usually detected more particles in the
clay fraction than what was measured by pipetting. Another possibility is that in SEM
studies such as this one, a loss of the finest fraction (mostly clays) occurs during the thin
section preparation process, which artificially lowers the number of small particles seen
by the SEM. This is a known difficulty for samples with high clay content.
Interestingly, though the elemental mapping results indicate more fines in the
cataclastic region of sample 1168A, its D-value is slightly lower than the cataclastic D-
value for sample 0307 (Table 4.1). On the other hand the break in slope, which is
considered an indication of a change in the breaking mechanism, occurs at 5 microns in
sample 1168A and in 2 microns in sample 0307. This difference reflects the larger
amount of fines in sample 1168A.
152
Fig. 4.11: SEM images. (a) Plagioclase breaking along cleavage planes into similar fragments
while neighboring quartz is shattered to various size fragments. BSE image taken from sample
0307, pulverized region. (b) Shattered garnet. BSE image taken from sample 0307, pulverized
region. (c) Poke-marked (altered) plagioclase next to K-feldspar. The plagioclase does not break
along cleavage. BSE image taken from sample 1168A, pulverized region. (d) A close up on a
shear zone, SE image taken from sample 0307, cataclastic region. Those are some of the smallest
particles observed in our samples. P – plagioclase, Q – quartz, G – garnet, K – K-feldpar.
Affects of Mineralogy on PSD and Shape
In general, higher D-value is considered an indication for more efficient
comminution, which produces a larger number of smaller particles. In the pulverized
regions the D-value of the lowest magnification was larger than the rest, most notably in
153
sample 0307, where D = 2.34 in the range of 63-500 micron. Consequently there must be
a non-fractal process that only affects the largest fractions, causing a reduction in the
largest particle size population, without shear and comminution. Cleavage in minerals
can be one such mechanism, being a plane of weakness that will tend to break under
stress more easily, as seen in the pulverized regions (Fig. 4.11a). This tendency will
possibly affect the PSD and shape analysis, because the breaking of cleavage planes will
create more evenly sized, rectangular particles. It is possible that because measurements
are not separated by mineralogy, the mixing of quartz and feldspar particle populations
create the larger D-value for the largest diameter fractions. This steepening of slope
toward the largest fractions is also observed by Wechsler et al. (2010) in their PSD
samples, which they ascribe to the result of mixing two Gaussian populations.
The affect of mineralogy on the shape of particles is also seen in the shape analysis
results (Fig. 4.7), where the L/S values of the lowest magnification in pulverized regions
have the lowest values. This is the result of plagioclase breaking along cleavage into
more regularly shaped, rectangular fragments. Heilbronner and Keulen (2006) observed
that quartz has larger L/S values than feldspar in experimentally deformed rocks.
In an isotropic, homogeneous rock, the initial breaking would be expected to
produce PSD with a comparatively low D-value, and slip will cause further breaking and
comminution, producing higher D-values. If there are planes of weakness in a mineral
that tend to break more easily, the D-value can be influenced by their spacing and the
initial breaking will produce smaller fragments compared to fracturing homogeneous
minerals. This difference in particle size is observed between minerals with and without
154
weakness planes. Quartz in pulverized regions is initially fragmented into angular
fragments of various sizes, and so is garnet (Fig. 4.11b), while plagioclase breaks into
more evenly sized and shaped fragments.
Processes of Particle Break-up
There have been various suggestions for the types of processes that act on fault
zone rocks and break them into smaller fragments. Our observations are mostly
consistent with the accepted model of a two-stage process, in which an initial rupturing
produces the pulverized textures, and further particle comminution occurs by slip,
attrition, shearing or grinding, producing cataclastic textures. There was no evidence of
particle size reduction in the larger fractions by clay alteration, as suggested by
Belnkinsop (1991). In sample 1168A the plagioclase alteration, evident by its scarred and
poke-marked appearance in BSE images, was not exclusive to the cataclastic regions and
did not cause it to break more easily (Fig. 4.11c). A significant increase in the amount of
fines and clays is observed next to a shear zone in sample 1168A, however. The fine
particles next to such a zone would be affected more readily by fluids and perhaps some
of the fining of particles seen in shear zones is also related to processes of alterations.
Fracture Surface Energy
It is possible to calculate the total surface area for both types of damage texture
from the PSD, which in turn can be used to estimate a part of the energy release during an
earthquake (Kanamori, 1994; Wilson et al., 2005; Chester et al., 2005; Keulen et al.,
2007). Keulen et al. (2007) had shown that for PSDs similar to this study, the amount of
energy necessary for the fragmentation is negligible. Considering that the smallest
155
particles observed in our samples are ~0.3 micron in diameter (Fig. 4.11d), which is
larger than the lowest bound on particle size used by Keulen et al., even lower surface
density values would be expected. An additional reduction in the energy estimates comes
from the fact that the energy per surface is usually calculated for quartz, while our
samples contain other minerals with cleavage planes that are easier to break. Further,
matrix composition next to shear zones is an indication of particle size reduction through
clay alteration, and not just by the creation of new surfaces by fragmentation.
On the other hand, our samples come from the damage zone and not from the fault
zone itself. The amount of energy spent in the creation and maintenance of the damage
zone is poorly constrained, in part due to incomplete knowledge about its geometry.
Assuming similar fragmentation over a damage zone width of ~500 m (Dor et al., 2006;
Wechsler et al., 2009) would add two orders of magnitude to the surface energy
calculations and put it on par with the amount calculated by Chester et al. (2005) for the
Punchbowl fault. There is still not enough data on how the damage intensity varies across
the fault zone, along the fault, or with depth. Therefore calculating surface areas using
only one point in the damage along the fault is quite futile.
Conclusions
In this study SEM images are used to estimate PSDs of fault zone rocks from a
shallow core, and elemental mapping methods are used to study their composition. Two
types of damage textures are defined, pulverized and cataclastic. We find that damage
zone rocks have similar PSD and D-values to those previously found for both
156
experimental and fault zone material. A break in slope is observed at around 2-5 micron
in the cataclastic regions only, to which several reasons are suggested, either a loss of the
clays in the sectioning process, or a difference in particle break-up processes at the
smallest scale which involves alteration at the cataclastic regions but not the pulverized
regions.
D-value observations in this study are mostly consistent with the accepted model of
a two-stage process of fragmentation. Nonetheless, it is demonstrated that non-fractal
processes can affect the PSD and the particle shape, such as the fracturing of minerals
along planes of weakness or clay alteration near shear zones.
The damage zone can be an important factor in the earthquake energy budget
calculations, mainly due to its width and amount of damage that is taken up by
pulverization. Although it is not yet clear how pulverized rocks are created, they are a
unique damage feature, different than cataclasite or gouge, with a wide occurrence in
granitic rocks along major faults.
157
Summary
This thesis presented field and lab observations from the damage zone of the San
Andreas and San Jacinto faults in southern California, with a focus on pulverized granitic
rocks. The study also presented new methods for analysis of fault zone data using GIS,
one for quantifying fault trace heterogeneities, another for studying the damage zone
using spatial analysis of drainage patterns. The main results and conclusions derived from
this work are summarized below.
In chapter 1, a systematic comparative analysis method for geometrical fault zone
heterogeneities using derived measures from digitized fault maps was developed. Several
parameters for quantifying fault trace heterogeneity were defined using statistics of
circular data and binning methods. Those parameters, which quantify the range or
dispersion in the data, provide effective measures of the fault zone disorder. The
cumulative slip and slip rate (or more generally slip rate normalized by healing rate)
represent the fault zone maturity. The fault zone misalignment from plate motion
direction does not seem to play a major role in controlling the fault trace heterogeneities.
The results in chapter 2 demonstrate that the damage zone along the central San
Jacinto Fault (SJF) zone can be resolved with remotely-sensed data in a quantitative
fashion. The use of high resolution topography from LiDAR data enabled focusing on the
main fault strand and removing influences from nearby faults. The LiDAR data proved
more useful for observations at the scales of the fault process zone. It was concluded that
there is a strong correlation between drainage density (Dd) and proximity to the fault,
with zones of structural complexity along the fault displaying the highest Dd. This was
158
interpreted as an effect of degree of rock damage, as these are areas that are expected to
be more damaged, and field observations support this contention.
The fault damage zone, as inferred from drainage properties, is more pronounced
near areas of complexities in the surface trace, such as kinks, double strands and fault
junctions. Highly pulverized granites along the San Jacinto fault are mostly located in
areas that are between two fault strands, as well as on the hanging wall of the fault where
it bends, resulting in areas where thrusting is the main slip component.
Chapter 3 results are divided into several subjects. Regarding particle size
distribution, the results agree with previous works that the pulverized granitoid PSD is
mostly fine sand and silt in size. All of the rock’s constituent minerals react similarly to
deformation and break into various size grains, with no observed size preference for the
mineral fractions. D-value results point to a mixed-mode deformation in the rocks, which
can be modeled using two Gaussians to describe two particle populations, one coarse and
one fine. The rocks contain abundant evidence of shear deformation, and the damage is
not homogeneous in its intensity and the distribution of particle sizes vary considerably
within the core.
Whole rock chemistry analysis indicates fluid-rock interactions and calcite
metasomatism occurring in spatial context with secondary faults. This supports the
contention that faults can be either conduits or barriers for fluids which could interact
with the rock and change its composition. Clay mineralogy analysis indicates that
although weathering does occur at the surface and contribute to the fine particle
population, evidence for it disappear rapidly with depth, and the observed clay minerals
159
are a result of faulting and damage. The fracturing and damage of the rock increase the
permeability and facilitate alteration processes.
Results of investigation into the PSD of different damage textures using SEM are
presented in chapter 4. It is concluded that the calculated D-values are mostly consistent
with the accepted model of a two-stage process of fragmentation, but that non-fractal
processes, such as the fracturing of minerals along planes of weakness or clay alteration
near shear zones, can affect the PSD, D-values and the particle shapes.
In chapters 3 and 4, the PSD and its relation to earthquake energy budget
calculations are discussed, and it is concluded that even taking into account pulverized
rocks, shear zones, and a wide damage zone, the amount of energy required for creating
and maintaining that damage zone is on the order of a few percent of the total released
seismic energy.
The implications of the results are as follows:
• The geometrical properties of fault zones cannot be described by a steady-state
process associated with universal scale-invariant functions, but rather exhibit clear
evolutionary trends. The progressive regularization of geometrical heterogeneities is
expected to lead to a more efficient mechanical process, associated with larger
dynamic weakening during failure, higher slip- and rupture-velocities and higher
seismic radiation. Those in turn would cause the widening of the surrounding
damage zone.
160
• The mapping of drainage density implies that the northeast side of the SJF is
generally more damaged. The implications of the observed asymmetry could be
geological evidence for a preferred rupture propagation direction, possibly a result
of material contrast at depth. A preferred propagation direction is predicted to
produce asymmetric damage structure that would be recorded in the volume of rock
surrounding a fault.
• Heterogeneities seen in the fault trace can create stress concentrations and are
correlated with observations of higher damage levels.
• The pulverized rocks bodies observed in this study contain shear, which may be
concurrent to pulverization. There is however a difference in the breaking
mechanisms that produce shear/cataclastic or pulverized textures. Whether shear
and pulverization occur simultaneously or not is not yet clear, but this issue has
implications for the theory of how pulverization occurs.
• At least part of the size reduction in the smallest particle sizes is not by mechanical
breaking, but is the result of clay alteration which is not related to surface
weathering processes. Clay alteration at shallow depths implies the presence of
fluids in the rocks. Fluids can significantly affect the mechanical properties of
rocks, but lab experiments are often performed on dry rocks. The ubiquitous
presence of clay minerals in fault zone observations are also sometimes ignored in
rock mechanics experiments. Both clays and fluids can reduce friction of faults, and
are often invoked as a reason for the weakness of faults.
161
• Different minerals fragment and deform differently under similar conditions,
according to their structural and mechanical properties. It was observed that schist
in contact with pulverized granite sustained much less observed damage. Care
should be taken when estimating the amount of damage in different rock types,
because they can react very differently under similar stresses.
• The pulverization of granitic rocks next to the SAF is observed and probably occurs
at or close to the surface, but there are no observations of similar damage along
exhumed faults. Assuming that pulverization is a product of high stain rates related
to dynamic rupturing, it is possible that pulverization is a shallow phenomenon
associated with dynamic generation of tensile stresses and characteristic only of the
damage zone above the seismogenic depth. It is important to consider the
geometrical and material properties of the fault. The SAF is a mature fault,
especially the Mojave section (as demonstrated in chapter 1), which separates
different rock bodies and is capable of generating very large earthquakes. The
material contrast across the fault may aid in the generation of high slip rates and
tensile stresses during the occurrence of earthquake ruptures. The currently known
exhumed faults may not have been as mature as the SAF is today and were not
capable of generating such high strain rates or tensile stresses. Pulverization may
possibly be observed along exhumed thrusts that are capable of generating very
large earthquakes, especially in the hanging block near the surface.
162
Bibliography
Ampuero, J.P., Ben-Zion, Y., 2008, Cracks, pulses and macroscopic asymmetry of
dynamic rupture on a bimaterial interface with velocity-weakening friction.
Geophysical Journal International 173, 674–692.
An, L.J., and Sammis, C.G., 1994, Particle-Size Distribution of Cataclastic Fault
Materials from Southern California - a 3-D Study: Pure and Applied Geophysics, v.
143, p. 203-227.
Anderson, J.L., Osborne, R.H., and Palmer, D.F., 1980, Petrogenesis of cataclastic rocks
within the San Andreas fault zone of Southern California, U.S.A: Tectonophysics, v.
67, p. 221-249.
Anderson, J.L., Osborne, R.H., and Palmer, D.F., 1983, Cataclastic rocks of the San
Gabriel fault: an expression of deformation at deeper crustal levels in the San
Andreas fault zone: Tectonophysics, v. 98, p. 209-251.
Ando, R., Shaw, B.E. and Scholz, C.H., 2009, Quantifying natural fault geometry:
statistics of splay fault angles, Bulletin of the Seismological Society of America, 99,
389–395.
Aochi, H., Fukuyama, E. and Matsu'ura, M., 2000, Spontaneous rupture propagation on a
non-planar fault in 3-D elastic medium, Pure and Applied Geophysics, 157, 2003-
2027.
Arrowsmith, J.R., Crosby, C.J., 2006, New looks at active faults: tectonic geomorphology
using Airborne Laser Swath Mapping (ALSM), Seismological Society of America
Annual Meeting, San-Francisco.
Bailey, I. W., Y Ben-Zion, T. W. Becker and M. Holschneider, 2010, Quantifying focal
mechanism heterogeneity for fault zones in Southern and Central California,
Geophys. J. Int, in review.
Bailey, I.W. and Ben-Zion, Y., 2009, Statistics of Earthquake Stress Drops on a
Heterogeneous Fault in an Elastic Half-Space, Bulletin of the Seismological Society
of America, 99, 1786-1800.
Barrows, A.G., Kahle, J.E., and Beeby, D.J., 1985, Earthquake Hazard and tectonic
history of the San Andreas Fault zone, Los Angeles County, California.
163
Barton, C.C., 1995, Fractal analysis of scaling and spatial clustering of fractures. in
Fractals in the Earth Sciences, pp. 141-178, ed. Pointe, C. C. B. P. R. L. Plenum
Press, New-York.
Becker, T.W., Hardebeck, J.L., Anderson, G., 2005, Constraints on fault slip rates of the
southern California plate boundary from GPS velocity and stress inversions.
Geophysical Journal International 160, 634–650.
Beeler, N.M. and Hickman, S.H., 2004, Stress-induced, time-dependent fracture closure
at hydrothermal conditions, Journal of Geophysical Research-Solid Earth, 109,
B02211, doi:10.1029/2002JB001782.
Ben-Zion, Y. and Andrews, D.J., 1998, Properties and implications of dynamic rupture
along a material interface, Bulletin of the Seismological Society of America, 88,
1085-1094.
Ben-Zion, Y. and Rice, J.R., 1993, Earthquake failure sequences along a cellular fault
zone in a 3-dimensional elastic solid containing asperity and nonasperity regions,
Journal of Geophysical Research-Solid Earth, 98, 14109-14131.
Ben-Zion, Y. and Rice, J.R., 1995, Slip patterns and earthquake populations along
different classes of faults in elastic solids, Journal of Geophysical Research-Solid
Earth, 100, 12959-12983.
Ben-Zion, Y., 1996, Stress, slip, and earthquakes in models of complex single-fault
systems incorporating brittle and creep deformations, Journal of Geophysical
Research-Solid Earth, 101, 5677-5706.
Ben-Zion, Y., 2001, Dynamic ruptures in recent models of earthquake faults: Journal of
the Mechanics and Physics of Solids, v. 49, p. 2209-2244.
Ben-Zion, Y., 2008, Collective behavior of earthquakes and faults: Continuum-discrete
transitions, progressive evolutionary changes, and different dynamic regimes:
Reviews of Geophysics, v. 46, p. 70.
Ben-Zion, Y., and Andrews, D.J., 1998, Properties and implications of dynamic rupture
along a material interface: Bulletin of the Seismological Society of America, v. 88, p.
1085-1094.
164
Ben-Zion, Y., and Huang, Y., 2002, Dynamic rupture on an interface between a
compliant fault zone layer and a stiffer surrounding solid: Journal of Geophysical
Research, v. 107.
Ben-Zion, Y., and Shi, Z.Q., 2005, Dynamic rupture on a material interface with
spontaneous generation of plastic strain in the bulk: Earth and Planetary Science
Letters, v. 236, p. 486-496.
Ben-Zion, Y., Dahmen, K., Lyakhovsky, V., Ertas, D. and Agnon, A., 1999, Self-driven
mode switching of earthquake activity on a fault system, Earth and Planetary Science
Letters, 172, 11-21.
Ben-Zion, Y., Peng, Z., Okaya, D., Seeber, L., Armbruster, J.G., Ozer, N., Michael, A.J.,
Baris, S. and Aktar, M., 2003, A shallow fault zone structure illuminated by trapped
waves in the Karadere-Duzce branch of the North Anatolian Fault, western Turkey,
Geophysical Journal International, 152, 699-717.
Ben-Zion, Y., Sammis, C.G., 2003, Characterization of fault zones. Pure and Applied
Geophysics 160, 677–715.
Bhat, H. S., M. Olives, R. Dmowska and J. R. Rice, 2007, Role of fault branches in
earthquake rupture dynamics, J. Geophys. Res., 112, B11309,
doi:10.1029/2007JB005027.
Biegel, R.L., Sammis, C.G., 2004, Relating fault mechanics to fault zone structure,
Advances in Geophysics, v. 47, p. 65–111.
Bjork, T.E., Mair, K., and Austrheim, H., 2009, Quantifying granular material and
deformation: Advantages of combining grain size, shape, and mineral phase
recognition analysis: Journal of Structural Geology, v. 31, p. 637-653.
Blenkinsop, T.G., 1991, Cataclasis and processes of particle size reduction: Pure and
Applied Geophysics, v. 136, p. 59-86.
Blythe, A.E., Burbank, D.W., Farley, K.A., and Fielding, E.J., 2000, Structural and
topographic evolution of the central Transverse Ranges, California, from apatite
fission-track, (U-Th)/He and digital elevation model analyses: Basin Research, v. 12,
p. 97-114.
165
Brietzke, G.B. and Ben-Zion, Y., 2006, Examining tendencies of in-plane rupture to
migrate to material interfaces, Geophysical Journal International, 167, 807-819.
Brietzke, G.B., Cochard, A., and Igel, H., 2009, Importance of bimaterial interfaces for
earthquake dynamics and strong ground motion: Geophysical Journal International,
v. 178, p. 921-938.
Brune, J.N., 2001, Fault-normal dynamic unloading and loading: an explanation for “non-
gouge” rock powder and the lack of fault-parallel shear bands along the San Andreas
Fault: EOS Transactions, v. 82, p. 47.
Brune, J.N., Brown, S., and Johnson, P.A., 1993, Rupture mechanism and interface
separation in foam rubber model of earthquakes: a possible solution to the heat flow
paradox and the paradox of large overthrusts: Tectonophysics, v. 218, p. 59-67.
Candela, T., Renard, F., Bouchon, M., Brouste, A., Marsan, D., Schmittbuhl, J. and
Voisin, C., 2009, Characterization of Fault Roughness at Various Scales:
Implications of Three-Dimensional High Resolution Topography Measurements,
Pure and Applied Geophysics, 166, 1817-1851.
Chakravarti, I.M., Laha, R.G., Roy, J., 1967, Handbook of methods of applied statistics.
Wiley, New York, pp. 392–394.
Chester, F.M., Chester, J.S., 1998. Ultracataclasite structure and friction processes of the
Punchbowl fault, San Andreas system, California. Tectonophysics 295, 199–221.
Chester, F.M., Chester, J.S., Kirschner, D.L., Schultz, S.E., and Evans, J.P., 2004,
Structure of large-displacement, strike-slip fault zones in the brittle continental crust,
in Karner, G.D., Taylor, B., Driscoll, N.W., and Kohlstedt, D.L., eds., Rheology and
Deformation in the Lithosphere at Continental Margins: New York, Columbia
University Press.
Chester, F.M., Evans, J.P., and Biegel, R.L., 1993, Internal Structure and Weakening
Mechanisms of the San-Andreas Fault: Journal of Geophysical Research, v. 98, p.
771-786.
Chester, J.S., Chester, F.M., and Kronenberg, A.K., 2005, Fracture surface energy of the
Punchbowl fault, San Andreas system: Nature, v. 437, p. 133-136.
166
Cowie, P.A., Scholz, C.H., 1992, Physical explanation for the displacement length
relationship of faults using a post-yield fracture-mechanics model, Journal of
Structural Geology, v. 14 (10), p. 1133–1148.
Dalguer, L.A., and Day, S.M., 2009, Asymmetric rupture of large aspect-ratio faults at
bimaterial interface in 3D: Geophysical Research Letters, v. 36.
de Joussineau, G. and Aydin, A., 2009, Segmentation along Strike-Slip Faults Revisited,
Pure and Applied Geophysics, 166, 1575-1594.
DeMets, C., Gordon, R.G., Argus, D.F. and Stein, S., 1994, Effect of recent revisions to
the geomagnetic reversal time scale on estimate of current plate motions, Geophys.
Res. Lett., 21, 2191-2194.
Densmore, A.L., Gupta, S., Allen, P.A., Dawers, N.H., 2007, Transient landscapes at
fault tips. Journal of Geophysical Research — Earth Surface 112 (F3), F03S08.
Dieterich, J.H. and Kilgore, B.D., 1996, Imaging surface contacts: Power law contact
distributions and contact stresses in quartz, calcite, glass and acrylic plastic,
Tectonophysics, 256, 219-239.
Doan, M.L., and Gary, G., 2009, Rock pulverization at high strain rate near the San
Andreas fault: Nature Geoscience, v. 2, p. 709-712.
Dokka, R.K. and Travis, C.J., 1990, Late Cenozoic strike-slip faulting in the Mojave
desert, California, Tectonics, 9, 311-340.
Dor, O., Ben-Zion, Y., Rockwell, T.K., and Brune, J., 2006a, Pulverized rocks in the
Mojave section of the San Andreas Fault Zone: Earth and Planetary Science Letters,
v. 245, p. 642-654.
Dor, O., Rockwell, T.K., Ben-Zion, Y., 2006b, Geological observations of damage
asymmetry in the structure of the San Jacinto, San Andreas and Punchbowl faults in
Southern California: a possible indicator for preferred rupture propagation direction.
Pure and Applied Geophysics 163, 301–349.
167
Dor, O., Yildirim, C., Rockwell, T.K., Ben-Zion, Y., Emre, O., Sisk, M., 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,
483–504.
Dorsey, R.J., Roering, J.J., 2006, Quaternary landscape evolution in the San Jacinto fault
zone, Peninsular Ranges of Southern California: transient response to strike-slip fault
initiation. Geomorphology 73, 16–32.
Douglas, D.H. and Peucker, T.K., 1973, Algorithms for the reduction of the number of
points required to represent a digitized line or its caricature, The Canadian
Cartographer, 10, 112-122.
Elliott, A.J., Dolan, J.F. and Oglesby, D.D., 2009, Evidence from coseismic slip gradients
for dynamic control on rupture propagation and arrest through stepovers, Journal of
Geophysical Research-Solid Earth, 114, B02312.
Evans, J.P., and Chester, F.M., 1995, Fluid-Rock Interaction In Faults of the San-Andreas
System - Inferences From San-Gabriel Fault Rock Geochemistry and
Microstructures, Journal of Geophysical Research, v. 100, p. 13007-13020.
Faulkner, D.R., Lewis, A.C., and Rutter, E.H., 2003, On the internal structure and
mechanics of large strike-slip fault zones: field observations of the Carboneras fault
in southeastem Spain, Tectonophysics, v. 367 (3–4), p. 235–251.
Faulkner, D.R., Mitchell, T.M., Healy, D., and Heap, M.J., 2006, Slip on ‘weak’ faults by
the rotation of regional stress in the fracture damage zone, Nature, v. 444, p. 922–
925.
Faulkner, D.R., Mitchell, T.M., Rutter, E.H., and Cembrano, J., 2008, On the structure
and mechanical properties of large strike-slip faults, In: Wibberley, C.A.J., Kurz, W.,
Imber, J., Holdsworth, R.E., Collettini, C. (Eds.), Structure of Fault Zones:
Implications for Mechanical and Fluid-flow Properties, Geological Society, London,
Special Publications, v. 299, p. 139–150.
Fialko, Y., 2006, Interseismic strain accumulation and the earthquake potential on the
southern San Andreas fault system. Nature 441 (7096), 968–971.
168
Finzi, Y., Hearn, E.H., Ben-Zion, Y., Lyakhovsky, V., 2009, Structural properties and
deformation patterns of evolving strike-slip faults: Numerical simulations
incorporating damage rheology. Pure and Applied Geophysics 166,
doi:10.1007/s00024-009-0522-1.
Fisher, N.I., 1993, Statistical Analysis of Circular Data, edn, Vol., pp. Pages, Cambridge
University Press, New-York.
Flinn, D., 1977, Transcurrent faults and associated cataclasis in Shetland: Journal of the
Geological Society of London, v. 133, p. 231-248.
Garfunkel, Z., 1974, Model for late Cenozoic tectonic history of Mojave-desert,
California, and for its relation to adjacent regions, Geological Society of America
Bulletin, 85, 1931-1944.
Girty, G.H., Biggs, M.A., and Berry, R.W., 2008, An unusual occurrence of probable
Pleistocene corestone within a Cretaceous dioritic enclave, Peninsular Ranges,
California: Catena, v. 74, p. 43-57.
Girty, G.H., Stephenson, D., Canfield, A., Marsh , J., Middleton, T., Rayburn, J., Sisk,
M., Verdugo, D., Moniz, R., Gunter, J., Nielsen, J., Gasca, C., Rowland-Smith, A.,
Smith, M., Lovering, K., Campbell, K., Castenada, F., Gray, R., and Campbell, C.,
2006, Geology of Picacho State Recreation Area, SE California: Implications for the
timing of the formation of the Chocolate Mountains anticlinorium, in Girty, G.H.,
and Cooper, J., eds., Using Stratigraphy, Sedimentology, and Geochemistry to
Unravel the Geologic History of the Southwestern Cordillera : A volume in honor of
Patrick L. Abbott, p. 77-96.
Graymer, R.W., Sarna-Wojcicki, A.M., Walker, J.P., McLaughlin, R.J. and Fleck, R.J.,
2002. Controls on timing and amount of right-lateral offset on the East Bay fault
system, San Francisco Bay region, California, Geological Society of America
Bulletin, 114, 1471-1479.
Hamiel, Y., Fialko, Y., 2007, Structure and mechanical properties of faults in the North
Anatolian Fault system from InSAR observations of coseismic deformation due to
the 1999 Izmit (Turkey) earthquake. Journal of Geophysical Research-Solid Earth
112 (B7), B07412.
169
Hancock, G.R., 2005, The use of digital elevation models in the identification and
characterization of catchments over different grid scales. Hydrological Processes 19,
1727–1749.
Harris, R.A. and Day, S.M., 1993, Dynamics of fault interaction: Parallel strike-slip
faults, Journal of Geophysical Research-Solid Earth, 98, 6665-6665.
Heilbronner, R., and Keulen, N., 2006, Grain size and grain shape analysis of fault rocks:
Tectonophysics, v. 427, p. 199-216.
Hickman, S.H. and Evans, B., 1992, Growth of grain contacts in halite by solution
transfer; Implications for diagenesis, lithification, and strength recovery, edn, Vol.,
pp. Pages, Academic, San Diego, Calif.
Hillers, G., Ben-Zion, Y. and Mai, P.M., 2006, Seismicity on a fault controlled by rate-
and state-dependent friction with spatial variations of the critical slip distance,
Journal of Geophysical Research-Solid Earth, 111, 23.
Hillers, G., Mai, P.M., Ben-Zion, Y. and Ampuero, J.P., 2007, Statistical properties of
seismicity of fault zones at different evolutionary stages, Geophysical Journal
International, 169, 515-533.
Horton, R.E., 1945, Erosional development of streams and their drainage basins —
hydrophysical approach to quantitative morphology. Geological Society of America
Bulletin 56, 275–370.
Jennings, C.W., 1977, Geologic map of California: geologic data map no. 2. California
Department of Conservation, Division of Mines and Geology.
Jennings, C.W., 1985, An explanatory text to accompany the 1:750,000 scale fault and
geologic maps of California, California Department of Conservation, Division of
Mines and Geology.
Jennings, C.W., 1994, Fault activity map of California and adjacent areas: geologic data
map no. 6. California Department of Conservation, Division of Mines and Geology.
Jones, T.A., 2006, MATLAB functions to analyze directional (azimuthal) data - I: Single
sample inference, Computers and Geosciences, 32, 166-117.
170
Kame, N. and Yamashita, T., 1999, Simulation of the spontaneous growth of a dynamic
crack without constraints on the crack tip path, Geophysical Journal International,
139, 345-358.
Kanamori, H., 1994, Mechanics of earthquakes, Annual Review of Earth and Planetary
Sciences, v. 22, p. 207-237.
Keulen, N., Heilbronner, R., Stunitz, H., Boullier, A.M., and Ito, H., 2007, Grain size
distributions of fault rocks: A comparison between experimentally and naturally
deformed granitoids: Journal of Structural Geology, v. 29, p. 1282-1300.
Kim, H., Arrowsmith, J.R., Crosby, C.J., Jaeger-Frank, E., Nandigam, V., Memon, A.,
Conner, J., Badden, S.B., Baru, C., 2006, An efficient implementation of a local
binning algorithm for digital elevation model generation of LiDAR/ALSM dataset.
Eos Transactions 87 (52).
King, G. and Nabelek, J., 1985, Role of fault bends in the initiation and termination of
earthquake rupture, Science, 228, 984-987.
King, G., 1983, The accommodation of large strains in the upper lithosphere of the earth
and other solids by self-similar fault systems: the geometrical origin of b-value, Pure
and Applied Geophysics, 121, 761-815.
Kirby, S.M., Janecke, S.U., Dorsey, R.J., Housen, B.A., McDougall, K., Langenheim, V.,
Steely, A., 2007, Pleistocene Brawley and Ocotillo formations: evidence for initial
strike-slip deformation along the San Felipe and San Jacinto fault zones, California.
Journal of Geology 115, 43–62.
Kirchner, J.W., 1993, Statistical inevitability of Horton laws and the apparent
randomness of stream channel networks. Geology 21, 591–594.
Le, K.N., Rockwell, T.K., Owen, L.A., Oskin, M., Lippincott, C., Caffee, M.W., Dortch,
J., in review, Late Quaternary slip-rate gradient defined using high-resolution
topography and 10Be dating of offset landforms on the southern San Jacinto Fault,
California. Journal of Geophysical Research.
Lewis, M.A., Ben-Zion, Y., McGuire, J.J., 2007, Imaging the deep structure of the San
Andreas fault south of Hollister with joint analysis of fault zone head and direct P
arrivals. Geophysical Journal International 169, 1028–1042.
171
Lewis, M.A., Peng, Z., Ben-Zion, Y., Vernon, F.L., 2005, Shallow seismic trapping
structure in the San Jacinto fault zone near Anza, California. Geophysical Journal
International 162, 867–881.
Li, Y.G., Aki, K., Vidale, J.E., Fei, X.U., 1999, Shallow structure of the Landers fault
zone from explosion-generated trapped waves. Journal of Geophysical Research —
Solid Earth 104, 20257–20275.
Lyakhovsky, V., Ben-Zion, Y. and Agnon, A., 2001, Earthquake cycle, fault zones, and
seismicity patterns in a theologically layered lithosphere, Journal of Geophysical
Research-Solid Earth, 106, 4103-4120.
Mair, K., Frye, K.M., and Marone, C., 2002, Influence of grain characteristics on the
friction of granular shear zones: Journal of Geophysical Research-Solid Earth, v.
107.
Matti, J.C., Morton, D.M., 1993, Paleogeographic evolution of the San Andreas fault in
Southern California: a reconstruction based on a new cross-fault correlation. In:
Powell, R.E., Weldon, R.J., Matti, J.C. (Eds.), The San Andreas Fault System:
Displacement, Palinspastic Reconstruction, and Geologic Evolution. Geological
Society of America Memoir. Geological Society of America, Boulder, CO, pp. 107–
159.
McCaffrey, R., 2005. Block kinematics of the Pacific-North America plate boundary in
the southwestern United States from inversion of GPS, seismological, and geologic
data, Journal of Geophysical Research-Solid Earth, 110.
Miller, S.A., Collettini, C., Chiaraluce, L., Cocco, M., Barchi, M., and Kaus, B.J.P.,
2004, Aftershocks driven by a high-pressure CO
2
source at depth, Nature, v. 427, p.
724–727.
Mitchell, T. M., Shimamoto, T., and Ben-Zion, Y., 2009, Pulverized Fault Rocks along
the Arima-Takatsuki Tectonic Line, Japan: Fault Structure, Damage Distribution
and Textural Characteristics, Abstract, European Geosciences Union, Vienna,
Austria. Geophys. Res. Abs. 11, EGU2009-11751-2, 2009
Mitchell, T.M., and Faulkner, D.R., 2009, The nature and origin of off-fault damage
surrounding strike-slip fault zones with a wide range of displacements: A field study
from the Atacama fault system, northern Chile: Journal of Structural Geology, v. 31,
p. 802-816.
172
Molnar, P., Anderson, R.S., Anderson, S.P., 2007, Tectonics, fracturing of rock, and
erosion. Journal of Geophysical Research — Earth Surface 112, F03014.
Montgomery, D.R., Dietrich,W.E., 1992, Channel initiation and the problem of landscape
scale. Science 255 (5046), 826–830.
Nakatani, M. and Scholz, C.H., 2004, Frictional healing of quartz gouge under
hydrothermal conditions: 1. Experimental evidence for solution transfer healing
mechanism, Journal of Geophysical Research-Solid Earth, 109.
Nesbit, H.W., and Young, G.M., 1982, Early Proterozoic climates and plate motions
inferred from major element chemistry of lutites: Nature, v. 299, p. 715-717.
Nesbit, H.W., Young, G.M., McLennan, S.M., and Keays, R.R., 1996, Effects of
Chemical Weathering and Sorting on the Petrogenesis of Siliciclastic Sediments,
with Implications for Provenance Studies: Journal of Geology, v. 104, p. 525-542.
Olsen, K.B., Day, S.M., Minster, J.B., Cui, Y., Chouasia, A., Moore, R., Maechling, P.,
Jordan, T., 2008, TeraShake2: simulation of Mw7.7 earthquakes on the southern San
Andreas fault with spontaneous rupture description. Bulletin of the Seismological
Society of America 98, 1162–1185.
Papageorgiou, A.S. and Aki, K., 1983, A specific barrier model for the quantitative
description of inhomogeneous faulting and the prediction of strong ground motion,
II, applications of the model, Bulletin of the Seismological Society of America, 73,
953-978.
Petersen, M.D, and Wesnousky, S.G., 1994, Fault slip rates and earthquake histories for
active faults in southern California, Bulletin of the Seismological Society of
America, 84, 1608-1649.
Poliakov, A.N.B., Dmowska, R. and Rice, J.R., 2002, Dynamic shear rupture interactions
with fault bends and off-axis secondary faulting, Journal of Geophysical Research-
Solid Earth, 107, 18.
173
Powell, R.E., 1993, Balanced palinspastic reconstruction of pre-late Cenozoic
paleogeography, southern California: geologic and kinematic constraints on
evolution of the San Andreas fault system. in The San Andreas Fault System:
Displacement, Palinspastic Reconstruction, and Geologic Evolution, pp. 1-106, eds.
Powell, R. E., Weldon, R. J. and Matti, J. C. Geological Society of America,
Boulder, CO.
Pucci, S., Pantosti, D., Barchi, M.R. and Palyvos, N., 2007, A complex seismogenic shear
zone: The Duzce segment of north anatolian fault (Turkey), Earth and Planetary
Science Letters, 262, 185-203.
Rice, J.R., Ben-Zion, Y. and Kim, K.S., 1994, 3-Dimentional perturbation solution for a
dynamic planar crack moving unsteadily in a model elastic solid, Journal of the
Mechanics and Physics of Solids, 42, 813-843.
Rockwell, T., Loughman, C., Merifield, P.,1990. Late Quaternary rate of slip along the
San- Jacinto Fault Zone near Anza, Southern California. Journal of Geophysical
Research- Solid Earth and Planets 95 (B6), 8593–8605.
Rockwell, T.K., 2000, Use of soil geomorphology in fault studies, in Noller, J.S., Sowes,
J.M., and Lettis, W.R., eds., Quaternary Geochronology: Methods and Applications:
AGU reference shelf 4: Washington D.C., American Geophysical Union, p. 273-292.
Rockwell, T.K., and Ben-Zion, Y., 2007, High localization of primary slip zones in large
earthquakes from paleoseismic trenches: Observations and implications for
earthquake physics: Journal of Geophysical Research, v. 112.
Rockwell, T.K., Seitz, G.G., 2008, Observations of mode-switching from long
paleoseismic records of earthquakes on the San Jacinto and San Andreas faults:
implications for making hazard estimates from short paleoseismic records, 33rd
International Geological Congress, Oslo, Norwey.
Rockwell, T.K., Seitz, G.G., Dawson, T.E., Young, J., 2006, The long record of San
Jacinto Fault paleoearthquakes at Hog Lake; implications for regional patterns of
strain release in the southern San Andreas Fault system. Seismological Research
Letters 77, 270.
174
Rockwell, T.K., Sisk, M., Girty, G., Dor, O., Wechsler, N., and Ben-Zion, Y., 2009,
Chemical and physical characteristics of pulverized Tejon Lookout Granite adjacent
to the San Andreas and Garlock faults: Implications for earthquake physics: Pure and
Applied Geophysics, v. 166, p. 1725-1746.
Sagy, A., Brodsky, E.E. and Axen, G.J., 2007, Evolution of fault-surface roughness with
slip, Geology, 35, 283-286.
Sammis, C., King, G., and Biegel, R., 1987, The Kinematics of Gouge Deformation: Pure
and Applied Geophysics, v. 125, p. 777-812.
Sammis, C.G., and King, G.C.P., 2007, Mechanical origin of power law scaling in fault
zone rock: Geophysical Research Letters, v. 34.
Sammis, C.G., Osborne, R.H., Anderson, J.L., Banerdt, M., and White, P., 1986, Self-
Similar Cataclasis in the Formation of Fault Gouge: Pure and Applied Geophysics, v.
124, p. 53-78.
Sanders, C.O., Kanamori, H., 1984, A seismotectonic analysis of the Anza seismic gap,
San- Jacinto Fault zone, Southern-California. Journal of Geophysical Research 89,
5873–5890.
Savage, H. M., Brodsky. E., 2010, Collateral Damage: The Evolution of Fracture
Distribution in Fault Damage Zones, in review.
Schleicher, A.M., Warr, L.N., and van der Pluijim, B.A., 2009, On the origin of mixed-
layered clay minerals from the San Andreas Fault at 2.5-3 km vertical depth
(SAFOD drillhole at Parkfield, California): Contributions to Mineral Petrology, v.
157, p. 173-187.
Scholz, C.H., Ando, R. and Shaw, B.E., 2009, The mechanics of first order splay
faulting: The strike-slip case, Journal of Structural Geology, 32, 118-126.
Sedlock, R.L. and Hamilton, D.H., 1991, Late Cenozoic tectonic evolution of southern
California, Journal of Geophysical Research-Solid Earth and Planets, 96, 2325-2351.
Sharon, E., Gross, S.P. and Fineberg, J., 1995, Local crack branching as a mechanism for
instability in dynamic fracture, Phys. Rev. Lett., 74, 5096-5099.
175
Sharp, R.V., 1967, San Jacinto Fault Zone in peninsular ranges of Southern California.
Geological Society of America Bulletin 78, 705.
Sibson, R.H., 1977, Fault Rocks and Fault Mechanisms: Journal of the Geological
Society of London, v. 133, p. 191-213.
Sibson, R.H., 1986, Rupture interactions with fault jogs. in Earthquake Source
Mechanics, pp. 157-167, eds. Das, S., Boatwright, J. and Scholz, C. American
Geophysical Union.
Sieh, K.E. and Williams, P.L., 1990, Behaviour of the southernmost San-Andreas fault
during the past 300 years, Journal of Geophysical Research-Solid Earth and Planets,
95, 6629-6645.
Sperazza, M., Moore, J.N., and Hendrix, M.S., 2004, High-resolution particle size
analysis of naturally occurring very fine-grained sediment through laser
diffractometry: Journal of Sedimentary Research, v. 74, p. 736-743.
Stepinski, T.F., Collier, M.L., 2004, Extraction of Martian valley networks from digital
topography. Journal of Geophysical Research-Planets 109 (E11).
Stierman, D.J., 1984, Geophysical and geological evidence for fracturing, water
circulation and chemical alteration in granitic-rocks adjacent to major strike-slip
faults. Journal of Geophysical Research 89, 5849–5857.
Stillings, M.P., 2007, Structural, Textural, and Geochemical Analyses of Fault Damage
Zones: Clark Strand, San Jacinto Fault Zone, Southern California [M. Sc. thesis]:
San-Diego, San Diego State University.
Stirling, M.W., Wesnousky, S.G. and Shimazaki, K., 1996. Fault trace complexity,
cumulative slip, and the shape of the magnitude-frequency distribution for strike-slip
faults: A global survey, Geophysical Journal International, 124, 833-868.
Storti, F., Billi, A., and Salvini, F., 2003, Particle size distributions in natural carbonate
fault rocks: insights for non-self-similar cataclasis: Earth and Planetary Science
Letters, v. 206, p. 173-186.
Strahler, A.N., 1952, Hypsometric (area–altitude) analysis of erosional topography.
Geological Society of America Bulletin 63, 1117–1141.
176
Strahler, A.N., 1964, Quantitative geomorphology of drainage basins and channel
networks. In: Chow, V.T. (Ed.), Handbook of Applied Hydrology. McGraw-Hill,
New York, pp. 4.39–4.76.
Tarboton, D.G., 1997, A new method for the determination of flow directions and
upslope areas in grid digital elevation models. Water Resources Research 33, 309–
319.
Tarboton, D.G., Bras, R.L., Rodrigueziturbe, I., 1991, On the extraction of channel
networks from digital elevation data. Hydrological Processes 5, 81–100.
Tarboton, D.G., Bras, R.L., Rodrigueziturbe, I., 1992, A physical basis for drainage
density. Geomorphology 5, 59–76.
Tchalenko, J.S., 1970, Similarities between shear zones of different magnitudes,
Geological Society of America Bulletin, 81, 1625-1639.
Tucker, G.E., Catani, F., Rinaldo, A., Bras, R.L., 2001, Statistical analysis of drainage
density from digital terrain data. Geomorphology 36, 187–202.
Turcotte, D.L., 1986, A fractal model for crustal deformation, Tectonophysics, 132, 261-
269.
Turcotte, D.L., 1997, Fractals and Chaos in Geology and Geophysics, edn, Vol., pp.
Pages, Cambridge University Press, New York, NY 10011, USA.
US Geological Survey, 2006. Quaternary fault and fold database for the United States.
January 2007, from USGS web site: http//earthquakes.usgs.gov/regional/qfaults/.
Wechsler, N., Allen, E. E., Rockwell, T. K., Girty, G., Chester, J. S. and Ben-Zion, Y.,
2010, Characterization of Pulverized Granitoids in a Shallow Core along the San
Andreas Fault, Little Rock, CA, manuscript in prep.
Wechsler, N., Rockwell, T.K., and Ben-Zion, Y., 2009, Application of high resolution
DEM data to detect rock damage from geomorphic signals along the central San
Jacinto Fault: Geomorphology, v. 113, p. 82-96.
Weertman, J., 1980, Unstable slippage across a fault that separates elastic media of
different elastic-constants. Journal of Geophysical Research 85, 1455–1461.
177
Wesnousky, S.G., 1988, Seismological and structural evolution of strike-slip faults,
Nature, 335, 340-342.
Wesnousky, S.G., 1994, The Gutenberg-Richter or Characteristic Earthquake
distribution, which is it, Bulletin of the Seismological Society of America, 84, 1940-
1959.
Wesnousky, S.G., 2008, Displacement and geometrical characteristics of earthquake
surface ruptures: Issues and implications for seismic-hazard analysis and the process
of earthquake rupture, Bulletin of the Seismological Society of America, 98, 1609-
1632.
Whipple, K.X., 2004, Bedrock rivers and the geomorphology of active orogens. Annual
Review of Earth and Planetary Sciences 32, 151–185.
Wibberley, C.A.J., Shimamoto, T., 2003, Internal structure and permeability of major
strike-slip fault zones: the Median Tectonic Line in Mie Prefecture, southwest Japan,
Journal of Structural Geology, v. 25 (1), p. 59–78.
Willgoose, G., Hancock, G., 1998, Revisiting the hypsometric curve as an indicator of
form and process in transport-limited catchment. Earth Surface Processes and
Landforms 23, 611–623.
Wilson, B., Dewers, T., Reches, Z., and Brune, J.N., 2005, Particle size and energetics of
gouge from earthquake rupture zones: Nature, v. 434, p. 749-752.
Wilson, J.E., Chester, J.S., and Chester, F.M., 2003, Microfracture analysis of fault
growth and wear processes, Punchbowl Fault, San Andreas system, California:
Journal of Structural Geology, v. 25, p. 1855–1873.
Zoback, M. D., Zoback, M. L., Mount, V. S., Suppe, J., Eaton, J. P., Healy, J. H.,
Oppenheimer, D., Reasenberg, P., Jones, L., Raleigh, C. B., Wong, I.G., Scotti, O.
and Wentworth, C., 1987, New evidence on the state of stress of the San Andreas
fault system, Science, v. 238, p. 1105–1111.
178
Zoller, G., Hainzl, S., Ben-Zion, Y. and HoIschneider, M., 2006, Earthquake activity
related to seismic cycles in a model for a heterogeneous strike-slip fault,
Tectonophysics, v. 423, p. 137-145, doi:10.1016/j.tecto.2006.03.007.
Zoller, G., Holschneider, M. and Ben-Zion, Y., 2005, The role of heterogeneities as a
tuning parameter of earthquake dynamics, Pure and Applied Geophysics, 162, 1027-
1049.
179
Appendix 1: Little Rock Creek logs
Fig. A-1: The complete log of the first Little Rock core (1.5 – 40.2 meters depth), including
lithology, shear zones and secondary fault locations and orientation (when available), sample
names and locations, chemical index of alteration (CIA) and median grain size (in microns) for
each sample.
180
Fig. A-1, Continued.
181
Fig. A-1, Continued.
182
Fig. A-1, Continued.
183
Fig. A-1, Continued.
184
Fig. A-1, Continued.
185
Fig. A-1, Continued.
186
Fig. A-1, Continued.
187
Fig. A-1, Continued.
188
Fig. A-1, Continued.
189
Fig. A-1, Continued.
190
Fig. A-1, Continued.
191
Fig. A-1, Continued.
192
Fig. A-2: The complete log of the second Little Rock core (35-42.7 meters depth), including
lithology, shear zones and secondary fault locations and orientation (when available), sample
names and locations, chemical index of alteration (CIA) and median grain size (in microns) for
each sample.
193
Fig. A-2, Continued.
194
Fig. A-2, Continued.
195
Fig. A-2, Continued.
196
Appendix 2: XRF Data
Table B-1: XRF major and trace elements data for surface and core samples, as well as CIA
values, median grain size in microns, depth in meters and lithology.
Sample
name DO01 DO02 DO03 DO04 DO05 DO06 DO07 DO08 DO09 DO10
SiO
2
71.01 72.47 63.75 75.22 67.61 70.28 65.07 71.53 70.25 55.21
Al
2
O
3
14.50 12.87 17.80 12.67 14.70 14.52 15.42 14.54 14.52 17.96
Fe
2
O
3
0.32 0.40 3.62 0.32 3.15 0.29 3.03 0.52 1.90 7.12
CaO 1.34 1.04 3.65 1.00 2.09 1.31 2.40 0.47 2.02 4.55
MgO 0.05 0.06 1.29 0.05 1.14 0.05 1.19 0.05 0.46 2.95
K
2
O 9.02 4.48 1.46 1.35 2.71 7.31 3.15 7.81 4.21 1.78
Na
2
O 2.39 4.20 5.10 5.95 4.18 3.53 4.34 3.18 3.51 3.65
MnO 0.04 0.01 0.07 0.02 0.07 0.02 0.07 0.07 0.03 0.09
TiO
2
0.03 0.04 0.66 0.04 0.55 0.03 0.53 0.03 0.21 1.20
P
2
O
5
0.02 0.02 0.18 0.01 0.17 0.01 0.15 0.02 0.07 0.30
LOI 0.96 2.98 1.45 0.38 1.29 1.03 1.99 0.47 0.66 3.32
Cu 8.1 24.4 6.6 5.8 9.9 7.1 11.1 9.4 24.8
Zn 5.4 87.8 61.7 6.9 64.8 6.6 63.0 30.7 107.5
Rb 587.2 120.7 78.1 77.1 141.1 516.4 141.7 143.7 101.7
Sr 50.7 401.4 727.6 30.1 442.4 36.0 576.1 610.0 620.5
Y 20.5 19.8 17.1 44.7 19.6 20.2 15.7 9.4 25.6
Zr 15.1 145.5 206.3 71.8 209.6 14.2 194.7 123.1 216.1
Nb 13.6 10.6 18.0 68.8 22.1 19.0 17.3 9.5 15.2
Ba 117.3 966.0 480.2 12.1 921.8 48.3 1517.7 2040.2 774.2
Pb 36.6 12.9 17.6 23.0 20.2 31.7 18.9 20.7 14.0
U 5.7 11.3 10.0 21.1 10.5 5.7 8.1 6.4 22.0
A 0.47 0.49 0.52 0.49 0.53 0.47 0.51 0.50 0.51 0.56
CN 0.21 0.33 0.43 0.45 0.37 0.27 0.37 0.21 0.33 0.38
K 0.32 0.18 0.05 0.06 0.11 0.26 0.11 0.29 0.16 0.06
A 0.47 0.47 0.43 0.48 0.42 0.47 0.42 0.49 0.45 0.37
CNK 0.52 0.50 0.39 0.50 0.38 0.52 0.39 0.49 0.43 0.29
FM 0.02 0.02 0.19 0.02 0.20 0.02 0.19 0.03 0.11 0.34
median
grain 456 53 185 108
210 118 140 385 106
Depth (m) 0 0 0 0 0 0 0 0 0 0
Lithology G gouge GD G GD G GD G G D
D value 2.54 2.89 2.59 2.66
2.55 2.66 2.65 2.58 2.78
197
Table B-1, Continued.
Sample
name DO11 DO12 DO13 DO14 LR01 LR02 LR03 LR04 LR05 LR06
SiO
2
56.77 73.62 68.80 67.23 66.62 61.83 66.27 74.28 72.99 70.26
Al
2
O
3
18.17 13.78 13.51 15.34 15.20 16.57 15.72 14.06 14.52 13.95
Fe
2
O
3
6.23 1.44 1.68 3.21 3.39 3.59 3.61 0.39 0.42 2.08
CaO 3.80 1.33 3.26 2.58 3.30 3.34 2.97 0.53 1.03 2.34
MgO 2.91 0.35 0.60 1.20 1.16 1.60 1.27 0.06 0.06 0.65
K
2
O 1.71 4.97 2.10 2.77 2.51 3.33 2.70 5.08 4.36 4.63
Na
2
O 4.08 3.36 5.28 4.09 3.86 3.42 4.04 4.34 4.98 3.66
MnO 0.09 0.04 0.06 0.05 0.05 0.06 0.06 0.03 0.08 0.05
TiO
2
1.07 0.19 0.21 0.57 0.63 0.60 0.67 0.04 0.04 0.27
P
2
O
5
0.27 0.05 0.06 0.18 0.19 0.19 0.21 0.02 0.02 0.07
LOI 3.24 0.48 2.60 1.64 1.00 3.15 1.12 0.22 0.78 1.68
Cu 12.5 7.6 9.4 5.9 6.4 8.6 5.8 7.6 6.8 7.9
Zn 98.0 34.3 26.5 69.1 70.2 79.0 67.5 8.9 10.5 41.8
Rb 141.6 169.0 127.3 100.7 90.0 166.1 98.4 347.5 319.3 191.1
Sr 572.6 264.9 103.3 568.3 643.7 511.1 592.9 39.5 33.5 268.3
Y 24.5 12.8 49.0 16.2 14.7 18.5 15.9 24.3 38.6 16.2
Zr 202.2 118.0 94.4 220.7 247.9 224.7 237.1 16.6 27.1 109.6
Nb 12.0 9.5 78.8 14.8 12.9 13.3 14.8 42.1 44.2 12.5
Ba 609.8 729.6 128.4 944.8 1005.9 1131.9 1136.8 84.4 36.2 675.6
Pb 14.0 29.1 23.0 20.2 20.0 20.4 20.5 27.4 25.9 29.1
U 16.9 6.0 29.2 10.7 5.7 9.4 6.9 7.4 10.2 10.4
A 0.55 0.51 0.45 0.52 0.51 0.53 0.52 0.51 0.50 0.48
CN 0.39 0.29 0.48 0.37 0.40 0.36 0.38 0.29 0.34 0.35
K 0.06 0.20 0.08 0.10 0.09 0.11 0.10 0.20 0.16 0.17
A 0.38 0.46 0.40 0.42 0.41 0.41 0.41 0.50 0.49 0.42
CNK 0.31 0.44 0.49 0.38 0.39 0.37 0.38 0.48 0.49 0.45
FM 0.32 0.09 0.11 0.20 0.20 0.22 0.21 0.02 0.02 0.13
median
grain 134 308 175 129 244 132 145 113 169 194
Depth (m) 0 0 0 0 1.6 1.7 2.0 2.3 2.7 2.9
Lithology D G G GD GD GD GD G G G
D value 2.77 2.52 2.71 2.62 2.52 2.78 2.62 2.66 2.54 2.65
198
Table B-1, Continued.
Sample
name LR07 LR08 LR09 LR10 LR11 LR12 LR13 LR14 LR15 LR16
SiO
2
73.66 65.37 72.85 74.22 71.94 74.82 72.05 68.83 66.99 72.55
Al
2
O
3
14.43 15.01 14.43 13.66 13.31 13.61 14.46 15.40 15.59 13.94
Fe
2
O
3
0.37 5.17 0.47 1.12 1.22 0.42 1.85 2.87 3.31 1.59
CaO 0.36 1.80 0.37 1.02 0.92 0.62 1.62 3.05 2.72 1.69
MgO 0.05 1.85 0.04 0.23 0.24 0.06 0.52 1.00 1.16 0.40
K
2
O 6.03 3.54 6.72 5.05 5.02 4.16 4.58 2.79 3.19 4.69
Na
2
O 4.04 3.91 3.73 3.53 3.39 4.64 3.63 3.96 4.04 3.41
MnO 0.05 0.10 0.07 0.04 0.03 0.04 0.04 0.04 0.05 0.04
TiO
2
0.04 0.70 0.03 0.14 0.15 0.05 0.28 0.53 0.60 0.22
P
2
O
5
0.03 0.18 0.02 0.03 0.04 0.02 0.09 0.17 0.18 0.06
LOI 0.17 1.19 0.15 0.28 1.01 0.13 0.46 0.62 1.15 0.51
Cu 6.7 13.1 7.8 7.3 7.6 5.9 8.1 8.0 6.7 8.7
Zn 8.0 112.4 7.8 31.3 37.2 9.7 39.7 56.5 68.0 38.0
Rb 402.6 230.6 447.9 225.7 241.7 264.0 186.3 83.9 107.2 177.5
Sr 27.1 365.3 36.6 210.9 216.2 42.3 325.0 645.8 594.5 293.3
Y 42.4 28.9 36.5 17.1 18.6 43.6 18.6 12.9 16.1 15.0
Zr 29.9 167.5 29.8 88.3 108.1 55.6 140.9 199.7 224.7 129.6
Nb 43.5 36.3 35.9 14.1 15.6 48.7 13.8 10.6 13.1 9.6
Ba 19.2 697.8 67.5 525.0 617.3 54.6 709.4 1101.7 1194.2 843.4
Pb 30.0 27.2 32.4 35.4 36.1 31.7 32.0 19.5 22.8 32.7
U 14.1 29.6 11.1 7.3 8.6 11.6 7.6 4.6 5.1 4.9
A 0.51 0.53 0.51 0.51 0.51 0.51 0.51 0.51 0.52 0.51
CN 0.26 0.33 0.24 0.28 0.28 0.33 0.31 0.39 0.37 0.31
K 0.23 0.14 0.26 0.20 0.21 0.17 0.18 0.10 0.11 0.18
A 0.50 0.38 0.49 0.48 0.47 0.49 0.45 0.42 0.42 0.46
CNK 0.48 0.33 0.48 0.45 0.45 0.48 0.43 0.40 0.39 0.45
FM 0.02 0.29 0.02 0.07 0.08 0.02 0.12 0.17 0.19 0.10
median
grain 151 28 143 135 208 159 199 331 297 341
Depth (m) 3.6 3.9 4.1 4.4 4.6 5.0 5.4 5.7 6.2 6.5
Lithology G GD G G G G G GD GD G
D value 2.59 2.92 2.59 2.66 2.53 2.59 2.59 2.57 2.53 2.6
199
Table B-1, Continued.
Sample
name LR17 LR18 LR19 LR20 LR21 LR22 LR23 LR24 LR25 LR26
SiO
2
66.83 71.64 68.18 65.64 66.40 75.01 73.49 72.17 64.98 65.86
Al
2
O
3
16.17 13.40 15.36 15.46 15.87 13.73 14.14 14.00 15.85 15.67
Fe
2
O
3
3.46 0.51 3.27 3.27 3.44 0.60 0.35 0.32 3.46 3.44
CaO 3.08 2.49 2.81 3.30 3.12 0.82 0.46 0.57 3.81 2.95
MgO 1.17 0.07 1.13 1.23 1.16 0.12 0.06 0.06 1.24 1.24
K
2
O 3.02 5.29 2.97 2.91 3.21 2.74 4.88 5.75 3.38 3.24
Na
2
O 4.16 3.86 3.92 3.92 3.97 5.31 4.52 3.97 3.71 3.89
MnO 0.05 0.04 0.06 0.06 0.06 0.04 0.02 0.02 0.05 0.05
TiO
2
0.63 0.06 0.59 0.59 0.63 0.08 0.05 0.05 0.65 0.63
P
2
O
5
0.19 0.02 0.19 0.18 0.19 0.03 0.02 0.02 0.20 0.19
LOI 0.84 1.67 0.87 1.98 1.08 0.33 0.26 0.45 1.33 1.35
Cu 5.8 8.8 6.1 7.1 6.9 7.0 6.9 7.0 7.3 11.4
Zn 70.7 9.3 66.6 64.1 70.5 12.6 8.1 7.9 71.4 71.5
Rb 108.5 332.1 107.6 107.1 129.6 194.9 343.3 404.7 102.7 95.5
Sr 610.3 65.6 569.6 549.4 576.4 59.7 35.3 52.0 694.8 660.4
Y 15.4 23.2 16.9 17.4 16.5 27.9 30.2 24.6 16.0 14.6
Zr 229.7 20.9 220.1 219.5 223.8 34.3 23.5 18.2 223.7 229.3
Nb 13.4 46.9 15.0 13.4 16.3 75.6 55.5 53.6 13.3 13.4
Ba 1197.0 110.2 1112.7 1044.0 1223.6 71.0 47.4 67.6 1509.4 1331.4
Pb 21.2 33.2 21.7 22.6 22.8 22.9 27.9 25.7 18.0 16.9
U 7.0 8.1 5.5 7.2 6.9 13.6 15.2 12.8 4.5 4.7
A 0.51 0.45 0.52 0.50 0.51 0.51 0.51 0.50 0.50 0.51
CN 0.38 0.36 0.37 0.39 0.38 0.38 0.30 0.27 0.38 0.37
K 0.10 0.19 0.11 0.10 0.11 0.11 0.19 0.22 0.12 0.12
A 0.42 0.44 0.42 0.41 0.41 0.49 0.50 0.49 0.40 0.41
CNK 0.39 0.54 0.39 0.40 0.40 0.47 0.48 0.49 0.40 0.39
FM 0.19 0.03 0.19 0.19 0.19 0.04 0.02 0.02 0.19 0.20
median
grain 165 468 150 132 170 145 113 103 280 229
Depth (m) 7.0 7.4 7.7 8.0 8.3 8.6 9.1 9.3 9.6 9.9
Lithology GD G GD GD GD G G G GD GD
D value 2.67 2.53 2.7 2.68 2.61 2.57 2.68 2.74 2.55 2.63
200
Table B-1, Continued.
Sample
name LR27 LR28 LR29 LR30 LR31 LR32 LR33 LR34 LR35 LR36
SiO
2
65.54 72.58 63.50 68.72 72.20 62.89 53.18 62.99 65.36 64.55
Al
2
O
3
15.88 14.03 16.24 15.06 14.36 15.61 17.85 15.45 15.34 16.02
Fe
2
O
3
3.84 1.17 4.54 1.64 0.44 4.09 7.04 4.14 2.88 3.09
CaO 3.24 1.01 2.38 1.55 0.85 5.76 6.84 4.21 3.08 2.76
MgO 1.27 0.30 1.47 0.61 0.11 1.53 2.94 1.55 0.97 1.04
K
2
O 3.25 3.13 3.82 4.26 5.56 1.29 1.75 2.38 3.74 4.40
Na
2
O 3.86 4.99 4.18 4.74 4.30 4.00 4.04 4.15 3.62 3.67
MnO 0.06 0.05 0.08 0.04 0.05 0.07 0.11 0.07 0.05 0.05
TiO
2
0.70 0.18 0.85 0.22 0.05 0.69 1.04 0.73 0.51 0.58
P
2
O
5
0.21 0.05 0.26 0.06 0.03 0.20 0.29 0.23 0.16 0.18
LOI 0.83 0.43 1.14 1.27 0.62 2.53 3.87 2.51 1.55 1.05
Cu 9.0 8.8 7.6 11.8 8.0 14.5 69.6 6.5 8.9 9.4
Zn 77.8 21.1 82.6 28.7 9.2 64.6 100.6 80.2 62.3 66.7
Rb 104.0 216.4 177.2 256.1 385.4 68.6 90.2 114.5 107.4 162.2
Sr 647.7 122.4 550.8 95.6 76.3 654.5 674.4 564.4 669.1 696.3
Y 16.7 24.1 22.6 22.0 33.3 18.7 26.4 19.1 13.7 15.7
Zr 245.8 66.3 295.2 51.9 37.9 212.6 211.1 261.1 192.7 207.0
Nb 15.9 48.5 36.7 54.0 46.4 13.1 13.2 17.2 11.4 13.2
Ba 1334.3 248.3 1698.0 200.6 128.0 413.6 692.8 1053.9 1530.6 2370.4
Pb 17.8 20.7 19.4 22.1 22.7 14.7 13.1 16.2 18.5 19.1
U 5.1 9.3 18.8 27.6 13.5 8.9 5.3 5.8 3.9 5.5
A 0.51 0.51 0.52 0.50 0.50 0.52 0.54 0.49 0.50 0.51
CN 0.38 0.36 0.34 0.35 0.30 0.44 0.40 0.43 0.37 0.34
K 0.11 0.12 0.13 0.15 0.21 0.05 0.06 0.08 0.13 0.15
A 0.40 0.47 0.40 0.45 0.48 0.40 0.36 0.38 0.42 0.42
CNK 0.39 0.45 0.36 0.45 0.49 0.37 0.31 0.40 0.41 0.41
FM 0.21 0.08 0.24 0.11 0.03 0.23 0.33 0.23 0.17 0.17
median
grain 200 192 287 38 255 267 110 210 367 234
Depth (m) 10.2 10.5 10.7 11.1 11.4 11.9 12.3 12.8 13.8 14.8
Lithology GD G GD G G GD D GD GD GD
D value 2.56 2.55 2.62 2.85 2.62 2.56 2.75 2.6 2.56 2.56
201
Table B-1, Continued.
Sample
name LR37 LR38 LR39 LR40 LR41 LR42 LR43 LR44 LR45 LR46
SiO
2
64.95 70.92 63.92 48.36 73.75 64.76 63.54 71.56 64.66 62.19
Al
2
O
3
15.49 14.30 15.61 16.86 13.10 15.55 16.18 13.31 15.86 15.47
Fe
2
O
3
3.64 0.91 3.44 9.47 0.40 2.97 3.48 0.76 3.66 3.20
CaO 3.71 2.05 3.49 7.27 1.05 3.63 3.24 1.52 2.41 4.74
MgO 1.25 0.24 1.25 4.41 0.06 1.09 1.37 0.16 1.33 1.15
K
2
O 3.12 5.67 2.94 1.72 4.53 3.17 2.82 5.10 3.41 2.19
Na
2
O 3.74 3.71 3.73 3.23 4.23 3.63 4.00 3.81 4.09 2.94
MnO 0.06 0.19 0.05 0.14 0.01 0.05 0.06 0.02 0.06 0.06
TiO
2
0.67 0.11 0.61 1.38 0.04 0.53 0.62 0.10 0.69 0.59
P
2
O
5
0.21 0.04 0.19 0.17 0.02 0.17 0.19 0.02 0.21 0.18
LOI 1.56 1.56 2.43 5.26 0.61 2.28 2.67 1.06 1.81 5.07
Cu 6.5 30.1 6.5 24.7 6.9 8.3 9.4 7.6 8.8 27.5
Zn 77.0 19.3 70.2 106.9 7.7 61.2 73.3 24.6 73.2 77.0
Rb 105.6 320.4 108.7 86.3 282.0 109.8 99.4 287.3 129.9 110.5
Sr 646.4 128.4 598.8 656.6 46.2 675.9 682.6 127.2 628.3 589.1
Y 16.1 34.9 15.4 25.3 27.8 14.5 16.6 13.6 17.8 14.6
Zr 245.1 55.4 224.9 142.4 45.6 203.6 226.5 78.7 245.7 212.7
Nb 14.3 57.7 13.9 7.5 42.9 11.5 12.9 18.3 15.0 16.5
Ba 1114.8 234.4 1068.5 508.6 85.3 1183.1 1120.6 280.9 1614.1 867.7
Pb 17.6 24.0 18.1 12.3 26.5 17.0 16.4 33.9 18.0 16.8
U 10.3 10.9 7.4 5.8 16.9 8.5 6.2 12.0 6.7 3.8
A 0.50 0.47 0.51 0.57 0.49 0.50 0.52 0.48 0.53 0.56
CN 0.39 0.32 0.39 0.36 0.33 0.39 0.38 0.32 0.35 0.35
K 0.11 0.20 0.10 0.06 0.18 0.11 0.10 0.20 0.12 0.09
A 0.40 0.45 0.41 0.32 0.48 0.41 0.41 0.46 0.42 0.45
CNK 0.40 0.50 0.40 0.24 0.50 0.41 0.38 0.50 0.37 0.35
FM 0.20 0.06 0.20 0.44 0.02 0.17 0.20 0.05 0.21 0.20
median
grain 206 444 145 97 159 137 120 108 143 159
Depth (m) 15.7 15.8 16.5 16.9 17.1 17.3 18.4 18.8 19.1 19.9
Lithology GD G GD D G GD GD G GD GD
D value 2.69 2.57 2.65 2.76 2.68 2.71 2.69 2.7 2.62 2.65
202
Table B-1, Continued.
Sample
name LR47 LR48 LR49 LR50 LR51 LR52 LR53 LR54 LR55 LR56
SiO
2
69.19 61.78 58.49 64.00 71.05 65.05 71.28 65.25 65.42 64.31
Al
2
O
3
13.97 15.21 16.59 14.15 11.58 15.40 13.06 15.81 15.75 15.11
Fe
2
O
3
1.00 3.14 3.92 2.47 0.55 3.02 0.75 3.38 3.50 3.61
CaO 2.34 4.49 6.43 5.47 3.93 2.65 2.10 3.03 3.32 3.24
MgO 0.37 1.19 1.78 0.90 0.12 1.09 0.18 1.14 1.21 1.28
K
2
O 5.25 2.80 1.88 2.81 3.95 3.53 4.02 3.50 3.00 2.99
Na
2
O 3.32 2.90 1.85 2.16 3.70 3.66 4.30 3.83 3.99 3.75
MnO 0.02 0.05 0.08 0.05 0.09 0.06 0.05 0.05 0.05 0.05
TiO
2
0.14 0.57 0.54 0.36 0.06 0.55 0.10 0.60 0.61 0.66
P
2
O
5
0.04 0.18 0.15 0.11 0.02 0.17 0.04 0.19 0.18 0.20
LOI 2.36 5.17 6.72 5.59 3.07 3.11 1.44 1.52 1.28 1.98
Cu 16.4 15.2 16.0 22.5 28.7 23.1 11.0 12.3 8.8 8.6
Zn 27.2 68.3 66.6 53.0 12.8 69.0 19.6 73.4 75.6 80.9
Rb 260.1 105.7 88.9 131.9 214.3 167.0 247.4 115.7 94.0 103.7
Sr 202.7 528.6 592.0 457.7 118.9 388.3 100.9 659.3 676.9 639.4
Y 12.9 15.9 25.9 20.4 30.9 34.3 42.6 16.2 14.0 16.7
Zr 94.5 214.6 157.7 156.6 45.5 219.0 62.9 221.3 219.2 239.8
Nb 15.7 12.8 24.0 20.9 32.3 30.2 23.3 14.4 11.5 12.7
Ba 370.8 976.0 346.5 614.1 238.0 731.6 176.4 1393.8 1239.2 1173.2
Pb 30.5 15.6 15.7 23.4 22.8 22.0 26.1 18.3 16.7 16.7
U 10.7 3.3 5.3 6.0 7.9 8.4 12.4 6.9 3.2 5.4
A 0.48 0.55 0.67 0.58 0.41 0.52 0.46 0.51 0.51 0.50
CN 0.33 0.34 0.25 0.29 0.43 0.35 0.38 0.37 0.39 0.39
K 0.19 0.11 0.08 0.13 0.15 0.13 0.15 0.12 0.10 0.11
A 0.44 0.44 0.48 0.48 0.40 0.42 0.44 0.41 0.41 0.40
CNK 0.49 0.36 0.24 0.34 0.57 0.39 0.51 0.40 0.40 0.39
FM 0.07 0.20 0.28 0.18 0.03 0.18 0.05 0.19 0.19 0.21
median
grain 132 98 37 44 191 32 54 63 234 115
Depth (m) 20.2 20.8 21.5 21.7 22.5 23.0 23.7 24.7 25.5 26.3
Lithology G GD gouge GD G GD G GD GD GD
D value 2.71 2.73 2.94 2.94 2.79 2.88 2.83 2.76 2.5 2.64
203
Table B-1, Continued.
Sample
name LR57 LR58 LR59 LR60 LR61 LR62 LR63 LR64 LR65 LR66
SiO
2
74.16 65.15 64.55 65.01 70.54 71.72 63.57 64.23 63.44 60.28
Al
2
O
3
13.84 15.82 15.27 15.37 14.28 14.43 15.50 14.64 15.54 16.11
Fe
2
O
3
0.42 3.54 3.54 3.51 0.95 0.52 3.63 4.02 3.10 3.87
CaO 0.89 3.56 3.55 2.91 1.86 0.83 3.39 3.51 3.18 2.91
MgO 0.04 1.23 1.26 1.24 0.23 0.07 1.27 1.37 0.97 1.28
K
2
O 3.56 2.82 2.82 3.22 5.82 4.98 3.16 2.17 4.86 4.64
Na
2
O 5.08 4.01 3.89 3.83 3.06 4.62 3.58 3.54 2.89 3.44
MnO 0.07 0.05 0.06 0.05 0.09 0.04 0.06 0.06 0.05 0.06
TiO
2
0.04 0.64 0.64 0.64 0.15 0.05 0.66 0.71 0.58 0.71
P
2
O
5
0.02 0.20 0.20 0.20 0.05 0.02 0.21 0.22 0.17 0.22
LOI 0.33 1.45 2.65 1.54 1.38 0.48 3.44 3.81 3.38 2.78
Cu 9.4 7.4 13.2 9.7 31.0 10.2 10.6 13.7 16.5 9.1
Zn 7.8 73.6 79.5 76.6 24.4 9.6 75.9 85.3 61.1 78.0
Rb 224.5 90.8 93.9 97.8 352.4 311.5 100.7 84.5 138.1 176.8
Sr 34.7 705.3 617.0 666.7 238.9 40.5 499.2 445.2 549.6 536.0
Y 48.6 16.3 17.0 15.1 36.9 23.6 18.2 17.3 17.0 21.7
Zr 58.0 239.1 239.9 234.8 64.9 23.1 244.4 259.5 214.5 254.0
Nb 69.2 12.4 12.6 12.4 25.9 63.9 15.8 14.3 12.1 15.1
Ba 27.9 1245.7 1105.0 1417.0 569.4 75.5 1153.3 637.1 2439.9 1873.6
Pb 25.7 16.6 16.8 16.4 24.0 23.1 16.4 17.0 18.6 18.7
U 23.9 5.3 4.2 2.6 7.9 24.7 10.3 7.0 4.5 5.0
A 0.50 0.50 0.50 0.51 0.49 0.50 0.51 0.51 0.51 0.51
CN 0.36 0.40 0.40 0.37 0.29 0.31 0.38 0.41 0.31 0.33
K 0.14 0.10 0.10 0.12 0.22 0.19 0.11 0.08 0.17 0.16
A 0.49 0.40 0.40 0.41 0.47 0.49 0.40 0.39 0.42 0.41
CNK 0.49 0.40 0.40 0.39 0.48 0.49 0.39 0.38 0.40 0.39
FM 0.02 0.20 0.20 0.20 0.06 0.03 0.20 0.23 0.17 0.21
median
grain 75 165 71 201 189 180 150 187 200 164
Depth (m) 26.9 27.3 28.4 29.2 29.9 30.8 31.3 32.5 33.5 33.6
Lithology G GD GD GD G G GD GD GD GD
D value 2.74 2.58 2.77 2.55 2.59 2.64 2.74 2.65 2.63 2.71
204
Table B-1, Continued.
Sample
name LR67 LR68 LR69 LR70 LR71 LR72 LR73 LR74 LR75 LR76
SiO
2
70.05 65.82 69.83 55.27 72.70 64.22 65.87 66.79 57.89 58.39
Al
2
O
3
14.43 15.69 14.10 17.29 14.21 14.95 15.02 12.85 16.53 15.80
Fe
2
O
3
1.58 2.84 1.36 6.76 0.66 3.50 3.27 0.69 5.93 6.21
CaO 1.29 2.25 1.44 5.61 0.79 3.26 3.04 4.25 3.92 4.80
MgO 0.44 0.93 0.38 2.97 0.23 1.22 1.14 0.18 3.22 3.20
K
2
O 5.52 4.36 5.49 2.60 6.24 2.79 3.26 4.69 1.64 1.46
Na
2
O 3.59 3.98 3.22 3.44 3.62 3.50 3.26 3.95 3.47 3.86
MnO 0.03 0.05 0.03 0.12 0.01 0.05 0.05 0.16 0.11 0.10
TiO
2
0.23 0.50 0.16 0.98 0.09 0.62 0.60 0.07 0.95 0.92
P
2
O
5
0.07 0.15 0.04 0.26 0.02 0.19 0.19 0.02 0.24 0.27
LOI 1.19 1.84 1.11 2.72 0.39 3.45 3.40 3.34 5.61 4.19
Cu 8.7 6.2 35.5 49.5 10.3 9.9 11.2 14.9 30.3 33.1
Zn 34.8 64.9 22.3 106.5 11.5 78.8 73.5 14.0 98.1 86.7
Rb 210.4 176.6 227.2 148.9 301.0 96.3 121.1 373.1 89.1 86.8
Sr 291.4 452.2 289.1 564.9 134.1 496.3 492.2 78.8 513.7 558.0
Y 19.6 18.4 14.1 23.4 19.5 18.4 17.8 39.9 23.9 24.5
Zr 117.4 191.4 95.1 190.3 29.6 238.3 214.7 33.1 195.2 206.8
Nb 10.0 12.7 15.5 12.9 19.4 13.8 13.9 56.9 14.3 13.8
Ba 737.8 1228.8 549.2 763.1 303.8 1131.7 1160.4 174.6 606.6 526.5
Pb 26.6 19.7 27.4 14.1 31.1 18.8 19.7 29.8 13.8 17.2
U 4.9 5.3 11.4 8.5 11.0 3.7 4.5 12.2 3.7 3.8
A 0.51 0.51 0.51 0.55 0.50 0.51 0.52 0.42 0.56 0.53
CN 0.28 0.33 0.28 0.36 0.26 0.38 0.36 0.42 0.38 0.42
K 0.21 0.15 0.21 0.09 0.24 0.10 0.12 0.16 0.06 0.05
A 0.46 0.43 0.46 0.36 0.48 0.41 0.42 0.40 0.36 0.34
CNK 0.44 0.41 0.45 0.30 0.47 0.39 0.39 0.56 0.29 0.31
FM 0.10 0.16 0.09 0.34 0.05 0.21 0.20 0.04 0.35 0.35
median
grain 67 37 179 74 53 73 79 60 23 166
Depth (m) 34.1 34.2 34.8 35.0 35.1 35.8 36.6 37.2 38.3 40.2
Lithology G GD G D G GD GD G D D
D value 2.79 2.85 2.63 2.78 2.82 2.79 2.77 2.91 2.99 2.77
205
Table B-1, Continued.
Sample
name LR77 LR78 LR79 LR80 LR81 LR82 LR83 LR84 LR85
LRG1
(gouge1)
SiO
2
53.18 54.31 55.27 51.71 62.87 51.31 55.02 49.96 56.84 66.32
Al
2
O
3
15.87 16.38 16.98 17.38 14.62 17.70 17.33 16.85 15.60 15.34
Fe
2
O
3
4.64 5.66 7.01 7.98 1.70 6.28 6.97 7.12 6.77 2.96
CaO 9.68 6.92 5.10 5.60 5.83 7.70 6.41 8.58 5.55 2.74
MgO 2.21 2.63 3.62 4.17 0.70 3.22 3.36 3.41 3.53 1.11
K
2
O 2.06 2.22 2.50 1.89 5.09 1.51 1.95 1.56 1.78 3.53
Na
2
O 2.67 2.67 3.08 3.38 2.65 3.09 3.57 3.38 3.53 3.56
MnO 0.09 0.10 0.13 0.13 0.17 0.11 0.12 0.15 0.10 0.05
TiO
2
0.63 0.81 1.04 1.15 0.19 0.99 1.03 1.13 1.11 0.53
P
2
O
5
0.17 0.22 0.23 0.28 0.05 0.23 0.27 0.30 0.21 0.17
LOI 7.46 7.15 4.44 5.81 5.33 7.40 3.48 6.69 3.89 2.07
Cu 24.5 43.7 9.7 14.1 11.0 11.9 11.1 30.7 13.1 11.7
Zn 79.2 94.0 104.0 117.7 27.5 96.9 96.0 105.9 90.9 66.8
Rb 117.4 107.6 131.1 108.0 256.0 86.6 87.4 74.1 77.3 141.8
Sr 498.2 656.4 902.2 564.4 326.3 539.8 608.0 662.2 528.5 566.3
Y 20.5 20.9 22.9 25.6 41.1 24.7 20.1 23.2 21.4 15.9
Zr 146.4 182.5 177.3 207.2 55.4 149.0 112.6 197.9 163.9 203.2
Nb 19.4 13.3 15.9 14.7 38.2 15.0 9.3 14.4 12.7 13.2
Ba 452.6 763.6 881.6 663.9 344.1 424.6 680.5 603.0 577.4 1173.9
Pb 19.3 18.6 15.7 16.2 24.4 13.6 17.4 16.2 15.5 30.4
U 5.8 7.5 3.6 3.0 4.2 3.4 2.9 3.7 3.7 12.5
A 0.59 0.59 0.57 0.57 0.51 0.60 0.56 0.57 0.54 0.52
CN 0.33 0.32 0.34 0.36 0.30 0.34 0.38 0.37 0.40 0.35
K 0.08 0.09 0.09 0.07 0.19 0.06 0.07 0.06 0.07 0.13
A 0.41 0.40 0.35 0.34 0.45 0.39 0.36 0.36 0.33 0.42
CNK 0.29 0.27 0.27 0.26 0.43 0.26 0.29 0.27 0.29 0.39
FM 0.30 0.33 0.38 0.40 0.12 0.35 0.36 0.37 0.38 0.18
median
grain 57 83 119 29 671 160
161 171 32
Depth (m) 35.9 36.0 37.0 38.0 38.7 39.1 40.1 41.5 42.6 24.5
Lithology GD D D D G D D D D gouge
D value 3.09 2.93 2.78 2.94 2.7 2.73
2.79 2.62 2.88
206
Table B-1, Continued.
Sample
name
LRG2
(gouge2)
LRG3
(gouge3)
LRG4
(gouge4)
TENS01
(outcrop)
AS-05
(outcrop)
SiO
2
59.12 53.55 57.10 71.61 67.14
Al
2
O
3
15.80 17.87 17.07 13.92 15.76
Fe
2
O
3
4.64 6.99 5.52 0.37 3.37
CaO 5.04 4.94 3.77 3.99 3.47
MgO 2.37 3.49 2.95 0.04 1.22
K
2
O 2.56 2.07 2.42 4.23 2.60
Na
2
O 3.11 3.38 2.52 3.99 4.09
MnO 0.09 0.12 0.10 0.02 0.06
TiO
2
0.74 1.01 0.89 0.03 0.62
P
2
O
5
0.20 0.27 0.22 0.01 0.19
LOI 5.19 5.33 5.76 2.30 0.64
Cu 18.5 43.0 24.1 8.4 9.3
Zn 80.6 124.5 96.8 10.3 69.3
Rb 150.4 111.4 250.2 139.0 90.6
Sr 580.7 522.2 538.6 167.8 663.4
Y 20.2 17.3 22.6 3.0 13.6
Zr 191.4 196.2 213.4 30.4 222.3
Nb 12.7 7.5 12.8 1.0 12.2
Ba 709.9 540.2 660.2 61.8 983.1
Pb 29.6 23.9 29.2 44.9 23.7
U 18.6 5.9 85.5 6.4 3.0
A 0.55 0.57 0.61 0.44 0.51
CN 0.36 0.36 0.30 0.42 0.40
K 0.10 0.07 0.09 0.14 0.09
A 0.39 0.36 0.40 0.43 0.41
CNK 0.32 0.27 0.26 0.55 0.40
FM 0.29 0.36 0.34 0.02 0.19
median
grain 3
Depth (m) 26.5 36.8 35.2 0 0
Lithology gouge gouge gouge G GD
D value 2.99
Abstract (if available)
Abstract
This thesis presents a compilation of results from studies of active fault zone geometry, structural properties, and macro- and micro-scale damage fabrics. Multi-scale observations using a wide range of techniques were made along the transform plate boundary of the Pacific and the North American plates. A new method for quantifying fault trace heterogeneity using Geographic Information System was outlined and used on the database of active faults in California. Several parameters were defined for quantifying fault trace heterogeneity and the range or dispersion in the data. The cumulative slip and slip rate proved effective measures of fault zone maturity.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Multi-scale imaging of major fault zones in Southern California
PDF
Symmetry properties, pulverized rocks and damage architecture as signatures of earthquake ruptures
PDF
Heterogeneity of earthquake stress drops, focal mechanisms and active fault zones
PDF
Seismicity distribution near strike-slip faults in California
PDF
Determining fault zone structure and examining earthquake early warning signals using large datasets of seismograms
PDF
Multi-scale imaging and monitoring of crustal and fault zone structures in southern California
PDF
Elements of seismic structures near major faults from the surface to the Moho
PDF
Towards an understanding of fault-system mechanics: from single earthquakes on isolated faults to millenial-scale collective plate-boundary fault-system behavior
PDF
Multi-scale imaging of the fault zone velocity structure: double-difference tomography, inversion of fault zone headwaves, and fault zone sensitivity kernels
PDF
Implications of new fault slip rates and paleoseismologic data for constancy of seismic strain release and seismic clustering in the eastern California shear zone
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
High-resolution imaging and monitoring of fault zones and shallow structures: case studies in southern California and on Mars
PDF
Microseismicity, fault structure, & the seismic cycle: insights from laboratory stick-slip experiments
PDF
Detection and modeling of slow slip events as creep instabilities beneath major fault zones
PDF
Spatiotemporal variations of stress field in the San Jacinto Fault Zone and South Central Transverse Ranges
PDF
Observations of temporal and spatial patterns of strain accommodation and earthquake occurrence along strike-slip faults of New Zealand and southern California, USA
PDF
Mapping the Moho in southern California using P receiver functions
PDF
Seismic velocity structure and spatiotemporal patterns of microseismicity in active fault zones of Southern California
Asset Metadata
Creator
Wechsler, Neta
(author)
Core Title
Multi-scale damage signatures across major strike-slip faults
School
College of Letters, Arts and Sciences
Degree
Doctor of Philosophy
Degree Program
Geological Sciences
Publication Date
06/14/2010
Defense Date
05/03/2010
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
drainage density,earthquake dynamics,earthquake ruptures,fault zone structure,fractures and faults,LiDAR,OAI-PMH Harvest,rock damage,self-organization
Place Name
California
(states),
faults: San Andreas
(geographic subject),
faults: San Jacinto
(geographic subject)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Ben-Zion, Yehuda (
committee chair
), Rockwell, Thomas K. (
committee member
), Sammis, Charles G. (
committee member
), Wilson, John P. (
committee member
)
Creator Email
neta.wechsler@sdsu.edu,wechsler@usc.edu
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-m3127
Unique identifier
UC1462870
Identifier
etd-Wechsler-3749 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-357482 (legacy record id),usctheses-m3127 (legacy record id)
Legacy Identifier
etd-Wechsler-3749.pdf
Dmrecord
357482
Document Type
Dissertation
Rights
Wechsler, Neta
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Repository Name
Libraries, University of Southern California
Repository Location
Los Angeles, California
Repository Email
cisadmin@lib.usc.edu
Tags
drainage density
earthquake dynamics
earthquake ruptures
fault zone structure
fractures and faults
LiDAR
rock damage
self-organization