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
/
Observations of temporal and spatial patterns of strain accommodation and earthquake occurrence along strike-slip faults of New Zealand and southern California, USA
(USC Thesis Other)
Observations of temporal and spatial patterns of strain accommodation and earthquake occurrence along strike-slip faults of New Zealand and southern California, USA
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
i
OBSERVATIONS OF TEMPORAL AND SPATIAL PATTERNS OF STRAIN
ACCOMMODATION AND EARTHQUAKE OCCURRENCE ALONG STRIKE-SLIP
FAULTS OF NEW ZEALAND AND SOUTHERN CALIFORNIA, USA
By
Alexandra Elise Hatem
______________________________________________________________________________
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 2019
ii
ACKNOWLEDGEMENTS
Despite only my name on the title page, the journey to get to this point would not be
possible without many people and animals.
First, I wish to dedicate this dissertation to my late grandmother, Dr. Eleanor Forsley
Shalhoup. As a widowed, single mother, she earned her doctorate in 1977 in Public Health from
Boston University, and went on to become the Dean of Health Professions at the University of
Massachusetts, Lowell. Her mantra was “persistence pays off.” It certainly did for her, and I
have tried to channel her energy in my work. I wouldn’t be in this spot without you, Sitto.
I want to acknowledge and thank my mother and sister for their unflinching love, encouragement
and support throughout the last five years, and the years before that too. I thank my dad for his
encouragement for me to play in the dirt (of the garden and the trails) from a young age. My
extended family has always been there to help raise me and encourage my growth all along the
way. It truly takes a village.
Thank you Claire for your unbelievable amount of love, for allowing me to rant and rave
about science to you, for supporting me when I was going crazy with science and otherwise, and
for reminding me that I could always start fresh the next day. Enormous thank you to our
amazing four-legged gals, Holly and La Fiamma, for all of the face licks.
Thank you to my adviser, James Dolan. The science we have worked on together has
only grown stronger with time, which has been an iterative process that I am quite grateful for. I
am thankful for his mentorship, editorial prowess, and willingness to make himself available to
me. I am better for having the opportunity to work with him as TA in Earthquakes and Structural
Geology & Tectonics, as I have learned a lot from him as an animated instructor (and learned
quite a few anecdotes over the years to boot). I’d also like to thank my committee members, Josh
iii
West of Earth Sciences and Steve Nutt of Viterbi Engineering, for their participation in my
defense and for guiding the revisions of this dissertation.
The work of the studies presented in this thesis have benefitted greatly from the
collaboration with many great scientists. I first need to thank my office mate and trusted field
partner, Rob Zinke. I couldn’t have done this work without him and our memorable many days
sitting in holes together. In particular, working with and learning from Russ Van Dissen of GNS
Science has been a total treat. I greatly enjoyed and appreciated the hospitality of his family on
our trips to New Zealand. Thank you to Rob Langridge, also of GNS Science, for being a good
partner in the field and for always playing devil’s advocate. Thank you to the Russ, Rob, Pilar
Villamor, Nicola Litchfield and Zoe Juniper of GNS, as well as Julie Rowland of University of
Auckland, for welcoming me and mentoring me when I had the opportunity to go to New
Zealand to map the 2016 Mw 7.8 Kaikōura earthquake. That was a trip of a lifetime, and I
learned so much from both that rupture and these people. Thank you to Ed Rhodes, Chris
McGuire and Nathan Brown for the numerous field explanations of luminescence dating and for
help preparing samples in the lab. Thank you to John Southon and Chanda Bertrand of UC Irvine
for their radiocarbon prep and analyses powers.
Thank you to the USC Earth Sciences office staff who keep us all afloat—Cindy, Karen,
Vadui, John McRaney, John Yu, Miguel, Barbara. You have made my time in this department
exceedingly painless from a bureaucratic point of view. Thank you to Frank Corsetti, Will
Berelson and John Platt for your continued support and dedication to help me accommodate my
needs within the department.
Two women have been instrumental mentors in my life and have guided me into who I
am becoming as a scientist. Michele Cooke, my MS adviser, has never stopped helping me out,
iv
even after I graduated from UMass. She has remained invested in my growth as a scientist, and
has encouraged me to punch above my weight time and time again. In 2013, Michele wanted me
to get out of the lab and into the field to see the faults I was modelling. She set me up on a
paleoseismic study of the San Andreas with her colleague Kate Scharer of the USGS. Kate has
been a huge factor in my success in Los Angeles and at USC. She has been a great friend and
mountain biking partner to me over the past five years, and has always been game to talk science
whenever the moment strikes, on the trails or otherwise. Michele and Kate—thank you for being
great examples for me to look up to.
Over my time in Los Angeles, I have made some incredible friends in and out of the
department. Your joy, willing ears, open arms, coffee breaks, dog parties and supreme levels of
support have made my life full and possible. Thank you to Maria for helping me find and making
me walk my path every day, and for believing in me every step of the way. Thank you to Al for
being the oldest person I know and always putting things into perspective, and for your grace and
guidance. Thank you to my old Wellesley crew who still comes through strong.
Finally, I need to acknowledge some creature comforts that have had serious impact on
me during the last five years. First, the Boston Red Sox. This team has given me both heart ache
and pure elation. They have come through when I needed them most. The joy I get from
honoring my Masshole heritage while rooting for this team is unparalleled. Witnessing the Game
5 win over the Dodgers in Reserve Section 16 at Dodger Stadium on October 28, 2018 was one
of the greatest treasures of my life. Finding happiness and relief in that team propelled my thesis
forward when I was in dark time. Second, the band Sleater-Kinney. Thank you to Carrie
Brownstein, Corin Tucker and Janet Weiss (and some drummers before Janet) for bringing me
familiarity, structure and power to nearly every day for me. Their music somehow always
v
stabilizes me, even just a little bit, no matter where my head is at. A fitting way to end this
section and get on with the big shebang of the thesis, I’ll quote from their song “Banned From
the End of the World” off of The Hot Rock (1999)— “throw me out when the party’s over.”
This dissertation includes parts of the following manuscripts:
Chapter 2:
Hatem, A.E., Dolan, J.F. (2018). A model for the initiation, evolution, and controls on seismic
behavior of the Garlock fault, California. Geochemistry, Geophysics, Geosystems, 19.
Chapter 3:
Hatem, A.E., Dolan, J.F., Zinke, R.W., Van Dissen, R.J., McGuire, C.P., Rhodes, E.J. (in review.)
A 2000-year paleoearthquake record along the Conway segment of the Hope fault: Implications
for patterns of earthquake occurrence in northern South Island and southern North Island, New
Zealand. Bulletin of the Seismological Society of America.
Chapter 4:
Hatem, A.E., Dolan, J.F., Langridge, R.M., Zinke, R.W., McGuire, C.P., Rhodes, E.J., Brown, N.,
Van Dissen, R.J. (in prep.) Variability observed in Holocene and latest Pleistocene incremental
slip rates along the Hope fault (Conway segment) at Hossack Station, Marlborough Fault System,
South Island, New Zealand. Geochemistry, Geophysics, Geosystems
vi
TABLE OF CONTENTS
ACKNOWLEDGEMENTS ......................................................................................................... ii
ABSTRACT ................................................................................................................................... x
CHAPTER 1: Introduction .......................................................................................................... 1
1.1 Elastic Rebound Theory: A classic solution to a complex problem ............................... 1
1.2 Problem identification: Elastic Rebound Theory is not perfect ..................................... 2
1.3 Problem solving: chasing earthquakes through time ...................................................... 4
1.4 Future problems: Earthquake cycles at depth ................................................................. 6
1.5 The current situation: what data to use moving forward ............................................... 7
CHAPTER 2: A model for the initiation, evolution, and controls on seismic behavior of the
Garlock fault, California .............................................................................................................. 8
2.1 Abstract ................................................................................................................................ 8
2.2 Introduction ......................................................................................................................... 8
2.3 The Garlock Fault ............................................................................................................... 9
2.4 Models of Garlock fault behavior .................................................................................... 11
2.5 A model for Garlock fault evolution ............................................................................... 12
2.6 Controls on current Garlock fault slip rates .................................................................. 16
2.6.1 Rotation of the Garlock fault due to ECSZ dextral shear ............................................ 16
2.6.2 Intracontinental transform contribution from Basin and Range extension .................. 19
2.6.3 Conjugate contribution from San Andreas (SAF) ....................................................... 20
2.7 Implications for fault interactions, earthquake recurrence, and plate boundary
evolution in southern California ............................................................................................ 21
2.8 Conclusions ........................................................................................................................ 25
CHAPTER 3: A 2000-year paleoearthquake record along the Conway segment of the Hope
fault: Implications for patterns of earthquake occurrence in northern South Island and
southern North Island, New Zealand ........................................................................................ 29
3.1 Abstract .............................................................................................................................. 29
3.2 Introduction ....................................................................................................................... 30
3.3 The Green Burn East and Green Burn West study sites .............................................. 33
3.4 Trench Results ................................................................................................................... 35
3.4.1 Green Burn East (GBE) trench observations ............................................................... 35
3.4.2 Evidence for paleo-surface ruptures ............................................................................ 39
3.4.3 Green Burn West (GBW) trench observations: Evidence for inferred landslides ....... 43
vii
3.4.4 Coseismic origin of colluvial wedges and landslides observed at GBW/GBE:
Observations of the Green Burn Reach following the 2016 Mw=7.8 Kaikōura earthquake . 46
3.5 Chronology of paleoseismic events observed at GBE and GBW .................................. 48
3.5.1 Age models and event boundary conditions ................................................................ 48
3.5.2 Green Burn East Paleo-Surface Rupture Ages ............................................................ 50
3.5.3 Age control for GBE northern marsh strata ................................................................. 53
3.5.4 Ages of GBW landslides .............................................................................................. 54
3.5.5 Combined Age Model for GBW and GBE sites .......................................................... 55
3.6 Discussion........................................................................................................................... 57
3.6.1 Plate Boundary System-Level Rupture Behavior ........................................................ 58
3.7 Conclusions ........................................................................................................................ 69
3.8 Figure Captions ................................................................................................................. 70
CHAPTER 4: Holocene to latest Pleistocene incremental slip-rates from the Hope fault
(Conway segment) at Hossack Station, Marlborough fault zone, South Island, New Zealand
....................................................................................................................................................... 74
4.1 Abstract .............................................................................................................................. 74
4.2 Introduction .................................................................................................................. 75
4.3 The Marlborough Fault System and Hope fault ....................................................... 75
4.4 Hossack study site ........................................................................................................ 77
4.5 Methods ......................................................................................................................... 78
4.5.1 Offset mapping........................................................................................................ 78
4.5.2 Excavation of paleo-channels ................................................................................. 78
4.5.3 Age determination ................................................................................................... 79
4.6 Description of offset features in landscape and trenches .............................................. 80
4.6.1 Offset A ........................................................................................................................ 80
4.6.2 Offset B ........................................................................................................................ 81
4.6.3 Offset C ........................................................................................................................ 83
4.6.4 Offset D ........................................................................................................................ 85
4.6.5 Offset E ........................................................................................................................ 86
4.6.6 Offset H ........................................................................................................................ 87
4.7 Age control ......................................................................................................................... 87
4.7.1 Age of S1 gravels (Offset A) ....................................................................................... 88
4.7.2 Age of Channel 1 initiation (Offset B) ........................................................................ 88
viii
4.7.3 Age of Channel 1 abandonment/Channel 2 initiation (Offset C) ................................ 89
4.7.4 Age of Channel 2 abandonment/Channel 3 initiation (Offset D) ................................ 89
4.7.5 Age of Channel 3 partial abandonment (Offset E) ...................................................... 90
4.8 Incremental slip rates ....................................................................................................... 90
4.9 Discussion........................................................................................................................... 90
4.9 Conclusions ........................................................................................................................ 93
4.10 Figure Captions ............................................................................................................... 94
CHAPTER 5: Holocene to latest Pleistocene incremental slip rates at the Sawyer’s Creek
site along the eastern Conway segment of the Hope fault, South Island, New Zealand:
Implications for along segment variations in strain accommodation .................................... 98
5.1 Abstract .............................................................................................................................. 98
5.2 Introduction ....................................................................................................................... 99
5.3 Sawyers Creek Site ......................................................................................................... 101
5.4 Offsets Measurements .................................................................................................... 103
5.5 Prior age results .............................................................................................................. 103
5.6 Slip rate calculations ....................................................................................................... 104
5.7 Discussion......................................................................................................................... 105
5.8 Conclusions ...................................................................................................................... 108
5.9 Figure Captions ............................................................................................................... 108
CHAPTER 6: Towards quantification of the three-dimensional deformation field
associated with the 1952 Mw 7.3 Kern County earthquake along the White Wolf fault,
southern California, USA ......................................................................................................... 111
6.1 Abstract ............................................................................................................................ 111
6.2 The 1952 Mw 7.3 Kern County earthquake .................................................................. 111
6.3 Off-fault deformation of coseismic ruptures ................................................................ 113
6.4 Existing correlation methodologies ............................................................................... 114
6.5 Pre- and post-event imagery .......................................................................................... 116
6.6 Point cloud generation and processing ......................................................................... 118
6.7 Preliminary results.......................................................................................................... 121
6.8 Future directions ............................................................................................................. 123
6.9 Conclusions ...................................................................................................................... 124
6.10 Figure Captions ............................................................................................................. 124
CHAPTER 7: Conclusions ....................................................................................................... 126
ix
REFERENCES .......................................................................................................................... 128
Chapter 2 Figures ..................................................................................................................... 156
Chapter 3 Figures ..................................................................................................................... 163
Chapter 4 Figures ..................................................................................................................... 174
Chapter 5 Figures ..................................................................................................................... 191
Chapter 6 Figures ..................................................................................................................... 197
APPENDICES ........................................................................................................................... 202
Appendix A: A 2000-year paleoearthquake record along the Conway segment of the
Hope fault: Implications for patterns of earthquake occurrence in northern South Island
and southern North Island, New Zealand........................................................................... 202
Appendix B: Towards quantification of the three-dimensional deformation field
associated with the 1952 Mw 7.3 Kern County earthquake along the White Wolf fault,
southern California, USA ..................................................................................................... 212
x
ABSTRACT
The spatial and temporal patterns of earthquake occurrence, despite theoretical
predictions, remains enigmatic. Quantification of these patterns has implications for fault
evolution processes, particularly variable fault strength at depth, as well as for probabilistic
seismic hazard assessments. Many methods are available to constrain the variability in
earthquake behavior over many time and length scales. Regional tectonic reconstructions can
illuminate loading patterns within complex fault systems, such as the understanding the role of
the sinistral Garlock fault in the largely dextral southern California Pacific-North American plate
boundary. Results from this study uncover direct contributions from the Eastern California Shear
Zone north and south of the Garlock, as well as from the southern San Andreas. The results of
this study (Chapter 2) imply that the Garlock fault aids in stabilization of the current plate
boundary configuration of southern California, and plays a major role in modulating strain
accommodation between various faults in the region. Smaller scale, site specific studies to obtain
primary data are needed to make compilations to address fault system behavior through time.
Paleoseismic studies show the temporal complexity of earthquake occurrence at a given site, and
can be synthesized with other paleoseismic records within the same fault system to assess the
presence of potential earthquake sequences versus isolated events, such as using the temporally
irregular paleoearthquake record of the Conway segment of the Hope fault to better understand
the Pacific-Australian plate boundary of the South Island of New Zealand (Chapter 3). In
addition to paleoseismic records, incremental slip rate records document the covariance of
incremental, discrete displacement along a fault coupled with the elapsed time between the ages
of offset features. Incremental slip rates are presented at two sites along the Conway segment of
the Hope fault, in the west (Chapter 4) and in the east (Chapter 5), showing variations in fault
xi
slip rate along strike of up to 200-300% over centennial and millennial time scales. Finally, on
the shortest time scale, coseismic displacements can reveal the amount of off-fault deformation
compared to on-fault, discrete slip. Such measurements are achieved using optical image
correlations methods, which I have improved upon by developing a novel workflow to obtain the
three-dimensional deformation field of the 1952 Mw 7.3 Kern County earthquake along the
White Wolf fault, southern California (Chapter 6). Together, these studies add to a growing,
global dataset describing how faults work over wide spatial and temporal skills. This thesis
shows that, no matter the temporal or spatial scale of observation, earthquake behavior is
complex. The results of these studies beg the question of why is earthquake occurrence in space
and time irregular? Answers, or hints of answers, likely like below kilometers below the surface,
far below the study area of the works presented in this thesis.
1
CHAPTER 1: Introduction
Earthquakes are amazing, highly energetic bursts of strain accommodation along faults,
which are planes of weakness in the Earth’s crust and upper mantle. These events shape our
planet as we know it, building mountains, opening basins, moving tectonic plates around the
globe, and causing massive disruptions to the built environment. While earthquakes typically
occur along plate boundaries, at the edges of the rigid interiors of plates, little is known about the
earthquake behavior of these faults through space and time.
1.1 Elastic Rebound Theory: A classic solution to a complex problem
According to classical Elastic Rebound Theory [Reid, 1911], an elastic body containing a
fault can accommodate applied stress without breaking, so long as the yield strength is not
exceeded. Deformation is therefore accommodated as warping in the far field (i.e., creep), with
the fault remaining locked. Once the yield strength is exceeded, the material breaks along the
fault (i.e., stick-slip), leaving a discrete offset along the fault that now matches the offset accrued
in the far field over the interseismic time. This insightful and simplistic model, based upon
Hooke’s Law, predict that earthquakes are characteristic, and their recurrence is periodic,
essentially creating the same, repeating earthquake over a constant recurrence interval. Elastic
Rebound Theory also implies that earthquakes must have a complete stress drop, as stress
applied to the system is linearly proportional to the displacement (i.e., strain) along the fault,
which includes no permanent off-fault deformation.
2
1.2 Problem identification: Elastic Rebound Theory is not perfect
While Elastic Rebound Theory informed our early understanding of the earthquake cycle,
there is much more nuance to earthquake behavior than what was proposed by Elastic Rebound
Theory. For instance, not all strain accommodation occurs on the fault plane proper during an
earthquake. Some of this strain is accommodated off of the main fault strand, either as ground
cracking, slip on subsidiary strands, folding, brecciation, dilation and mineral twinning. This
deformation can span ones to hundreds of meters away from the main fault trace. Many
observations of off-fault deformation have been made within the longer-term, rock record (i.e.,
finite) [e.g., Little and Jones, 1998; Shelef and Oskin, 2010; Titus et al., 2011], as well as the
shorter-term, individual earthquake record (i.e., infinitesimal) [Oskin et al., 2012; Zinke et al.,
2014, 2016, 2019; Gold et al., 2015; Milliner et al., 2015, 2016a; Scott et al., 2018]. Because off-
fault deformation is typically distributed across the broad fault zone, it can be difficult to
measure without optical image correlation methodologies. Off-fault deformation varies with fault
maturity and the ability of a fault to accommodate plate motion, meaning that younger parts of
faults and/or parts of faults that have much structural complexity (i.e., restraining bends) [Dolan
and Haravitch, 2014; Hatem et al., 2015]. However, even structurally mature faults have been
shown to have up to 20% of the applied plate displacement accommodated as off-fault
deformation [Hatem et al., 2017]. Furthermore, slip distributions have high-resolution (sub-meter
scale) variability in slip accommodated on and off the main fault, indicating that a single
measurement of slip along a rupture may not be representative of the total slip accommodated in
that earthquake, and instead may depend on the thickness of sediment (depth to bedrock) [e.g.,
Milliner et al., 2016]. Observations of permanent strain accommodation off of the main fault
make clear that any amount of strain we may measure as a discrete offset along the main fault
3
trace is a mere underestimate of the total strain accommodated by that patch of fault within the
fault zone.
Another difference between Elastic Rebound Theory and observed earthquake behavior is
the lack of temporal predictability. Multiple types of earthquake recurrence patterns have been
documented along strike-slip faults. Earthquake recurrence models are classified based on the
coefficient of variation of their recurrence intervals [Kagan and Jackson, 1991]. For example,
the southern Alpine fault has a periodic recurrence interval [Berryman et al., 2012a], the
southern San Andreas has a quasi-periodic recurrence interval [Scharer et al., 2007], and the
southern Dead Sea fault has a clustered recurrence interval [Marco et al., 1996]. Assuming these
paleoseismic records are complete, meaning they do not omit or add events, these major, plate
boundary, strike-slip faults are all accommodating large amounts of applied plate motion in very
different ways. These observations underscore that either strain accommodation is not occurring
at a constant rate or magnitude, implying that there must be some mechanism not yet directly
observed that stores energy within a fault zone throughout the earthquake cycle. Or, conversely,
the accumulation of elastic strain energy may not be constant through time as plate rates may
accelerate or decelerate following enormous subduction zone earthquakes [i.e., Anderson, 1975].
These two deviations from Elastic Rebound Theory obfuscate the simplicity of the model
and color just some of the complexity of understanding the earthquake cycle. The earthquake
cycle evidently relies on more parameters than just loading rate and time since the last
earthquake. Other parameters and processes must be considered to explain our observations, such
as: structural maturity of the fault, lithology and other structural contrasts juxtaposed across the
fault (i.e., bi-material interfaces), changes in pore fluid pressure, changes in ambient stress and
rotations within the local stress field in three dimensions, static and dynamic triggering of
4
earthquakes within a regional fault system, changes in strengthening and weakening of the lower
crust, and strain-dependent delocalization of strain within the fault zone.
1.3 Problem solving: chasing earthquakes through time
In order to address this litany of possibilities contributing to irregularity within the
earthquake cycle, we must continue to add to the growing datasets on the timing of earthquakes
with their associated displacements. Different methodologies can address different space and
time scales of the earthquake cycle.
For instance, optical image correlations and geodetic modelling can account for much of
the short-time scales (i.e., less than the past 100 years) over a far range of spatial scales (from
millimeters to hundreds of kilometers). Correlation methods utilize high-resolution aerial
photography, lidar topography and radar data collected before and after an event. Once the pre-
event and post-event data are co-registered, they can be correlated and yield a two- or three-
dimension deformation field, yielding centimeter resolution displacement data. One such
example of development of a new correlation workflow is presented in Chapter 6 of this
dissertation. These optical correlation data represent the far-field coseismic deformation recorded
at the Earth’s surface (i.e., strain accommodation). In contrast, geodesy can inform our
understanding of strain accumulation over the tens of years’ time scale. Modeling of GPS data
can yield geodetic strain deficit rates, which indicate how much strain should be accommodated
on each fault based on the motion between modeled fault blocks given the total plate rate. While
both of these numerical methods are extremely helpful at elucidating the strain accumulation and
accommodation, they typically do not answer questions about the earthquake cycle through time
for large, surface rupturing, stick-slip events.
5
In order to gain information about the longer term record of the earthquake cycle, we can
complete geologic field studies of active faults, such as geologic slip rate and paleoearthquake
studies. Geologic slip rates studies require a displacement measurement of an offset feature (i.e.,
a channel), along with an age of that offset feature (i.e., age of initial incision of that channel).
Given these two pieces of data, a rate can be calculated. This type of study is particularly
effective at highlighting changes in the earthquake cycle over time when multiple offset features
can be mapped and dated at a given site, allowing for the calculation of multiple incremental slip
rates, modeled between successive displacement-time measurements, instead of only a single,
finite slip rate fit between a single displacement-time measurement and the occurrence of the
most recent earthquake. Two such studies are presented in Chapters 4 and 5 of this dissertation.
These studies yield rates of earthquake occurrence, but do not indicate the actual age of the
earthquakes themselves. Paleoseismic studies, however, can determine the age range of
paleoevents using fault perpendicular trenching to observe deformation in sediment layers.
Precise dating of these paleoevent horizons yields patterns of strain accommodated in large,
surface rupturing, strike-slip events. One such study is presented in Chapter 3 of this manuscript.
Together, these data sets have the potential to yield a dated path of events through time, which
combines single event displacements with paleoearthquake ages. This method can identify
periods of faster than average strain accommodation (i.e., earthquake clusters) and periods of
slower than average strain accommodation (i.e., earthquake lulls). These deviations from
“average” earthquake behavior, especially when observed on multiple faults within a complex
fault system, beg the question: how does the distribution of strain oscillate between faults?
In order to answer these questions, studies need to be conducted over moderate (1,000—
10,000 years) to long (1,000,000—10,000,000 years) time scales and broad spatial scales (100’s
6
of km). Tectonic reconstructions and fault evolution studies can highlight the importance of
broad scale fault interactions that inform how plate boundaries become more kinematically
efficient over time. Plate boundaries typically contain multiple faults working together to
accommodate plate loading. Their coevolution, including changes in fault geometry made over
millions of years, inform how faults are tectonically loaded on centennial time scales. As such,
patterns of fault evolution within a plate boundary fault system may control the previously
discussed alternations between seismic highs and lows. These concepts are further explored in
Chapter 2 of this dissertation.
1.4 Future problems: Earthquake cycles at depth
Up until this point in this introduction, I have only addressed observations of the
earthquake cycle in the upper crust, and, truthfully, the upper three meters of the upper crust at
most. Simply put, the data accessible on currently active faults lie primarily at the surface (save
for seismic data and geodetic inversions). The kilometers of shear zone below the surface is
where fault is actually accumulating strain, and this portion of the fault and how it relates to the
earthquake cycle measured at the surface is not directly addressed in the studies presented in this
thesis. To date, there has been limited study on looking for evidence of variability within
earthquake cycle over time preserved in deeper shear zones. One example of recent work of an
exhumed shear zone from mid-to-lower crustal levels shows evidence of geometrically complex
ruptures using psuedotachylyte orientations and interactions, indicating that the sub-surface
geometry of faults may influence how much strain is accommodated throughout the shear zone
up to the Earth’s surface [Rowe et al., 2018]. Continued work in the mid-to-lower crust will
7
hopefully yield a physical understanding of how or why faults can effectively speed up or slow
down, perhaps due to discrete changes in fault strength.
1.5 The current situation: what data to use moving forward
Identifying the physics (likely in the lower crust) that control the irregularity of
earthquake occurrence over time is critical for earthquake simulations and probabilistic seismic
hazard analysis (PSHA). PSHA is used to develop maps of seismic hazard for California [Field
et al., 2013] and for the United States [Petersen et al., 2015]. Such modelling efforts are
concerned with making products and predictions on human time scales, on the order of 50 years,
rather than 500 or 5,000 years. Currently, such modelling utilizes long-term, finite geologic slip
rates. Variability in the earthquake cycle is therefore not presently reflected in these input data.
This might be a reasonable assumption to make. How does millennial variability in slip rate
translate to 50-year periods? Moreover, recurrence intervals of most major faults are on the order
of 100’s of years, not 10’s of years. In order to model this long-term variability over shorter-term
human time scales requires a better understanding of earthquake physics and strength variations
with depth, which would enhance the input parameters of fault sources and could therefore create
natural, realistic variability within the earthquake cycle in these earthquake simulators.
8
CHAPTER 2: A model for the initiation, evolution, and controls on seismic behavior of the
Garlock fault, California
This chapter is based upon the published manuscript:
Hatem, A.E., Dolan, J.F. (2018). A model for the initiation, evolution, and controls on seismic
behavior of the Garlock fault, California. Geochemistry, Geophysics, Geosystems, 19.
2.1 Abstract
We develop a model for the evolution and activity of the Garlock fault that combines
elements of three previously proposed mechanisms: (1) conjugate slip to the San Andreas fault
(SAF); (2) extension in the Basin and Range; and (3) bending from oblique shear in the Eastern
California Shear Zone (ECSZ). Conjugate slip is greatest in the west, and decreases eastward.
Conversely, extension-induced slip increases westward from the eastern termination of the fault,
reaching a maximum at and to the west of the intersection with the Sierra Nevada frontal fault.
Oroclinal bending provides only a small contribution to Garlock slip that increases eastward
from the east-central segment. These spatio-temporally complex loading patterns may explain
alternating periods of fault activity along the Garlock and neighboring faults. Moreover, these
complex kinematic relationships demonstrate that the Garlock fault acts as an efficient
mechanical “bridge” linking slip on the northern ECSZ and SAF that may have delayed or even
obviated the long-hypothesized development of a new Pacific-North America plate boundary
along the ECSZ-Walker Lane.
2.2 Introduction
Several different models have been proposed to explain the role of the sinistral northeast-
to east- striking Garlock fault in accommodating relative plate motion within the primarily
dextral, NNW- to SSE-striking Pacific-North America (Pac-NAm) plate boundary in southern
9
California. None of these models by themselves, however, provide a unified mechanical
explanation for the evolution and continued activity of this major fault [Hatem and Dolan, 2015].
Such understanding is critical to untangling the mechanics of fault system behavior in the
kinematically complex Pac-NAm plate boundary, as well as similar plate boundaries around the
world.
In this paper, we consider the geometry of the Garlock fault, together with geologic slip
rates and fault initiation timing constraints, to assess the relative contributions of three different
driving mechanisms suggested in previous conceptual models. We use these results to devise a
unified model that encompasses elements of all three models and quantifies the contributions
from each along strike. We discuss these results in light of Garlock fault evolution, the complex
fault interactions that drive Garlock slip, and controls on earthquake occurrence in southern
California.
2.3 The Garlock Fault
The sinistral Garlock fault extends for ~255 km in a broad NE to EW arc that spans more
than half the width of California (Figure 1). Along its length, the Garlock exhibits notable
changes in strike that have been used to separate the fault into three segments. The 100-km-long
western segment, which extends from the western intersection of the Garlock fault with the San
Andreas fault (SAF) to the Koehn Lake transtensional stepover, strikes ~060˚. The 070˚-
080˚striking central segment spans the ~90-km-long distance between Koehn Lake and the
eastern end of Pilot Knob Valley, near the southern end of Panamint Valley. The 65-km-long,
090˚-095˚-striking, eastern segment extends from the eastern end of Pilot Knob Valley to the
eastern end of the Garlock fault at the southern end of the Death Valley fault system. Offsets of
bedrock and structural features show that the Garlock fault has accommodated a cumulative
10
displacement of 48-64 km [Smith, 1962; Smith and Ketner, 1970; Jahns et al., 1971; Davis and
Burchfiel, 1973; Monastero et al., 1997]. Surface displacement likely began in late Miocene time
[Eaton, 1932], sometime after 17 Ma [Monastero et al., 1997], and likely c. 11 Ma [Burbank and
Whistler, 1987; Frankel et al., 2008; Andrew et al., 2014].
Although the Garlock fault has not generated any significant earthquakes during the
historic period, there is abundant evidence for recent activity [Clark, 1973; Burke and Clark,
1978; Roquemore et al., 1982; McGill and Sieh, 1991, 1993; Dawson et al., 2003; McGill et al.,
2009; Ganev et al., 2012; Madugo et al., 2012; Rittase et al., 2014; Dolan et al., 2016].
Specifically, geologic slip rates averaged over early Holocene-late Pleistocene time (7-100 ka)
show that the western and central segments of the fault are among the faster-slipping faults in
southern California. These rates reveal a consistent eastward decrease in Garlock fault rate, from
7.6+3.1/−2.3 mm/yr on the western segment [McGill et al., 2009], to ~5-6 mm/yr on the central
segment [Clark and Lajoie, 1974; McGill and Sieh, 1993; Ganev et al., 2012; Crane, 2014;
Dolan et al., 2015], to 1
+ 1 . 5
− 0 . 5
mm/yr on the eastern segment [Crane, 2014]. Paleoseismologic
studies demonstrate that the Garlock fault has generated large-magnitude, late Holocene surface-
rupturing earthquakes [McGill, 1992; McGill and Rockwell, 1998; Madugo et al., 2012].
Collectively, these data suggest that the Garlock experiences super-cycles of strain release,
characterized by variability in both earthquake occurrence and slip rates [Dawson et al., 2003;
Dolan et al., 2007; 2016; Ganev et al., 2012; Rittase et al., 2014], that are potentially correlated
with the behavior of other regional faults [Peltzer et al., 2001; Dawson et al., 2003; Dolan et al.,
2007; 2016; Oskin et al., 2008; Ganev et al., 2012; McAuliffe et al., 2013], underscoring the
importance of considering regional tectonics when devising a comprehensive model for the
origin and continued activity of the Garlock fault.
11
2.4 Models of Garlock fault behavior
Three kinematic models have been proposed to explain the origin and evolution of the
Garlock fault: (1) conjugate slip with the SAF [Hill and Dibblee, 1953]; (2) intracontinental
transform faulting from Basin and Range (BR) extension [Troxel et al., 1972; Davis and
Burchfiel, 1973]; and (3) oroclinal bending from eastern California shear zone (ECSZ) dextral
shear [Garfunkel, 1974; Guest et al., 2003] (Figure 2). Although each of these models highlights
important aspects of the origin and current behavior of the Garlock fault, none of them alone
provides a complete mechanical explanation of the Garlock fault. Garfunkel [1974], for example,
postulated that there may be more than one mechanism driving the Garlock. Here, we quantify
these previously proposed mechanisms into a unified model for the origin and behavior of the
Garlock fault.
In the first model, the Garlock fault serves as a conjugate to the San Andreas fault to
accommodate the north-south shortening and east-west extrusion resulting from the mechanically
inefficient 295° strike of the SAF southeast of the Big Bend, relative to the ~325° strike [DeMets
et al., 2010] of Pac-NAm relative plate motion [Hill and Dibblee, 1953; Stuart, 1991; King et
al., 2004] However, only the western Garlock is at a mechanically efficient conjugate (R’) shear
orientation with respect to the San Andreas today [McGill et al., 2009].
The second model posits that the Garlock fault behaves as an intracontinental transform
fault that accommodates east-west extension of the Basin and Range (BR) north of the Garlock,
relative to the non-extending Mojave block to the south [Davis and Burchfiel, 1973]. Although
BR extension-induced sinistral shear on the Garlock effectively explains some of the slip on the
12
central and eastern segments of the Garlock, this model alone cannot explain the fast slip rates of
the western and central Garlock fault [McGill et al., 2009].
The third model invokes clockwise oroclinal rotation of the Garlock fault in response to
NNE-SSW, dextral shear in the ~80-km-wide ECSZ [Garfunkel, 1974; Luyendyk et al., 1980;
Dokka and Travis, 1990a; Luyendyk, 1991; Humphreys and Weldon, 1994; Schermer et al.,
1996; Guest et al., 2003]. These models postulate that the entire Garlock fault initiated at the
060˚ present-day strike of the western Garlock, and that the central and eastern segments have
rotated clockwise to their current orientations [Guest et al., 2003; McGill et al., 2009; Andrew et
al., 2014; Andrew and Walker, 2017] [Guest et al., 2003; Andrew et al., 2014; Andrew and
Walker, 2017]. A key limitation of this model, as discussed below, is that it applies primarily to
the eastern Garlock, and minimally to the central Garlock, without contributing any slip to the
fast-slipping western Garlock, which lies to the west of any ECSZ dextral shear [McGill et al.,
2009].
2.5 A model for Garlock fault evolution
We propose a comprehensive model for the evolution and ongoing activity of the Garlock
fault that encompasses elements of all three earlier models and that is constrained by the current
geometry and recent slip rates of the Garlock, as well as by regional structural and
geochronologic data. Sinistral Garlock fault slip likely began c. 11 Ma [Andrew et al., 2014]
(Figure 2a). We assume that sinistral slip initiated on the eastern and central Garlock first,
primarily to accommodate intra-continental extension within the Basin and Range and ECSZ
north of the Garlock, as proposed by Troxel et al., [1972] and Davis and Burchfiel, [1973].
13
The original strike of the currently active, sinistral Garlock fault is not well constrained.
If the intracontinental transform fault model is correct, then the original strike of the Garlock
would be expected to be approximately parallel to the BR extension direction north of the fault
[e.g., Menard and Atwater, 1968, 1969; Atwater, 1970]. Regional palinspastic reconstructions
indicate that, prior to initiation of the Garlock fault c. 11 Ma, extension in the region of the future
Garlock fault c. 14-16 Ma was oriented approximately ~075° [McQuarrie and Wernicke, 2005;
their figure 10E]; this SW-NE extension direction may have generated a WNW-trending regional
fabric along which nascent faults could develop. In contrast, at 10-12 Ma, at the time of Garlock
fault initiation, the BR extension direction north of the fault was approximately E-W, rotating to
~N75°W by 8-10 Ma [Wernicke and Snow, 1998; Snow and Wernicke, 2000; McQuarrie and
Wernicke, 2005]. Following a different line of reasoning, some authors have suggested that the
original orientation of the Garlock was 060°, parallel to the current orientation of the western
segment of the Garlock [e.g., Guest et al., 2003; McGill et al., 2009]. Thus, various proposals
for the initial orientation of the Garlock fault have ranged from 090° to 060°.
An original strike of 090° for the central and eastern Garlock fault is problematic for a
number of reasons. For example, this orientation would require that the currently ~080°-striking
central segment would have rotated CCW within the ~N-S, nearly Garlock-perpendicular dextral
ECSZ strain field, which seems kinematically unlikely. Moreover, an original 090° orientation
would suggest that the current 090°-striking eastern segment has not rotated at all, despite being
subjected to at least 3 My of dextral ECSZ shear, as well as evidence for significant CW
rotations south of the eastern Garlock fault in the eastern Mojave region [Schermer et al., 1996;
Guest et al., 2003]. Finally, an original 090° orientation for the central and eastern Garlock
would leave no room for subsequent northward motion of the southern Sierra Nevada, which is
14
required by palinspastic reconstructions; these reconstructions suggest a more ENE original
strike of the central and eastern Garlock (~060°-075°) [McQuarrie and Wernicke, 2005]. We
therefore suggest that an original 090° strike for the central and eastern segments of the Garlock
fault can be ruled out. In the following, we therefore only consider possible original orientations
for the central and eastern Garlock fault of 060° and 075°, either of which is consistent with
currently available constraints
One of the most striking features of the Garlock fault is its arcuate trace (Figure 1). We
suggest that as BR extension north of the Garlock continued after 11 Ma, the central Garlock
fault grew southwestward along a pre-existing zone of weakness, possibly along a pre-existing
normal fault that aided in Late Cretaceous gravitational collapse of the Sierra Nevada Batholith
[Malin et al., 1995; Wood and Saleeby, 1997; Chapman et al., 2012; Blythe and Longinotti,
2013] (Figure 2C). This southwestward propagation was likely enhanced by the orientation of
the stress field generated at the western end of the nascent fault (Figure 2B). Specifically,
Coulomb failure function modeling indicates that a propagating fault will curve into the
extensional stress lobe of the fault tip [e.g., Bowman et al., 2003; Armijo et al., 2004; Flerit et
al., 2004] (Figure 2B). Thus, the western end of the central Garlock would propagate with
progressively more southwesterly strikes relative to an inferred original ~075° strike of the east-
central and eastern sections of the fault (Figure 2B-D). Eventually, the southwestward-
propagating Garlock intersected the SAF (Figure 2D), likely along the pre-existing normal fault
(Figure 2C), establishing the mechanically efficient conjugate fault pair that prevails today.
The timing of this event remains poorly constrained, but the geometry of the SAF and the
slip rate of the western segment of the Garlock fault can be used to infer a minimum-possible age
of Garlock-SAF intersection. If the SAF through the future region of the Big Bend initially had a
15
relatively linear NW local strike [Matti and Morton, 1993], then the portion of the SAF north of
its intersection with the Garlock fault has been warped 25 ± 5 km to the southwest by sinistral
motion along the Garlock fault [Davis and Burchfiel, 1973; Garfunkel, 1974; Bohannon et
al.,1982; Stuart, 1991] (Figure 2D-E). Assuming ~25 ± 5 km of bending of the SAF by
southwestward motion of the Sierra Nevada-Great Valley block with the current ~7.5 mm/yr slip
rate [McGill et al., 2009], suggests that Garlock fault-induced warping of the SAF could have
begun as recently as c. 3 Ma. But if the early phases of slip along the newly established,
mechanically immature western Garlock occurred at slower rates than the current 7.5 mm/yr,
then the ~25 km of SAF bending could have taken much longer than 3 Myr, and the Garlock
could have reached the SAF at an earlier date. Indeed, earlier contractional deformation manifest
in early Pliocene (4-6 Ma) exhumation documented by Apatite (U-Th)/He thermochronology
within the Big Bend region [Niemi et al., 2013] is consistent with an earlier initiation of the Big
Bend. This onset of accelerated exhumation rates in the Big Bend area c. 4-6 Ma [Niemi et al.,
2013] suggests that the current efficient conjugate pairing of the SAF and the Garlock fault had
been established prior to the onset of the earliest well-developed NNW-trending faults in the
currently active ECSZ c. 3-4 Ma, when the Death Valley, Panamint Valley, and Owens Valley
fault systems all began to accommodate rapid dextral shear [Burchfiel et al., 1987; Monastero et
al., 2002; Lee et al., 2009; Norton, 2011] and dextral slip began in the southern ECSZ on the
Blackwater fault [Oskin and Iriondo, 2004; Andrew et al., 2014; Andrew and Walker, 2017].
Whether the Garlock fault reached the SAF at 3 Ma or somewhat earlier, the full extent
of the Garlock fault developed with minimal contribution due to dextral slip within the northern
and southern ECSZ fault rotation from ECSZ dextral shear (Figure 2D). Palinspastic
reconstructions suggest that there may have been distributed dextral shear in the southern ECSZ
16
as early as 8-12 Ma [McQuarrie and Wernicke, 2005],which could have resulted in some CW
rotation of the central and eastern Garlock fault prior to the initiation of well-developed, high-
slip rate faults in the ECSZ. We note that any such distributed dextral shear prior to the ca. 3-4
Ma onset of the currently active ECSZ dextral fault system would render the rotation-induced
sinistral slip rate estimates for the central and eastern Garlock that we calculate in the following
section as maximum values – the true, full 11 Ma rotation-induced sinistral slip rates must be
slower if there was any distributed dextral shear across the Garlock prior to 3-4 Ma.
2.6 Controls on current Garlock fault slip rates
Geologic slip rates determined along the Garlock fault provide key kinematic
observations that help to constrain a comprehensive mechanical model for the ongoing activity
of the Garlock fault. We use geologic slip rates that are at least of mid-Holocene to latest
Pleistocene age as constraints (Data Repository Table 1). This age requirement intentionally
excludes some young (<4 ka), fast rates observed on the central segment [Rittase et al., 2014;
Dolan et al., 2016] because these rates encompass a four-event earthquake cluster at 0.5-2 ka
during which the fault was slipping far faster than its long-term average rate [Dawson et al.,
2003; Dolan et al., 2007; 2016; Ganev et al., 2012]. Integration of the long-term, mid-Holocene
to latest Pleistocene fault slip rates with our model of fault evolution allows us to assess the
relative along-strike contributions of each of the three kinematic models that have been
proposed to drive the Garlock fault.
2.6.1 Rotation of the Garlock fault due to ECSZ dextral shear
17
We calculate the contribution to sinistral slip rate along the Garlock fault due to oroclinal
bending of the fault by NNW-oriented ECSZ dextral shear by quantifying the difference in
length between the initial and present-day traces (Fig. 3). The present-day Garlock fault must
have lengthened from its initial, un-deflected state as dextral shear in the ECSZ north and south
of the Garlock fault has rotated the fault clockwise, and in the process, extended the length of the
Garlock fault. These locations mark deviations from the preferred initial proposed fault strike of
075°, which correspond to increases in dextral shear rate across the Garlock fault, and are thus
pivot points, or fulcrums, that mark changes in the rate of ECSZ dextral slip-induced clockwise
rotation of the Garlock fault (Figure 3A). The initial geometry of the Garlock fault is shown as a
gray, dashed line, whereas the current, rotated geometry of the Garlock fault is shown by the
solid purple line (Figure 3A).
As discussed in the previous section, the preferred initial ENE geometry of the central
and eastern Garlock is approximately perpendicular to the NNW orientation of the main northern
ECSZ faults. Hence, as shown in Figure 3B, we can envisage a triangle with a right angle
between the trend of the ECSZ and the initial Garlock fault, and with an angle theta (θ) between
the initial and present-day strikes of the Garlock fault. The leg of the triangle adjacent to theta
represents the initial Garlock fault, and the hypotenuse represents the present-day Garlock fault
(Figure 3B). The opposite side of this triangle represents the amount of dextral ECSZ slip that
has occurred since initiation of dextral shear in the ECSZ to produce the Garlock fault rotation.
As the ECSZ accommodates dextral shear over time, the leg of the triangle opposite theta in
Figure 3B must grow. As that leg grows, the hypotenuse must also lengthen, showing that the
Garlock fault must lengthen to accommodate continued dextral shear in the ECSZ. Under these
assumptions, the present-day Garlock must be longer than the initial Garlock. We use the
18
following equation to calculate how much the present-day Garlock fault must have grown
relative to the initial fault:
g = L/cosθ – L
where g is the amount of growth, or lengthening, of the Garlock fault due to ECSZ dextral shear-
induced clockwise rotation of the Garlock, L is the length of the initial trace of the Garlock
(adjacent leg of the triangle in Figure 3B—the initial length of the Garlock fault), and θ is the
angle between the initial and present day traces of the Garlock fault.
For this exercise, we assume spatially uniform strikes of the present-day east-central and
eastern segments of the Garlock of ~078° and ~092°, respectively. Because total dextral shear
rate increases progressively eastward at each intersection with major ECSZ faults, the amount of
total southward deflection of the Garlock is greatest at the eastern end. The change in length of
the Garlock fault between the inferred initial strike and the longer, present-day fault represents
the “growth” of the Garlock fault required for lengthening the fault; this lengthening of the
Garlock fault denotes the resulting amount of sinistral fault slip on the Garlock fault as it was
oroclinally rotated by ECSZ dextral shear (Figure 3B). The resulting contributions to Garlock
fault slip due to rotation will increase at each of the fulcrum points at each distinct change in
present-day strike. Specifically, these pivot points (denoted by purple dots in figure 3B) are
located at the western end of the ECSZ-induced rotation field along the western end of the
central segment of the Garlock fault at the western edge of significant dextral ECSZ shear, and at
the central to eastern Garlock fault segment boundary for the eastern segment.
We use the c. 3 Ma initiation of significant ECSZ dextral shear [Burchfiel et al., 1987;
Monastero et al., 1997; Oskin and Iriondo, 2004; Lee et al., 2009; Norton, 2011; Andrew et al.,
2014; Andrew and Walker, 2017] to calculate the rotation-induced component of sinistral
19
Garlock fault slip rate. As noted above, the following estimates do not consider any pre-3 Ma
clockwise rotation of the Garlock that may have occurred in response to an earlier phase of
distributed dextral shear in the Mojave prior to establishment of the well-developed, currently
active ECSZ faults by c. 3 Ma. Thus, the rates we calculate will be maxima, as we are assuming
that all of the rotation has occurred within the minimum 3 My period since the establishment of
discrete NNW faulting within the ECSZ.
The resulting rotation-induced sinistral Garlock rate using a preferred initial strike of
075° and an initial fault length of 25 km (the length of the eastern part of the Garlock fault
central segment that is oriented at strikes greater than 075°) is extremely slow on the east-central
Garlock fault, with a maximum rate of 0.01 mm/yr due to the small amount of rotation (~3°)
(Figure 4). However, the 50-km-long eastern segment has rotated more (~17° relative to an
initial 075° strike), yielding a larger rotational contribution to sinistral Garlock slip that reaches a
maximum of ~0.8 mm/yr at the eastern end of the fault. It is noteworthy that this rotation-
induced sinistral rate is similar to the preferred ~1 mm/yr slip rate documented for the eastern
Garlock fault by Crane [2014]. Alternatively, for an initial fault strike of 060°, the maximum
sinistral slip rate of the east-central Garlock increases to 0.4 mm/yr, and to 2.7 mm/yr at the
eastern end of the Garlock, at or faster than the extreme high end of the observed eastern Garlock
slip rate of 1
+ 1 . 5
− 0 . 5
mm/yr [Crane, 2014].
2.6.2 Intracontinental transform contribution from Basin and Range extension
Because the rotation model does not generate sufficient sinistral slip to explain the
behavior of the entire Garlock, especially for the faster-slipping central and western segments,
we explore the slip contribution due to BR extension north of the fault. This component increases
20
westward in an additive sense as each successive BR fault is crossed, from a minimum at the
southern end of Death Valley to a maximum at the Sierra Nevada Frontal fault [Davis and
Burchfiel, 1973; McGill et al., 2009].
We used extension rates from BR normal and ECSZ oblique-dextral faults north of the
Garlock, specifically the Sierra Nevada Frontal, Panamint Valley, and Death Valley dextral-
normal faults, to calculate Garlock-parallel extension rates. We projected these extension rates
(Data Repository Table 1) onto the local strike of the eastern and central segments of the Garlock
fault. The resulting Garlock-parallel sinistral slip rates are 0.5-0.7 mm/yr from the Death Valley
fault, 0.9-1.5 mm/yr from the Panamint Valley fault, and 0.1-0.2 mm/yr from the Sierra Nevada
Frontal fault [Le et al., 2007; Hoffman, 2009; Frankel et al., 2015]. The maximum cumulative
BR extension contribution to Garlock fault slip is therefore 1.5-2.4 mm/yr, at and extending to
the west of the Sierra Nevada Frontal fault (Figure 4).
2.6.3 Conjugate contribution from San Andreas (SAF)
The combined BR extensional and ECSZ rotational components of Garlock slip in our
model match the slow (1
+ 1 . 5
− 0 . 5
mm/yr) slip rate documented by Crane [2014] for the eastern
Garlock fault (Figure 4), especially for an initial inferred fault orientation of 075°, suggesting
that the current slip rate of the eastern segment is adequately explained by these mechanisms.
However, these summed slip-rate components are insufficient to explain the faster rates on the
central and western segments. This slip rate discrepancy indicates that conjugate slip must
contribute the difference between the combined extension-rotation contributions and the
measured slip rates on the western and central segments (Figure 3). For example, the preferred
slip rate of the western segment is ≤7.5 mm/yr [McGill et al., 2009; McGill written
21
communication 2018]. Because the western Garlock is not deformed by dextral shear in the
ECSZ (i.e., the rotational contribution to slip is 0 mm/yr), and BR extension contributes only
~1.5-2.5 mm/yr of sinistral slip to the western Garlock, indicating that conjugate motion
accounts for the remaining ~5-6 mm/yr of the preferred ~7.5 mm/yr western Garlock fault slip
rate measured by McGill et al. [2009]. Similarly, the slip rate of the central Garlock, which is
well constrained from a number of sites at ~5-6 mm/yr [Clark and Lajoie, 1974; McGill and
Sieh, 1993; Ganev et al., 2012; Crane, 2014; Dolan et al., 2015], exceeds the combined
contributions of sinistral slip due to BR extension and ECSZ shear of ~1.5-2.5 mm/yr, indicating
that conjugate slip contributes the remaining ~3-4 mm/yr to the central Garlock slip rate.
Because the current slip rate of the eastern segment of the Garlock fault can be explained solely
by contributions from the ECSZ rotational and BR extensional mechanisms, we suggest that little
to none of the current slip on the eastern segment is due to conjugate behavior.
2.7 Implications for fault interactions, earthquake recurrence, and plate boundary
evolution in southern California
Because it is driven by a combination of SAF conjugate shear, BR extension, and ECSZ-
induced rotation, the Garlock fault can store elastic strain energy under multiple loading regimes
depending on which of the three driving mechanisms were most active at any given time.
Understanding these relationships will thus provide insight into spatio-temporal patterns of
earthquake occurrence and fault slip on the Garlock and neighboring faults. Moreover, these
complex kinematic relationships suggest that the Garlock fault acts as a mechanical “bridge”
connecting slip on the Mojave SAF (mSAF) and southern SAF (sSAF) with slip on faults in the
BR and ECSZ north of the Garlock (NECSZ). The NECSZ-GF-mSAF-sSAF fault network
22
creates a closed kinematic loop that acts as an efficient, mechanically complementary alternative
to N-S dextral shear in the ECSZ-Walker Lane-sSAF system. As a result, strain transfer via the
Garlock fault enables the mSAF to accommodate dextral shear that would otherwise be
accommodated on the ECSZ south of the Garlock (SECSZ).
The occurrence of large historic earthquakes (1872 Mw~7.6 Owens Valley, 1992 Mw=7.3
Landers, 1999 Mw=7.1 Hector Mine) and several recent pre-historic surface ruptures on the
Panamint Valley and Death Valley faults, as well as on several faults in the SECSZ [Rockwell et
al., 2000a; Frankel et al., 2008; Ganev et al., 2012; McAuliffe et al., 2013], attest to recent rapid
strain release on the through-going ECSZ sub-system. Conversely, the Garlock has generated
only one surface rupture in the past 1000-1200 years, and none in the past c. 500 years [Dawson
et al., 2003; Madugo et al., 2012], despite a relatively rapid average long-term slip rate.
Moreover, most models of geodetic data suggest present-day elastic strain accumulation rates
that are much slower than the average Garlock fault slip rate [e.g., McClusky et al., 2001; Peltzer
et al., 2001; Meade and Hager, 2005; Loveless and Meade, 2011; Evans et al., 2016]. These data
suggest that the sub-system including the Garlock fault is currently storing and releasing strain
energy at much slower than average rates while the ECSZ sub-system is releasing energy at
faster than average rates. The recognition of millennia-long strain supercycles along the Garlock
and related faults suggests that the apparent alternation between these two sub-systems may
persist over longer time scales [Dolan et al., 2016].
Further complicating this behavior is the fact that the SAF stores and releases strain much
faster than any other fault in the plate boundary system. Thus, because the western and central
segments of the Garlock fault are loaded primarily by conjugate shear-related strain
accumulation, the Garlock will accumulate and accommodate strain faster when the SAF is
23
slipping faster. Conversely, the eastern Garlock is loaded primarily by ECSZ dextral shear-
induced rotation, and will therefore accumulate and accommodate strain primarily in response to
the behavior of the ECSZ. The central segment of the Garlock, which is loaded by both
conjugate shear and BR extension, might be expected to experience very complicated patterns of
strain accumulation and release controlled by the relative rates of the two driving mechanisms.
In general, available paleoearthquake and incremental fault slip rate data are too
incomplete to fully evaluate our proposed multi-mechanism Garlock loading model. However,
where the data are of sufficient density and precision, as along the mSAF and central Garlock,
they support our model results. For example, as noted by Dolan et al., [2016], the two most-
recent central Garlock earthquakes observed at El Paso Peaks (c. 0.5 and 1.0 ka; [Dawson,
2003]), correlate closely in time with an anomalously large-magnitude mSAF earthquake and a
period of highly accelerated mSAF slip rate observed at the Wrightwood paleoseismic site
[Weldon et al., 2004], consistent with the predominant, conjugate-loading mechanism we suggest
for the western and central Garlock fault. Unfortunately, detailed incremental slip rate data are
not available for mSAF prior to c. 1.5 ka, and thus we cannot evaluate whether the two earlier
earthquakes in the 0.5-2.0 ka central Garlock cluster (c. 1.5-1.75 ka and 1.75-2.0 ka; [Dawson,
2003]) were related to either anomalously rapid slip and/or particularly large SAF earthquakes.
The location of the 1992 Mw 7.3 Landers earthquake led to the suggestion that this
earthquake was a harbinger of an eventual eastward jump in the Pac-NAm plate boundary from
the SAF system to east of the Sierra Nevada along the ECSZ-Walker Lane [Nur et al., 1993; Du
and Aydin, 1996; Spotila and Anderson, 2004; Faulds et al., 2005; Yang and Hauksson, 2013;
Thatcher et al., 2016]. This inference is based on the assumption that events such as Landers,
which connected parts of six different ECSZ faults together to form an overall more linear north-
24
trending rupture, will eventually create a through-going N-S zone of dextral shear that could
more efficiently accommodate Pac-NAm plate motion by bypassing the mechanical complexities
of the SAF system in southern California (e.g., the Big Bend and San Gorgonio “knot”) [Nur et
al., 1993].
Although this model is attractive in its simplicity, and earthquakes such as Landers
suggest that such a plate boundary switch may eventually occur, a key observation is that no
ECSZ faults extend across the Garlock fault [e.g., Davis and Burchfiel, 1973; Garfunkel, 1974;
Dokka and Travis, 1990b; Glazner et al., 2002; Oskin and Iriondo, 2004; Andrew et al., 2014;
Andrew and Walker, 2017], despite at least 3 million years of N-S dextral shear and current
geodetic data suggesting that dextral shear extends across the Garlock fault [Gan et al., 2000;
McClusky et al., 2001; Peltzer et al., 2001; Monastero et al., 2002; Lee et al., 2009; McGill et
al., 2009; Norton, 2011]. Given these observations, the question arises, why has such a through-
going fault system not yet developed?
We propose that the mechanical efficiency of the Garlock fault, in providing an
alternative pathway for dextral ECSZ shear, reduces the need for this postulated eastward shift.
Continued activity along the Garlock thus maintains the long-term stability of the current plate
boundary configuration, including the mechanical inefficiency of the Big Bend of the San
Andreas. Loading of the Garlock fault by both conjugate SAF slip and BR extension will
continue to be accommodated effectively by Garlock fault slip, at least whenever the NECSZ-
Garlock-mSAF-sSAF sub-system is active. However, ongoing sinistral Garlock slip will also
continue to amplify the curvature of the Big Bend, making this part of the SAF less and less
mechanically efficient in accommodating Pac-NAm plate boundary dextral shear. If, eventually,
increasing misalignment of the SAF at the Big Bend caused by continued Garlock fault slip
25
renders the SAF too inefficient to effectively accommodate Pac-NAm motion, and if fault
growth and linkage similar to those that occurred in the Landers earthquake continue, a
mechanically efficient, through-going N-S zone of dextral shear east of the Sierra Nevada may
yet become established. Until then, however, the Garlock fault will continue to provide an
alternative “stairstep” means of north-south transfer of Pac-NAm shear via conjugate fault pairs
for the foreseeable future.
2.8 Conclusions
Slip along the Garlock fault is driven by a combination of conjugate shear from the San Andreas,
intra-continental transform motion within the Basin and Range, and oroclinal bending due to
dextral shear in the Eastern California Shear Zone. Specific contributions to Garlock fault slip
from different drivers were determined using geologic slip rates. We find that the Garlock fault is
a mechanical “bridge” that stabilizes the primary plate boundary configuration.
2.9 Figure Captions
Figure 1: A) Location map showing major faults in black; other Quaternary faults shown in pale
gray [Jennings, 1994]; SAF: San Andreas fault; SGK: San Gorgonio Knot; SSAF: Southern
SAF; KL: Koehn Lake; PKV: Pilot Knob Valley; PMF: Pinto Mountain fault. ECSZ is eastern
California shear zone. B) Conjugate shear model in which slip on the Garlock fault is driven by
interactions with the San Andreas fault. Large gray arrows show relative motion directions of
key regions. C) Extension model driven by Basin and Range extension north of the Garlock
fault. D) Oroclinal bending model driven by ECSZ dextral shear across the Garlock fault.
26
Dashed pink line shows inferred original orientation of Garlock fault at initiation, and arrow
shows inferred clockwise rotation associated with NNW-trending dextral ECSZ shear.
Figure 2: Proposed incremental evolution of the Garlock fault from c. 11 Ma (A), c. 8 Ma (B), c.
6 Ma (C), c. 3 Ma (D), c. 0 Ma (E), showing our preferred 075° initial strike for the central and
eastern Garlock. Red lines indicate changes in Garlock fault length or geometry from previous
time step; blue lines denote portions of Garlock fault unchanged from previous panel. Dark gray
arrows show extension direction within the current time step; lighter gray arrows show extension
direction in the previous time step, in order to highlight this change. Red lobes show positive
changes in Δ σf (Coulomb stress on optimally oriented faults) that result from stresses at the tip
of a propagating fault and indicate area where failure is promoted; blue lobes show negative
changes in Δ σf, which do not promote fault propagation.
Table 1: Geologic slip rates used to constrain slip rate contributions on Garlock fault.
Figure 3: Schematic diagrams detailing rotational contribution calculations. A) The Garlock
fault (blue line segments) is progressively deformed by dextral shear along NNW-trending ECSZ
faults at ~10 mm/yr. We assume that the Garlock fault bends at pivot points (fulcrums)
associated with intersections between the Garlock and individual ECSZ faults; these fulcrums are
located east of the Blackwater fault (1) and the Panamint Valley fault zone near Pilot Knob
Valley (2), and represent the changes in fault strike along the central and eastern segments of the
Garlock fault. The rate of dextral shear increases eastward at each successive intersection of the
Garlock fault with an ECSZ fault, commensurately increasing the amount of clockwise rotation
27
of the Garlock eastward. Purple lines denote western and eastern boundaries of NNW-directed
dextral ECSZ. No dextral shear is assumed to occur outside of these boundaries. Purple dots
show key positions along the Garlock fault at which major changes in cumulative rotation are
observed. Purple dot 1 shows the location of a slight clockwise change in strike of central
segment of Garlock fault, and purple dot 2 shows the more pronounced change in Garlock fault
strike at the central-eastern segment boundary. B) Schematic illustration of how we estimate the
component of sinistral Garlock fault slip related to ECSZ-induced clockwise rotation of the
Garlock. The inferred original strike of the Garlock fault us shown by the blue line. The purple
line represents total ECSZ dextral shear across the entire system. This dextral shear, which
decreases westward to zero at the western boundary of the ECSZ, induces clockwise rotation of
the Garlock fault (shown schematically by purple arrows). As the Garlock fault rotates, it will
lengthen. The rotated, current Garlock fault trace (thin red line) is longer than the initial trace
projection (blue line) by a finite amount g (thick red line), which is the difference between the
hypotenuse (red) and adjacent (blue) legs in this triangle. As discussed in the text, we use g to
estimate the component of Garlock fault slip that is driven by ECSZ dextral shear-induced
clockwise rotation and lengthening of the Garlock fault.
Figure 4: Garlock fault loading contributions constrained by existing slip rate data. Top Panel
(A): Purple lines plot the relative contributions to Garlock fault loading from ECSZ rotation
(dash-dot), BR extension (dash), and combined (solid) contribution from rotation and extension
(using the mean value of Garlock-parallel extension rates from ranges reported in text). Rotation-
induced sinistral rates include end-member cases of 060° and 075° initial orientation of the
Garlock fault, and rotation rate ranges and consequent extension and rotation combined
28
contributions are represented with purple shaded polygons. Available Garlock geologic slip rates
(a-l; see Table 1). Locations of slip-rate sites are shown on Figure 1 and on lower panel of this
figure. Slip rates are plotted with reported full ranges (transparent, narrow bars) and preferred
ranges (opaque, wide bars), and are colored according to color bar on y-axis. Vertical gray
dashed bars represent slip rate deficit between combined rotation + extension contribution and
the preferred slip rate at each site, showing the assumed contribution to slip rate from conjugate
shear. Broad gray curve shows idealized slip rate distribution along strike. Lower panels (B-E):
Garlock fault map colored by preferred geologic slip rate; same slip-rate color scale as in upper
panel. (B), ECSZ-driven rotation contribution BR (C), extensional contribution (D), and
conjugate contribution to the preferred geologic slip (E). Stars representing slip rates are CW:
Clark Wash; KL: Koehn Lake; SR: Summit Range; SL: Searles Lake; QM: Quail Mountains;
AM: Avawatz Mountains.
Figure 5: Schematic map showing the two proposed strain accumulation and accommodation
systems in southern California, highlighting alternation between fault activities between faster
(black) or slower (gray) rates in each sub-system relative to long-term geologic strain
accommodation rate. Modified from Dolan et al. [2007]. A) Fast, high strain accommodation
Garlock fault. B) Slow, low strain accommodation Garlock fault.
29
CHAPTER 3: A 2000-year paleoearthquake record along the Conway segment of the Hope
fault: Implications for patterns of earthquake occurrence in northern South Island and
southern North Island, New Zealand
This chapter is based upon the submitted manuscript:
Hatem, A.E., Dolan, J.F., Zinke, R.W., Van Dissen, R.J., McGuire, C.P., Rhodes, E.J. (in review.)
A 2000-year paleoearthquake record along the Conway segment of the Hope fault: Implications
for patterns of earthquake occurrence in northern South Island and southern North Island, New
Zealand. Bulletin of the Seismological Society of America.
3.1 Abstract
Paleoseismic trenches excavated at two sites reveal ages of late Holocene earthquakes
along the Conway segment of the Hope fault, the fastest-slipping fault within the Marlborough
fault system in northern South Island, New Zealand. At the Green Burn East site (GBE), a fault-
perpendicular trench exposed gravel colluvial wedges, fissure fills, and upward fault
terminations associated with five paleo-surface ruptures. Radiocarbon age constraints indicate
that these five earthquakes occurred after 36 BCE, with the four most recent surface ruptures
occurring during a relatively brief period (550 years) between c. 1290 CE and the beginning of
the historical earthquake record c.1840 CE. Additional trenches at the Green Burn West site
(GBW) site 1.4 km to the west reveal four likely co-seismically generated landslides that
occurred at approximately the same times as the four most recent GBE paleoearthquakes,
independently overlapping with age ranges of events GB1, GB2, and GB3 from GBE.
Combining age constraints from both trench sites indicates that the most recent event (GB1)
occurred between 1731–1840 CE, the penultimate event GB2 occurred between 1657-1797 CE,
GB3 occurred between 1495-1611 CE, GB4 occurred between 1290-1420 CE, GB5 between 36
BCE and 1240 CE. These new data facilitate comparisons with similar paleoearthquake records
from other faults within the Alpine-Hope-Jordan-Kekerengu-Needles-Wairarapa (Al-Hp-JKN-
30
Wr) fault system of through-going, fast slip rate (i.e., ≥10 mm/yr) reverse-dextral faults that
accommodate a significant portion of Pacific-Australia relative plate boundary motion. These
comparisons indicate that combinations of the faults of the Al-Hp-JKN-Wr system may
commonly rupture within relatively brief, ≤~100-year-long sequences, but that full “wall-to-
wall” rupture sequences involving all faults in the system are rare over the span of our
paleoearthquake data. Rather, the data suggest that the Al-Hp-JKN-Wr system may generally
rupture in sub-sequences that do not involve the entire system, and potentially, at least
sometimes, in isolated events.
3.2 Introduction
Documenting patterns of large-magnitude earthquake occurrence in space and time is of
critical importance for both seismic hazard assessment and a deeper understanding of the
mechanics of plate boundaries. Earthquake recurrence on individual faults has been shown to
exhibit a wide variety of behaviors, from periodic [e.g., Berryman et al., 2012], to quasi-periodic
[e.g., Weldon et al., 2004; Scharer et al., 2007] to clustered [e.g., Marco et al., 1996; Rockwell et
al., 2000; Dawson, 2003; Ferry et al., 2011; Wechsler et al., 2014, 2018]. But, most plate
boundaries exhibit multiple major faults that collectively operate as mechanically integrated fault
systems. Thus, to understand the mechanics of earthquake occurrence along a plate boundary, it
is necessary to document the spatial and temporal earthquake behavior of the primary, fast-
slipping faults that make up the plate boundary.
To complete plate boundary system earthquake behavior analysis, we study the fastest-
slipping strike-slip faults of the Australian-Pacific plate boundary of the northern South Island
and southern North Island of New Zealand, where previous studies have documented patterns of
31
earthquake recurrence on many of these major faults [e.g., Cooper and Norris, 1990; Wells et
al., 1999; Langridge et al., 2003, 2013; Mason et al., 2006; Sutherland et al., 2007; Little et al.,
2009; Van Dissen and Nicol, 2009; Berryman et al., 2012a, 2012b; De Pascale and Langridge,
2012; Howarth et al., 2012, 2014, 2016; Clark et al., 2013; Nicol et al., 2016; Khajavi et al.,
2016; Cochran et al., 2017; Nicol and Dissen, 2018]. The Hope fault, the subject of this
manuscript, is the central link between high slip-rate faults to the southwest (Alpine fault) and
northeast (Jordan-Kekerengu-Needles fault and the Wairarapa fault in southern North Island).
This >850-km-long system (Alpine fault through to Wairarapa fault) of dextral strike-slip and
oblique reverse-dextral faults accommodates the majority of the ~39 mm/yr of relative Pacific-
Australia plate boundary motion in central and northern South Island [DeMets et al., 2010;
Wallace et al., 2012], with the exception of the Wairarapa fault, which, together with the
BooBoo and Needles faults, serves to connect the Marlborough Fault System from South Island
to North Island (Figure 1). Although the Wairarapa fault slips at about half the rate of the South
Island faults, the Wairarapa carries the predominant portion of onshore slip of the plate boundary
on the southern North Island.
In central South Island, much of the relative plate motion is accommodated on the
oblique reverse-dextral Alpine fault, with a right-lateral strike-slip rate of 23-27 mm/yr
[Berryman, 1992; Norris and Cooper, 2001; Sutherland et al., 2006]. At ~42.8°S, 171.5°E the
Alpine fault splays northeastward into multiple parallel, predominantly dextral strike-slip faults,
referred to collectively as the Marlborough Fault System (MFS). The four major faults of the
MFS are, from north to south, the Wairau, Awatere, Clarence and Hope faults. Within the MFS,
the southernmost Hope fault, with a slip rate of ~20-25 mm/yr, accommodates more than half of
the total relative plate motion [Van Dissen, 1989; McMorran, 1991; Van Dissen and Yeats, 1991;
32
Langridge et al., 2003; Stirling et al., 2012; Hatem et al., 2016], with most of the remaining ~15-
20 mm/yr occurring on the other main MFS faults [Van Dissen, 1989; Van Dissen and Yeats,
1991; Holt and Haines, 1995; Walcott, 1998; Wallace et al., 2007, 2012; Litchfield et al., 2014;
Reyners, 2018] . Along the east-central part of the Hope fault, the 65-km-long Conway segment,
the focus of this study, is structurally bounded between the transtensional Hanmer Basin to the
west [Wood et al., 1994] and the transpressional Jordan fold and thrust belt to the east [Van
Dissen, 1989; Van Dissen and Yeats, 1991]. At the northeastern end of the Conway segment, the
Hope fault transfers slip northeastward onto the Jordan thrust system and its northeastward
extension, the Kekerengu fault, which has a dextral slip rate of ~ 25 mm/yr [Van Dissen et al.,
2016]. Farther to the northeast, the Kekerengu fault extends offshore into Cook Strait as the
Needles fault [Barnes and Audru, 1999; Kearse et al., 2017], transferring dextral slip northward
onto the BooBoo and Wairarapa faults, the latter of which has a slip rate of ~11 mm/yr, in
southern North Island [Little et al., 2009; Pondard and Barnes, 2010]. Thus, the Alpine, Hope,
Jordan, Kekerengu, Needles, and Wairarapa faults (Al-Hp-JKN-Wr) constitute a >850-km-long,
through-going system of fast-slip rate (≥ 10 mm/yr) dextral and oblique reverse-dextral faults
that collectively serve to accommodate the majority of Australia-Pacific relative plate motions at
their respective locations. Of the major faults comprising the Al-Hp-JKN-Wr system, only the
Alpine fault and the Conway segment of the Hope fault have not generated a surface-rupturing
earthquake during the historical period, which began with European settlement c.1840 CE.
Whereas a >2,000-year-long paleoearthquake record has been documented for the Alpine
fault [Berryman et al., 2012b; Howarth et al., 2012, 2014, 2016; Clark et al., 2013; Cochran et
al., 2017; see Howarth et al., 2018 for review], the paleoearthquake record of the Conway
segment of the Hope fault is not well-documented beyond the age of the most recent event
33
[Langridge et al., 2003]. In this study, we document and provide age constraints for at least the
five most recent events along the Conway segment. These new data facilitate comparisons of
earthquake occurrence in time and space on other faults within the Al-Hp-JKN-Wr fault system,
providing insight into the system-level behavior of these major plate-boundary faults. Such
comparisons have important implications for understanding seismic hazard in New Zealand, and
more generally, for understanding the spatial and temporal earthquake behavior of similar fault
systems around the world.
3.3 The Green Burn East and Green Burn West study sites
The Green Burn study area is located on the eastern part of the Conway segment of the
Hope fault (Figure 1). Along the Green Burn reach, the Hope fault is generally expressed as a
linear, single fault trace that, along much of this stretch, extends along the northern base of a
sequence of ~4-8-m-high shutter ridges that are located ~50 m to the south of and sub-parallel to
the main, south-facing mountain front (Figure 2). The presence of these shutter ridges,
particularly at our study sites, causes sediment to pond to the north, resulting in fault-parallel
marshes between the fault and the mountain front to the north. The shutter ridges at our trench
sites appear to be long-lived features, as reconstruction of ~200 m of right-lateral Hope fault slip
restores a prominent, NNW-trending reach of a stream that has deeply incised through the shutter
ridge at Green Burn East (see figure S1 available in the electronic supplement to this article for
reconstructions of this offset).
Langridge et al. [2003] conducted paleoseismic investigations on this stretch of the Hope
fault at the Green Burn Stream (GBS) site (-42.395914°, 173.392075°) (Figure 2A). They used
greywacke cobble weathering-rind thickness age estimates (a semi-quantitative geochronometer
34
specific to New Zealand [Knuepfer, 1988]) to suggest that the most recent event (MRE) at their
site occurred c. 1780 ± 60 CE. Additionally, radiocarbon ages show that the penultimate surface
rupture occurred after 1295 CE and before the beginning of the historic era in this part of New
Zealand (c. 1840).
In the current study, we excavated trenches at two localities, one to the east (Green Burn
East [GBE]) (Figure 2B), and another to the west (Green Burn West [GBW]) (Figure 2C, figure
S2 available in the electronic supplement) of the original Langridge et al. [2003] GBS
excavations. We selected the GBE and GBW sites using air photo analyses and reconnaissance
field mapping. These new excavations allow us to extend the paleoearthquake record further
back in time and to place tighter constraints on the timing of late-Holocene Conway segment
Hope fault surface ruptures.
At the GBE site (-42.393212°, 173.405528°), we excavated a 14-m-long, 1.5-m-deep,
fault-perpendicular trench across the Hope fault that extended from the northern slope of the
local shutter ridge/scarp northward into the ponded marshy area to the north (Figure 2B). We
selected GBE as a paleoseismic trench site hypothesizing that during surface ruptures, colluvial
wedges would be shed northward off the scarp and deposited downslope across the surface
rupture trace and into the marsh deposits to the north, possibly with interfingering relationships
between colluvial and organic-rich marsh deposits, dateable by radiocarbon, that would help
refine event ages (Figure 3). The GBE trench was field logged on grid paper at a scale of 1:20.
We also created high-resolution photomosaics of the trench walls using Agisoft Photoscan
photogrammetry software [Bemis et al., 2014], which not only provides an archive of the trench
exposures (see Supplementary Information Figure S2), but also facilitated detailed mapping of
35
the finer-scale features once out of the field; this additional mapping focused on documenting
clast size and distribution within the colluvial wedges observed at the GBE site.
At the GBW site (-42.396560°, 173.388838°), we excavated a 16-m-long, 1.5-m-deep
trench (T-1) that extended northward from the local shutter ridge into a flat, marshy area at the
base of a steep, landslide- prone slope. We also excavated a short (1.7-m-long) 1.5-m-deep
trench (T-2) located ~25 m north of the fault at the base of the steep, landslide-prone slope, about
7 m NNW of the northern end of the T-1 trench (Figure 2C). The GBW T-1 trench was designed
to capture colluvial wedge deposition along with primary surface rupture indicators within the
fault zone, as well as any possible long-runout landslides exposed in the northern end of the
trench. GBW T-2 pit was excavated closer to the base of the landslide-prone slope to intercept
more proximal paleo-landslide events that might have been shed off the mountain front during
Hope fault surface rupturing earthquakes. Supplementary Figure 1 shows the shutter ridge to the
south (left in the image) and landslide-prone slope of the main south-facing mountain front (right
in the image). Previous reconnaissance mapping and trenches [e.g., Langridge et al., 2003]
demonstrated that much of the Green Burn reach of the Conway segment is affected by
moderate- to shallow-seated landsliding, where the mountain front can collapse toward the fault
zone. Such co-seismic, fault-controlled, landsliding was a common feature in northeastern South
Island during to the 2016 Mw =7.8 Kaikōura earthquake [Langridge et al., 2018; Massey et al.,
2018]. As with the GBE trench, we mapped the GBW trench exposures in detail, creating a 1:20
scale log of T-1 and a graphic strat column of T-2 in the field, and also created high-resolution
photomosaics of the trench walls (Supplementary Information Figure 4).
3.4 Trench Results
3.4.1 Green Burn East (GBE) trench observations
36
The northern part of the GBE trench revealed a sequence of organic-rich silts, clays, and
layers of compressed grasses/plants that we interpret as having been deposited in a marsh
environment (M units) (Figure 3). In the southern part of the trench, clastic sediments comprising
the scarp (CW units) overly basal silts and clays (B units) (Figure 3). The upper part of the scarp
sequence consists of a series of pebble to large cobble gravel-rich units interpreted to be colluvial
wedges, referred to as CW1 (youngest) to CW5 (oldest),as well as an older, potential colluvial
wedge (CW6) that could not be confidently attributed to a specific paleoearthquake at GBE.
These wedges were shed northward down the slope of the scarp from the main exposures of the
Hope fault zone near the southern end of the trench. The colluvial gravel clasts, which are
generally set within a silt to medium-grained sand matrix, consist almost exclusively of Torlesse
greywacke typical of bedrock exposures in this part of New Zealand [Rattenbury et al., 2006].
The gravel clasts are typically sub-angular to sub-rounded and range in size from an average of
~5-11 cm to a maximum diameter of 20 cm. The colluvial gravels were derived from older
alluvial gravels that locally mantle the shutter ridge scarp a few meters above and south of the
southern end of the GBE trench. The colluvial wedges were differentiated from each other on the
basis of variations in predominant clast size, weathering, abundance and type of matrix, and, in
several instances, the presence of a prominent basal cobble layers that we interpret as mantling
the ground surface at the time of deposition of each wedge. In general, the overall clast size and
density in the colluvial wedges increase downslope, reaching a maximum at their distal ends
adjacent to the marsh deposits. The distal, northernmost toes of some of the colluvial wedges
locally interfinger with marsh deposits (M units – described below) beneath the southern toe of
the scarp. Table S1 (available in the electronic supplement to this article) provides all
stratigraphic descriptions made in the field.
37
Beneath the colluvial wedges, the scarp is composed of highly sheared, locally highly
indurated clay to clayey-silt units (B units). These basal clays are virtually clast free and
typically massive, with limited discernible internal bedding. We differentiated three distinct units
(B1 [youngest] to B3 [oldest]) on the basis of clay to silt ratio, color, and degree of induration.
B3, the most sheared and indurated of these units, may be deformed Torlesse bedrock. All of the
scarp-derived colluvial wedges were deposited atop the basal clays; the base of the marsh (M)
units in the northern part of the trench was not reached, and B units were not exposed north of ~
m 4 (Figure 3).
The generally massive marsh (M) units in the northern part of the trench consist of
organic-rich silts, peats, and compressed marsh plant layers. In general, the marsh deposits dip
gently to the south (i.e., towards the scarp), consistent with a long-term, minor, down-to-the-
north (i.e., mountain side down) component of vertical motion along the predominatly dextral
Hope fault exposed in the scarp. The marsh deposits contained individual seeds, grass blades in
growth formation, and plant leaves/fronds and other macroflora indicative of in situ deposition
within the marsh, as well as detrital charcoal and wood fragments. The marsh deposits were
generally clast free, and showed no obvious sedimentary structures. Stratigraphic delineations
were made on the basis of color, wetting characteristics, firmness, silt content, and the presence
or absence of plant material.
Although some of the scarp-derived colluvial wedges do locally interfinger with the
southern ends of the marsh strata, especially near the top and base of the trench (e.g., CW2 and
M1; CW5 and M5-6), in general the northern marsh deposits and the southern scarp-derived
deposits are separated into distinct stratigraphic and depositional sequences by a ~1.5- to 2-m-
wide zone (~m 5–m 7 on the trench logs) of complex stratigraphy associated with a wood mass
38
that may be a paleo-tree or trees that were either growing at the base of the scarp or fell along the
base of the scarp (Figure 3). We could not correlate stratigraphic units across this wood-rich
zone, and as we discuss below, upon dating the northern marsh section using apparently in situ
seeds, leaves, and other plant material, we observed a large mismatch in age between the ages of
the marsh units north of the wood-rich zone and the younger colluvial and marshy deposits to the
south at similar depths (see Age Control section). Because we cannot correlate units either
stratigraphically or chronologically across this mass of wood, we do not utilize the northern
marsh stratigraphy in the age determinations of scarp-preserved events (GB1-GB5), as all of the
sedimentological and structural evidence for the five most-recent paleo-surface rupture at GBE
comes from the southern part of the trench, south of the wood-rich zone. We present the detailed
logs for the southern eight meters of the trench in Figure 3 to highlight these relationships. North
of the section of the trench exposure shown in Figure 3, the marsh units became massive and
increasingly difficult to log with any certainty. Photomosaics of the full trench exposures of GBE
are presented in S3, available in the electronic supplement to this article.
The Hope fault through the GBE trench is expressed as a 5-m-wide zone comprising five
main fault strands, denoted as F1 (farthest south) to F5 (farthest north) between m 0 and m 5.
The faults extend upward through the basal units and locally extend through (or are overlain by)
the colluvial wedge gravels. Fault F1 dips steeply to the north, F2 is near vertical, F3 generally
dips steeply to the south, F4 dips variably northward, and F5 dips more shallowly to the north.
Most of these main faults exposed at the base of the trench splay upwards into subsidiary strands.
The steeply north-dipping, southernmost, scarp-bounding fault (F1), separates the pervasively
sheared, highly indurated local clay bedrock (B3) from the somewhat less-indurated basal
clayey-silt layers B1-2, along with the colluvial wedges that overlie B1-3. Several of the Hope
39
fault strands, particularly F1-F3, exhibit large, upward-opening fissure fills and local graben-like
down-dropped blocks. In addition, basal unit B2 is locally tightly folded between strands F1 and
F2, best observed on the east wall. The fault zone between meters 1.5 and 3.5, encompassing
strands F2-F5, represents a wide, somewhat distributed shear zone, in which stratigraphy is less
well expressed in the basal units which exhibit a locally pervasive, steeply dipping shear fabric,
interpreted here as being indicative of relatively distributed shearing.
3.4.2 Evidence for paleo-surface ruptures
In the following, we describe structural and stratigraphic evidence for the five most-
recent surface ruptures (events GB1-GB5, from youngest to oldest), as well as sedimentological
evidence for a possible older, sixth event, (GB6).
Event 1 (GB1)
Event GB1, the most recent event (MRE) observed in the GBE trench, is marked by both
the deposition of colluvial wedge CW1 and the upward terminations of faults F1B and F1C at the
base of CW1 on the eastern wall. Most of the strike-slip in event GB1 likely occurred along
fault F1a, but this fault does not directly interact with CW1; instead, fault F1a terminates at the
base of the A horizon on the East Wall near m 0. Faults F1b and F1c form a small graben into
which CW1 was deposited either cosesimically or soon after slip in GB1. Another small graben
was formed between a potential additional splay fault near m 0 and fault F1b. Both of these
grabens were filled with by CW1 gravels. Colluvial wedge CW1 is a pebbly gravel, with a black
to dark brownish-gray silt to medium-grained sand matrix, which is overlain by the generally
clast-free, thin, active surface soil A horizon. The unit extends downslope from between m 0 and
1 to the base of the scarp at m 5, where it interfingers marsh unit M1 (Figure 3).
Event 2 (GB2)
40
Event GB2, the penultimate event observed in GBE trench, is recorded by deposition of
colluvial wedge unit CW2 and upward termination of fault F4 at m 2.8 on west wall at or near
the base of CW2 (Figure 3). Unit CW2 is a clast-supported, pebble (clast size 0.5-2 cm) gravel
consisting of sub-rounded graywacke clasts in a gray to black silt to fine-grained sand matrix.
CW2 is markedly different from overlying CW1, and is distinguished from CW1 by its greater
clast content. Additionally, CW2 is marked by a predominance of distinctive orange, highly
weathered, friable clasts that were not observed in any other unit in the GBE trench. These
distinctive clasts also had a higher sand content than the generally finer-grained Torlesse
greywacke clasts found in all of the other colluvial wedges we observed. The presence of these
distinctive clasts in CW2, and their absence in all other scarp-derived colluvial gravels in the
GBE trench, suggests that a small exposure of this source rock was first exposed in the fault
scarp during event GB2, and was moved right laterally away from the trench locality by strike
slip during GB1, leaving only typical older gray-colored Torlesse pebble and cobble alluvium
exposed along the top of the shutter ridge above the trench during the MRE. In contrast to
colluvial wedge CW1, colluvial wedge CW2 is only exposed on the lower, distal (northern) part
of the slope. In addition to the presence of the distinctive colluvial wedge, event GB2 is marked
by the presence of a small fault block with CW2 colluvium associated with the upward
termination of fault F4 (splays 4a and 4b), best expressed at ~ m 2.8 on the eastern wall (Figure
3).
Event 3 (GB3)
Event GB3 is marked by not only the deposition of colluvial wedge CW3, but also the
opening of a large fissure between faults F1a and F1b that was filled with CW3 colluvium at the
south end of the trench between ~ m 0 and 1, and folding and faulting of older units (B2, CW4,
41
CW4a) onto which colluvial wedge CW3 was deposited. Unit CW3 is a clast-supported gravel
characterized by numerous large (8-10 cm), sub-angular cobbles within a gray to pale brown,
medium-grained sand to clay matrix. Unit CW3 does not contain any of the sandy orange clasts
that characterize CW2. In the northern, downslope extent of CW3, we differentiate a subunit
(CW3a), which is similar to CW3 in matrix composition but is relatively clast poor. Unit CW3 is
the largest colluvial wedge observed in this trench, and was sourced from a scarp created during
slip on the well-defined, southernmost fault F1. A cobble layer extends along and defines the
base of CW3. Unit CW3 is not continuous along the length of the trench exposure, pinching out
between m 2 and m 4, where colluvial wedge CW2 was deposited directly on underlying unit
CW4 (Figure 3). This preserved geometry of CW3 may be due to discontinuous lateral
deposition of CW3, and then subsequent strike-slip of CW3 along fault F4 in event GB2, which
could juxtapose different portions of the CW3 deposit, therefore yielding the observed pinch-out
of CW3. At the base of the scarp, the distal part of unit CW3 locally interfingers with marsh
deposits M1-M3 near m 5. CW3 gravel also fills a small fissure formed by faults 2c and 2d
terminating at the top of unit CW4/4a near m 1.8 on the east wall. Additionally, folding of units
CW4 and CW4a between ~ m 0.5 and 1.5 likely also occurred during surface rupture GB3.
Collectively, these observations underscore the extensive structural disruption and the large
volume of the CW3 colluvial wedge relative to the other colluvial wedges observed at GBE.
Event 4 (GB4)
Event GB4 is marked by the deposition of colluvial wedge CW4, by infilling of fissures
that opened in the GB4 surface rupture with CW4 cobble colluvium, and by upward termination
of fault F2a at the base of the CW4 colluvial wedge. Unit CW4 is a clast-supported colluvial
gravel, with a dark brown to gray, medium-grained sand to silt matrix among sub-angular to
42
angular pebble to cobble clasts (max 12 cm). In the southern, upslope part of CW4, we
differentiate a subunit (CW4a) that is generally similar to CW4, with a similar matrix, but which
has far fewer cobbles and is slightly paler in color.
The distal, downslope end of unit CW4 terminates against marsh units M4 and M5, as
best observed on the east wall. Although in the eastern wall exposure this contact suggests minor
interfingering between the CW4 wedge and M3/5, on the western wall the contact is marked by a
near-vertical stone line where the colluvial wedge is juxtaposed with the marsh deposits. This
relationship is markedly different than the distal end of underlying unit CW5 (described below),
which extends farther out into the marsh.
Unit CW4 is folded and faulted in multiple places, and several fissures opened in this
event. Specifically, faulting from event GB4 opened fissures near m 1.3 and m 2.6 that were
filled with CW4 colluvium. The large fissure that opened along fault F2 at m 1.3 has an
accumulation of large cobbles (max 20 cm) exposed near the base of the fissure fill on the
eastern wall, consistent with filling of an open cavity. Similarly, the smaller fissure that opened
between faults 3c and 3d at ~m 2.6 also has larger clasts near the base of the fissure, although
these clasts were smaller (large pebbles) than the clasts at the base of the m 1.3 fissure fill. In
both fissure fills, the clasts exhibited sub-vertical alignments sub-parallel to the fault-formed
free-faces that once bounded the fissures.
Event 5 (GB5):
Event GB5 is marked by the deposition of the colluvial wedge CW5, which is a clast-
supported pebble to cobble gravel with a medium brown, fine-grained sand to silt matrix, as well
as pervasive shearing of CW5 not affecting younger units (e.g., CW4). This distributed shear
zone is preserved at ~m 2.5 - 3.5 on the eastern wall, where the stratigraphy becomes difficult to
43
differentiate between CW5 and undifferentiated sheared silty gravels. The shape of unit CW5 is
similar to the overlying colluvial wedges, consistent with these gravels having been shed
northward down the scarp and out into the marsh. Unit CW5 appears to have been shed off the
northernmost fault zone (F4 on the eastern wall), and cannot be traced further southward towards
the top of the scarp. At its distal end, the CW5 colluvial wedge is deposited on top of thinly
bedded, organic-rich rich silt unit M6. This is particularly clear on the east wall of the trench,
where the CW5 gravel extends northward beneath the peat-like, compressed grasses that make
up unit M5, and overlies the older marsh unit M6. These relationships indicate that deposition of
CW5 pre-dates deposition of unit M5 and post-dates deposition of M6.
3.4.3 Green Burn West (GBW) trench observations: Evidence for inferred landslides
At the Green Burn West (GBW) site 1.4 km west of the GBE trench, fault-perpendicular
trench GBW T-1 extended 16 m from the lower part of the north-facing fault scarp northward
across a marshy flat to near the base of the steep, south-facing mountain front (Figure 4, Figure
S4, available in the electronic supplement to this article). The trench exposed a gently north-
dipping fault with an ~10 cm-thick gouge zone (fault F1 on Figure 4) that juxtaposes basal
sheared pale-gray siltstone bedrock (unit B1) against overlying moderately indurated, massive
silt to gravelly silt (unit S1), which becomes progressively more clast-rich to the south. We did
not observe any structural evidence of individual surface ruptures in this trench, although the
southern end of the trench is marked by two gravel colluvial wedges (units CWa and CWb)
composed of sub-angular to sub-rounded pebbles in a sandy silt, dark gray matrix that overlie ~1
meter of massive gravelly silt. Neither of these colluvial gravels was observed in relationship to
the fault, so we therefore cannot directly attribute surface rupturing events to these colluvial
44
wedges. As in the GBE trench, we interpret these colluvial wedges to have been sourced from
the scarp at the south end of the trench, although this relationship was not exposed in GBW T-1.
We did not recover any datable material from these colluvial gravels.
Trench GBW T-1 also exposed a tan-to-orange mottled, matrix-supported gravelly silt
with local minor sub-rounded to sub-angular pebbles and rare cobbles (unit L1). This gravel
overlies a buried, organic-rich paleosol A horizon (unit P1). The paleosol is a dark brown to
black, organic-rich, sandy silt, similar to other marshy soils we observed at GBE. The gravelly,
thinning southward towards the fault, is consistent with deposition in a landslide originating at
the steep bedrock slope to the north of the trench, with the distal end of this potential
paleolandslide pinching out between m 10 and m 13, <10 m south of the base of the steep south-
facing mountain front to the north (Figure 2C; Supplementary Figure 2). Consistent with this
interpretation, the basal depositional contact of this paleolandslide atop an organic rich paleosol
is extremely sharp and undulatory over short (<10’s of cm) wavelength, with evidence of rip-up
of the basal paleosol, indicating potentially high energy, in, for instance, a landslide event. The
underlying paleosol is flat, indicating that the gravel is not a tilted channel. Overlying this
potential paleolandslide landslide deposit is a matrix-supported gravelly silt (unit G1) that has
considerably fewer clasts than the inferred landslide deposit. Additionally, the contact between
the interpreted landslide and the overlying deposit is diffuse in some places (e.g., m 13 & m 14
denoted with the number 5 on the contact, indicating the contact is diffuse over a width of 5 cm).
The shorter GBW T-2 trench was excavated closer to the base of the steep, S-facing
bedrock slope, ~ 8 m northwest of the northern end of GBW T-1 (Figure 2C). Trench GBW T-2
revealed four gravel layers that we interpret as paleo-landslides (units L1, L2, L3 and L4, from
youngest to oldest) (Figure 5, Figure S5 available in the electronic supplement to this article).
45
These matrix-supported gravels consist of sub-angular pebbles (1-8 cm diameter clasts), that are
separated from one another by organic-rich, clast-free, silty buried paleosols (P1, P2 and P3). L1
is the thinnest gravelly silt deposit (5-10 cm thick), with small sub-angular clasts (~1 cm) within
an orange sandy silt matrix. The upper contact is gradational whereas the basal contact is locally
sharp and locally gradational, potentially due to bioturbation. The L2 deposit was ~ 15 cm thick
with a similar makeup to L1, but containing slightly larger (rare) clasts and with areas of local
reducing matrix as evinced by blue coloration of the deposit. The basal contact of L2 against the
underlying P2 paleosol is sharp along the whole contact, whereas the upper contact between L2
and overlying P1 is gradational. This observation supports our interpretation that L2 was
deposited rapidly on top of P3, and that P2 gradually accumulated atop L2 over a longer period
of time. Although L3 is a gravelly silt, this deposit is relatively clast poor compared to L1, L2
and L4. The upper and lower contacts of L3 are both diffuse, potentially due to relict
bioturbation when this deposit was near the active A horizon. L4 is the most clast rich of the four
gravelly silts described in GBW T-2. We could not expose the base of L4 due to the shallow
groundwater table at this site. The top contact of L4 against the overlying P3 paleosol is also
diffuse, again perhaps due to relict bioturbation. This alternation in GBW T-2 between coarse-
grained gravelly silt deposition and fine-grained deposition punctuated by periods of soil
development is consistent with episodic deposition of the gravelly silts, potentially during
landslides, with intervening periods of organic-rich silt accumulation and pedogenesis. We
observe three full cycles of this behavior, ending at the development of the modern marshy
organic-rich soil exposed at the surface.
Because the exposure of T-2 is considerably smaller than T-1, we could not observe
changes in lateral thickness of the unit. But, the presence of these gravels at the base of a steep,
46
bedrock slope that has experienced multiple landslides as observed by the multiple landslide
scars above T-2, is consistent with their deposition as paleolandslides. In the following section,
we discuss the rationale behind our suggestion that each of the four landslides in this trench
records co-seismic landslide deposition during the four most recent large, surface-rupturing Hope
fault earthquakes on the Conway segment along the Green Burn reach, and that the intervening
paleosols represent the periods of soil development between major Conway segment surface
ruptures. As we describe below, this interpretation is supported by the similar, but completely
independent ages we determined for the past three events at GBW and GBE.
3.4.4 Coseismic origin of colluvial wedges and landslides observed at GBW/GBE: Observations
of the Green Burn Reach following the 2016 Mw=7.8 Kaikōura earthquake
Our inference that the GBW landslides and the GBE colluvial wedges were only
deposited during prior Hope fault surface ruptures is supported by our field observations of the
Green Burn reach of the Hope fault following the 2016 Mw=7.8 Kaikōura earthquake, which
occurred nine months after our trench studies were conducted. We visited our by-then backfilled
trenches at both the GBE and GBW sites during our reconnaissance mapping following the
earthquake to investigate whether any colluvium had accumulated at either the base of the steep
slope we trenched at the GBW site or the fault scarp we trenched at the GBE site. Both of our
trench sites experienced very strong ground motions during the 2016 earthquake. Specifically,
the Green Burn trench sites likely experienced a shaking intensity of VII-VIII on the Modified
Mercali Intensity scale, with peak ground accelerations ~ 25% g and peak ground velocities of
~45 cm/s (KIKS station) Despite the strong shaking that affected our GBE and GBW sites, we
observed no colluviation in either location, or anywhere else along the Green Burn reach of the
47
fault. At the GBE site, our filled-in trench was found intact beneath a newly sprouted cover of
grass. Similarly, we observed no landsliding at the GBW site, with the filled-in GBW T-1 and T-
2 trenches undisturbed and already re-vegetated by grasses and thistle. The only colluviation
observed along the Green Burn reach were small slope failures along a dirt roadcut, and several
small slides on steep stream banks.
During our reconnaissance mapping of the Hope fault following the Kaikōura event, we
noted no definitive surface rupture along the Green Burn reach, in keeping with more extensive
field and helicopter mapping of the entire Conway segment [Litchfield et al., 2018]. In
subsequent weeks of mapping, one area of potentially disturbed ground on a slope 4 km east of
the GBE site suggested local ground cracking of 0.9-1.4 m of net (reverse-dextral) slip
[Litchfield et al., 2018], but most of the Conway segment did not experience any surface rupture.
Similarly, we did not observe any newly exposed, un-vegetated slopes. The south-facing
slopes north of GBE and GBW are both vegetated. In fact, the slope north of T-1 and T-2 at
GBW has trees growing on it, and has not been disturbed for at least 50 years or more given the
size of the trees, indicating that slope failures at this site are rare and not events typically
triggered by rainfall. The creation of a scarp free-face during surface rupture, including extreme
peak ground accelerations, on the Conway segment of the Hope fault thus appears to be
necessary for colluviation or landsliding along the Green Burn reach. In addition to the
supporting sedimentologic and geomorphic arguments indicating that these colluvial wedge and
paleolandslide deposits likely originated during Conway segment surface rupturing earthquakes,
we now present compelling age data showing that these deposition events are essentially coeval.
Such age results indicate that GBW paleolandslide age ranges can be used to help constrain the
timing of paleo-surface ruptures observed at the GBE trench.
48
3.5 Chronology of paleoseismic events observed at GBE and GBW
3.5.1 Age models and event boundary conditions
To provide age control on event horizons observed in the GBE and GBW trenches, we
radiocarbon dated 53 samples, consisting of detrital charcoal, wood, seeds, and plant material
(Table 1). The samples were inspected under a microscope to ensure that no young roots were
included, and individual organic fragments including leaves and seeds were used to date marsh
samples. All samples were prepared with a standard acid-base-acid pre-treatment protocol, and
analyzed at the W.M. Keck accelerator mass spectrometer (AMS) lab at the University of
California, Irvine. The resulting radiocarbon ages were calibrated using OxCal 4.3.2 [Bronk
Ramsey, 2017] and the most up-to-date southern hemisphere calibration curve, SHCal 13 [Hogg
et al., 2013]. We observed that many of these samples were reworked. We created our age model
using the philosophy that detritral charcoal included in a colluvial wedge must be older than the
depositional age of the colluvial wedge itself. Therefore, to get as close as possible to the true
deposition age of a colluvial wedge, we select the youngest detrital charcoal age from a given
colluvial wedge and discard the older, reworked ages. The only way that detrital charcoal
included in a colluvial wedge could be younger than the depositional age of the wedge itself is if
the charcoal was bioturbated into its stratigraphic position. That situation is unlikely along the
Conway segment at GBE because there is a lack of burrowing organisms in New Zealand,
especially those who could bioturbate coarse gravels of colluvial wedges. After completing this
analysis, the remaining ages were then used as inputs to stratigraphic ordering models in OxCal
to create a Bayseian age model for each exposure [Lienkaemper and Bronk Ramsey, 2009]. All
ages reported herein are calibrated, calendric ages in terms of BCE/CE.
49
We constructed six age models to provide timing constraints on the paleo-earthquakes at
our Green Burn study sites (Figure 6; Table 1): one using ages only from the GBE trench scarp-
derived colluvium (Figure 6A); one using ages only from the GBE scarp-derived colluvium and
the GBE marsh deposits south of the wood mass (Figure 6B); one using ages only from the GBE
marsh deposits north of the wood mass (Figure 6C); one using ages from only GBW T-2 (Figure
6D); one using ages from GBW T-1 and T-2 (Figure 6E); and a final, preferred model combining
age constraints from the GBE scarp-derived colluvium and southern marsh section, as well as
GBW T-1 and T-2 (Figure 6F). We excluded from consideration in our age models all samples
that exhibited anomalously old ages indicative of inheritance (i.e., those samples with ages that
are much older than underlying samples). In addition to the exclusion of a number of samples
from the faulted, southern part of the GBE trench, we did not use the vertical profile of
radiocarbon ages we collected from the marsh deposits in the northern part of the GBE trench
north of the wood mass near m 5-6 (Figure 6C), as these ages were all significantly older at all
stratigraphic levels than correlative scarp-derived colluvial deposits (Figure S6, available in the
electronic supplement to this article). Moreover, as noted above, none of these deposits can be
correlated confidently with the scarp-derived colluvial section that contains all of our
stratigraphic and structural evidence for the five most recent earthquakes recorded at GBE
(Figure 3; Figure 6A-C; Figure S6, available in the electronic supplement to this article).
Complete documentation of all radiocarbon age data and associated metadata from the GBE and
GBW trenches, including those ages that were not included in our age models, is presented in the
Table 1.
We present the results of our GBE-only age models first (Figure 6A-C), with detailed
reference to all radiocarbon ages that were used to directly constrain the five well-constrained
50
surface ruptures observed in the GBE trench. Following this discussion, we present our GBW
age models (Figure 6D) and then the combined GBW and GBE model (Figure 6E), which uses
the additional age constraints from the GBW landslide deposits to independently test and
corroborate the ages of the GBE surface ruptures, and to more tightly constrain the age of the
penultimate surface rupture (GB2) observed in the GBE trench. All 2 σ event age ranges for each
model are listed in Table 2.
3.5.2 Green Burn East Paleo-Surface Rupture Ages
All of the paleoearthquake event stratigraphy recorded in the GBE trench (i.e., fault
terminations, folding, fissure fills) is contained within the scarp-derived colluvial units in the
southernmost portion of the trench. In the following section, we present event ages based on
samples collected only from the units that record the events (Figure 6A). Where possible (e.g.,
sample SF-5), we further constrain these event age ranges using samples from the marsh south of
the wood mass at ~ m 5 (Figure 6B). We report individual sample ages as calibrated yet
unmodeled ages, which are also included (for all dated samples) Table 1.
In contrast to evidence for historical surface rupture of the Hope fault farther west during
the 1888 Mw~7-7.3 Amuri earthquake [McKay, 1890; Cowan, 1991; Khajavi et al., 2016], there
is no record of historical rupture of the Conway segment of the Hope fault through the Green
Burn sites. Thus, the most recent surface rupture we observe (GB1) must have occurred prior to
European settlement, which began c. 1840 CE. Surface rupture GB1 is younger than detrital
charcoal samples SF-33 (1680—1723 CE, or post-1802 CE) and SF-34 (1691—1728 CE, or
post-1805 CE) included in colluvial wedge CW1, which we interpret as having been shed
northward off the scarp during and soon after event GB1. Combining the historical constraint
51
with the age constraints from the GBE scarp-derived, colluvial wedge ages-only OxCal model
indicates that event GB1 occurred between 1722—1840 CE. This age range is similar to the
1780 ± 60 yBP age of the most recent event suggested by Langridge et al., [2003] on the basis of
weathering rind age estimates from their GBS trench.
Surface rupture GB2 is younger than the ages of the detrital charcoal samples collected
from CW2 itself, as the material contained in the colluvial wedge must have existed higher on
the scarp prior to deposition of the colluvial wedge. Thus, event GB2 is younger than samples
SF-15 (1462—1627 CE) and SF-16 (1505—1643 CE), which were collected from within unit
CW2. GB2 is stratigraphically older than CW1, and we make the assumption that this surface
rupture occurred before the material included in CW1 was generated and then included in CW1.
Thus, the ages of samples SF-33 and SF-34, collected from unit CW1, provide a minimum age
for GB2. Taken together, our GBE scarp-derived colluvial samples-only age model indicates that
the penultimate GBE surface rupture occurred between 1558 and 1724 CE. These data
significantly narrow the previous post-1295 CE constraint for the occurrence of GB2 from
Langridge et al., [2003].
We constrain the age of GB3 using a similar rationale as used for dating the previous
events. Specifically, charcoal samples SF-1 (1394—1425 CE), SF-2 (1496—1636 CE), SF-3
(1320—1410 CE) and SF-28 (1396—1436 CE) were collected from the CW3 colluvial wedge
and therefore pre-date event GB3. However, sample SF-2 is significantly younger than SF-1, SF-
3 & SF-28, indicating that the three older charcoal samples were likely incorporated into the soil
and gravel mantle atop the shutter ridge/fault scarp about 100-150 years prior to incorporation of
the SF-2 charcoal sample, and/or that these three samples were significantly older than SF-2
when they were all incorporated into the CW3 colluvial wedge. We infer that the three older
52
charcoal samples were generated during an earlier brush fire (or fires) that occurred c. 1400 CE,
whereas the sample SF-2 charcoal fragment was produced during a separate, younger brush fire
during the late 1400s or 1500s CE. Given the apparent inheritance of samples SF-1, SF-3 & SF-
28, we use the age of sample SF-2 as a maximum age for GB3. Charcoal samples SF-15 and SF-
16, which were collected from the overlying colluvial wedge CW2, post-date GB3, as CW2 is
deposited atop CW3. Using these constraints, the GBE scarp-derived colluvial sample-only age
model (Figure 6A) produces an age range of GB3 as 1495—1610 CE.
As with the previous events, assuming material from within a colluvial wedge is older
than the coseismic deposition of that wedge itself indicates that event GB4 is younger than
charcoal sample SF-41 (1273-1380 CE), which was collected from CW4. Using sample SF-41 as
a maximum age of GB4 with sample SF-2 from CW3 as minimum age, we model the age of
GB4 as 1288—1532 CE.
To determine the maximum age for event GB5, we assume as in the case of the younger
colluvial wedges that charcoal samples collected from the colluvial wedge are older than the
coseismic deposition of that wedge. We therefore use sample SF-21 (195—52 BCE), collected
from CW5, as a maximum age for event GB5. Knowing that CW4 was deposited atop CW5, and
is therefore younger than CW5, we use sample SF-41 collected from CW4 as a minimum age
constraint on event GB5. Using these two scarp-derived charcoal samples as constraints results
in a modeled event range for event GB5 of 61 BCE—1277 CE.
We can further constrain the age of event GB5 by using charcoal ages collected from the
coseismic colluvial wedge in combination with ages on plant material collected from the
southern part of the marsh section, south of the wood mass, atop which the wedge was deposited.
Specifically, the distal, downslope toe of colluvial wedge CW5 was deposited onto a thinly
53
bedded organic silt/peat succession (units M6-M7), best observed at ~m 5 on the east wall
(Figure 3). We collected wood sample SF-5 (99 BCE-114 CE) from peat layer M7, underlying
CW5. We sub-sampled SF-5 as three separate pieces, and these yielded similar radiocarbon ages
on all splits (SF-5a: 42 BCE—115 CE; SF-5b: 96 BCE—25 CE; SF-5c: 99 BCE—23 CE).
Because sample SF-5 was a piece of wood, with an unknown amount of age inheritance, we can
only use these ages as maximum ages because the wood deposited within the layer could
potentially be much older than the deposit itself. Combining the ages of samples both from the
scarp-derived colluvial deposits and from the marsh units south of the wood-rich section yields a
modeled age range for event GB5 of 36 BCE—1240 CE.
3.5.3 Age control for GBE northern marsh strata
The ages determined from the vertical sampling profile we collected north of the wood
mass at ~m 7 (Figure 3 east wall), are presented in Figure 6C. These 11 ages are primarily based
on short-lived plant material, mainly leaves and seeds. The resulting ages and OxCal age model
(Figure 6C) indicate that all samples are in correct stratigraphic order, recording semi-continuous
deposition in the northern marsh from c.100-300 BCE at 1.5 m depth (sample SF-14 [350—104
BCE]) to c. 700-800 CE at 30 cm depth (sample SF-6 [681—862 CE]). Interestingly, these ages
are significantly older at all depths relative to the scarp-derived colluvial section south of the
wood-rich zone. Moreover, the fact that c. 1200- to 1300-year-old strata are exposed at only ~30
cm depth in the marsh north of the wood mass suggests that either there has been little deposition
in the northern marsh over the past 1,000-plus years, and/or that the northern marsh section has
experienced significant erosion during the same time period when the colluvial wedges marking
that five most recent Hope fault surface ruptures were being deposited south of the wood mass.
54
This mismatch in ages suggests that the mass of wood acted as a barrier to sediment
accumulation, effectively separating the southern, scarp-derived colluvial section from the
northern marsh section for much of the time recorded in the GBE trench (Figure S5).
Consequently, we cannot use the radiocarbon dates from the northern marsh section to constrain
the ages of paleo-earthquakes, evidence for which is derived exclusively from the scarp-related
section south of the wood-rich zone. Rather, we use only those radiocarbon ages collected from
the scarp-derived, southern section to constrain the ages of the five most recent Green Burn
surface ruptures.
3.5.4 Ages of GBW landslides
The radiocarbon ages from GBW T-2 provide constraints on the ages of the four influxes
of clastic sediment that we interpret as paleo-landslides observed at that site (Figure 6D). We
collected two radiocarbon samples from the shallowest paleosol (P1) beneath the shallowest
gravelly silt L1, a plant leaf sample from ~23 cm depth [LS2-6 (Modern)], and another plant leaf
from ~65 cm depth [LS2-4 (1672—1743 CE, or post 1772 CE)]. Given the shallowness of
sample LS2-6, and lack of clast-supported gravel deposits observed at GBE, we suspect that this
sample may have been bioturbated into position from which it was collected. We therefore use
sample LS2-4 to provide a maximum age for the overlying L1 interpreted landslide deposit.
Alternatively, if this Modern plant sample LS2-6 from beneath the L1 gravelly silt was not
bioturbated in to, then the L1 deposit must be historical, likely mid-to-late 20
th
century following
the production of bomb-generated radiocarbon testing in 1945 CE. This alternative explanation
for the timing of L1 deposition makes no difference in the interpretation of the sample LS2-4, as
55
that sample still post-dates inferred-landslide L2, which must pre-date any subsequent landslide
following L2.
We bracket the timing of deposition of inferred-landslide L2, we use the age of sample
LS2-4 from paleosol 1 above the penultimate landslide L2 to post-date the L2 deposit, as well as
the age of sample LS2-5 (1665—1895 CE) from paleosol 2 beneath L2. These ages indicate that
the landslide occurred between 1668 and 1806 CE. The two wood samples that we radiocarbon
dated from within the L3 landslide (LS2-2 [693-891 CE] and LS2-11 [1032-1151 CE] are much
older than underlying samples, and are not considered further. The age of L3 is, however,
constrained by charcoal samples LS2-5, collected from paleosol P2 above the L3 deposit, and
LS2-9 (1184—1267 CE), collected from paleosol P3 beneath L3. These ages bracket the timing
of L3 deposition to 1225-1685 CE. We can further refine this age range by incorporating
charcoal sample HL16-04 (1400-1439 CE) from GBW trench T-1, which was collected from the
paleosol that was over-ridden by perhaps the only landslide observed in that trench, which we
correlate with L3 based on the age correlation of L3 with GB3 (discussed in the subsequent
section) (Figure 6D). This additional constraint narrows the age range for L3 to 1414-1694 CE.
Using sample HL16-04 to pre-date L3 deposition and LS2-5 to post-date L3 deposition, we
arrive at a revised age range for L3 deposition as 1415—1711 CE. We were unable to collect any
samples from beneath the fourth landslide back (L4), but the age of charcoal sample HL16-04
from paleosol directly below the inferred landslide deposit in T-1, as well as the age of sample
LS2-9 collected from P3 in T-2 indicates that L4 was deposited before 1400—1439 CE,
providing the youngest possible age for event GB4.
3.5.5 Combined Age Model for GBW and GBE sites
56
The ages of the five event horizons we identified at GBE based on fault terminations,
fissure fills, folding, and colluvial wedge deposition overlap in time with the deposition of the
four inferred-landslide silty gravel units in GBW T-2 (Supplemental Figure S7). As discussed
earlier, we observed no evidence of colluviation or landsliding along the Green Burn Reach
following the Mw=7.8 Kaikōura earthquake, providing support for our inference that landslide
deposition at GBW occur only during surface rupturing events along the Green Burn Reach of
the Hope fault. We therefore combine age models from GBE and GBW using independent age
constraints to more precisely determine the ages of the five surface rupturing events observed at
GBE. One could arrive at these combined event ages by averaging together the probability
density functions from GBE and GBW, an approach similar to that of Biasi and Weldon [2009].
We present those results, along with comparisons of the GBE and GBW probability density
functions and the OxCal combined age results, in the electronic supplement to this article in
Figure S7.
Using the preferred OxCal age combination approach, inclusion of the 1682—1836 CE
age range of sample LS2-4 from the youngest paleosol P1 at GBW as an additional constraint on
the maximum age of the most recent surface rupture observed in the GBE trench overlaps with
the age of event GB1, and slightly tightens the possible age range from 1722—1840 CE in the
GBE-only age model to 1731-1840 CE. Similarly, addition of sample LS2-5 from paleosol P2 in
the combined GBE-GBW age model narrows the possible age of event GB2 to between 1657
and 1797 CE (1558—1724 CE for GBE-only model). Specifically, inclusion of the ages of GBW
samples LS216-5, collected from the paleoseol P2 above the third-most-recent inferred-landslide
(L3) at GBW, and samples HL16-04 & LS216-9, collected from below L3, in the combined
GBE–GBW age model yields a nearly identical age range for event GB3 of 1496-1611 CE
57
(1495—1610 CE for GBE only model). The two age models produce similar age ranges from
GB3 because the additional sample from GBW T-1 of HL16-04 is slightly older than the sample
SF-2 from GBE (Figure 6B vs 6F). Similarly, the age range of event GB4 is shortened markedly
by incorporating the 1400-1439 CE age of sample HL16-04, which was collected from GBW T-
2 paleosol P3, above the fourth-oldest inferred-landslide L4, which we interpret to be
contemporaneous with deposition of colluvial wedge CW4 in the GBE trench during event GB4.
Although sample LS2-9 (1184—1267 CE) was also collected from paleosol 3 in GBW T-2, we
do not use the LS2-9 radiocarbon date in further age modeling of event GB4 because this sample
is slightly older than sample SF-41, which was collected from CW4 at GBE, and which therefore
must pre-date event GB4. Thus, the older age of sample LS2-9 suggests that this sample had
some inherited, pre-event GB4 age before it was incorporated into the paleosol overlying
inferred-landslide L4. Including the age of sample HL16-04 into the combined GBE-GBW age
model yields a revised age range for event GB4 of 1290-1420 CE, somewhat older than the
1288—1532 CE age range from the GBE-only age model.
3.6 Discussion
The Green Burn trenches reveal the occurrence of five surface rupturing earthquakes on
the Conway segment of the Hope fault during the past c. 2000 years. The more tightly
constrained ages for the past four GBE events suggest potentially irregular earthquake
occurrence. Specifically, whereas the two most recent events (GB1—GB2) occurred within a
relatively brief, <183-year period between 1657 and 1840 CE (mean RI between GB1 and GB2
= 58 years), they were preceded by events GB3 and GB4, which occurred over a maximum of
321 years from 1290-1611 CE (mean RI between GB3 and GB4 = 198 years). The oldest event,
58
GB5, has a long age range, therefore making the recurrence interval calculated less informative
than the younger events; although we did not observe evidence of events between GB4 and GB5,
or events older than GB5 in the GBE trench, we may have an incomplete event record prior to
GB4. Event GB5 aside, the younger two events may thus represent a temporal cluster during
which earthquake recurrence was more frequent than average. Interestingly, event GB3, the third
earthquake back, which precedes these two events, exhibited much more significant structural
disruption in the GBE trench and resulted in deposition of a much more extensive colluvial
wedge than previous or more recent Conway segment surface ruptures, suggesting that it may
have been a larger-displacement surface rupture at the GBE site. The large displacements
suggested by these observations are consistent with possible time-predictable behavior
[Shimazaki and Nakata, 1980] of the Hope fault, with the large inferred displacement at the
Green Burn sites in GB3 being followed by a period of time with shorter than average recurrence
intervals. Analysis of small offsets in lidar and ground-penetrating radar data on the Conway
segment, however, suggests that the past three earthquakes have each produced ~3-4 m of
displacement [Beauprêtre et al., 2012], and thus that the inferred larger displacements in the
Green Burn trenches may have been a local feature of that event and are not necessarily
indicative of GB3 being a larger-magnitude earthquake. \
3.6.1 Plate Boundary System-Level Rupture Behavior
The new Green Burn data add to a growing body of paleo-earthquake age constraints
from multiple sites along the Alpine-Hope-Jordan-Kekerengu-Needles-Wairarapa (Al-Hp-JKN-
Wr) system of major dextral strike-slip and oblique reverse-dextral faults that collectively
accommodate significant portions of Pacific-Australia relative plate motion in South Island and
59
southern North Island [Pondard and Barnes, 2010; Robinson et al., 2011; Litchfield et al., 2014].
Specifically, paleoseismologic records are now available from the Hope fault along the Hurunui
segment from the Matagouri Flats (MF) [Langridge et al., 2013] and Hope Shelter (HS) [Khajavi
et al., 2016] sites ~100 km west of the GBE site, from the Kekerengu (EK) fault at a site ~ 100
km northeast of the Green Burn sites [Little et al., 2018], from the Cross Creek (CC) site on the
Wairarapa fault – an extension of this fault system northward into southern North Island [Little et
al., 2009], and from multiple sites along the central Alpine fault (A) with dendrochronologically
dated records of tree disturbance [Wells et al., 1999] and records of strong ground shaking from
paleo-seismite records in lakes in the footwall of the Alpine fault on the coastal plain of the
Southern Alps [Howarth et al., 2012, 2014, 2016]. It is worth pointing out, in contrast to
paleoseismic results from trenches of the active fault traces, the dendrochronology and paleo-
seismite data record strong ground shaking at the site off of the active fault traces, and thus could
record earthquakes generated by other faults. We summarize these paleoseismic records in Table
3. It should be noted that we considered all events described in the original sources, and used the
preferred paleo-event age ranges of the original authors. In addition to these records, we discuss
the 2016 Kaikōura earthquake, and its potential implications, in a separate section below. We can
use all of these data to address basic questions about earthquake occurrence in the Al-Hp-JKN-
Wr fault system
Most basically, although the faults are linked, do they rupture independently in isolated
events? Or, do they rupture in brief temporal sequences along strike? Although hampered by the
long possible allowable age ranges of some events at some sites (e.g., GB5, HS2, EK4), the
available data allow us to examine the system-level behavior of the Al-Hp-JKN-Wr fault over
60
the past 1,000-plus years. In Figure 7, we show available paleoseismic constraints on the faults at
the specific paleoseismic sites discussed above.
In an attempt to assess the possibility that large parts of the Al-Hp-JKN-Wr system
rupture together in brief sequences, we interrogate 100-year-long intervals where there is overlap
between the 2 σ age ranges of more than two ruptures along the different faults. Specifically, if
there is overlap between events, in Figure 7 we show a pink bar, labeled Sx, across all sites that
could potentially have ruptured within a ≤100-year-long sequence. Although the 100-year time
window is arbitrary, it was chosen because it is shorter than the average recurrence intervals at
all sites, and helps to bring into focus possible brief sequences involving rupture of large sections
of this fault system. We attempt to minimize the number of sequences within the Al-Hp-JKN-Wr
system. That is to say, we select the temporal placement of the 100-year-long possible-sequence
“bar” show in Figure 7 across as many faults as is allowable within the given 100-year time
window. This analysis is designed to highlight possible multi-fault sequences, and does not
necessarily indicate that all paleoearthquake ruptures occurred in the given 100-year time
windows. Conversely, this analysis can point out the occurrence of an isolated event in the case
of a lack of paleoeathquakes on neighboring faults.
For example, one issue we explore is whether the record requires that the 65-km-long
Conway segment, which is bounded on both the east and west ends by major structural
complexities [Van Dissen, 1989; McMorran, 1991; Wood et al., 1994], may commonly rupture
by itself in isolated Mw~7 earthquakes, or whether it typically ruptures within a short period of
time with other parts of the Al-Hp-JKN-Wr plate-boundary fault system. We define isolated
events as rupture of a fault segment without rupture defined on adjacent fault segments. We
61
denote any potentially isolated earthquakes with blue horizontal bars labeled Isx on Figure 7
(e.g., GB4, A3 & A5).
We investigate the most recent possible multi-fault rupture sequence (S1) by comparing
the age of the most recent event at Green Burn (GB1), which occurred sometime between 1730
and before the period of European settlement began c. 1840 CE, with ages from other sites along
the fault system to the northeast and southwest. The 1730-1840 CE age range of the GB1 is
similar to the 1700-1840 CE time range of the most-recent surface rupture documented on the
Kekerengu fault by Little et al. [2018], indicating that the Conway segment and the Kekerengu
fault likely both ruptured within a <~100-year-long time window just prior to the beginning of
European settlement. Subsequently, the historical 1855 Mw~8.1 Wairarapa earthquake ruptured a
~160-km-long section of the Wairarapa fault extending into Cook Strait [Grapes and Downes,
1997; Rodgers and Little, 2006], and the 1888 Mw~7-7.3 Amuri earthquake ruptured the Hurunui
and Hope River sections of the central Hope fault [McKay, 1890; Cowan, 1990, 1991; Cowan
and McGlone, 1991; Khajavi et al., 2016]. Thus, if the most recent events on the Conway
segment (GB1) and the Kekerengu fault (EK1) occurred relatively late during their allowable
time ranges, the events observed at all four sites could record a temporally brief sequence of
large-magnitude earthquakes that ruptured the entire fault system northeast of the Alpine fault
during the late 18
th
and 19
th
centuries [Little et al., 2018]. Alternatively, if the prehistoric GB1
and EK1 most recent events occurred early in their allowable time ranges (i.e., as early as 1730
CE and 1700 CE, respectively), they might have occurred within a short period of time of the
most-recent, c. 1717 CE earthquake on the Alpine fault, which ruptured a ≥375-km-long section
of that fault as far north as the Alpine-Hope fault intersection [Wells et al., 1999; De Pascale and
Langridge, 2012; Howarth et al., 2018]. If so, then the Alpine fault, the Hope fault Conway
62
segment, and the Kekerengu fault ruptures may have occurred long before the historical 1855
and 1888 earthquakes, and thus these events may not have been part of a brief sequence.
Earthquake occurrence on this system over the past c. 300 years has may therefore have been
more random in time and space, with ruptures occurring piecemeal over the entire length of the
Al-Hp-JKN-Wr system. However, it seems less likely that GB1 ruptured early in the allowable
1730—1840 CE age range, given that the 2 σ age range of the penultimate surface rupture GB2
(1657—1797 CE) significantly overlaps with the GB1 range (Supplementary Figure S7). For this
reason, we suggest that GB1 likely occurred late in its allowable time range, just prior to the
beginning of the historic era, suggesting the possibility that the entire Hp-JKN-Wr part of the
system may have ruptured in a brief sequence beginning just prior to the historic era and ending
with the 1888 earthquake.
The next-older possible-sequence (sequence S2) includes the c. 1717 CE most recent
event on the Alpine fault, MF2, HS2, and GB2. We note that the c. 1717 CE Alpine event
occurred within a brief period of time with ruptures along the Hurunui, Hope River and Conway
segments of the Hope fault, but without rupture of the Kekerengu fault. As noted above, based
on the occurrence of the 1855 Mw~8.1 Wairarapa rupture to the northeast of the Kekerengu fault,
and the occurrence of GB1 to the southwest of the Kekerengu fault, we assume that EK1 likely
occurred during the most-recent, possible-sequence 1. If so, then the data suggest that possible-
sequence 2 did not extend northeastward beyond the Hope fault. This would be consistent with
the idea that portions of the Al-Hp-JKN-Wr system rupture in sub-sequences that involve only
part of the system, rather than as system-wide, “wall-to-wall” sequences.
Possible-sequence S3 encompasses ruptures on all faults in the system except for the
Wairarapa, including events A2, MF3, HS2, GB3 and EK2. Howarth et al., [2014, 2016, 2018]
63
have called into question the fault origin of event A2, which is marked in sediment cores by
submarine slope failure in all three examined lakes, but did not include a strong signature of
post-seismic landsliding [Howarth et al., 2014]. Event A2 is therefore equivocal with respect to
an Alpine fault source—either A2 occurred on another nearby fault in the Southern Alps, or the
event occurred on the Alpine fault and only weakly shook the region (MMI ~VI as opposed to
IX) [Howarth et al., 2014, 2018]. Given the apparent weaker shaking intensity during event A2,
it is possible that this event A2 did not occur on an Alpine fault source and instead occurred on a
smaller fault neighboring the Alpine fault. Alternatively, given the “bimodal” rupture model of
DePascale et al. [2014], the Alpine fault may rupture most of its length in Mw 8+ events, or may
rupture in parts in Mw ~6-6.5 events. These latter events would not be recorded in paleoseismic
trenches, but may be recovered in off-fault records of lake seismites [DePascale et al., 2014;
Howarth et al., 2018]. Given the fact that these records of lake seismites are off-fault records of
Alpine fault seismicity, it remains possible that these seismities represent earthquakes with
sources on faults adjacent to the Alpine fault, of which many have been documented [e.g., Cox et
al., 2012; DePascale et al., 2016]. We denote this uncertainty as a more transparent box for
possible-sequence S3 on Figure 7. If event A2 did occur on the Alpine fault, possible-sequence 3
could potentially represent a near-complete sequence of events that ruptured the Al-Hp-JKN
faults. However, notably, this rupture sequence did not cross Cook Strait onto the Wairarapa
fault. This observation likely reflects the fact that the Wairarapa fault exhibits a slower slip rate
fault of 11 ± 3 mm/yr [Little et al., 2009], much slower than the fast slip rates of the Kekerengu,
Hope, and Alpine faults (~20-25+ mm/yr) [Van Dissen and Yeats, 1991; Berryman, 1992; Norris
and Cooper, 2001; Hatem et al., 2016; Van Dissen et al., 2016] in South Island as slip is
transferred northeastward onto the BooBoo fault [Robinson et al., 2011], with a modeled slip rate
64
of 11 mm/yr [Pondard and Barnes, 2010], and ultimately onto the underlying Hikurangi
megathrust fault beneath North Island [Rodgers and Little, 2006; Wallace et al., 2012].
We observe a long (c. 400 year) lull in potential sequence activity between possible-
sequences S3 and S4, with two temporally isolated earthquakes (Is1 & 2) occurring on the
Alpine and Hope faults, respectively. Specifically, events A3 (1388—1407 CE), which
potentially ruptured the central Alpine fault, and GB4 (1230—1420 CE), which ruptured the
Conway segment of the Hope fault, do not overlap with the 2 σ age ranges of any other events
that have not already been plausibly assigned to a possible earthquake sequence. Although the
age range of A3 overlaps with the age range of GB4, we do not include these events as part of a
larger sequence because no faults with available paleoseismic data ruptured on either side of the
Alpine or Conway fault during this time period in surface-rupturing earthquakes that have not
already been included in sequence S3. For example, event GB4 does overlap in time with HS2,
but HS2 has been previously assigned to possible-sequence 3. Although events GB4 and HS2
could have ruptured within a brief time of one another as part of a sequence, our preferred
interpretation is that HS2 occurred in the same sequence as MF3, as these sites are only ~ 30 km
apart along strike of the Hurunui segment of the Hope fault, and thus likely record the same
earthquake. Events A3 and GB4 may represent a discontinuous sequence along the plate
boundary. Our preferred interpretation, however, is that events A3 and GB4 represent isolated
events because of the lack of spatial continuity and paucity of faults that ruptured during this
time period. Alternatively, it is possible that both A3 and GB4 were part of a brief, continuous
sequence of events that involved rupture along the southern Kakapo strand of the central Hope
fault system, and potentially bypassing the Matagouri Flats and Hurunui Shelter paleoseismic
sites of Langridge et al. [2013] and Khajavi et al. [2016], which are located on the northern
65
Hurunui and Hope River segments of the Hope fault. Currently, there are no paleoseismic data
available for the Kakapo strand with which to constrain this possibility.
This possible lull in potential sequence behavior was preceded by sequence S4, the only
inferred possible “wall-to-wall” rupture of the entire Al-Hp-JKN-Wr plate boundary system
during the past 2000 years. Specifically, between 1000 and 1100 CE ruptures along the Alpine
(A4), Hurunui (HS4), Conway (GB5), Kekerengu (EK3) and Wairarapa (CC2) are all
permissible, suggesting the possibility that the entire >850-km-long fault system may have
ruptured during a brief sequence of large-magnitude events. Such a wall-to-wall sequence
involving rupture faults of different recurrence intervals, with the Wairarapa hosting events
about every c. 1000 years [Little et al., 2009] and other faults in the system hosting events about
every ~300 years or less [Langridge et al., 2013; Khajavi et al., 2016; Howarth et al., 2018;
Little et al., 2018, this study], highlights the importance of understanding fault connectivity and
potential rupture patterns, such as those that occurred in the Mw=7.8 Kaikōura earthquake
[Litchfield et al., 2018].
Although the age range of GB5 is quite long and could possibly belong to another, older
rupture sequence, we think it unlikely that a rupture sequence rupturing from the Alpine to the
Wairarapa would bypass the Conway segment, given its central role in transferring relative plate
motion through northeastern South Island. Moreover, Coulomb failure function modeling shows
that rupture on the Jordan fault system increases the likelihood of rupture on the Conway
segment by 30% [Robinson, 2004], highlighting the strong relationship between these two
faults. Given these kinematic arguments, we favor placing GB5 in sequence S4.
In the above interpretation of possible-sequence S4, we assume that events HS3, GB5,
and EK3 ruptured within a short time of Alpine fault rupture A4 and Wairarapa fault rupture
66
CC2, and in Figure 7 we show our preferred interpretation of this sequence, with preceding
Alpine fault event A5 marked as an isolated event. However, it is equally allowable that events
HS4, GB5, and EK3, rather than rupturing as part of a brief sequence involving A4, ruptured as
part of a slightly older sequence involving A5, in which case A4 was likely an isolated event. If
this slightly older sequence did occur c. 900-950 CE, it cannot have involved rupture of the
Wairarapa fault in southern North Island, as event CC3 significantly post-dates event A5.
Whichever scenario is correct, the Alpine fault paleoseismic constraints for event A4 and A5
allow only one of these possibilities to be correct.
If sequence S4 did occur as suggested in Figure 7, it appeared to have been preceded by a
several hundred-year-long lull in potential sequence-like behavior. Specifically, although the
Alpine fault ruptured in A5 (915—961 CE), potentially as an isolated, Alpine fault-only rupture,
the preceding Alpine fault rupture A6 occurred between 592—646 CE. The Hope fault Hurunui
segment record [Khajavi et al., 2016] also suggests a long-duration lull during this interval prior
to event HS4. The record is less clear for the remaining parts of the system to the northeast, as
the age ranges of events on those faults permit multiple possible interpretations. For example, the
long possible age range of GB5 spans the occurrence of possible-sequence S4, possible-sequence
S5, and the intervening apparent lull. The only possible 100-year-long period during which the
Alpine fault could potentially have ruptured during a brief sequence together with the Hope and
Kekernegu faults occurred between 525 and 625 CE, encompassing A6, HS4, and EK4. As noted
above, however, the potential age ranges of EK4 is quite long, yielding relatively low confidence
in the occurrence of this possible-sequence S5. Given that the age range of GB5 is so long, and
the fact that the eastern Kekerengu and Hope Shelter sites record multiple events over this time
period, it remains a possibility that we are missing an additional event over the GB5 time
67
interval. However, because we have not documented a separate GB event, we cannot assign an
event at Green Burn to sequence S5 (note spatial gap in S5 on Figure 7).
Possible-sequence S6 is marked by rupture A7 on the Alpine fault and rupture HS5 on
the Hurunui segment of the Hope fault. It is perhaps noteworthy that all of the possible later
sequences encompass ruptures of both the Conway and Hurunui/Hope River segments of the
Hope fault. Thus, while it is possible that there was an as-yet unrecorded surface-rupturing
earthquake at Green Burn during possible-sequence S6, the Green Burn paleoseismic record does
not preserve a separate event during this time, so this possibility must remain speculative.
Although the paleoseismic timing constraints are too imprecise in many instances to
prove sequence-like behavior, the data are consistent with the possibility that large parts of the
Al-Hp-JKN-Wr fault system commonly rupture in brief (i.e., ≤ 100 year) sequences of large-
magnitude events. The available historical and paleoseismic records for the Hope, Kekerengu,
and Wairarapa faults, however, indicate that such possible-sequences are not always simple,
along-strike progressions of large-magnitude events. For example, the observation that the most
recent surface ruptures along the Conway segment of the Hope fault and the Kekerengu fault
occurred prior to European settlement, whereas the historical 1855 Mw~8.1 Wairarapa and 1888
Mw~7-7.3 Amuri earthquakes occurred to the northeast and southwest, respectively, indicates
complex spatial patterns of earthquake occurrence. An obvious possible complicating factor in
the occurrence of individual events during any possible-sequence is the occurrence of major
earthquakes on other nearby faults, such as the close temporal relationship between the 1855
Mw~8.1 Wairarapa earthquake and the 1848 Mw~7.4-7.5 Awatere earthquake, which ruptured
~105 km of the Awatere fault north of the Hope fault [Grapes et al., 1998]. Coulomb stress
modeling of these two events, for example, indicates that stresses related to the 1848 earthquake
68
elevated failure stresses along the future rupture plane of the 1855 event [Pondard and Barnes,
2010].
The fact that the c. 1000-1100 CE possible-sequence S4 is the only possible “wall-to-
wall” sequence of its kind over the past >1,000 years suggests that while such system-wide
behavior is possible, it is uncommon. Two of the past possible-sequences (S3 and S5) appear to
have encompassed rupture of all faults from the Alpine fault in the southwest to the Kekerengu
fault on the northeast, but neither of these possible-sequences extended across the Cook Strait
onto the Wairarapa fault. This could simply reflect the slower rate of elastic strain accumulation
and accommodation on the Wairarapa fault, evinced by its long recurrence interval compared to
the South Island faults [e.g., Little et al., 2009; Litchfield et al., 2018]. In contrast to possible-
sequences S2, S3, S4, as noted above, sequence S1 encompassed ruptures of the Hope,
Kekernegu, and Wairarapa faults, but did not include rupture of the Alpine fault. Thus, the
individual sequences do not always conform to the same pattern of ruptures. Whereas the
occurrence of individual events on the Al-Hp-JKN-Wr system is likely modulated by the
occurrence of earthquakes on other faults, leading to different patterns of ruptures and rupture
locations, the basic observation is that most large-magnitude earthquakes in the Al-Hp-JKN-Wr
system over the past >1,000 years appear to have occurred as parts of relatively brief (≤~100-
year-long) sequences.
The most recent earthquake generated within this plate boundary system – the 2016
Mw=7.8 Kaikōura earthquake – ruptured most of the Jordan-Kekerengu-Needles fault [Litchfield
et al., 2018; Kearse et al., 2017], as well as other faults to the south of the Hope fault, and likely
parts of the subduction megathrust beneath northeastern South Island [Duputel and Rivera,
2017; Hamling et al., 2017; Hollingsworth et al., 2017; Kaiser et al., 2017; Litchfield et al.,
69
2018; Wen et al., 2018]. Interestingly, no other major ruptures have occurred on the plate-
boundary fault system since the 1888 Mw~7-7.3 Amuri earthquake ruptured the Hurunui and
Hope River segments of the central Hope fault [McKay, 1890; Khajavi et al., 2016]. Thus, the
2016 event was preceded by a c. 128-year-long lull in which the entire Alpine-Hope-Kekerengu-
Wairarapa fault system remained dormant. It remains to be seen whether the complex, multi-
fault 2016 Kaikōura event is a harbinger of a near-future sequence of large-magnitude
earthquakes on the Al-Hp-JKN-Wr system, as occurred during previous possible-sequences S3
(c. 1500—1600 CE) and S5 (c. 525—625 CE). Alternatively, we have shown that some
sequences may have included temporally and spatially isolated large-magnitude earthquakes (Is
1&3 on the Alpine fault, Is 2 on the Conway segment), and it is possible that the 2016 Kaikōura
event is an isolated rupture. However, although simultaneous rupture of the specific faults that
occurred during the Kaikōura event was a rare occurrence, due to involvement of the slow slip-
rate faults in the North Canterbury District [Nicol et al., 2018], as well as the Papatea fault
[Langridge et al., 2018], large displacements on the Jordan-Kekerengu-Needles fault system
[Kearse et al., 2017], coupled with modeled 2016 Coulomb stress changes along the major plate
boundary faults in northeastern South Island [Hamling et al., 2017], suggest that the 2016
earthquake may presage another sequence along the Al-Hp-JKN-Wr to begin (So).
3.7 Conclusions
We present new paleoearthquake ages using primary event evidence from the Green Burn
reach of the Hope fault. We document the occurrence of five surface ruptures along the Conway
segment of the Hope fault at two sites along the Green Burn (GB) reach of the fault. These
earthquakes have occurred during the past c. 2000 years, with event ages as follows: GB1:
70
1731—1840 CE, GB2: 1657—1797 CE, GB3: 1495—1611 CE, GB4: 1290-1420 CE, GB5 36
BCE–1240 CE. The new Green Burn data, together with other previously documented on-fault
and off-fault paleo-earthquake age constraints from the various faults of the >850-km-long
Alpine-Hope-Jordan-Kekerengu-Needles-Wairarapa system of fast-slipping plate-boundary
faults in South Island and southern North Island, are consistent with the possibility that several of
the Green Burn surface ruptures could have occurred during relatively brief (≤100 years)
sequences that involved rupture of large sections of the fault system. However, the available data
indicate that “wall-to-wall” rupture of the entire Alpine-Hope-Jordan-Kekerengu-Needles-
Wairarapa system during brief sequences that ruptured all faults in the system must be a rare,
event. Indeed, the only possible such sequence occurred c. 1,000-1100 CE, during which all
faults in the system, from the Alpine fault in the southwest, to the Wairarapa fault in the
northeast, allowably ruptured during the same brief time interval. Partial rupture of the Alpine-
Hope-Jordan-Kekerengu-Needles-Wairarapa fault system in the Mw=7.8 2016 Kaikōura
earthquake may be a harbinger for future events within a potential new rupture sequence along
the plate boundary (So), potentially involving the Conway segment of the Hope fault adjacent to
the Jordan-Kekerengu-Needles system.
3.8 Figure Captions
Figure 1: (a) Map of New Zealand with plate motion vectors [DeMets et al., 2010] WLG-
Wellington, CHC-Christchurch. Red lines delineate major active faults of northern South Island
and southern North Island. (b) Regional fault map showing Alpine fault, Marlborough fault
system, and North Island faults. Conway segment of Hope fault is shown in yellow; yellow star
denotes Green Burn study site (GB). Hope fault system includes Kelly fault, Hurunui segment,
71
Hope River segment, Conway segment, and Seaward segment. KF-Kakapo fault, HB-Hanmer
Basin, EF-Elliott Fault, JT-Jordan thrust, PF-Papatea fault, OhF- Ohariu fault, ClF-Cloudy fault,
VnF-Vernon fault, WgF-Wellington fault, WrF-Wairarapa fault. Fault maps adapted from
Langridge et al., [2016].
Figure 2: Location maps generated using lidar digital elevation model (DEM) collected by GNS
Science/LINZ following the Mw=7.8 2016 Kaikōura earthquake. (a) Hillshaded DEM of Green
Burn stretch of Conway segment of Hope fault. GBW-Green Burn West (this study) (-
42.396560°, 173.388838°), GBS-Green Burn Stream (Langridge et al., [2003]) (-42.395914°,
173.392075°), GBE-Green Burn East (this study) (-42.393212°, 173.405528°). (b) Hillshaded
DEM with 50 cm contours at the Green Burn East site, showing the fault-perpendicular trench.
Small landslides are denoted with gray outlines (ls). (c) Hillshaded DEM with 50 cm contours at
the Green Burn West site, showing T-1 and T-2. Small landslides are denoted with gray outlines
(ls).
Figure 3: Composite field and photomosaic logs of GBE (a) East wall (inverted) and (b) West
wall. Unadulterated photomosaic is presented in Figure S2, available in the electronic
supplement to this article. Colluvial wedge deposits are denote in shades of purple, clay units are
shades of brown, shear zones are shades of red, and marsh units are shades of blue. Pebbles and
cobbles and distinctive orange clasts from unit CW2 were logged on photomosaics after field
work. Radiocarbon ages are colored yellow for samples included in age models; gray samples
were not included in the age model, but results are listed in Table 1.
72
Figure 4: Log of west wall of GBW T1 atop photomosaic. Unadulterated photomosaic is
presented in Figure S3, available in the electronic supplement to this article. Note landslide tip
(opaque purple) atop paleosol (green) at northern end of the trench near between m 11 and 16.
Radiocarbon sample used in age models are shown in yellow.
Figure 5: Log of GBW T2. Unadulterated photomosaic is presented in Figure S4, available in
the electronic supplement to this article. Landslide deposits are shown in light gray with purple
outlines, and paleosol is gradational from blue to purple. Radiocarbon samples used in GBW age
models are showing in yellow.
Table 1: Radiocarbon sample data for all dated samples along Green Burn reach. Bold typeface
indicate inclusion of a sample into an age model (Figure 6).
Figure 6: OxCal derived age models. (a) GBE colluvial samples only, (b) GBE colluvial and
south marsh samples, (c) GBE north marsh samples only, (d) GBW T2 only, (e) GBW T1 & T2.
(f) Preferred age model, which incorporates GBE, GBW T1 and T2.
Table 2: Paleoearthquake age ranges for all age models presented for Green Burn record.
Negative ages represent BCE.
Table 3: Two-sigma age ranges of plate boundary paleo-event plotted in Figure 7.
73
Figure 7: Events through time along the north-central Alpine, Hurunui (Hope), Conway (Hope)
Kekerengu-Needles and Wairarapa faults. Top map shows faults with study sites labeled; see
Table 3 for citations and age information. Individual event names are indicated with a capital
letter for each site with a number as shown in Table 3. Bottom panel shows temporal length of
events (2σ age range) with vertical gray bars. Horizontal pink-shaded boxes represent 100-year-
long potential “clustered event” sequences (see text for explanation), and are label SX near each
box. Horizontal blue-shaded boxes represent isolated earthquakes, and are labeled IsX near each
box. Thin, horizontal, red bars represent known surface rupture earthquakes on the Al-Hp-JKN-
Wr system, either in the historical period [McKay, 1890; Little et al., 2009; Khajavi et al., 2016;
Kearse et al., 2017] or using tree ring disturbance analysis [Wells et al., 1999]. Paleoseismites
recovered at Lake Ellery after 370 CE have poorly constrained rupture limits and may not have
occurred on the central-northern Alpine fault [Howarth et al., 2016].
74
CHAPTER 4: Holocene to latest Pleistocene incremental slip-rates from the Hope fault
(Conway segment) at Hossack Station, Marlborough fault zone, South Island, New Zealand
The work presented in this chapter is based upon the following manuscript in preparation:
Hatem, A.E., Dolan, J.F., Langridge, R.M., Zinke, R.W., McGuire, C.P., Rhodes, E.J., Brown, N.,
Van Dissen, R.J. (in prep.) Variability observed in Holocene and latest Pleistocene incremental
slip rates along the Hope fault (Conway segment) at Hossack Station, Marlborough Fault System,
South Island, New Zealand. Geochemistry, Geophysics, Geosystems
4.1 Abstract
Geomorphic field and aerial lidar mapping, coupled with fault-parallel trenching, reveals four
progressive offsets of a stream channel and one offset of an older bedrock-gravel contact at
Hossack Station along the Conway segment of the Hope fault, the fastest-slipping fault within
the Marlborough fault system (MFS) in northern South Island, New Zealand. Radiocarbon and
luminescence dating of aggradational surface deposition and channel initiation/abandonment
event horizons yields incremental slip rates for five different time periods during Holocene to
latest Pleistocene time. These incremental rates are highly variable through time, ranging from a
latest Holocene (0.5 ka –present) rate of 28.2 +15.4/-3.8 mm/yr, to a 1.6—0.5 karate of 14.6
+2.8/-2.1 mm/yr, to a rate of 19.1 ± 0.8 mm/yr between 5.4 and 1.6 ka, to a 9.4—5.4 ka rate of
12.1 +1.1/-0.8
mm/yr, to a 13.8—9.4 ka rate of 16.5 +2.0/-2.1 mm/yr. No significant slip has
occurred on the fault since at least 1840 CE (0.179 ka). The youngest, fastest rate correlates with
a temporally brief sequence of 4 earthquakes documented at a trench site ~40 km to the east.
This fastest incremental rate is also averaged over the shortest period (≤ 500 years), suggesting
that the factor of two variations in Hope fault slip rate observed in the four, older, longer-
75
duration (3—4 ky) incremental rates may mask even greater temporal variations in rate over
shorter time scales.
4.2 Introduction
Understanding the rate at which faults store and release elastic strain energy is of fundamental
importance for a wide range of issues, from seismic hazard assessment, to the proper
interpretation of geodetic data, to the strength and evolution of crustal faults. Previous analyses
of incremental fault slip rates indicate a wide range of behaviors, from rates that are seemingly
constant over a wide range of time scales [e.g., Weldon and Sieh, 1985; Noriega et al., 2006;
Kozaci et al., 2007; Gold and Cowgill, 2011; Van Der Woerd et al., 2017; Salisbury et al., 2018]
to examples in which strain release is markedly non-constant [e.g., Friedrich et al., 2003;
Weldon et al., 2004; Mason et al., 2006; Gold and Cowgill, 2011; Onderdonk et al., 2015; Dolan
et al., 2016; Zinke et al., 2017, 2018]. Despite a growing number of studies, the overall dearth of
these slip-rate data from major faults globally hampers our ability to understand the causes of
such behavior. In this study, we document incremental slip rates over five different Holocene to
latest Pleistocene time intervals on the Hope fault, one of the fastest-slipping strike-slip faults in
the Australian-Pacific plate boundary in northern South Island, New Zealand. We discuss these
results in light of their implications for plate-boundary strain accommodation, fault mechanics,
and probabilistic seismic hazard analysis (PSHA).
4.3 The Marlborough Fault System and Hope fault
New Zealand straddles the Pacific-Australian plate boundary, and is sandwiched between
subduction zones of opposing motion, with the Hikurangi megathrust dipping NW off the south
76
coast of the North Island and the Puysegar megathrust dipping SE off the northwest coast edge of
the South Island (Figure 1A). In northern South Island, most relative plate motion is
accommodated by the Marlborough fault system (MFS), a system of sub-parallel right-lateral
strike-slip faults that splay northeastward from the Alpine fault, the main plate-boundary fault to
the southwest (Figure 1B). From north of south, the four main faults of the MFS, which
collectively accommodate ~80-90% of the total plate motion of ~39 mm/yr [DeMets et al., 2010;
Wallace et al., 2012; Litchfield et al., 2014], are the Wairau, Awatere, Clarence and Hope faults.
Within the MFS, the southernmost Hope fault is thought to accommodate the fastest slip rate,
estimated by previous workers to be greater than 10 mm/yr along the entire Hope fault, and
likely ~20-25 mm/yr over the Holocene and latest-Pleistocene on the Conway segment itself
[Cowan, 1990; Cowan and McGlone, 1991; McMorran, 1991; Van Dissen and Yeats, 1991;
Langridge et al., 2003; Langridge and Berryman, 2005; Khajavi et al., 2018].
This study focusses on the Conway segment of the Hope fault, which is a structural
segment bounded at its southwest end by the ~7-km-wide transtensional Hanmer Basin [Wood et
al., 1994], and to the northeast by a complex structural transition in which most slip is
transferred northeastward from of the Hope fault onto the oblique-reverse faults of the Jordan-
Kekerengu fault system [Van Dissen, 1989; Kearse et al., 2017] (Figure 1). The Conway
segment has not generated a surface-rupturing earthquake within the historical period (c. 1840
CE). Although minor ground deformation was observed at one location ~ 45 km east of the
Hossack study site following the 2016 Mw=7.8 Kaikōura earthquake, the majority of the Conway
segment did not exhibit any signs of surface rupture [Litchfield et al., 2018.; J. Pettinga,
unpublished data]. In contrast to the Conway segment, the Hurunui and Hope River segments of
the Hope fault immediately west of the Hanmer pull-apart basin ruptured together in the 1888
77
Mw~7.2 Amuri earthquake [McKay, 1890; Cowan, 1990, 1991; Cowan and McGlone, 1991;
Khajavi et al., 2016].
4.4 Hossack study site
The slip rate study site at Hossack Station is located ~3 km east of the eastern end of the
Hanmer Basin (Figure 1). At the study site, the Hope fault is single-stranded and strikes ~ 075°,
extending across a broad valley filled with a pervasive, low topographic gradient surface, that we
refer to as S1, and has been subsequently incised by channels flowing westward to the Hanmer
River (Figure 2). The site was previously studied by McMorran [1991], who noted progressive
offsets of this channel and calculated a late Holocene slip rate of 18 ± 8 mm/yr.
As will be discussed more in-depth when describing trench and pit observations, this
surface is primarily a clast-supported silty-gravel with fist-sized cobbles overlain by a layer of
silt ~20 cm thick. This sequence of stratigraphy has been noted elsewhere in this region of New
Zealand, including the Tophouse Road [Zinke et al., 2018], Saxton River [Zinke et al., 2017] and
Branch River [Lensen, 1968] sites along the Clarence, Awatere and Wairau faults, respectively,
of the MFS. Observations made by Bull and Knuepfer [1987] indicated that these pervasive
surfaces within valleys and piedmonts of the Hope fault are aggradation surfaces, and represent a
phase of valley filling following the Last Glacial Maximum. This surface has then been incised
(i.e., degraded) during episodes of progressive channelization. These channels isolated a slightly
lower (~3 m) surface, denoted as S2 (Figure 2A). The youngest deposit at the Hossack site is a
cover of Holocene alluvium, which is deposited as a fan northeast of the main site.
At the Hossack site, we observe a modern-day channel (Channel 1; C1) with seasonal
flow, informally referred to as the Loops Stream, as well as remnants of two abandoned,
78
beheaded channels (Channel 2; C2 and Channel 3; C3) (Figure 2B).Channel 3 largely flows
westward south of the fault, guided by a linear bedrock ridge (Figure 2A). Once Channel 3
reaches the fault, the channel flows WSW along the fault towards the Hanmer River (Figure 2A).
In contrast to the along fault flow of Channel 1, Channel 2 “loops” NNW north of the fault.
There is only a small remnant of Channel 2 preserved south of the fault, which is immediately
adjacent to the Channel 3 western bank (see section on Offset C for a more detailed analysis of
this channel remnant). This Channel 2 remnant thalweg trends NW, in comparison to the
Channel 3 thalweg south of the fault that trends more northerly (NNW). The oldest channel,
Channel 1, is preserved by a “loop” with a longer wavelength along-strike than the Channel 2
loop. The loops of Channels 1 and 2 merge nearly due north of Pit 1 (Figure 2B), indicating that
Channel 2 occupied and further incised a portion of the abandoned Channel 1. A prominent yet
small (~100 m) shutter ridge long separates the extents of Channel 1 and Channel 2 north of the
fault. This shutter ridge is continuous with no signs of incision into the ridge, indicating that
there was no channel active temporally or spatially between Channel 1 and Channel 2.
4.5 Methods
4.5.1 Offset mapping
To document the progressive offset of the Loops Stream at the Hossack site and constrain
the ages of these incremental offsets, we used geomorphic mapping and field excavations. The
geomorphic mapping was based on a combination of field work, study of air photos, and analysis
of high-resolution (>12 hits/m
2
) lidar data (http://dx.doi.org/10.5069/G9G44N75 available at
www.opentopography.org).
4.5.2 Excavation of paleo-channels
79
To observe these channels in three-dimensions, and sample the stratigraphy for age
dating, we excavated a total of 10 pits and trenches. Pit 5 was dug to determine an age on S1, the
oldest surface at the site. Trench 7 and Trench 10 were dug to constrain the geometry of Channel
1 flowing around the adjacent shutter ridge, and to date the initiation and abandonment of that
channel. Trench 4 was dug to determine the orientation of Channel 2 near the fault, and to date
the initiation and abandonment of that channel, and Trench 3 captured any along fault channel
flow of Channel 2. Pit 9 was dug to date the abandonment of the slackwater deposits northeast of
active flow in Channel 3.
Trench 6 and Trench 8 were excavated to document any paleoearthquake surface
rupturing, and will not be described in this paper. Pit 1 was excavated into S2 to complete
luminescence dating of the surface isolated at the incision of Channel2. Pit 2 was excavated on a
slightly lower surface inset by ~1 m than S1, but was later reinterpreted to still be part of S1, and
was therefore not sampled. We did not run any samples collected from Pit 1 or Pit 2, and
therefore will not present any results from these pits in this paper.
Trench 3, Trench 4, Trench 7, and Trench 10 were gridded and logged using logging
paper at 1:20 scale. These trench walls were photographed in detail, and the photographs were
used to generate high-resolution photomosaics using Agisoft Photoscan [Bemis et al., 2014]. Pit
1, Pit 5 and Pit 9 were also photographed and logged.
4.5.3 Age determination
The excavations discussed in this manuscript were sampled for radiocarbon and
luminescence age determination. Radiocarbon samples, including detrital charcoal, wood, seeds
and other plant matter, were prepared using standard acid-base-acid pre-treatment and analyzed
80
at University of California, Irvine W.M. Keck accelerator mass spectrometer facility.
Radiocarbon age results were then calibrated to calendric years using the most up-to-date
southern hemisphere calibration curve, SHCal13 [Hogg et al., 2013] using the program OxCal
4.4.2 [Bronk-Ramsey, 2017]. Luminescence samples were prepared and analyzed at the
University of California, Los Angeles, using the newly developed post-IR-IRSL225 single grain
aliquot method (Huntley and Baril, 1997; Rhodes 2015; Lewis et al, 2017). All age data were
modeled using Bayseian statistics in OxCal.
4.6 Description of offset features in landscape and trenches
4.6.1 Offset A
We observe a relatively flat, pervasive surface within the valley at the Hossack. Pit 5 was
excavated into this surface, and revealed a clast-supported pebble gravel with rare cobbles,
which is defined as Surface 1 (S1) (Figure 3). The matrix of S1 was composed of medium
grained sand to silt. S1 was deposited on top of a silt, and was overlain by a silty gravel is then
overlain by a relatively clast-free silt. These upper two deposits likely represent sporadic
reoccupation of S1 as an active surface, likely in flooding events, after S1 deposition had largely
ceased. The composition of S1 in the sub-surface, coupled with the observation of a nearly
flat/low topographic gradient across the surface within the valley, supports the hypothesis that
gravel aggradation within valleys was accelerated following the Last Glacial Maximum [i.e.,
Bull and Knuepfer, 1987].
As S1 was being deposited, it lapped onto the bedrock at the outer edges of the E-W trending
valley at the Hossack site (Figure 4). Hence, the deposition of S1 at the Hossack marks the end
of any carving of the valley walls, meaning the geometry of the valley walls is representative of
81
the geometry at the time of S1 deposition. Therefore, any offset of bedrock features or other
features with the onlap contact of S1 onto bedrock must be coeval with or post-date deposition of
S1.
East of the Loops Stream, a bedrock promontory is offset along the fault. South of the fault, a
NE-SW trending bedrock ridge is juxtaposed against a young alluvial fan; north of the fault, a
NNW-SSE trending bedrock ridge abuts into the linear ridge south of the fault (Figure 4A & B).
S1 onlaps onto the bedrock ridge south of the fault (Figure 4B). Restoration of the shape and
western extent of the ridges yields a preferred offset of 230 m (Figure 4C; note red triangles on
this figure for piercing points). This 230 m offset also restores the headwaters of the Loops
Stream to its original source. Fault-related topography (pink unit in Figure 4B & C) has dammed
the headwaters of the Loops Stream, causing this drainage to aggrade northward into the range.
The present-day source of water for the Loops Stream is likely a spring at the base of this fault
related topography. This restoration of 230 m represents both our maximum and also preferred
configuration for this offset. A minimum restoration provides the smallest plausible restoration
of the bedrock knob, particularly the eastern edge of the bedrock feature. This minimum offset
also provides the smallest sedimentologically plausible restoration of the Loops Stream to its
headwaters, which we determine to be 218 m (Supplemental Material).
4.6.2 Offset B
In the present-day topography, Channel 1 is beheaded north of the fault and has no upstream
reach south of the fault (Figure 2). The correlative piece of Channel 1 south of the fault has been
subsequently eroded away with repeated flow through that thalweg in the present-day
configuration, as Channel 3 is currently utilizing the thalweg initially cut by flow through
82
Channel 1. Because the channel banks of Channel 1 are weakly exposed in the present-day
topography, we excavated Trench 7 (Figure 5) and Trench 10 (Figure 6) to better constrain the
orientation of the channel walls as Channel 1 flowed across the fault. These trenches also
exposed the stratigraphy of the channel, as well as the abandonment and cut facies of Channel 1.
Trench 7 and Trench 10 exposed Channel 1 as a brown, organic rich channel, sub-angular, well-
sorted, clast-supported silty, sandy gravel. Channel 1 was incised into an angular-to-sub-angular,
closely-packed, clast-supported blue-gray-green gravel. The Channel 1 deposit was overlain by
series of organic rich silts with ~10 cm long (max 50 cm) pieces of wood, with some peat layers
near the surface. We interpret these fine-grained, organic-rich deposits as being deposited after
flow through Channel 1 had ceased in favor of flow through Channel 2.
In addition to stratigraphic observations, Trenches 7 and 10 constrained the fault-
proximal geometry of Channel 1. Trench 7 revealed the western incised bank of Channel 1,
while Trench 10 exposed both banks of Channel 1 (Figures 5, 6, 7). The orientations of Channel
1 preserved in Trench 7 and Trench 10 are plotted as navy blue lines on Figure 7A & B. These
observations of Channel 1 north of the fault correspond to remnants the initial incision of
Channel 1 into S1 south of the fault. Channel 1 has subsequently been reoccupied and further
incised by younger Channels 1 and 2; gray shading on Figure 7 indicates modern activity and
erosion within this channel.
To describe the minimum offset of Channel 1, we restore the trend of the incised eastern
bank exposed in three-dimensions in Trench 10 (blue line south of yellow triangle north of the
fault on Figure 7A) relative to the NNW-trending, linear trend of the initial Channel 1 incision
into the S1 surface south of the fault (highlighted in light blue shading north of yellow triangle
south of the fault on Figure 7A). The offset of the eastern bank of Channel 1 is 146 m.
83
The maximum offset is defined by restoring the western bank of the channel, as exposed
in three-dimensions in both Trench 7 and Trench 10 (blue line south of the orange triangle north
of the fault on Figure 7B), to the western bank of Channel 1 initial incision into the S1 terrace
tread south of the fault (note near N-S topographic contour north of the orange triangle south of
the fault on Figure 7B). The offset of the western bank of Channel 1 is 152 m.
The “preferred” offset (149 m) is simply the average of the minimum and maximum
estimates of offset, as subsequent Channel 3 incision and erosion has destroyed any further
evidence of Channel 1 in this area.
4.6.3 Offset C
As observed with Channel 1, the downstream portion of Channel 2 north of the fault is
beheaded (Figure 2). Unlike Channel 1, however, Channel 2 has a NW-trending thalweg
direction as opposed to the NNW-trending flow direction of the Channel 1 (Figure 2). South of
the fault, Channel 2 has a small remnant bank incised into S1 to the west of the initial incision of
Channel 2.
To better expose the three-dimensional orientation of the Channel 2 southwestern channel
bank north of the fault, as well as the stratigraphy associated with flow through Channel 2, we
excavated Trench 4 (Figure 8). The Channel 2 deposit was composed of a sub-angular, well-
sorted, clast-supported sandy-to-silty gravel (blue units on Figure 8), which was cut into the
valley fill sequence (yellow units on Figure 8) and overlain by organic rich, fine-grain sediments
(green units on Figure 8). Again, this accumulation of fine-grained, organic material represents
the abandonment of flow through this once-active Channel 2 in favor of flow through the modern
Channel 3 thalweg. Using both walls of Trench 4, which exposed the southwestern bank of
84
Channel 2 (pale blue line in Figure 9), as well as the channel morphology exposed in previous a
auger profile and trench [McMorran, 1991] (purple rectangles in Figure 9), we were able to
measure a trend of Channel 2 initial incision. The trend of all these features is ~285° (Figure 9).
Using this trend, we restore the geometry of the initial incision of Channel 2. In addition
to the sub-surface observations of Channel 2 incision, the western bank of Channel 2 is preserved
in the landscape NW of Trench 4 (light blue shading north of the fault on Figure 9). The Channel
2 bank north of Trench 4 are extremely similar to the banks of Channel 2 exposed in Trench 4 as
well as the McMorran [1991] excavations. The trend of all these sub-aerial Channel 2 western
bank features is ~285°--295°, which matches the trend measured from the sub-surface exposures
of the southwestern Channel 2 wall. This agreement between sub-aerial and sub-surface trends
affirms the usage of the preserved, elevated NW-flowing channel south of the fault in Channel 2
offset restorations, as opposed to using the NNW-flowing thalweg utilized for restorations of
Channel 1 and 3. Therefore, in our depositional model, NW-trending Channel 2 incision begins
when the NNW-flowing Channel 1 south of the fault is finally strike-slipped southwest far
enough that flow can begin to incise NW around the shutter ridge (beige coloring on Figure 9),
marking the initial, NW-flowing incision of Channel 2.
Restoration of these southwestern Channel 2 wall measurements observed in the present-
day geomorphology and excavations requires 101 m of backslip to align the preferred geometry
of Channel 2 incision (Figure 9B). The thalweg of the Channel 2 provides another, secondary
constraint on this offset (dashed darker blue line north of the fault; shaded as indigo on Figure 9),
which was weakly exposed in Trench 4 and is inferred from the lowest elevation of the NW-
oriented, interpreted Channel 2 remnant south of the fault. The uncertainty of Offset C
encompasses the full height of the western bank preserved south of the fault (Figure 9A & C).
85
This error range takes into account any measurement error when the western bank was sighted in
the field from the excavations. The error range is well-constrained (±3 m) due to the precision of
the projection of multiple points along the western edge of Channel 2, as well as the high-
resolution of lidar-collected topography data.
4.6.4 Offset D
Channel 3, the modern-day channel, initiated after Channel 2 was abandoned. Channel 3
utilizes NNW-flowing thalweg south of the fault, and then flows nearly parallel to the fault north
of (and along) the fault (Figure 2). The NNW-oriented eastern bank of Channel 3 north of the
fault therefore likely corresponds to the NNW-oriented eastern bank of Channel 3 south of the
fault. The pre-existing eastern bank orientation south of the fault guided the incision north of the
fault to match the same NNW-orientation. Orange triangles on Figure 10 represent the piercing
points used in the restoration, demarcating the eastern bank of Channel 3 at the time of Channel
3 initiation north and south of the fault.
The preferred measurement of this offset is 29 m, with a maximum offset of 30.5 m and
maximum-minimum of 27.5 m (Figure 10). This ± 1.5 m lateral uncertainty in the offset
measurement encompasses ± 1 m of potential elevation change in the eastern bank elevation
across the fault (lateral uncertainty encompasses 2 contour lines on the 50 cm contour interval
maps presented in Figure 10). These error bounds provide the maximum and minimum values of
sedimentologically plausible offset. Any offset value greater or smaller than these will create an
“S” bend in the eastern bank of Channel 3, and there is no preserved evidence, nor any reason to
suspect, that initial incision of Channel 3 occurred in an “S” shape.
86
4.6.5 Offset E
In the present-day geometry of Channel 3, the eastern bank of Channel 3 has a
pronounced curve (as previously discussed with Figure 10 with Offset D). If water were to flow
in along the northern/eastern bank here, water would need to flow NE (i.e., up gradient and away
from the downstream Hanmer River to the SW). Because this flow-path is sediementologically
implausible, we interpret this area west of the curved eastern bank both north and south of the
fault as having little-to-no flow. Due to the lack of Channel 3 flow here, this could potentially be
an area of deposition (i.e., slack water deposition in a low-energy environment).
To test this hypothesis, we excavated Pit 9 within this interpreted slack water area north
of the fault (Figure 11). This pit exposed Channel 3 gravels, which was a sub-angular, clast-
supported, well-sorted clayey pebble gravel. The channel gravel was overlain by a waterlogged
clay with sparse gravels. The observation of the gravelly clay implies that occasional flow
(perhaps seasonal flooding) still occurs in this area of Channel 3. However, the fine-grain nature
of the deposit suggests that flow is largely abandoned from this part of Channel 3.
The curved, eastern banks of Channel 3 across the fault can be restored by a small
amount of backslip (Figure 12). When offset started accruing here, Channel 3 flow through this
area was probably already slow to ephemeral due to the curved eastern bank of Channel 3.
Hence, we use the lowest topographic contour to restore the “S”-bend in the slack water deposit,
as this contact likely represents the last amount of incision that occurred in this area (orange lines
on Figure 12). The last vestige of Channel 3 incision preserved in the slackwater area is now
likely buried under sediment deposited since incision occurred, and that piercing point is
therefore impossible to be restored. However, when excavating Pit 9, we observed only <50 cm
of fine grained deposits on top of the Channel 3 gravels, indicating that the amount of surficial
sediments is less than the contour interval on the topographic maps presented in Figure 12,
87
meaning that the exposed contour is a good approximation of the true elevation of last incision
here. This restoration ignores the contour lines showing colluviation of the eastern bank into the
slackwater area, as identified by field observations. Our preferred restoration of Offset E is 12 m.
The ± 2 m of lateral uncertainty take into account the small range of sedimentologically plausible
orientations of this natural curve in the channel.
4.6.6 Offset H
Because we know there has been no historical rupture of the Conway segment through
Hossack Station, we can make the assumption that there has been 0 m of offset since the onset of
the historical period (ca. 1840 CE). We refer to this lack of offset during the historical period
Offset H.
4.7 Age control
We dated a total of 63 radiocarbon samples and five infra-red luminescence samples
collected over two field seasons conducted at the Hossack site. Radiocarbon data for all samples,
including reworked samples that were not incorporated into our final age model, are presented in
Table 1. All ages listed on trench log figures and in the text are in units of thousands of years
before the year 2019 CE (ka).
We present a single depositional model across the site, utilizing samples from Pit 5,
Trench 7, Trench 10, Trench 4 and Pit 9, excavations which all record stratigraphy critical to
date events preserved as offset features at the Hossack (Figure 13). This model includes all ages
from these excavations that were not reworked or out of stratigraphic order. These ages were
grouped into depositional phases pervasive at the Hossack (i.e., S1 deposition; or, Channel 1
88
inactivity/Channel 2 activity). Because we are confident that the five presented offset
measurements capture all channel offsets at this specific location, and we have made
morphochronologic correlations across these trenches, we therefore combine ages from each
trench exposure into one age model. For instance, pebble-cobble gravel deposits of Channel 1
exposed in Trench 7 and Trench 10 will be capped by fine grained deposits, which represent the
abandonment of Channel 1 locally. The fine grained deposits in Trench 7 and Trench 10 are
therefore contemporaneous with coarse gravel deposits of Channel 2 exposed in Trench 4.
4.7.1 Age of S1 gravels (Offset A)
The age of S1 gravel deposition is determined by IRSL samples collected from Pit 5
(Figure 3). Sample HS15-23-L provides our preferred age of 13.9 ± 0.9 ka for the age of
deposition of S1, which dates Offset A.
4.7.2 Age of Channel 1 initiation (Offset B)
The temporal gap between the deposition of S1 (HS15-23-L; 13.9 ± 0.9 ka) and flow
through Channel 1, which incised into S1, exposed in Trench 7 and Trench 10 (i.e., HS15-13;
9.2—9.4 ka) spans nearly 5 kyr. Moreover, the substrate Channel 1 is carved into in both Trench
7 and Trench 10 is even older than the S1 deposit exposed in Pit 5 (blue gravel age i.e., HS17-
01-L; 23.1—25.0 ka). In order to narrow the wide age range on Channel 1 initiation, we use the
oldest sample preserved in the Channel 1 deposit (HS15-13; 9.2—9.4 ka) as a proxy age for
Channel 1 incision. The age of true Channel 1 initiation is perhaps slightly older than this age,
but likely not by much. This sample was collected a few centimeters above the base of the
channel, indicating that this sample was deposited early on during Channel 1 activity. Other
89
older samples were collected from the Channel 1 deposits (i.e., HS15-P7-11; 11.4—11.7 ka), but
stratigraphically above sample HS15-13, indicating that samples such as HS15-P7-11 are
reworked. Our preferred age of Channel 1 initiation is 9.2—9.4 ka, which dates the initial
incision of Channel 1, which is restored in Offset B.
4.7.3 Age of Channel 1 abandonment/Channel 2 initiation (Offset C)
Samples collected from Trench 4 and Trench 7 bracket the timing of initial incision of
Channel 2. We combine ages between these trenches to narrow the age range of this event by
combining the abandonment of Channel 1 with the initiation of Channel 2, which are likely
contemporaneous events. We use sample HS15-3 (5.3—5.7 ka) from the abandonment facies of
Channel 1 exposed in Trench 7 as the youngest possible age of flow through Channel 1, and
sample HS15-50 (4.8-5.2 ka) from the terrace (channel cut) facies of Channel 2 exposed in
Trench 4 as the oldest possible age of incision of Channel 2. The boundary age (achieved by
using the “Date” function in OxCal) between these two samples is c. 5.3—5.4 ka. The age of
Channel 2 initial incision corresponds to Offset C.
4.7.4 Age of Channel 2 abandonment/Channel 3 initiation (Offset D)
Because Channel 3 is the active channel, and as such, we did not trench into Channel 3,
we can only constrain the age of Channel 3 initiation via Channel 2 abandonment. Fine grain
facies overlying Channel 2 deposits in Trench 4 provide both the youngest possible time (HS15-
8; 1.4—1.5 ka) and oldest possible time HS15-24 (1.6—1.7 ka) of flow through C2. This
boundary age yields an age of Channel 2 abandonment/Channel 3 initiation of 1.6—1.7 ka,
which dates the displacement of Offset D.
90
4.7.5 Age of Channel 3 partial abandonment (Offset E)
The age of Offset E is associated with the partial abandonment/slack water initiation of
Channel 3. This event is dated using the fine-grained deposits above Channel 3 gravel exposed in
Pit 9. Samples HS15-18A, 18B and 20 yield similar results (ages are all within 25 calibrated
years of each other), which are then combined in OxCal to yield a minimum age on Channel 3
partial abandonment of c. 0.3—0.5 ka.
4.8 Incremental slip rates
We calculate incremental slip rates over different time spans between channel initiation
and abandonment events by combining the offset measurements with age results. We utilize a
Markov Chain-Monte Carlo approach for calculating slip rates and quantifying their associated
errors [Gold and Cowgill, 2011; Zinke et al., 2017, 2018]. The iteration of the Monte Carlo
approach presented in this paper calculates and reports errors in 2σ.
Using this method, we calculate the slip rate between Offset A and Offset B as 16.5
+2.0/-2.1 mm/yr, slip rate BC (between offsets B and C) as 12.1 +1.1/-0.8 mm/yr, slip rate CD as
19.1 ± 0.8mm/yr, slip rate DE as 14.6 +2.8/-2.1 mm/yr, and slip rate EH as 28.2 +15.4/-3.8
mm/yr (2σ error) (Figure 14). Incremental rate EH is a minimum rate, because we do not know
the exact calendric date of the most recent event on the Conway segment, and instead only
constrain the youngest possible age of the most recent slip along the Conway segment [e.g.,
Styron, 2018].
4.9 Discussion
91
Variability between the five incremental slip rates along the Conway segment measured at
the Hossack site indicate non-steady state strain accommodation along the fault at this location
(Figure 14). The greatest disparity between these incremental slip rates are between the fastest
rate, EH, (c. 0.5 ka—1840 CE) at 24.4—46.3 mm/yr and the slowest rate, BC, (5.4—9.3 ka) at
11.3—13.2 mm/yr. Incremental slip rates at Hossack Station vary by a factor of 1.5—3.
The statistical significance of variability between incremental slip rate distributions can be
tested using the Kolmogorov-Smirnov test [Styron, 2018]. This test, conducted on the cumulative
density functions of each incremental slip rate that resulted from the Monte Carlo-Markov Chain
calculations, indicates that all incremental slip rate distributions are significantly different than
one another.
To describe all five displacement-time measurements and associated incremental slip rates
with a single, finite (absolute) slip rate requires an extremely broad error range (green and blue
swaths on Figure 15). The minimum absolute rate that describes all of the displacement-time
data (Offsets A-H) is 29.2 ± 14.2 mm/yr (green swath on Figure 15). However, in order to
encompass all of the slip rates, in addition to the displacement-time measurements themselves,
the absolute rate must have an even broader error of 27.3 ± 15 mm/yr (blue swath on Figure 15).
Despite the variability in incremental slip rates, some displacement-time measurements are
fit by similar absolute slip rates (pink swaths on Figure 15). An older absolute slip rate of 15.0—
17.0 mm/yr can be fit through displacement-time measurements A and B, while a younger slip
rate of 18.1—19.6 mm/yr can be fit through displacement time measurements B and C. The
absolute slip rate fit through displacement-time box E is much faster than any other absolute rate,
suggesting that earthquake occurrence was faster than average during this time period.
92
Additionally, the absolute rates plotted in pink swaths on Figure 15 do not match the incremental
rate BC, suggesting that earthquake occurrence was slower than average during this time period.
We observe marked variability between rates DE and EH during the young part of the record.
Propagation of the variability expressed between the youngest two incremental slip rates at the
Hossack (rates DE and EH) overestimate slip rates in the older portion of the record (i.e., rates
AB and BC (Figure 16). This indicates that the young variability in incremental slip rates occurs
over a shorter time interval than the later variability, or that the younger part of the record simply
records more surface rupturing earthquakes. Propagating rates DE and EH from displacement
time measurement BC to AB shows a moderate fit to the incremental slip rates. This highlights
incremental rate BC as an outlier with respect to the rest of the Hossack data set, as the rest of
the incremental slip rates can more or less be fit by the c. 2 kyr variability expressed by rates DE
+ EH. Rate BC samples a time period when the Conway segment of the Hope fault was almost
certainly slipping overall at a much slower rate than the rest of the recorded 14 kyr history. This
slower rate between 5.4 and 9.4 ka may be a result of longer than average recurrence intervals
between surface rupturing earthquakes (i.e., earthquake lulls). This observation is in contrast to
the fast rate EH (<0.5 ka), which likely represents a time period with shorter than average
recurrence intervals between surface rupturing earthquakes (i.e., earthquake clusters).
This correlation between variable slip rate and earthquake occurrence can be tested by
comparing incremental slip rate records (i.e., Hossack) to paleoearthquake timing and single
event displacement data along the Conway segment of the Hope fault (Figure 17). Previous work
along the Green Burn reach of the Conway segment at the far eastern end of the segment
recorded the occurrence of five surface rupturing earthquakes in the past 2 ka [Hatem et al., in
review]. Additionally, offset restorations at the Terako site at approximately the mid-point of the
93
Conway segment reveal single event displacements of ~ 3 m per event [Beauprêtre et al., 2012].
Combining these two datasets, we reconstruct a dated path of earthquakes through time (pink
boxes and crosses on Figure 17) and compare that dated path with our incremental slip rates EH
and DE. The dated path of earthquake occurrence through time matches our incremental slip rate
record, with two earthquakes occurring in close succession ~0.5—0.6 ka, constituting an
earthquake cluster. The agreement between the dated path with data from the eastern and central
portions of the Conway segment with the incremental slip rate data from the western portion of
the Conway segment strongly suggests that this portion of the Hope fault ruptures together in
every surface rupturing earthquake.
Geologic slip rates remain a powerful input into earthquake hazard modeling [Field et al.,
2013; Petersen et al., 2015]. However, given records like the one developed at Hossack Station
that indicate a variable slip rate over time, the question arises: what slip rate, measured over
hundreds or thousands of years, is representative of the earthquake hazard posed over the next 50
years? Currently, probabilistic seismic hazard analyses do not consider temporally-variable
geologic slip rates, although some methods have been proposed [Zheng, 2018]. In a dense
network of faults, such as the Australian-Pacific plate boundary of the northern South Island and
southern North Island of New Zealand, fault interactions must also be taken into consideration
when deducing the seismic hazard of a given fault through time, especially along the Hope fault
which is a part of both the plate boundary system of the Alpine-Hope-Jordan-Kekerengu-
Needles-Wairarapa faults, as well as the Marlborough Fault System.
4.9 Conclusions
94
The Conway segment of the Hope fault exhibits non-constant strain accommodation through
time, as observed by new incremental Holocene-latest Pleistocene slip rate records. These slip
rates vary by a factor of ~1.5—3 times, similar to other faults in the Marlborough Fault System.
This work highlights that complex processes occur either on the scale of individual faults or fault
systems that govern the irregularity and deviation from constant slip rate and periodic earthquake
occurrence.
4.10 Figure Captions
Figure 1: (a) Map of New Zealand with plate motion vectors [DeMets et al., 2010] WLG-
Wellington, CHC-Christchurch. Red lines delineate major active faults of northern South Island
and southern North Island. (b) Regional fault map showing Alpine fault, Marlborough fault
system, and North Island faults. Conway segment of Hope fault is shown in yellow; yellow star
denotes Green Burn study site (GB). Hope fault system includes Kelly fault, Hurunui segment,
Hope River segment, Conway segment, and Seaward segment. KF-Kakapo fault, HB-Hanmer
Basin, EF-Elliott Fault, JT-Jordan thrust, PF-Papatea fault, OhF- Ohariu fault, ClF-Cloudy fault,
VnF-Vernon fault, WgF-Wellington fault, WrF-Wairarapa fault. Fault maps adapted from
Langridge et al., [2016]. To view this figure in color, the reader is directed to the online version
of this manuscript.
Figure 2: (a) Oblique view of topography looking NW at the Hossack site with prominent
geomorphic features colored. Contour interval is 50 cm. Background color is hillshaded digital
elevation model. (b) Oblique view of topography at smaller scale to show channels (C1—C3)
and trench and pit excavations. Contour interval is 50 cm. Background color is hillshaded digitial
elevation model. (c) Hillshade of entire Hossack study site. (d) Topographic map at nadir of
outlined area in (c). Trench and pit locations are shown in blue. Contour interval is 50 cm.
Figure 3: Annotated picture of Pit 5 wall with prominent clasts and layers outlined.
95
Figure 4: (a) Uninterpretted hillshade image showing extent of Offset A. (b) Geomorphic
mapping of map extent of (a). Restoration of Offset A. Red triangles denote piercing points for
restoration.
Figure 5: Line log of Trench 7 north wall. Blue units denote active channel deposits. Green
colors denote channel abandonment facies. Orange samples are radiocarbon, and pink samples
are IRSL.
Figure 6: Line log of Trench 10 north wall. Blue units denote active channel deposits. Green
colors denote channel abandonment facies. Orange samples are radiocarbon, and pink samples
are IRSL.
Figure 7: Restoration of Offset B at 146 m (a) and 152 m (b). Blue boxes are T7 (northern box)
and T10 (southern box). Blue lines show channel edges of C1 exposed in the trenches. Gray
shading shows areas that have been eroded since activity of C1. Beige shading shows the shutter
ridge. Yellow (a) and orange (b) triangles show piercing points for restorations. Topographic
contour interval is 50 cm.
Figure 8: Line log of southeastern wall of Trench 4. Yellow units denote pre-incision facies (i.e.,
alluvial surface sequence). Blue units denote active channel deposits. Green colors denote
channel abandonment facies.
Figure 9: Restorations of Offset C at 98 m (a), 101 m (b) and 104 m (c). Blue boxes indicate
trench excavations (T4 is indicated; see Figure 2 for other trench numbers). NE-SW trending
purple rectangle shows McMorran [1991] auger profile; NNW-SSE purple rectangle shows
McMorran [1991] trench. Light blue shading indicates channel banks at time of initial incision of
Channel 2. Indigo shading represents channel floor. Light blue solid line connects observations
of SW channel edge of Channel 2; darker blue dashed line aligns thalweg features. White circle
represents NE channel edge exposed in McMorran [1991] auger profile. Gray shading indicates
erosion and channel incision younger than Channel 2 activity. Beige shading shows the shutter
ridge. Topographic contour interval is 50 cm.
96
Figure 10: Restorations of Offset D at 27.5 m (a), 29 m (b), and 30.5 m (c). Blue boxes show
trench and pits (see Figure 2 for trench numbers). Orange triangles highlight piercing points.
Topographic contour interval is 50 cm.
Figure 11: Annotated image of Pit 9 exposure. Blue colors show active channel facies, and
green colors show abandonment facies.
Figure 12: Restorations of Offset E at 10 m (a), 12 m (b) and 14 m (c). Blue boxes show trench
and pits (see Figure 2 for trench numbers). Orange lines highlight natural curvature of channel
walls at last local incision of Channel 3. Topographic contour interval is 50 cm.
Figure 13: Bayseian age model of radiocarbon and luminescence ages generated in OxCal
showing ages from across the site combined into a single age model based on the site-wide
depositional history.
Figure 14: Displacement versus time plot of dated offset measurements made at Hossack Station
(blue boxes). Incremental slip rates between these boxes are shown graphically as gray lines
which appear black with increasing density of slip rate probability. Two-sigma range of
incremental slip rates are labeled along the curve. This plot is generated using Markov Chain-
Monte Carlo modeling of displacement-time data using a program built by Zinke et al., [2017].
Figure 15: Figure 14 replotted with the addition of finite (non-incremental) slip rates. Green
swath encompasses all displacement-time measurements. Blue swath encompasses all possible
incremental slip rates. Purple swaths represent finite slip rates from each displacement time
measurement. Note that no (purple) finite slip rate captures the slow period between 5.4 and 9.4
ka or the fast period between 0.1 and 0.5 ka.
Figure 16: Figure 14 replotted with the addition of propagation of rates EH and DH through
time, starting at the historical period (green rates) and at displacement-time measurement B. Not
that the young variability between incremental rates EH and DH over-estimates cumulative slip
at the Hossack site over the latest Pleistocene (inferred from the mismatch between the green
97
polygons and displacement-time measurement A). However, applying the EH + DE variability
starting at displacement-time measurement B reduces this over-estimate of cumulative slip,
indicating that the time period between 5.4 and 9.4 was perhaps anomalously seismically
quiescent.
Figure 17: Figure 14 replotted on shorter axes with the addition of earthquake ages from
Chapter 3 of this manuscript from the Green Burn site from the western Conway segment and
single event displacement estimates by Beaupretre et al., [2012] at the Terako site from the
central Conway segment (purple crosses show median values of both measurements). Purple
boxes surround the total possibilities of displacement and earthquake ages from these data sets.
Paleoearthquake slip and timing create a dated path for the last five events on the Conway
segment, and fits the incremental slip rate data well.
98
CHAPTER 5: Holocene to latest Pleistocene incremental slip rates at the Sawyer’s Creek
site along the eastern Conway segment of the Hope fault, South Island, New Zealand:
Implications for along segment variations in strain accommodation
The work presented in this chapter written as an incremental progress report reflecting years of
ongoing work conducted in collaboration with J.F. Dolan, R.W. Zinke, R.J. Van Dissen, E.J.
Rhodes, C.P. McGuire and N.D. Brown
5.1 Abstract
The Sawyers Creek slip rate site, located at the far eastern end of the Conway segment of
the Hope fault, preserves at an aggradational and three degradational terrace surfaces. Risers
between the terraces present offset piercing lines used to restore the risers across the fault at the
time of upper terrace abandonment. Van Dissen [1989] initially described and mapped the site
using air photos and obtained weathering rind age estimates on two of the offset surfaces. In this
chapter, I present a reinterpretation the site using recently acquired lidar topography data. I
sampled the site for infrared-stimulated single grain luminescence age dating; those age results
remain forthcoming. Re-measuring the offsets resulted in a 130% increase of the older
restoration, while the younger measurement remained largely unchanged. Using weathering rind
age estimates from Van Dissen [1989] couple with the larger, re-measured offset results in an
extremely fast slip rate in excess of the known plate rate. If the weathering rind ages are
representative of the true ages of the fluvial terraces here, slip rates measured at Sawyers Creek,
at the far eastern end of the Conway segment, are faster than the slip rates measured at the
Hossack site, at the far western end of the Conway segment (rates which were presented in a
Chapter 4 of this thesis). The Hossack site may have slower slip rates due to the transtensional
99
Hanmer Basin <5 km to the west of the site, whereas the Sawyers Creek site may have slightly
faster slip rates due to the transpressional Kowhai-Jordan-Kekerengu fault system immediately
to the east of Sawyers Creek.
5.2 Introduction
Geologic slip rates, the rate at which faults accommodate strain as discrete slip
measurable at the Earth’s surface, are one of the primary sources data we can collect to describe
the behavior of a faults. Such data are used as inputs and calibrations for fault evolution studies,
geodetic modelling and probabilistic seismic hazard analyses. Geologic slip rates can be either
constant [i.e., Weldon and Sieh, 1985; Noriega et al., 2006; Gold and Cowgill, 2011; Van Der
Woerd et al., 2017; Salisbury et al., 2018] or not constant [Friedrich et al., 2003; Weldon et al.,
2004; Gold and Cowgill, 2011; Dolan et al., 2016; Zinke et al., 2017, 2018] over centennial to
multi-millennial time scales. Studies of incremental slip rates can uncover periods of time when
slip rates vary by up to 500% [Zinke et al., 2018], variability that otherwise would have been
averaged over by calculating a traditional, absolute (i.e., finite) slip rate.
Despite the global dearth of incremental slip rate records, there is a growing body of such
studies in the Marlborough Fault System (South Island, New Zealand). Here, four sub-parallel
dextral strike-slip faults, named the Wairau, Awatere, Clarence and Hope faults from north of
south, accommodate the bulk (80-90%) of the ~39 mm/yr of Australian-Pacific plate motion at
this latitude [DeMets et al., 2010; Wallace et al., 2012; Litchfield et al., 2014]. Within the
Marlborough Fault System, the Hope fault accommodates about half of that motion with the
fastest slip rate of the four faults [Van Dissen, 1989; Cowan, 1990; Cowan and McGlone, 1991;
100
McMorran, 1991; Van Dissen and Yeats, 1991; Langridge et al., 2003; Langridge and
Berryman, 2005; Khajavi et al., 2018; Chapter 4 of this thesis].
To date, incremental slip rate sites across the Marlborough, at Saxton River along the
Awatere fault [Zinke et al., 2017], at Tophouse Road along the Clarence fault [Zinke et al.,
2018], and at Hossack Station along the Hope fault (Chapter 4 of this thesis). However, these are
single sites investigations from which we extrapolate the fault slip history of that fault segment
or entire fault. By applying a slip rate from one site across the whole fault, one makes the
assumption that the entire stretch of fault is behaving in the same way through both space and
time. The validity of this assumption depends on each fault. For instance, the Hope fault is
segmented into at least five sections along strike: Western Hope, Hurunui, Hope River, Conway,
and Seaward [e.g., Langridge et al., 2003]. Some of these segment boundaries are weakly
defined (i.e., Hurunui-to-Hope River), where earthquakes are known to propagate across the
boundary, such as the Mw 7.0—7.3 Amuri earthquake ruptured along both the Hurunui and Hope
River segments. This earthquake did not continue from the Hope River segment to the Conway
segment, likely due the arrest of rupture at the broad, transtensional Hanmer Basin separating the
Hope River and Conway segments [McKay, 1890; Wood et al., 1994; Khajavi et al., 2016].
Variable slip rate along strike of a fault can elucidate patterns of fault growth, evolution
and current tectonic loading sources [Hatem and Dolan, 2018; Chapter 2 of this thesis]. Thus,
when possible, multiple slip rate sites should be developed to uncover whether or not a particular
fault or segment ruptures, if the fault ruptures in semi-characteristic increments along slip, along
its entire length, or if there is variability of the spatio-temporal occurrence of earthquakes along
strike, which may be expressed as variable cumulative slip along strike. Such variability could be
potentially due to a discontinuity in the fault at depth across a potential segment boundary
101
unobservable in the geomorphology or fault trace geometry. Furthermore, slip likely tapers off at
the ends of faults or at overlapping fault segments [e.g., Manighetti et al., 2005; Milliner et al.,
2016]. The Sawyers Creek site is located near the western extent of the Kowhai fault just 3 km to
the north.
To determine the along-strike incremental slip rate records of the Conway segment of the
Hope fault, I present another incremental slip rate site along the Conway segment of the Hope
fault at Sawyers Creek. In this Chapter, I discuss the Sawyers Creek site geomorphologic
mapping, associated offsets, and potential incremental slip rates using dates from [Van Dissen,
1989; Van Dissen and Yeats, 1991], leading to an along-strike comparison of incremental slip
rates over at least two time intervals within the Holocene and latest Pleistocene using additional
data presented in Chapter 4 of this thesis.
5.3 Sawyers Creek Site
The Sawyers Creek site is located near the far eastern end of the Conway segment of the
Hope fault. Sawyers Creek is ~2 km from the Green Burn reach of the Conway segment, which
has been the subject of prior paleoseismic studies [Langridge et al., 2003; Chapter 3 of this
thesis] (Figure 1). Sawyers Creek has been the subject of geologic slip rate investigation by Van
Dissen [1989], followed upon by Van Dissen and Yeats [1991]. Using air photos and field
surveying techniques, Van Dissen [1989] completed geomorphic mapping of the site and
identified multiple terraces. Here, there is an aggradational surface, likely deposited during the
end of a glacial event, is the highest preserved terrace and named T1 (Figure 2), with younger
terraces representing successive incision invites into the older alluvial surface (T2-T4). Van
102
Dissen [1989] identified two offsets of these terrace risers across the Hope fault at Sawyers
Creek.
This Chapter improves upon these previous mapping efforts by adding high-resolution
topography data to interpret the terrace risers and associated offsets. This dataset was freely
available and downloadable from www.opentopography.org [DOI: 10.5069/G93J3B2J]. These
data provide a refined interpretation of the locations and geometries of the terrace risers, which
can reduce epistemic uncertainty in the slip rate calculations. Using these topography data, I
identified the similar terrace surfaces and risers as Van Dissen [1989] (Figure 3). I mapped
terraces T1 and T3 on both sides of the fault. There is an intermediary terrace T2 that is only
preserved north of the fault. While Van Dissen [1989] mapped two surfaces of T2 (T2a and T2b),
I identify only one main T2 surface (his T2a), with a broad riser between T2 and T3 (this T2/T3
riser being Van Dissen [1989]’s T2b). This identification of the broad riser was only made
possible by lidar topography in this densely vegetated area of the site. The youngest surface, T4,
is only preserved south of the fault. Surface T4 north of the fault has been eroded away because
T4 was right-laterally offset into the active Sawyers Creek drainage.
The origin of the “orphan” terrace T2 north of the fault, with no correlative terrace
intermediate between T1 and T3 south of the fault, remains enigmatic. The south side of the fault
is protected from fluvial erosion, as the north side of the fault is progressively right-laterally
strike-slipped northeastward into the active Sawyers Creek drainage. Although the north side of
the fault is laterally exposed to fluvial erosion, the vertical sense of motion locally on the Hope
fault is north side up, leaving the north side of the fault elevated and therefore isolated. As such,
one possible explanation for this preferential preservation of T2 north of the fault is coseismic
avulsion of Sawyers Creek south of the down-dropped southern block. Such avulsion could
103
potentially erode any remnant of T2 south of the fault. Such observations of extreme avulsion
were made along the Clarence River crossing the Papatea fault in the 2016 Mw 7.8 Kaikōura
earthquake [Langridge et al., 2018]. While the Clarence River has a much higher flow than
Sawyers Creek, this avulsion hypothesis remains a possible explanation of the mismatch of
surfaces across the Hope fault at this site.
5.4 Offsets Measurements
Van Dissen [1989] identified an offset between the T1/T3 riser south of the fault with the
T2b/T3 riser north of the fault. That study measured this offset as 150 ± 20 m. Using the lidar
data, I measured this offset at 200 ± 5 m (Figure 4). My measurement of the offset increased
likely due the utilization of high-resolution, bare-earth topographic data, as the T2b/T3 riser is
heavily vegetated and difficult to traverse. A smaller offset was measured by Van Dissen [1989]
between the T3/T4 riser south of the fault to the T3/active drainage riser north of the fault at a
minimum of 78 ± 2 m. I have re-measured this offset at a minimum of 80 m ± 3 m (Figure 5).
The T3/T4 riser south of the fault has been trimmed by young, riser-parallel incision (note small
channel immediately east of riser on Figure 5). The T3/T4 riser south of the fault is therefore
non-linear. This restoration therefore utilizes the southeastern edge of the riser (~ 80 m south of
the fault, as this portion of the riser minimizes the width of T4 south of the fault and therefore
removes at least some of the modern modification of this riser. This ~80 m offset is considered a
minimum measurement because the T3/T4 riser has been completely eroded north of the fault,
and T3 itself has been eroded by the active Sawyers Creek drainage.
5.5 Prior age results
104
Van Dissen [1989] obtained age estimates of these offset terrace surfaces using
weathering rind thicknesses, which is approximate absolute dating method calibrated to New
Zealand [Knuepfer, 1988]. Using this technique, Van Dissen [1989] reports an age of 2780 ± 560
years before 1989 CE for the T4 surface offset ~80 m, and an age of 4570 ± 910 years before
1989 CE for the T3 surface offset ~200 m.
While in the field over two seasons at Sawyers Creek, three pits were excavated for
luminescence sample collection. Single grain infrared-stimulated luminescence dating has
greatly reduced the epistemic uncertainty in dating alluvial and fluvial (organic-poor, sand-rich)
deposits [Rhodes, 2015], and we hope to improve the accuracy and precision of the dating efforts
made by Van Dissen [1989]. By dating terrace gravels exposed in a pit on terrace T2a, there will
be a maximum (oldest possible) age of T3 incision; T2a was abandoned as T3 became the active
surface. Pits excavated on the T3 surface itself will provide a depositional age of this surface.
The pit excavated on T4, as well as the sampled natural exposure of T4, will provide depositional
ages of T4, which will younger than T3 deposition. The boundary age between T3 and T4
depositional ages, calculated using Bayseian age modelling within the OxCal 4.3 program
[Bronk Ramsey, 2017], will provide an age of the T3/T4 riser. Likewise, the boundary age
between T2a and T3 will provide an age on the transition between T2 and T3 activity at Sawyers
Creek.
5.6 Slip rate calculations
Combining his offset measurements and age results, Van Dissen [1989] calculated two
finite (i.e., absolute) slip rates at Sawyers Creek (Figure 6). The T4 slip rate is reported as a
minimum rate of 28 ± 8 mm/yr and the T3 slip rate is reported as 33 ± 13 mm/yr (dextral), with
105
2.4 ± 0.6 mm/yr vertical (NW up). These rates are calculated by essentially fitting a line between
each of these two displacement-time measurements and the origin of displacement-time space.
Van Dissen [1989] did not calculate an incremental slip rate between these the T3 and T4
displacement-time measurements. I calculate the rate between these two displacement-time
measurements 40 + ~∞/-11 mm/yr. The positive, near infinite error on this rate is due to the near
overlap in ages of the T3 and T4 surfaces.
Due to the increase in offset measurement of the T1/T3 to T2/T3 restoration, the older
slip rate at Sawyers Creek would increase if using Van Dissen [1989]’s age of the T3 surface.
The recalculation of the T4 slip rate changes minimally, because the offset measurement of Van
Dissen [1989] agrees well with the lidar measurement.
5.7 Discussion
Slip rates T3 and T4 as described by Van Dissen [1989] are quite fast, and are even
approaching and equaling slip rates of the Alpine fault [e.g., Norris and Cooper, 2001]. The re-
measurement of the T1/T3 to T2/T3 offset even increases this rate, which exceeds the Pacific-
Australian plate rate at this location (~39 mm/yr; [DeMets et al., 2010; Wallace et al., 2012]). To
explain this fast rate, the weathering rind age calculated by Van Dissen [1989] could be perhaps
too young, which is making this rate apparently fast. This hypothesis will be tested upon
obtaining age data from the pits sampled for luminescence dating. Alternatively, the fast slip rate
could be representative of Hope fault activity, in which case strain accumulation rates (i.e., plate
rates) may change over centennial to millennial time scales. Or, strain accumulation and
accommodation may change locally if faults are able to store elastic strain energy at depth over
106
many earthquake cycles, which could then be accommodated as larger or more frequent
earthquakes along that fault.
In the case that the weathering rind ages are representative of the true terrace ages, the
slip rates calculated at Sawyers Creek, at the far eastern end of the Conway segment, are
generally faster than the slip rates calculated at Hossack Station, at the far western end of the
Conway segment (all Hossack slip rates reported herein are from Chapter 4 of this thesis) (Figure
6). In particular, a comparison of incremental rates from the Hossack over 0.5—1.6 ka at 14.6
+2.8/2.1 mm/yr to incremental rates from Sawyers Creek over 0.1—2.7 ka at 28 ± 8 mm/yr show
a factor of 2 difference in slip rate. Additionally, comparing the Hossack slip rate from 1.6—5.4
ka at 19.1 ± 0.8 mm/yr to incremental rates from Sawyers Creek over and from 2.7—4.6 ka at
29—40 mm/yr highlights a factor of 1.5—3+ difference in slip rate. These variations between
slip rates measured over similar time intervals suggest that there is a variability in millennial
scale earthquake behavior along strike of the Conway segment.
Despite the general observation that Hossack incremental slip rates being slower than
Sawyers Creek incremental slip rates over similar time scales, there is agreement between the
two sites at the youngest part of the records. The fastest incremental slip rate calculated from
Hossack data occurs between 0.1 and 0.5 ka (28.2 + 15.4/-3.8 mm/yr), which overlaps with the
28 ± 8 mm/yr rate from Sawyers Creek. This indicates that the Hossack has slipped at a rate as
fast as Sawyers Creek over some time period over the past ~5 kyr. However, this fast period
recorded at the Hossack is on the centennial scale, representing a short-lived cluster of 3-4
earthquakes (see Chapter 3 of this thesis for Conway segment earthquake timing). Furthermore,
when coupled with single event displacement estimates by Beauprêtre et al., [2012], these
earthquake ages provide a dated path that fits well within the error of incremental slip rates from
107
the Hossack (Figure 17 of Chapter 4 of this thesis). The agreement between the dated path
produced by combination of earthquake ages from the eastern Conway with single event
displacements from the central Conway and incremental slip rates from both the western and
eastern Conway segment suggest at least a short-term segment-long unified earthquake history.
This is evidently not the case at all times, however, as the gap between Sawyers Creek
incremental rates and Hossack incremental rates widens with increasing age.
The incongruence in incremental slip rates along strike of the Conway segment could be
due to several factors. One of which is simply, again, that the weathering rind age, an
approximate dating tool, of T3 is too young. An older age of T3 would reduce the incremental
slip rate between T3 and T4, which would reduce the difference between western and eastern
Conway rates from ~3.8 –5.8 ka. Another possibility, which is not mutually exclusive from the
former possibility, is that the far western Conway segment (Hossack site; Chapter 4 of this
thesis) generally has smaller single event displacements than the rest of the Conway segment.
This could be due to the 6-km-wide, 25-km-long, transtensional Hanmer Basin less than 5 km
west of the Hossack slip rate site [Wood et al., 1994]. This basin likely arrested the 1888 Mw
~7.0—7.3 Amuri earthquake, which ruptured along the Hurunui and Hope River segments to the
west of the Hanmer Basin, preventing rupture from continuing eastward onto the Conway
segment. Using numerical modeling of synthetic slip distributions along irregular fault
geometries, Resor et al., [2018] showed that long-term slip rates should decrease closer to large,
dialtional stepovers along the fault. In contrast to the western end, the eastern end of the Conway
segment meets the Kowhai-Jordan-Kekerengu fault system, which forms a restraining bend
accommodating oblique-slip motion [Van Dissen, 1989; Van Dissen and Yeats, 1991]. Given that
there is currently no incremental slip rate data from the central Conway segment, it remains to be
108
seen if the Sawyers Creek site can host greater cumulative displacement from being near this
oblique-slip system (both from ruptures along the Conway segment itself and, additionally, from
triggered slip in restraining bend events), or if, in fact, slip rate is highest in the central part of the
fault as would be expected with slip tapering towards the ends [e.g., Scholz, 2012].
5.8 Conclusions
I have re-examined the Sawyers Creek site, initially described by Van Dissen [1989],
using recently-acquired lidar topography data and luminescence age dating. Re-measuring the
offset terrace risers increased the T1/T3 to T2/T3 riser restoration from 150 m to 200 m; re-
measuring the smaller T3/T4 offset resulted in little change. Using the weathering rind age
estimates from Van Dissen [1989], in the absence of pending results from luminescence age
sampling at the site, coupled with the new T3 offset measurement, I calculated an incremental
slip rate from 2.7—4.5 ka of 29—40+ mm/yr, a factor of 1.5—3 faster than incremental slip rates
at the western end of the Conway segment. The incremental rate from 0.1—2.7 ka at Sawyers
Creek is faster than the 0.5—1.6 ka rate from the western part of the fault by a factor of 2. This
discrepancy in incremental slip rate along strike may be a result of a skewed slip distribution due
to transtensional and transpressional structures bounding the ends of the Conway segment.
5.9 Figure Captions
Figure 1: Satellite image showing location of Sawyers Creek slip rate site (yellow box and star;
SC = Sawyers Creek). Green Burn Reach of Conway segment is directly west of the Sawyers
Creek site, and is outlined in a purple box. Green Burn West (GBW) and Green Burn East (GBE)
109
are discussed in Chapter 3 of this thesis; Green Burn Stream (GBS) was studied by Langridge et
al., [2003].
Figure 2: Modified from Figure 4 of Van Dissen [1989]. Map shows original interpretation of
this site using air photos and field mapping. Fault scarps are shown in red, and terrace risers are
shown in blue.
Figure 3: Lidar derived digital elevation model (hillshaded) with terrace risers indicated. Pit
locations used for luminescence sampling are shown in blue. Fault is shown in red; terrace risers
are shown in black.
Figure 4: Topographic contour map (contour interval 50 cm) backslipped to restore T1/T3 and
T2/T3 riser at 195 m (a), 200 m (b), and 205 m (c).
Figure 5: Topographic contour map (contour interval 50 cm) backslipped to restore T3/T4 riser
at 77 m (a), 80 m (b) and 83 m (c).
Figure 6: Displacement versus time plot showing offsets measured by Van Dissen [1989] with
weathering rind ages with incremental slip rates in orange. The updated measurement of the T3
offset surface and incremental rates shown in red. Given the larger offset following the re-
measurement using the lidar data, the weathering rind age of T3 calculated by Van Dissen [1989]
appears to be too young; gray arrow indicates that the IRSL age results may be older than prior
weathering rind age estimates and therefore reduce this slip rate calculation. Blue displacement-
110
time measurements with incremental slip rates were measured at Hossack Station at the western
end of the Conway segment of the Hope fault (presented in Chapter 4 of this thesis).
111
CHAPTER 6: Towards quantification of the three-dimensional deformation field
associated with the 1952 Mw 7.3 Kern County earthquake along the White Wolf fault,
southern California, USA
The work presented in this chapter written as an incremental progress report reflecting years of
ongoing work conducted in collaboration with J.F. Dolan with initial consultation from C.W.D.
Milliner and J. Hollingsworth.
6.1 Abstract
The 1952 Mw 7.3 Kern County earthquake ruptured the White Wolf fault of southern California,
USA. The earthquake was extremely damaging to structures, particularly to those in the hanging
wall of the southeastern dipping fault. The rupture produced both thrust uplift south of the fault
as well as lateral (sinistral) motion. Most displacements were recorded as occurring in broad
fault zones, as opposed to discrete rupture planes. To quantify the full, three-dimensional
deformation field associated with this earthquake, including this distributed deformation, I have
developed and continue to improve a novel workflow that derives three-dimensional
displacements associated with this mainshock using pre- and post-earthquake, stereo air photos.
In the preliminary analysis presented here, the workflow has produced a coherent signal of
motion in the east-west direction, and encouraging results in the north-south and vertical
directions. This method remains a work in progress. Future work will focus on improvement of
co-registration between pre-event and post-event point clouds.
6.2 The 1952 Mw 7.3 Kern County earthquake
112
The July 21, 1952 Mw 7.3 Kern County earthquake was, after the 1906 Mw 7.9 San Francisco
earthquake, the second most damaging earthquake in California history at that time, causing an
estimated $60,000,000 worth of damage [Jenkins and Oakeshott, 1955]. At the time of the event,
the California Institute of Technology Seismological Laboratory calculated a Richter magnitude
of 7.7 for this event, and a focus of about 10 miles (~16 km) depth [Oakeshott, 1955]. The
epicenter was under Wheeler Ridge, an actively growing anticline, and propagated northeastward
along the northern base of the Tehachapi Mountains [Oakeshott, 1955] (Figure 1). Displacement
along the rupture was oblique-slip, with a larger thrust component in the west, where rupture was
nearly entirely blind, and a larger sinistral component in the east, which had more of a daylighted
rupture; overall, the rupture is described as a thrust event with a sinistral component [Buwalda
and St. Amand, 1955].
Geomorphic and topographic effects of the rupture were noted by Dibblee [1955], including
mole tracks, fence line deformation, ground cracking (particularly in water-saturated muds and
alluvium), and mud volcanoes. More detailed surface rupture measurements were made by
Buwalda and St. Amand [1955], who produced a surface rupture map composed of 62
observations of either vertical or lateral motion, or in some cases, both at the same location (their
summary table on pages 50-51 of their work). Some of these measurements include only
observations of en echelon cracking, and are not described with a dominant sense of motion
(thrust or sinistral). The maximum vertical component of slip along the fault was measured at 4
feet (1.2 m); the majority of vertical measurements had SE (mountain) side up motion [Buwalda
and St. Amand, 1955]. The minimum value of thrust motion is 0.25 feet (7.6 cm), and some sites
observed had no thrust motion at all [Buwalda and St. Amand, 1955]. The sinistral component of
slip was mostly preserved by en echelon cracking, with very little discrete surface rupture
113
[Buwalda and St. Amand, 1955]. A handful of offset fences provided more discrete
measurements of sinistral slip; these measurements were recorded as > 1 foot or ± 1 foot
[Buwalda and St. Amand, 1955]. In one location, near the NE end of the rupture, Buwalda and
St. Amand [1955] noted that en echelon cracking accommodates lateral motion over a zone of
200 feet (61 m); the maximum fault zone width observed was half a mile wide (1.6 km) (location
of this broad fault zone width estimate was not listed by the original authors).
6.3 Off-fault deformation of coseismic ruptures
Based on the reconnaissance mapping of Buwalda and St. Amand [1955], the Kern County
surface rupture could be characterized as diffuse and distributed. More specifically, this rupture
evidently had a large proportion of off-fault deformation. Off-fault deformation is any
accommodation of strain away from the primary fault strand. In many cases, it seems that there
was not even a primary fault strand identified by Buwalda and St. Amand [1955]. Off-fault
deformation, and, in particular, fault zone width, has been shown to be larger when ruptures
travel through alluvium versus bedrock [Milliner et al., 2015]. True surface rupture was rare
along the SW portion of the rupture that crossed the southern end of the San Joaquin Valley, with
only some ground cracking observed. The eastern two thirds of the 1952 White Wolf fault
rupture, however, juxtaposes sediments south of the fault against the bedrock of the hanging wall
Tehachapi Mountains north of the fault. Off-fault deformation was found to sharply decrease at
the surficial bi-material interface of sediment to bedrock in the Landers rupture [Milliner et al.,
2016b]. While off-fault deformation may have decreased from the sediment-sediment interface
of the San Joaquin Valley to the alluvium-bedrock contact further eastward, the rupture along the
114
base of the mountains still appears quite diffuse according to the field mapping by Buwalda and
St. Amand [1955].
An accurate slip distribution measured at the surface, along with fault zone width, can be
used as inputs for slip inversion models, which elucidate not only the slip patterns at depth, but
also the width of the damage zone about the fault at depth [e.g., Stein and Thatcher, 1981; Pollitz
et al., 2011; Xu et al., 2016; Hollingsworth et al., 2017]. Most importantly, measurements of off-
fault deformation at the surface helps quantify the lack of coseismic slip in the shallow sub-
surface (upper ~3 km of a fault zone), identified by many geodetic slip inversions as the
“shallow slip deficit” [Fialko et al., 2005]. This shallow slip deficit is thought to be a
manifestation of coseismic deformation accommodated off of the main fault (i.e., off-fault
deformation) within the damage zone surrounding the main fault strand [e.g., Cochran et al.,
2009; Dolan and Haravitch, 2014]. Using the exact amount of off-fault deformation, coupled
with the slip inversions, we can quantify the actual shallow slip deficit, which may be
accommodated as after slip or interseismic creep [Xu et al., 2016].
6.4 Existing correlation methodologies
In order to obtain such high-resolution, high-fidelity slip distributions of discrete slip
versus off-fault deformation, we require a more accurate method than estimating displacements
across sets of ground fractures, as was done immediately after the earthquake. While “boots on
the ground” mapping is an important way to collect data on the fault, this method has its pitfalls
and can miss, in some cases, up to 50% of the full deformation field [Milliner et al., 2015]. The
field of view while field mapping is quite small (on the order of meters to tens of meters). Off-
fault deformation can occur locally on the scale of centimeters and be distributed over widths of
115
100’s to 1000’s of meters. Therefore, the accurate determination of coseismic slip distributions
requires a method with a wider aperture.
Optical imagery acquired before and after an earthquake can be co-registered and
correlated, yielding the full, horizontal deformation field accrued between the acquisition of the
imagery. The data used for such analyses can be either aerial photography or satellite sensed
imagery. This process can be carried out using the program COSI-Corr, which utilizes a sub-
pixel correlator to difference images in two dimensions in the near field once the far field is held
fixed after being co-registered [Leprince et al., 2007; Ayoub et al., 2009]. This method, and other
similar algorithms, has successfully been used to describe the deformation fields of the 1992 M w
7.3 Landers earthquake [Milliner et al., 2015], 1999 Mw 7.1 Hector Mine earthquake [Milliner et
al., 2016a], 2005 Mw 7.6 Kashmir earthquake [Avouac et al., 2006], and 2013 Mw 7.7
Balochistan earthquake [Zinke et al., 2014; Gold et al., 2015], for example. The slip in most of
these events was accommodated within the fault zone as horizontal (lateral) motion, with
minimal uplift or subsidence. In the case of the 1952 Mw 7.3 Kern County event, however, most
of the deformation was uplift along the southeast dipping White Wolf fault. As such, we require
a method that takes the third dimension into consideration.
Only a few studies have thus far completed such analyses of the full three-dimensional
deformation field, such as the 2010 Mw 7.2 El Mayor-Cucapah earthquake [Oskin et al., 2012]
and the 2008 Mw 6.9 Iwate-Miyagi and 2011 Mw 7.1 Fukushima-Hamadori earthquakes [Nissen
et al., 2012], using pre- and post-event lidar data. Unlike COSI-Corr, which correlates two-
dimensional, co-registered pixel textures of air photos (recent advances have been made in
COSI-Corr to derive the three dimensional deformation field using satellite imagery [Zinke et al.,
2019], but not yet air photos), these methods employ either lidar differencing (i.e., simple
116
subtraction), which does not account for lateral north-south and east-west motion) [Oskin et al.,
2012], or iterative closest point matching (ICP) of lidar point clouds, a method that rotates and
stretches a deformed point cloud to fit the reference point cloud [Nissen et al., 2012]. The latter
method measures deformation in north-south, east-west and vertical directions, yielding vectors
of displacement in XYZ space.
These methods discussed so far in deriving two-dimensional and three-dimensional fields
rely on data that have been georeferenced and have metadata associated with their collection.
While the 1952 Kern County event has air photos collected before and after the event, similar in
quality to the air photos collected before and after Landers and Hector Mine events [Milliner et
al., 2015, 2016a], the Kern County data are not georeferenced and do not have important
information included with the imagery, such as camera parameters, distortion measurements,
nadir angle or fiducial point locations. These metadata are required as input data into the COSI-
Corr algorithim. Metadata aside, the air photos before and after the 1952 event were captured in
stereo (60% overlap between photos in both sets of imagery). Hence, the photos can be aligned
and constructed such that they reveal the three-dimensional topography of the overlapped
imagery in stereo.
In this chapter, I provide a progress report on work completed thus far to derive the three-
dimensional deformation field of the 1952 event by first describing the air photos data collected
before and after the earthquake, and then outline the novel method I have tried to implement to
solve this problem.
6.5 Pre- and post-event imagery
117
Pre-event air photos (flight C-17790) were collected on May 23, 1952, almost exactly
two months before the earthquake, by Fairchild Aerial Surveys, Inc. The scale of the photos is 1:
31,680, and were captured at an altitude of 21,780 feet (6638 m). The lens focal length was 8.25
inches and spectral range of the photos is 400—700 nm. The post-event air photos (flight C-
18160) were collected between August 1 and September 30 of 1952, immediately following the
July 21 mainshock. The purpose of this air photo flight was to document the earthquake damage;
the flight index is labeled “San Joaquin Valley Quake Project.” This photo series was flown at an
altitude of 6,600 feet (2011 m), resulting in a much higher resolution (1: 9,600) than the pre-
event air photos. The focal length and spectral length of the post-event series is the same as the
pre-event series, suggesting that the same or similar camera was used for both missions. Both
flights have physical images that are 9 inches x 9 inches (22.86 cm x 22.86 cm). Neither the pre-
event nor post-event imagery was georeferenced or contained any metadata beyond what is listed
above.
The imagery was accessed through the University of California, Santa Barbara Map and
Imagery Library. Air photos scanned (on a machine with unknown specifications) at 1200 dots
per inch. The imagery used was expressly asked to not be post-processed in Adobe Photoshop,
so as to not change the spectral quality of the pixels. Given the scan resolution of the imagery,
the scale of the pre-event imagery (C-17790) is ~2 feet/pixel (~60 cm/pixel) and the scale of the
post-event imagery is (C-18160) is ~0.6 feet/pixel (18 cm/pixel). These scales are for the
unrectified (raw) imagery.
118
6.6 Point cloud generation and processing
To generate the point clouds that would later be used for iterative closest point matching
(ICP), I utilized the program Agisoft Photoscan. This program is widely used in the geosciences
for creating high-resolution photomosaics of outcrops and trench walls, as well as digital
elevation models from drone photography and the creation of three-dimensional models of hand
samples [Bemis et al., 2014]. Agisoft can reconstruct the camera look angle using many
thousands of common tie points between images. In the best-case scenario of the pre- and post-
event imagery of the Kern County event, each point on the ground is covered by four to six
separate camera looks and can be independently located by multiple camera pairs. Prior to
camera alignment, each image must be individually masked around the black edges and the
White Out used to label each image. If this masking process is not done, tie point registration
will decrease in fidelity and may lead to spurious results. Tie point generation is strongest in the
center of the image, and decays away towards the distorted edges of images; distortion at the
edges of the individual images compounds at the edges of a stitched photomosaic, enhancing the
so-called “bowl effect” where the edges of the photomosaic curl up or down relative to the true
topography. I did not specify the amount of tie points to search for, and instead allowed Agisoft
to determine the appropriate number of tie points (done by entering “0” in the dialogue box for
tie point number).
If the photos were previously georeferenced, Agisoft would have aligned them based
upon their external reference frame. However, in the case of the Kern County event imagery, the
user must supply the geographic coordinates. To georeference the photos, I use a combination of
satellite imagery (maximum pixel resolution of this imagery is 1 m) in the basemap of ArcGIS
(which has a higher resolution than Landsat data), or other high-resolution orthophotos available
119
through www.earthexplorer.usgs.gov. These satellite imagery products were used to obtain X
and Y coordinates. To obtain elevation (Z), I use 1/3 second (10 m) digital elevation model
(DEM) from The National Map (https://viewer.nationalmap.gov/basic/). Ground control points
are selected at road crossings, buildings, prominent geomorphic markers far from suspected fault
displacement. Some images have no anthropogenic markers at all, while others have many.
Shadows within the canyons and those cast across trees may distort the edges of features, adding
or subtracting feet of error in the all directions. Georeferencing the imagery can present an
enormous source of error if referencing a single image. However, the alignment of images in
three dimensions can reduce the error associated with the location of points because all images
are geographically tied to that location. The distribution of ground control points is critical.
Placing many points in one region of the cluster of photos to stretches the rest of the point cloud
around it towards the congregation of ground control points. Furthermore, in the case of applying
ICP to the pre- and post-event point clouds, only one side of the fault should be referenced, to
allow the other side to deform. These points should not be near the fault itself, as the fault zone is
likely broad about the mapped trace of the fault. Sometimes, this is unavoidable, but future work
should be directed at finding optimal ground control points to allow for co-registration while still
allowing the fault zone to freely deform. Ground control points provide the best photo alignment
when they are added after the initial tie point generation. The ground control points simply add a
reference frame to the alignment that Agisoft has created external of a geographic coordinate
system; I find that this order of operations introduces less error into the alignment of photos.
Following photo alignment by tie point generation and then by ground control points, the
next step in the work flow is to calculate a dense cloud array. In this step, Agisoft picks more
matching points, oriented in three dimensions. The user can select many levels of detail to go
120
through for dense point cloud generation, increasing or decreasing the point density. Increasing
the point density may introduce artifacts to the data due to extreme gradients between points. In
my experience, every set of images (over a reach of the fault a few kilometers long) responds
differently to different levels of point cloud density; I believe this inconsistency is due to, among
other factors, the variations in the image texture, including pixel saturation (of the scan or the
original exposure), amount of vegetation, topographic gradient, ground control point density and
interconnectedness of shadows.
An orthorectified photomosaic can then be created in Agisoft, and individual images can
be orthorectified and exported. This photomosaic is helpful for reference, but is not required for
the point cloud differencing methodology. In order to use COSI-Corr, the input imagery must be
orthorectified; these orthorectified images from Agisoft can then be used as inputs to COSI-Corr
if desired.
The dense cloud created by Agisoft can be exported as a georeferenced .LAS file, which
is then imported to a separate program called CloudCompare for ICP analysis. CloudCompare
allows the user to import two point clouds that ought to be registered with similar, or even the
same, ground control points using the same geographic coordinate scheme (i.e., UTM Zone 11N,
WGS 1984). However, there is some mysterious offset of reference frames that occurs upon
import to CloudCompare. CloudCompare will prompt the user with a dialogue box to shift the
imported point cloud to a different (native) reference frame; results are typically better if the shift
is declined. However, it is worthwhile to try importing the point clouds both ways to see which
import method provides the best registration. In addition to this shift prompt, CloudCompare
offers alignment tools within the user interface to arrange the point clouds as nearly to exact far-
field co-registration as possible. Once the point clouds are aligned as best as possible, which will
121
almost certainly not be exact (which is the most vexing source of error in this workflow, and
should be explored in a systematic way in the future), ICP calculations can commence. The
output of the ICP calculations are four separate deformation fields: C2C (cloud to cloud)
absolute distances (includes X, Y, and Z), C2C X, C2C Y and C2C Z.
6.7 Preliminary results
Because of the extensive amount of time it has taken to develop this workflow over many
iterations, definitive results have not yet been produced. This method is hopeful, though, and
shows promise if the error in point cloud co-registration (in particular in the Z and Y (north-
south) directions) can be reduced. In the following section, I present the most successful results
of the workflow in progress. The results discussed below look at a segment of the White Wolf
fault where both lateral and vertical surface rupture was noted in field mapping by [Buwalda and
St. Amand, 1955]. This fault segment shown in Figure 2-5 is captured in three dimensions by
alignment of 22 (post-event) images using 11 ground control points, which resulted in a dense
cloud of 5.4 x 10
7
points (Figure 2 shows a two-dimensional projection of the point cloud). For a
detailed report built by Agisoft on the generation of this post-event point cloud, please refer to
the Appendix of this thesis.
Utilizing the method outlined above has yielded promising results in the X direction
(east-west) of displacement between point clouds (Figure 3). Although there are large errors of
some points (range of C2C distances ranges from -41 to 72 m), there appears to be some
coherence to the signal on opposite sides of the fault. In Figure 3, the Tehachapi Mountains south
of the fault have moved in the positive direction (to the right), and the alluvium and associated
hills have moved in the negative direction (to the left). This pattern is consistent with left-lateral
122
displacement. There is prominent noise throughout the image, as shown with the yellow coloring
on both sides of the fault (yellow on this color bar represents -2 to 0 m of motion towards the
west). Based on this observation, there is evidently a mis-registration of the two point clouds on
the order of 2 m; the block south of the fault did not slip both sinistrally and dextrally, so the left
slip south of the fault must represent either noise or mis-registration. Although the displacement
results are unreasonably high in some cases, particularly of the topographic features, they are
consistent on either side of the fault, indicating that there is at least some reliable signal coming
from this workflow with respect to east-west displacement.
There is apparently more mis-registration of the same two point clouds in the y direction
(north-south) (Figure 4). Here, the signal is not coherent across the fault. One side of a canyon
has a positive y displacement (moving north), while the other side of the canyon has a negative y
displacement (moving south). A coherent signal would be manifest with a negative y
displacement north of the fault, with a positive y displacement south of the fault. Instead,
landforms both north and south show this “two-way” displacement. Because this “two-way”
displacement is seemingly impossible (there is likely no shear parallel to each canyon within this
point cloud), and the y displacements do not match the pattern of the x displacement, I believe
that these two point clouds are either severely mis-registered in the y direction (rotated) or
physically distorted in that direction. The latter possibility could be due to uneven placement of
ground control points during the Agisoft photo alignment phase. The questionable patterns of y
displacement remain an active area of research.
Similar to the y displacement field, the z displacement field (vertical) is mis-registered in
a seemingly non-tectonic way (Figure 5). There appears to be no difference in mis-registration
between vegetated and non-vegetated regions of the point cloud. However, areas of high
123
topography appear to have exaggerated z displacements. Likewise, the valleys appear to be mis-
registered in the opposite direction. This suggests that the point cloud could be rotated in slightly
the wrong direction (i.e., tilted). If that were true, the mis-registration patterns may be coherent
across the fault. However, that does not appear to be the case. It seems as if the point cloud
generation in Agisoft has stretched a point cloud in the z direction in a non-linear way. The cause
of this stretch is a topic of further research.
6.8 Future directions
Based on the results presented above, I remain hopeful that this workflow can yield
results with continued, careful work. Many more combinations of settings and point cloud
alignment methods, both manually in CloudCompare and with varying alignments of ground
control points in Agisoft, need to be systematically tested. This process takes many hours of
computer processing time, as well as user input to determine the best, most reliable array of
ground control points. Point cloud alignment may benefit from erasure of features above the
ground surface, such as trees and buildings, using tools within CloudCompare. Removing such
features may yield a closer match between ground surfaces, given that the height and lateral
extent of vegetation may have changed over the summer months between aerial photo collection.
I will continue to investigate the effects on the low resolution of the pre-event imagery
compared to the post-event imagery. It is possible that the interpolation between known points
during creation of the dense cloud of pre-event imagery is introducing artifacts into that point
cloud, which is enhancing the mis-registration of the point clouds. However, working at the
resolution of the pre-event imagery will severely hamper my ability to discern any reliable signal
124
of discrete displacements, given that the individual measured displacements were so small
[Buwalda and St. Amand, 1955].
6.9 Conclusions
The 1952 Mw 7.3 Kern County earthquake was a damaging oblique-slip earthquake with
components of thrust and sinistral motion. Field mapping immediately following the mainshock
indicated primarily distributed deformation within a broad fault zone. I have attempted to
constrain this broad fault zone using a novel workflow deriving the three dimensional
deformation field using legacy stereo air photos. This method remains a work in progress but has
produced a semi-coherent signal in at least one of the three dimensions examined. Future work
will focus on improved co-registration of the pre-event and post-event point clouds.
6.10 Figure Captions
Figure 1: Satellite image with active faults in red [Jennings, 1994]. Epicenter of 1952 event is
shown with a yellow star at Wheeler Ridge. Rupture propagated to the northeast. Yellow box
outlines region of interest for Figures 2-5.
Figure 2: Post-event dense point cloud overlain on satellite imagery. Red lines indicate mapped
active trace of White Wolf fault that is thought to have ruptured in the 1952 event.
Figure 3: Cloud to cloud distances in east-west (X) direction. Note the consistent signal of
negative X (west) motion north of the fault, and positive X (east) south of the fault.
125
Figure 4: Cloud to cloud distances in the north-south (Y) direction.
Figure 5: Cloud to cloud distances in the vertical (Z) direction.
126
CHAPTER 7: Conclusions
This thesis has made significant additions to the understandings of the earthquake
behavior of faults along the Pacific plate boundary with the Australian plate (New Zealand) and
the North American plate (southern California). In Chapter 2, I dissected the southern California
plate boundary with respect to the Garlock fault. I used pre-existing geologic slip rate data,
coupled with the 11 Myr evolution of the fault and surrounding region, to determine the present-
day loading of the Garlock fault. This analysis highlighted the importance of the Garlock fault in
maintaining the present-day geometry of the southern California Pacific-North America plate
boundary, despite the obvious kinematic inefficiencies of this configuration (i.e., the Big Bend of
the San Andreas). Chapter 3, the paleoseismic study of the Hope fault along the Green Burn
reach of the Conway segment, revealed a chronology of 5 events over the past 2000 years, with
the four most recent of those events occurring over c. 600 years. This study also successfully
utilized a complementary event dating approach using an independent record of secondary
earthquake evidence (coseismic landslides). Chapter 4 developed the longest record of
incremental geologic slip rates on the Hope fault (Conway segment) at Hossack Station.
Incremental slip rates measured here fluctuate up to a factor of 200%. Chapter 5 complemented
the record developed at the western end of the Conway segment with an incremental record of
slip from the eastern end of the Conway segment at the Sawyers Creek site. Here, reinterpreted
offsets using previously determined ages of surfaces using the weathering rind technique yield
slip rates 300+% faster than those over similar time periods determined at Hossack Station.
Finally, Chapter 6 reviewed promising advances in a novel workflow to determine the full, three-
dimensional deformation field of the 1952 Mw 7.3 Kern County earthquake. Altogether, these
studies add to the growing dataset of complex behavior of faults in space and time, and have
127
potential to impact probabilistic seismic hazard assessments, either with direct data input
(Chapters 3-6) or understanding sub-system fault interactions of a larger plate boundary (Chapter
2).
128
REFERENCES
Anderson, D. L. (1975), Accelerate Plate Tectonics, Science (80-. )., 187(4181), 1077–1079.
Andrew, J. E., and J. D. Walker (2017), Path and amount of dextral fault slip in the Eastern
California shear zone across the central Mojave Desert, Geol. Soc. Am. Bull., 129(7–8),
855–868, doi:10.1130/B31527.1.
Andrew, J. E., J. D. Walker, and F. C. Monastero (2014), Evolution of the central Garlock fault
zone, California: A major sinistral fault embedded in a dextral plate margin, Bull. Geol. Soc.
Am., 127(1–2), 227–249, doi:10.1130/B31027.1.
Anon (n.d.), Biasi and Weldon 2009.pdf,
Armijo, R., F. Flerit, G. King, and B. Meyer (2004), Linear elastic fracture mechanics explains
the past and present evolution of the Aegean, Earth Planet. Sci. Lett., 217(1–2), 85–95,
doi:10.1016/S0012-821X(03)00590-9.
Atwater, T. (1970), Implications of Plate Tectonics for the Cenozoic Tectonic Evolution of
Western North America, Geol. Soc. Am. Bull., 81(12), 3513, doi:10.1130/0016-
7606(1970)81[3513:IOPTFT]2.0.CO;2.
Avouac, J. P., F. Ayoub, S. Leprince, O. Konca, and D. V. Helmberger (2006), The 2005, Mw
7.6 Kashmir earthquake: Sub-pixel correlation of ASTER images and seismic waveforms
analysis, Earth Planet. Sci. Lett., 249(3–4), 514–528, doi:10.1016/j.epsl.2006.06.025.
Ayoub, F., S. Leprince, and J. P. Avouac (2009), Co-registration and correlation of aerial
photographs for ground deformation measurements, ISPRS J. Photogramm. Remote Sens.,
64(6), 551–560, doi:10.1016/j.isprsjprs.2009.03.005.
Barnes, P. M., and J. C. Audru (1999), Recognition of active strike-slip faulting from high-
resolution marine seismic reflection profiles: Eastern Marlborough fault system, New
129
Zealand, Bull. Geol. Soc. Am., 111(4), 538–559, doi:10.1130/0016-
7606(1999)111<0538:ROASSF>2.3.CO;2.
Beauprêtre, S. et al. (2012), Finding the buried record of past earthquakes with GPR-based
palaeoseismology: A case study on the Hope fault, New Zealand, Geophys. J. Int., 189(1),
73–100, doi:10.1111/j.1365-246X.2012.05366.x.
Bemis, S. P., S. Micklethwaite, D. Turner, M. R. James, S. Akciz, S. T. Thiele, and H. A.
Bangash (2014), Ground-based and UAV-Based photogrammetry: A multi-scale, high-
resolution mapping tool for structural geology and paleoseismology, J. Struct. Geol.,
69(PA), 163–178, doi:10.1016/j.jsg.2014.10.007.
Berryman, K., A. Cooper, R. Norris, P. Villamor, R. Sutherland, T. Wright, E. Schermer, R.
Langridge, and G. Biasi (2012a), Late Holocene rupture history of the alpine fault in South
Westland, New Zealand, Bull. Seismol. Soc. Am., 102(2), 620–638,
doi:10.1785/0120110177.
Berryman, K. R. (1992), The Alpine fault, New Zealand: variation in Quaternary structural style
and geomorphic expression, Ann. Tectonicae Supple to Vol VI, VI, 126–163.
Berryman, K. R., U. A. Cochran, K. J. Clark, G. P. Biasi, R. M. Langridge, and P. Villamor
(2012b), Major Earthquakes Occur Regularly on an Isolated Plate Boundary Fault, Science
(80-. )., 336(June 29), 1690–1694.
Blythe, A. E., and N. Longinotti (2013), Exhumation of the southern Sierra Nevada – eastern
Tehachapi Mountains constrained by low-temperature thermochronology : Implications for
the initiation of the Garlock fault, Lithosphere, 5(3), 321–327, doi:10.1130/L252.1.
Bowman, D., G. King, and P. Tapponnier (2003), Slip partitioning by elastoplastic propagation
of oblique slip at depth., Science, 300(5622), 1121–1123.
130
Bronk Ramsey, C. (2017), OxCal Program, Version 4.3,
Bull, W. L., and P. L. K. Knuepfer (1987), Adjustments by the Charwell New Zealand, to uplift
and climatic changes, Geomorphology, 1, 15–32.
Burbank, D. W., and D. P. Whistler (1987), Temporally constrained tectonic rotations derived
from magnetostratigiraphic data: Implications for the initiation of the Garlock fault,
California, Geology, 15(12), 1172–1175, doi:10.1130/0091-
7613(1987)15<1172:TCTRDF>2.0.CO;2.
Burchfiel, B. C., K. V. Hodges, and L. H. Royden (1987), Geology of Panamint Valley - Saline
Valley Pull Apart System, California: Palinspastic Evidence for Low-Angle Geometry of a
Neogene Range-Bounding Fault, J. Geophys. Res., 92(B10), 10422–10426.
Burke, D. B., and M. M. Clark (1978), Late Quaternary activity along Garlock fault at Koehn
Lake, Fremont Valley, California, in Transactions-American Geophysical Union, p. 1126,
American Geophysical Union, Washington D.C.
Buwalda, J. P., and P. St. Amand (1955), Geological effects of the Arvin-Tehachapi earthquake,
in Earthquakes of Kern County, California during 1952, pp. 41–56.
Chapman, A., J. B. Saleeby, D. J. Wood, A. Piasecki, S. Kidder, M. N. Ducea, and K. A. Farley
(2012), Late Cretaceous gravitational collapse of the southern Sierra Nevada batholith,
California, Geosphere, 8(2), 314, doi:10.1130/GES00740.1.
Clark, K. J. et al. (2013), Deriving a long paleoseismic record from a shallow-water holocene
basin next to the Alpine fault, New Zealand, Bull. Geol. Soc. Am., 125(5–6), 811–832,
doi:10.1130/B30693.1.
Clark, K. J. et al. (2017), Highly variable coastal deformation in the 2016 Mw7.8 Kaikōura
earthquake reflects rupture complexity along a transpressional plate boundary, Earth Planet.
131
Sci. Lett., 474, 334–344, doi:10.1016/j.epsl.2017.06.048.
Clark, M. M. (1973), Map showing recently active breaks along the Garlock fault and associate
faults, California, scale 1:24,000, U.S. Geol. Surv., Map 1-741, 3 sheets.
Clark, M. M., and K. R. Lajoie (1974), Holocene behavior of the Garlock fault, in Geological
Society of America Abstracts with Programs (Cordilleran Section), volume 6.
Cochran, E. S., Y.-G. Li, P. M. Shearer, S. Barbot, Y. Fialko, and J. E. Vidale (2009), Seismic
and geodetic evidence for extensive, long-lived fault damage zones, Geology, 37(4), 315–
318, doi:10.1130/G25306A.1.
Cochran, U. A., K. J. Clark, J. D. Howarth, G. P. Biasi, R. M. Langridge, P. Villamor, K. R.
Berryman, and M. J. Vandergoes (2017), A plate boundary earthquake record from a
wetland adjacent to the Alpine fault in New Zealand refines hazard estimates, Earth Planet.
Sci. Lett., 464, 175–188, doi:10.1016/j.epsl.2017.02.026.
Cooper, A. F., and R. J. Norris (1990), Estimates for the timing of the last coseismic
displacement on the alpine fault, northern fiordland, new zealand, New Zeal. J. Geol.
Geophys., 33(2), 303–307, doi:10.1080/00288306.1990.10425688.
Cowan, H. A. (1990), Late Quaternary displacements on the Hope fault at Glynn Wye, north
Canterbury, New Zeal. J. Geol. Geophys., 33(2), 285–293,
doi:10.1080/00288306.1990.10425686.
Cowan, H. A. (1991), The North Canterbury earthquake of September 1, 1888, J. R. Soc. New
Zeal., 21(1), 1–12, doi:10.1080/03036758.1991.10416105.
Cowan, H. A., and M. S. McGlone (1991), Late Holocene displacements and characteristic
earthquakes on the Hope River segment of the Hope fault, New Zealand, J. R. Soc. New
Zeal., 21(4), 373–384, doi:10.1080/03036758.1991.10420834.
132
Cox, S. C., M. W. Stirling, F. Herman, M. Gerstenberger, and J. Ristau (2012), Potentially active
faults in the rapidly eroding landscape adjacent to the Alpine Fault , central Southern Alps ,
New Zealand, Tectonics, 31(Figure 1), 1–24, doi:10.1029/2011TC003038.
Crane, T. M. (2014), Qualitative comparison of offset surfaces between the central and eastern
Garlock fault, California State University, San Bernardino.
Davis, G. A., and B. C. Burchfiel (1973), Garlock Fault: An intracontinental transform structure,
Southern California, Geol. Soc. Am. Bull., 84, 1407–1422.
Dawson, T. E. (2003), Irregular recurrence of paleoearthquakes along the central Garlock fault
near El Paso Peaks, California, J. Geophys. Res., 108(B7), n/a-n/a,
doi:10.1029/2001JB001744.
Dawson, T. E., S. F. McGill, and T. K. Rockwell (2003), Irregular recurrence of
paleoearthquakes along the central Garlock fault near El Paso Peaks, California, J. Geophys.
Res. Solid Earth, 108(B7), doi:10.1029/2001JB001744.
DeMets, C., R. G. Gordon, and D. F. Argus (2010), Geologically current plate motions,
Geophys. J. Int., 181(1), 1–80, doi:10.1111/j.1365-246X.2009.04491.x.
DePascale, G. P., M. C. Quigley, and T. R. H. Davies (2014), Lidar reveals uniform Alpine fault
offsets and bimodal plate boundary rupture behavior , New Zealand, Geology, 42(5), 411–
414, doi:10.1130/G35100.1.
DePascale, G. P., N. Chandler-Yates, F. Dela Pena, P. Wilson, E. May, A. Twiss, and C. Cheng
(2016), Active tectonics west of New Zealand’s Alpine Fault: South Westland Fault Zone
activity shows Australian Plate instability, Geophys. Res. Lett., 43, 3120–3125,
doi:10.1002/2016GL068233.Received.
Dibblee, T. W. J. (1955), Geology of the southeastern margin of the San Jaoquin Valley,
133
California, in Earthquakes of Kern County, California during 1952, pp. 23–36.
Van Dissen, R. J. (1989), Late Quaternary faulting in the Kaikoura region, southeastern
Marlborough, New Zealand, Oregon State University.
Van Dissen, R. J., T. A. Little, R. M. Burke, P. J. Tonkin, K. P. Norton, S. N. Bacon, R. Bowers,
H. L. Goldstein, J. R. Redwine, and D. G. Sutherland (2016), Late Quaternary dextral slip
rate of the Kekerengu fault: New Zealand’s third fastest on-land fault, in New Zealand
GeoSciences Societ Conference, p. 89 pp., GeoSciences Society of New Zealand,
Miscellaneous Publication 142A, Wanaka, New Zealand.
Van Dissen, R., and A. Nicol (2009), Mid-late holocene paleoseismicity of the Eastern Clarence
Fault, Marlborough, New Zealand, New Zeal. J. Geol. Geophys., 52(3), 195–208,
doi:10.1080/00288300909509886.
Van Dissen, R., and R. S. Yeats (1991), Hope Fault, Jordan Thrust, and uplift of the Seaward
Kaikoura Range, New Zealand, Geology, 19(4), 393–396, doi:10.1130/0091-
7613(1991)019<0393:HFJTAU>2.3.CO;2.
Dokka, R. K., and C. J. Travis (1990a), Late Cenozoic strike-slip faultsing in the Mojave Desert,
California, Tectonics, 9(2), 311–340, doi:10.1029/TC009i002p00311.
Dokka, R. K., and C. J. Travis (1990b), Role of the Eastern California Shear Zone in
accommodating Pacific North American Plate motion, Geophys. Res. Lett., 17(9), 1323–
1326, doi:10.1029/GL017i009p01323.
Dolan, J. F., and B. D. Haravitch (2014), How well do surface slip measurements track slip at
depth in large strike-slip earthquakes? The importance of fault structural maturity in
controlling on-fault slip versus off-fault surface deformation, Earth Planet. Sci. Lett., 388,
38–47, doi:10.1016/j.epsl.2013.11.043.
134
Dolan, J. F., D. D. Bowman, and C. G. Sammis (2007), Long-range and long-term fault
interactions in Southern California, Geology, 35(9), 855–858, doi:10.1130/G23789A.1.
Dolan, J. F., S. F. McGill, and E. J. Rhodes (2015), A new mid-Holocene slip rate for the central
Garlock fault: Implications for the constancy of fault slip rates and system-level fault
behavior, in Southern California Earthquake Center Annual Meeting Proceedings Volume
XXV, p. 144.
Dolan, J. F., L. J. McAuliffe, E. J. Rhodes, S. F. McGill, and R. Zinke (2016), Extreme multi-
millennial slip rate variations on the Garlock fault, California: Strain super-cycles,
potentially time-variable fault strength, and implications for system-level earthquake
occurrence, Earth Planet. Sci. Lett., 446, 123–136, doi:10.1016/j.epsl.2016.04.011.
Du, Y., and A. Aydin (1996), Is the San Andreas big bend responsible for the Landers
earthquake and the eastern California shear zone?, Geology, 24(3), 219–222,
doi:10.1130/0091-7613(1996)024<0219:ITSABB>2.3.CO;2.
Duputel, Z., and L. Rivera (2017), Long-period analysis of the 2016 Kaikoura earthquake, Phys.
Earth Planet. Inter., 265, 62–66, doi:10.1016/j.pepi.2017.02.004.
Eaton, J. E. (1932), Decline of Great Basin, Southwestern United States, Am. Assoc. Pet. Geol.
Bull., 16(1).
Evans, E. L., W. R. Thatcher, F. F. Pollitz, and J. R. Murray (2016), Persistent slip rate
discrepancies in the eastern California (USA) shear zone, Geology, 44(9), 691–694,
doi:10.1130/G37967.1.
Faulds, J. E., C. D. Henry, and N. H. Hinz (2005), Kinematics of the northern Walker Lane : An
incipient transform fault along the Pacific – North American plate boundary, Geology,
33(6), 505–508, doi:10.1130/G21274.1.
135
Ferry, M., M. Meghraoui, N. A. Karaki, M. Al-Taj, and L. Khalil (2011), Episodic behavior of
the Jordan Valley section of the Dead Sea fault inferred from a 14-ka-long integrated
catalog of large earthquakes, Bull. Seismol. Soc. Am., 101(1), 39–67,
doi:10.1785/0120100097.
Fialko, Y., D. Sandwell, M. Simons, and P. Rosen (2005), Three-dimensional deformation
caused by the Bam, Iran, earthquake and the origin of shallow slip deficit, Nature,
435(7040), 295–299, doi:10.1038/nature03425.
Field, E. H. et al. (2013), Uniform California earthquake rupture forecast, version 3
(UCERF3) —The time-independent model.
Flerit, F., R. Armijo, G. King, and B. Meyer (2004), The mechanical interaction between the
propagating North Anatolian Fault and the back-arc extension in the Aegean, Earth Planet.
Sci. Lett., 224(3–4), 347–362, doi:10.1016/j.epsl.2004.05.028.
Frankel, K. L. et al. (2008), Active tectonics of the eastern California Shear Zone, Geol. Soc.
Am., Field Guid(09), 27–38, doi:10.1130/2008.fl.
Frankel, K. L., L. A. Owen, J. F. Dolan, J. R. Knott, Z. M. Lifton, R. C. Finkel, and T.
Wasklewicz (2015), Timing and rates of Holocene normal faulting along the Black
Mountains fault zone, Death Valley, USA, Lithosphere, 8(1), 3–22, doi:10.1130/L464.1.
Friedrich, A. M., B. P. Wernicke, N. A. Niemi, R. A. Bennett, and J. L. Davis (2003),
Comparison of geodetic and geologic data from the Wasatch region, Utah, and implications
for the spectral character of Earth deformation at periods of 10 to 10 million years, J.
Geophys. Res. Solid Earth, 108(B4), 1–23, doi:10.1029/2001JB000682.
Gan, W., J. L. Svarc, J. C. Savage, and W. H. Prescott (2000), Strain accumulation across the
Eastern California Shear Zone at latitude 36 30’N, J. Geophys. Res., 105(B7), 16299–
136
16236.
Ganev, P. N., J. F. Dolan, S. F. Mcgill, and K. L. Frankel (2012), Constancy of geologic slip rate
along the central Garlock fault: Implications for strain accumulation and release in southern
California, Geophys. J. Int., 190(2), 745–760, doi:10.1111/j.1365-246X.2012.05494.x.
Garfunkel, Z. (1974), Model for the Late Cenozoic Tectonic History of the Mojave Desert ,
California , and for Its Relation to Adjacent Regions, Bull. Geol. Soc. Am., 85(December),
1931, doi:10.1130/0016-7606(1974)85<1931.
Glazner, A. F., J. D. Walker, J. M. Bartley, and J. M. Fletcher (2002), Cenozoic evolution of the
Mojave block of souther California, in Geologic Evolution of the Mojave Desert and
Southewestern Basin and Range Geological Society of America Memoir 195, edited by A.
F. Glazner, J. D. Walker, and J. M. Bartley, pp. 19–42.
Gold, R. D., and E. Cowgill (2011), Deriving fault-slip histories to test for secular variation in
slip, with examples from the Kunlun and Awatere faults, Earth Planet. Sci. Lett., 301(1–2),
52–64, doi:10.1016/j.epsl.2010.10.011.
Gold, R. D., N. G. Reitman, R. W. Briggs, W. D. Barnhart, G. P. Hayes, and E. Wilson (2015),
On- and off-fault deformation associated with the September 2013 Mw 7.7 Balochistan
earthquake: Implications for geologic slip rate measurements, Tectonophysics,
660(September 2013), 65–78, doi:10.1016/j.tecto.2015.08.019.
Grapes, R., and G. Downes (1997), The 1855 Wairarapa, New Zealand, earthquake--Analysis of
historical data, Bull. New Zeal. Natl. Soc. Earthq. Eng., 30, 271–369.
Grapes, R., T. Little, and G. Downes (1998), Rupturing of the Awatere Fault during the 1848
October 16 Marlborough earthquake, New Zealand: Historical and present day evidence,
New Zeal. J. Geol. Geophys., 41(4), 387–399, doi:10.1080/00288306.1998.9514818.
137
Guest, B., T. L. Pavlis, H. Golding, and L. Serpa (2003), Chasing the Garlock: A study of
tectonic response to vertical axis rotation, Geology, 31(6), 553–556, doi:10.1130/0091-
7613(2003)031<0553.
Hamling, I. J. et al. (2017), Complex multifault rupture during the 2016 Mw7.8 Kaikōura
earthquake, New Zealand, Science (80-. )., 356(6334), doi:10.1126/science.aam7194.
Hatem, A. E., and J. F. Dolan (2015), Garlock Fault : What is your deal, in Southern California
Earthquake Center Annual Meeting, poster 233, Palm Springs.
Hatem, A. E., and J. F. Dolan (2018), A Model for the Initiation, Evolution, and Controls on
Seismic Behavior of the Garlock Fault, California, Geochemistry, Geophys. Geosystems, 19,
1–13, doi:10.1029/2017GC007349.
Hatem, A. E., M. L. Cooke, and E. H. Madden (2015), Evolving efficiency of restraining bends
within wet kaolin analog experiments, J. Geophys. Res. B Solid Earth, 120(3),
doi:10.1002/2014JB011735.
Hatem, A. E., J. F. Dolan, R. M. Langridge, R. Zinke, C. P. McGuire, E. J. Rhodes, and R. J.
Van Dissen (2016), Incremental slip rate and paleoseismic data from the eastern Hope fault,
New Zealand: the Hossack and Green Burn sites, in American Geophysical Union, Fall
General Assembly 2016, p. abstract T32B-07.
Hatem, A. E., M. L. Cooke, and K. Toeneboehn (2017), Strain localization and evolving
kinematic ef fi ciency of initiating strike-slip faults within wet kaolin experiments, J. Struct.
Geol., 101, 96–108, doi:10.1016/j.jsg.2017.06.011.
Hill, M. L., and T. W. J. Dibblee (1953), San Andreas, Garlock and Big Pine faults, California,
Geol. Soc. Am. Bull., 64, 443–468.
Hoffman, W. R. (2009), Late Pleistocene slip rates along the Panamint Valley fault zone, eastern
138
California, Pennsylvania State University.
Hogg, A. G. et al. (2013), SHCal13 Southern Hemisphere Calibration, 0–50,000 Years cal BP,
Radiocarbon, 55(04), 1889–1903, doi:10.2458/azu_js_rc.55.16783.
Hollingsworth, J., L. Ye, and J. P. Avouac (2017), Dynamically triggered slip on a splay fault in
the Mw7.8, 2016 Kaikoura (New Zealand) earthquake, Geophys. Res. Lett., 44(8), 3517–
3525, doi:10.1002/2016GL072228.
Holt, W. E., and A. J. Haines (1995), The kinematics of northern South Island, New Zealand,
determined from geologic slip rates, J. Geophys. Res., 100(B9), 17991–18010.
Howarth, J. D., S. J. Fitzsimons, R. J. Norris, and G. E. Jacobsen (2012), Lake sediments record
cycles of sediment flux driven by large earthquakes on the Alpine fault, New Zealand,
Geology, 40(12), 1091–1094, doi:10.1130/G33486.1.
Howarth, J. D., S. J. Fitzsimons, R. J. Norris, and G. E. Jacobsen (2014), Lake sediments record
high intensity shaking that provides insight into the location and rupture length of large
earthquakes on the Alpine Fault, New Zealand, Earth Planet. Sci. Lett., 403, 340–351,
doi:10.1016/j.epsl.2014.07.008.
Howarth, J. D., S. J. Fitzsimons, R. J. Norris, R. Langridge, and M. J. Vandergoes (2016), A
2000 yr rupture history for the Alpine fault derived from Lake Ellery, South Island, New
Zealand, Bull. Geol. Soc. Am., 128(3–4), 627–643, doi:10.1130/B31300.1.
Howarth, J. D. et al. (2018), Past large earthquakes on the Alpine Fault : paleoseismological
progress and future directions, New Zeal. J. Geol. Geophys., 61(3), 309–328,
doi:10.1080/00288306.2018.1464658.
Humphreys, E. D., and R. J. Weldon (1994), Deformation acdross the western United States: A
local estimate of Pacific-North America transform deformation, J., 99(B10), 19975–20010.
139
Jahns, R., B. W. Troxel, and L. A. Wright (1971), Some structural implication of a late
Precambrian-Cambrian section in the Avawatz Mountains, California, in Geological Society
of America Abstracts with Programs (Cordilleran Section) 3(2), p. 140.
Jenkins, O. P., and G. B. Oakeshott (Eds.) (1955), Earthquakes in Kern County California
During 1952.
Jennings, C. W. (Compiler) (1994), Fault activity map of California and adjacent areas, Calif.
Dep. Conserv. Div. Mines Geol. Geol. Data Map No. 6, scale 1750,000.
Kagan, Y. Y., and D. D. Jackson (1991), Long-term earthquake clustering, Geophys. J. Int.,
104(1), 117–134, doi:10.1111/j.1365-246X.1991.tb02498.x.
Kaiser, A. et al. (2017), The 2016 Kaikōura, New Zealand, Earthquake: Preliminary
Seismological Report, Seismol. Res. Lett., 88(3), 727–739, doi:10.1785/0220170018.
Kearse, J. et al. (2017), Onshore to Offshore Ground-Surface and Seabed Rupture of the Jordan –
Kekerengu – Needles Fault Network during the 2016 Mw 7.8 Kaikoura Earthquake , New
Zealand, Bull. Seism, 108(3B), 1573–1595, doi:10.1785/0120170304.
Khajavi, N., R. M. Langridge, M. C. Quigley, C. Smart, A. Rezanejad, and F. Martín-González
(2016), Late Holocene rupture behavior and earthquake chronology on the Hope fault, New
Zealand, Bull. Geol. Soc. Am., 128(11–12), 1736–1761, doi:10.1130/B31199.1.
Khajavi, N., A. Nicol, M. C. Quigley, and R. M. Langridge (2018), Temporal slip-rate stability
and variations on the Hope Fault, New Zealand, during the late Quaternary, Tectonophysics,
738–739(May), 112–123, doi:10.1016/j.tecto.2018.05.001.
King, G. C., A. Friedrich, R. Armijo, Y. Gaudemer, and D. Bowman (2004), Speculations on the
Evolution of the East California Shear Zone and Associated Structures by Fault
Propagation: Comparison with the Mechanics of Anatolia and the Aegean, in American
140
Geophysical Union, Fall Meeting 2004, abstract #T31A-1258.
Knuepfer, P. L. K. (1988), Estimating ages of late Quaternary stream terraces from analysis of
weathering rinds and soils, Geol. Soc. Am. Bull., 100(8), 1224–1236, doi:10.1130/0016-
7606(1988)100<1224:EAOLQS>2.3.CO;2.
Kozaci, Ö., J. Dolan, R. Finkel, and R. Hartleb (2007), Late Holocene slip rate for the North
Anatolian fault, Turkey, from cosmogenic36CI geochronology: Implications for the
constancy of fault loading and strain release rates, Geology, 35(10), 867–870,
doi:10.1130/G23187A.1.
Langridge, R., J. Campbell, N. Hill, V. Pere, J. Pope, J. Pettinga, B. Estrada, and K. Berryman
(2003), Paleoseismology and slip rate of the Conway Segment of the Hope Fault at
Greenburn Stream, South Island, New Zealand, Ann. Geophys., 46(5), 1119–1140,
doi:10.4401/ag-3449.
Langridge, R. M., and K. R. Berryman (2005), Morphology and slip rate of the hurunui section
of the hope fault, South Island, New Zealand, New Zeal. J. Geol. Geophys., 48(1), 43–57,
doi:10.1080/00288306.2005.9515097.
Langridge, R. M., P. C. Almond, and R. P. Duncan (2013), Timing of late holocene
paleoearthquakes on the Hurunui segment of the hope fault: Implications for plate boundary
strain release through South Island, New Zealand, Bull. Geol. Soc. Am., 125(5–6), 756–775,
doi:10.1130/B30674.1.
Langridge, R. M. et al. (2018), Coseismic Rupture and Preliminary Slip Estimates for the
Papatea Fault and Its Role in the 2016 Mw 7.8 Kaikōura, New Zealand, Earthquake, Bull.
Seismol. Soc. Am., 108(3B), 1596–1622, doi:10.1785/0120170336.
Le, K., J. Lee, L. A. Owen, and R. Finkel (2007), Late Quaternary slip rates along the Sierra
141
Nevada frontal fault zone, California: Slip partitioning across the western margin of the
Eastern California Shear Zone-Basin and Range province, Bull. Geol. Soc. Am., 119(1–2),
240–256, doi:10.1130/B25960.1.
Lee, J., D. F. Stockli, L. A. Owen, R. C. Finkel, and R. Kislitsyn (2009), Exhumation of the Inyo
Mountains, California: Implications for the timing of extension along the western boundary
of the Basin and Range Province and distribution of dextral fault slip rates across the
eastern California shear zone, Tectonics, 28(1), doi:10.1029/2008TC002295.
Lensen, G. J. (1968), Analysis of Progressive Fault Displacement During Downcutting at the
Branch River Terraces, South Island, New Zealand, Geol. Soc. Am. Bull., 79(May), 545–
556.
Leprince, S., F. Ayoub, Y. Klinger, and J. P. Avouac (2007), Co-Registration of Optically
Sensed Images and Correlation (COSI-Corr): An operational methodology for ground
deformation measurements, Int. Geosci. Remote Sens. Symp., 1943–1946,
doi:10.1109/IGARSS.2007.4423207.
Lienkaemper, J. J., and C. Bronk Ramsey (2009), OxCal: Versatile Tool for Developing
Paleoearthquake Chronologies--A Primer, Seismol. Res. Lett., 80(3), 431–434,
doi:10.1785/gssrl.80.3.431.
Litchfield, N. J. et al. (2014), A model of active faulting in New Zealand, New Zeal. J. Geol.
Geophys., 57(1), 32–56, doi:10.1080/00288306.2013.854256.
Litchfield, N. J. et al. (2018), Surface Rupture of Multiple Crustal Faults in the M w 7 . 8 2016
Kaikōura Earthquake , New Zealand, Bull. Seismol. Soc. Am., XX(Xx),
doi:10.1785/0120170300.
Little, T. A., and A. Jones (1998), Seven million years of strike-slip and related off-fault
142
deformation, northeastern Marlborough fault system, South Island, New Zealand, Tectonics,
17(2), 285–302.
Little, T. A., R. Van Dissen, E. Schermer, and R. Carne (2009), Late Holocene surface ruptures
on the southern Wairarapa fault, New Zealand: Link between earthquakes and the uplifting
of beach ridges on a rocky coast, Lithosphere, 1(1), 4–28, doi:10.1130/L7.1.
Little, T. A., R. Van Dissen, J. Kearse, K. Norton, A. Benson, and N. Wang (2018), Kekerengu
fault, New Zealand: timing and size of Late Holocene surface ruptures, Bull. Seismol. Soc.
Am., 108(3B), 1556–1572, doi:10.1785/0120170152.
Loomis, D. P., and D. W. Burbank (1988), The stratigraphic evolution of the El Paso basin,
southern California: implications for the Miocene development of the Garlock Fault and the
Sierra Nevada, Geol. Soc. Am. Bull., 100(1), 12–28, doi:10.1130/0016-
7606(1988)100<0012.
Loveless, J. P., and B. J. Meade (2011), Stress modulation on the San Andreas fault by
interseismic fault: System interactions, Geology, 39(11), 1035–1038,
doi:10.1130/G32215.1.
Luyendyk, B. P. (1991), A model for Neogene crustal rotations, transtension, and transpression
in southern California, Geol. Soc. Am. Bull., 103(11), 1528–1536, doi:10.1130/0016-
7606(1991)103<1528:AMFNCR>2.3.CO;2.
Luyendyk, B. P., M. J. Kamerling, and R. Terres (1980), Geometric model for Neogene crustal
rotations in southern California Geometrie model for Neogene crustal rotations in southern
California, Geol. Soc. Am. Bull., 91, 211–217, doi:10.1130/0016-7606(1980)91<211.
Madugo, C. M., J. F. Dolan, and R. D. Hartleb (2012), New paleoearthquake ages from the
western Garlock fault: Implications for regional earthquake occurrence in Southern
143
California, Bull. Seismol. Soc. Am., 102(6), 2282–2299, doi:10.1785/0120110310.
Malin, P. E., E. D. Goodman, T. L. Henyey, Y. G. Li, D. A. Okaya, and J. B. Saleeby (1995),
Significance of seismic reflections beneath a tilted exposure of deep continental crust,
Tehachapi Mountains, California, J. Geophys. Res. Solid Earth, 100(B2), 2069–2087,
doi:10.1029/94JB02127.
Manighetti, I., M. Campillo, C. Sammis, P. M. Mai, and G. King (2005), Evidence for self-
similar, triangular slip distributions on earthquakes: Implications for earthquake and fault
mechanics, J. Geophys. Res. Solid Earth, 110(5), 1–25, doi:10.1029/2004JB003174.
Marco, S., M. Stein, A. Agnon, and H. Ron (1996), Long-term earthquake clustering: A 50,000-
year paleoseismic record in the Dead Sea Graben, J. Geophys. Res., 101(B3), 6179–6191.
Mason, D. P. M., T. A. Little, and R. J. Van Dissen (2006a), Rates of active faulting during later
Quaternary fluvial terrace formation at Saxton River, Awatere fault, New Zealand, Bull.
Geol. Soc. Am., 118(11–12), 1431–1446, doi:10.1130/B25961.1.
Mason, D. P. M., T. A. Little, and R. J. Van Dissen (2006b), Refinements to the paleoseismic
chronology of the eastern awatere fault from trenches near upcot saddle, Marlborough, New
Zealand, New Zeal. J. Geol. Geophys., 49(3), 383–397,
doi:10.1080/00288306.2006.9515175.
Massey, C. et al. (2018), Landslides Triggered by the 14 November 2016 Mw 7.8 Kaikoura
Earthquake, New Zealand, Bull. Seismol. Soc. Am., 108(3), 1630–1648,
doi:10.1785/0120170305.
Matti, J. C., and D. M. Morton (1993), Paleogeographic evolution of the San Andreas fault in
southern California: A reconstruction based on a new cross-fault correlation, in Geological
Society of American Memoirs 178, edited by R. E. Powell, R. J. Weldon, and J. C. Matti, pp.
144
107–160, GSA Memoirs.
McAuliffe, L. J., J. F. Dolan, E. Kirby, J. C. Rollins, B. Haravitch, S. Alm, and T. M. Rittenour
(2013), Paleoseismology of the southern Panamint Valley fault: Implications for regional
earthquake occurrence and seismic hazard in southern California, J. Geophys. Res. Solid
Earth, 118(9), 5126–5146, doi:10.1002/jgrb.50359.
McClusky, S. C., S. C. Bjornstad, B. H. Hager, R. W. King, B. J. Meade, M. M. Miller, F. C.
Monastero, and B. J. Souter (2001), Present day kinematics of the Eastern California shear
zone from a geodetically constrained block model, Geophys. Res. Lett., 28(17), 3369–3372,
doi:10.1029/2001GL013091.
McGill, S., and T. Rockwell (1998), Ages of Late Holocene earthquakes on the central Garlock
fault near El Paso Peaks, California, J. Geophys. Res., 103(B4), 7265–7279,
doi:10.1029/97JB02129.
McGill, S. F. (1992), Paleoseismology and neotectonics of the central and eastern Garlock fault,
California, California Institue of Technology.
McGill, S. F., and K. Sieh (1991), Surficial offsets on the central and eastern Garlock fault
associated with prehistoric earthquakes, J. Geophys. Res., 96(B13), 21597–21621.
McGill, S. F., and K. Sieh (1993), Holocene slip rate of the central Garlock fault in Southeastern
Searles Valley , California, J. Geophys. Res., 98, 14217–14231.
McGill, S. F., S. G. Wells, S. K. Fortner, H. A. Kuzma, and J. D. McGill (2009), Slip rate of the
Western Garlock fault, at Clark Wash, near lone tree Canyon, Mojave desert California,
Bull. Geol. Soc. Am., 121(3–4), 536–554, doi:10.1130/B26123.1.
McKay, A. (1890), On the earthquakes of September 1888 in the Amuri and Marlborough
districts of the South Island, New Zeal. Geol. Surv., 20, 1–16.
145
McMorran, T. J. (1991), The Hope Fault at Hossack Station east of Hanmer Basin, North
Canterbury, University of Canterbury.
McQuarrie, N., and B. P. Wernicke (2005), An animated tectonic reconstruction of southwestern
North America since 36 Ma, Geosphere, 1(3), 147–172, doi:10.1130/GES00016.1.
Meade, B. J., and B. H. Hager (2005), Block models of crustal motion in southern California
constrained by GPS measurements, J. Geophys. Res., 110(B03403), 1–19,
doi:10.1029/2004JB003209.
Menard, H. W., and T. Atwater (1968), Changes in direction of sea floor spreading, Nature, 219,
463–467.
Menard, H. W., and T. Atwater (1969), Origin of fracture zone topography, Nature, 222, 1037–
1040.
Milliner, C. W. D., J. F. Dolan, J. Hollingsworth, S. Leprince, F. Ayoub, and C. G. Sammis
(2015), Quantifying near-field and off-fault deformation patterns of the 1992
M
w
7.3 Landers earthquake, Geochemistry, Geophys. Geosystems, 16(5), 1577–
1598, doi:10.1002/2014GC005693.
Milliner, C. W. D., J. F. Dolan, J. Hollingsworth, S. Leprince, and F. Ayoub (2016a),
Comparison of coseismic near-field and off-fault surface deformation patterns of the 1992
Mw7.3 Landers and 1999 Mw7.1 Hector Mine earthquakes: Implications for controls on the
distribution of surface strain, Geophys. Res. Lett., 43(19), 10,115-10,124,
doi:10.1002/2016GL069841.
Milliner, C. W. D., C. Sammis, A. A. Allam, J. F. Dolan, J. Hollingsworth, S. Leprince, and F.
Ayoub (2016b), Resolving Fine-Scale Heterogeneity of Co-seismic Slip and the Relation to
Fault Structure, Sci. Rep., 6(November 2015), 1–9, doi:10.1038/srep27201.
146
Monastero, F. C., A. E. Sabin, and J. D. Walker (1997), Evidence for post-early Miocene
initiation of movement on the Garlock fault from offset of the Cudahy Camp Formation,
east-central California, Geology, 25(3), 247–250, doi:10.1130/0091-
7613(1997)025<0247:EFPEMI>2.3.CO;2.
Monastero, F. C., J. D. Walker, A. M. Katzenstein, and A. E. Sabin (2002), Neogene evolution
of the Indian Well Valley, east-central California, in Geologica Evolution of the Mojave
Desert and Southwestern Basn and Range: Geological Society of America Memoir 195,
edited by A. F. Glazner, J. D. Walker, and J. M. Bartley, pp. 199–228.
Nicol, A., and R. Van Dissen (2018), A 6000-year record of surface-rupturing paleoearthquakes
on the Wairau Fault, New Zealand, New Zeal. J. Geol. Geophys., 61(3), 341–358,
doi:10.1080/00288306.2018.1498360.
Nicol, A., R. Robinson, R. J. Van Dissen, and A. Harvison (2016), Variability of recurrence
interval and single-event slip for surface-rupturing earthquakes in New Zealand, New Zeal.
J. Geol. Geophys., 59(1), 97–116, doi:10.1080/00288306.2015.1127822.
Nicol, A. et al. (2018), Preliminary Geometry , Displacement , and Kinematics of Fault Ruptures
in the Epicentral Region of the 2016 M w 7 . 8 Kaikoura , New Zealand , Earthquake, Bull.
Seismol. Soc. Am., 108(3B), 1521–1539, doi:10.1785/0120170329.
Niemi, N. A., J. T. Buscher, J. A. Spotila, M. A. House, and S. A. Kelley (2013), Insights from
low-temperature thermochronometry into transpressional deformation and crustal
exhumation along the San Andreas fault in the western Transverse Ranges, California,
Tectonics, 32(6), 1602–1622, doi:10.1002/2013TC003377.
Nissen, E., A. K. Krishnan, J. R. Arrowsmith, and S. Saripalli (2012), Three-dimensional surface
displacements and rotations from differencing pre-and post-earthquake LiDAR point
147
clouds, Geophys. Res. Lett., 39(16), 1–6, doi:10.1029/2012GL052460.
Noriega, G. R., J. R. Arrowsmith, L. B. Grant, and J. J. Young (2006), Stream channel offset and
Late Holocene slip rate of the San Andreas fault at the Van Matre Ranch site, Carrizo Plain,
California, Bull. Seismol. Soc. Am., 96(1), 33–47, doi:10.1785/0120050094.
Norris, R. J., and A. F. Cooper (2001), Late Quaternary slip rates and slip partitioning on the
Alpine Fault, New Zealand, J. Struct. Geol., 23(2–3), 507–520, doi:10.1016/S0191-
8141(00)00122-X.
Norton, I. (2011), Two-stage formation of Death Valley, Geosphere, 7(1), 171–182,
doi:10.1130/GES00588.1.
Nur, A., H. Ron, and G. C. Beroza (1993), The Nature of the Landers-Mojave Earthquake Line,
Science (80-. )., 261(5118), 201–203.
Oakeshott, G. B. (1955), The Kern County earthquakes in California’s geologic history, in
Earthquakes in Kern County, California during 1952, pp. 15–22.
Onderdonk, N. W., S. F. McGill, and T. K. Rockwell (2015), Short-term variations in slip rate
and size of prehistoric earthquakes during the past 2000 years on the northern San Jacinto
fault zone, a major plate-boundary structure in southern California, Lithosphere, 7(3), 211–
234, doi:10.1130/L393.1.
Oskin, M., and A. Iriondo (2004), Large-magnitude transient strain accumulation on the
Blackwater fault, Eastern California shear zone, Geology, 32(4), 313–316,
doi:10.1130/G20223.1.
Oskin, M., L. Perg, E. Shelef, M. Strane, E. Gurney, B. Singer, and X. Zhang (2008), Elevated
shear zone loading rate during an earthquake cluster in eastern California, Geology, 36(6),
507–510, doi:10.1130/G24814A.1.
148
Oskin, M. E. et al. (2012), Near-Field Deformation from the El Mayor-Cucapah Earthquake
Revealed by Differential LIDAR, Science (80-. )., 335(6069), 702–705,
doi:10.1126/science.1213778.
De Pascale, G. P., and R. M. Langridge (2012), New on-fault evidence for a great earthquake in
A.D. 1717, central alpine fault, New Zealand, Geology, 40(9), 791–794,
doi:10.1130/G33363.1.
Peltzer, G., F. Crampe, S. Hensley, and P. Rosen (2001), Transient strain accumulation and fault
interaction in the Eastern California shear zone, Geology, 19(11), 975–978,
doi:10.1029/2000JB000127.
Petersen, M. D. et al. (2015), The 2014 United States National Seismic Hazard Model, Earthq.
Spectra, 31(December), S1–S30, doi:10.1193/120814EQS210M.
Pollitz, F. F., R. Bürgmann, and P. Banerjee (2011), Geodetic slip model of the 2011 M9.0
Tohoku earthquake, Geophys. Res. Lett., 38(17), 6–11, doi:10.1029/2011GL048632.
Pondard, N., and P. M. Barnes (2010), Structure and paleoearthquake records of active
submarine faults, Cook Strait, New Zealand: Implications for fault interactions, stress
loading, and seismic hazard, J. Geophys. Res. Solid Earth, 115(12), 1–31,
doi:10.1029/2010JB007781.
Rattenbury, M. S., D. B. Townsend, and M. R. Johnston (2006), Geology of the Kaikoura area.
Resor, P. G., M. L. Cooke, S. T. Marshall, and E. H. Madden (2018), Influence of Fault
Geometry on the Spatial Distribution of Long‐Term Slip with Implications for Determining
Representative Fault‐Slip Rates, Bull. Seismol. Soc. Am., XX(Xx), doi:10.1785/0120170332.
Reyners, M. (2018), Impacts of Hikurangi Plateau subduction on the origin and evolution of the
Alpine Fault, New Zeal. J. Geol. Geophys., 0(0), 1–12,
149
doi:10.1080/00288306.2018.1454481.
Rhodes, E. J. (2015), Dating sediments using potassium feldspar single-grain IRSL: Initial
methodological considerations, Quat. Int., 362, 14–22, doi:10.1016/j.quaint.2014.12.012.
Rittase, W. M., E. Kirby, E. McDonald, J. D. Walker, J. Gosse, J. Q. G. Spencer, and A. J. Herrs
(2014), Temporal variations in Holocene slip rate along the central Garlock fault, Pilot
Knob Valley, California, Lithosphere, 6(1), 48–58, doi:10.1130/L286.1.
Robinson, R. (2004), Potential earthquake triggering in a complex fault network: The northern
South Island, New Zealand, Geophys. J. Int., 159(2), 734–748, doi:10.1111/j.1365-
246X.2004.02446.x.
Robinson, R., R. Van Dissen, and N. Litchfield (2011), Using synthetic seismicity to evaluate
seismic hazard in the Wellington region, New Zealand, Geophys. J. Int., 187(1), 510–528,
doi:10.1111/j.1365-246X.2011.05161.x.
Rockwell, T. K., S. Lindvall, M. Herzberg, D. Murbach, T. Dawson, and G. Berger (2000a),
Paleoseismology of the Johnson Valley , Kickapoo , and Homestead Valley Faults :
Clustering of Earthquakes in the Eastern California Shear Zone, Bull. Seismol. Soc. Am.,
90(5), 1200–1236.
Rockwell, T. K., S. Lindvall, M. Herzberg, D. Murbach, T. Dawson, and G. Berger (2000b),
Paleoseismology of the Johnson Valley , Kickapoo , and Homestead Valley Faults :
Clustering of Earthquakes in the Eastern California Shear Zone, , (October), 1200–1236.
Rodgers, D. W., and T. A. Little (2006), World’s largest coseismic strike-slip offset: The 1855
rupture of the Wairarapa Fault, New Zealand, and implications for displacement/length
scaling of continental earthquakes, J. Geophys. Res. Solid Earth, 111(12), 1–19,
doi:10.1029/2005JB004065.
150
Roquemore, G. R., P. E. Smith, and E. W. Banks (1982), Holocene earthquake activity of the
eastern Garlock fault in Christmas Canyon, San Bernardino County, California, in
Geological Society of America Abstracts with Programs, p. 228.
Rowe, C. D. et al. (2018), Geometric Complexity of Earthquake Rupture Surfaces Preserved in
Pseudotachylyte Networks, J. Geophys. Res., 123(2012), 7998–8015,
doi:10.1029/2018JB016192.
Salisbury, J. B., J. R. Arrowsmith, N. Brown, T. Rockwell, S. Akciz, and L. G. Ludwig (2018),
The age and origin of small offsets at Van Matre ranch along the San Andreas fault in the
Carrizo Plain, California, Bull. Seismol. Soc. Am., 108(2), 639–653,
doi:10.1785/0120170162.
Scharer, K. M., R. J. Weldon, T. E. Fumal, and G. P. Biasi (2007), Paleoearthquakes on the
southern San Andreas Fault, Wrightwood, California, 3000 to 1500 B.C.: A new method for
evaluating paleoseismic evidence and earthquake horizons, Bull. Seismol. Soc. Am., 97(4),
1054–1093, doi:10.1785/0120060137.
Schermer, E. R., B. P. Luyendyk, and S. Cisowski (1996), Late Cenozoic structures and tectonics
of the northern Mojave Desert, Tectonics, 15(5), 905–932.
Scholz, C. H. (2012), The Mechanics of Earthquaks and Faulting, Cambridge University Press.
Scott, C. P., J. R. Arrowsmith, E. Nissen, L. Lajoie, T. Maruyama, and T. Chiba (2018), The M7
2016 Kumamoto, Japan, Earthquake: 3D deformation along the fault and within damage
zone constrained from differential lidar topography, J. Geophys. Res. Solid Earth,
doi:10.1029/2018JB015581.
Shelef, E., and M. Oskin (2010), Deformation processes adjacent to active faults: Examples from
eastern California, J. Geophys. Res. Solid Earth, 115(5), 1–24, doi:10.1029/2009JB006289.
151
Shimazaki, K., and T. Nakata (1980), Time-predictable recurrence model for large earthquakes,
Geophys. Res. Lett., 7(4), 279–282.
Smith, G. I. (1962), Large lateral displacement on Garlock fault, California, as measured from
offset dike swarm, Am. Assoc. Pet. Geol. Bull., 46(1).
Smith, G. I., and K. B. Ketner (1970), Lateral displacement on the Garlock fault, southeastern
California, suggested by offset sections of similar metasedimentary rocks, USGS.
Snow, J. K., and B. P. Wernicke (2000), Cenozoic tectonism in the central basin and range:
Magnitude, rate, and distribution of upper crustal strain, Am. J. Sci., 300(9), 659–719,
doi:10.2475/ajs.300.9.659.
Spotila, J. A., and K. B. Anderson (2004), Fault interaction at the junction of the Transverse
Ranges and Eastern California shear zone: A case study of intersecting faults,
Tectonophysics, 379(1–4), 43–60, doi:10.1016/j.tecto.2003.09.016.
Stein, R. S., and W. Thatcher (1981), Seismic and aseismic deformation associated with the 1952
Kern County, California, earthquake and relationship to the Quaternary history of the White
Wolf Fault, J. Geophys. Res., 86(6), 4913–4928.
Stirling, M. et al. (2012), National seismic hazard model for New Zealand: 2010 update, Bull.
Seismol. Soc. Am., 102(4), 1514–1542, doi:10.1785/0120110170.
Stuart, W. D. (1991), Cause of the Garlock fault, in Geological Society of America, p. 198.
Styron, R. (2018), The impact of earthquake cycle variability on neotectonic and paleoseismic
slip rate estimates, Solid Earth Discuss., (May), 1–18.
Sutherland, R., K. Berryman, and R. Norris (2006), Quaternary slip rate and geomorphology of
the Alpine fault: Implications for kinematics and seismic hazard in southwest New Zealand,
Bull. Geol. Soc. Am., 118(3–4), 464–474, doi:10.1130/B25627.1.
152
Sutherland, R. et al. (2007), Do Great Earthquakes Occur on the Alpine Fault in Central South
Island, New Zealand?, A Cont. Plate Bound. Tectonics South Island, New Zeal., 235–251,
doi:10.1029/175GM12.
Thatcher, W., J. C. Savage, and R. W. Simpson (2016), The Eastern California Shear Zone as the
northward extension of the southern San Andreas fault, J. Geophys. Res. Solid Earth, 121,
1–14, doi:10.1002/2016JB013070.Received.
Titus, S. J., S. Crump, Z. McGuire, E. Horsman, and B. Housen (2011), Using vertical axis
rotations to characterize off-fault deformation across the san andreas fault system, central
California, Geology, 39(8), 711–714, doi:10.1130/G31802.1.
Troxel, B. W., L. A. Wrigth, and R. H. Jahns (1972), Evidence for differential displacement
along the Garlock fault zone, California, in Geological Society of America Abstracts with
Programs, p. 250.
Walcott, R. I. (1998), Modes of oblique compression: Late cenozoic tectonics of the South Island
of New Zealand, Rev. Geophys., 36(1), 1–26, doi:10.1029/97RG03084.
Wallace, L. M., J. Beavan, R. McCaffrey, K. Berryman, and P. Denys (2007), Balancing the
plate motion budget in the South Island, New Zealand using GPS, geological and
seismological datas, Geophys. J. Int., 168(1), 332–352, doi:10.1111/j.1365-
246X.2006.03183.x.
Wallace, L. M., P. Barnes, J. Beavan, R. Van Dissen, N. Litchfield, J. Mountjoy, R. Langridge,
G. Lamarche, and N. Pondard (2012), The kinematics of a transition from subduction to
strike-slip: An example from the central New Zealand plate boundary, J. Geophys. Res.
Solid Earth, 117(2), doi:10.1029/2011JB008640.
Wechsler, N., T. K. Rockwell, Y. Klinger, P. Štěpančíková, M. Kanari, S. Marco, and A. Agnon
153
(2014), A paleoseismic record of earthquakes for the dead sea transform fault between the
first and seventh centuries C.E.: Nonperiodic behavior of a plate boundary fault, Bull.
Seismol. Soc. Am., 104(3), 1329–1347, doi:10.1785/0120130304.
Wechsler, N., T. K. Rockwell, and Y. Klinger (2018), Variable slip-rate and slip-per-event on a
plate boundary fault: The Dead Sea fault in northern Israel, Tectonophysics, 722(October
2017), 210–226, doi:10.1016/j.tecto.2017.10.017.
Weldon, R., K. Scharer, T. Fumal, and G. Biasi (2004), Wrightwood and the earthquake cycle:
What a long recurrence record tells us about how faults work, Geol. Soc. Am. Today, 14(9),
4–10, doi:10.1130/1052-5173(2004)014<4.
Weldon, R. J., and K. E. Sieh (1985), Holocene rate of slip and tentative recurrence interval for
large earthquakes of the San Andreas fault, Cajon Pass, southern California., Geol. Soc. Am.
Bull., 96(6), 793–812, doi:10.1130/0016-7606(1985)96<793:HROSAT>2.0.CO;2.
Wells, A., M. D. Yetton, R. P. Duncan, and G. H. Stewart (1999), Prehistoric dates of the most
recent Alpine fault earthquakes, New Zealand, Geology, 27(11), 995–998,
doi:10.1130/0091-7613(1999)027<0995:PDOTMR>2.3.CO;2.
Wen, Y.-Y., K.-F. Ma, and B. Fry (2018), Multiple-Fault, Slow Rupture of the 2016 Mw 7.8
Kaikoura, New Zealand, Earthquake: Compelementary Insights from Teleseismic and
Geodetic Data, Bull. Seismol. Soc. Am., XX(April), 1–10, doi:10.1785/0120170285.
Wernicke, B., and J. K. Snow (1998), Cenozoic tectonism in the central basin and range: Motion
of the Sierran Great Valley block, Int. Geol. Rev., 40(5), 403–410,
doi:10.1080/00206819809465217.
Van Der Woerd, J., P. Tapponnier, F. J. Ryerson, A. Meriaux, B. Meyer, Y. Gaudemer, R. C.
Finkel, M. W. Caffee, Z. Guoguang, and X. Zhiqin (2017), Uniform postglacial slip-rate
154
along the central 600 km of the Kunlun Fault (Tibet), from 26Al, 10Be, and 14C dating of
riser offsets, and climatic origin of the regional morphology, Geophys. J. Int.,
148(December), 356–388.
Wood, D. J., and J. B. Saleeby (1997), Late cretaceous-paleocene extensional collapse and
disaggregation of the southernmost sierra nevada batholith, Int. Geol. Rev., 39(11), 973–
1009, doi:10.1080/00206819709465314.
Wood, R. A., J. R. Pettinga, S. Bannister, G. Lamarche, and T. J. McMorran (1994), Structure of
the Hanmer strike-slip basin , Hope fault , New Zealand Structure of the Hanmer strike-slip
basin , Hope fault , New Zealand, , (November), 1459–1473, doi:10.1130/0016-
7606(1994)106<1459.
Xu, X., X. Tong, D. T. Sandwell, C. W. D. Milliner, J. F. Dolan, J. Hollingsworth, S. Leprince,
and F. Ayoub (2016), Refining the shallow slip deficit, Geophys. J. Int., 204(3), 1867–1886,
doi:10.1093/gji/ggv563.
Yang, W., and E. Hauksson (2013), The tectonic crustal stress field and style of faulting along
the Pacific North America plate boundary in southern California, Geophys. J. Int., 194(1),
100–117, doi:10.1093/gji/ggt113.
Zinke, R., J. Hollingsworth, and J. F. Dolan (2014), Surface slip and off-fault deformation
patterns in the 2013 MW 7.7 Balochistan, Pakistan earthquake: Implications for controls on
the distribution of near-surface coseismic slip, Geochemistry, Geophys. Geosystems, 15(12),
5034–5050, doi:10.1002/2014GC005538.
Zinke, R., J. F. Dolan, R. Van Dissen, J. R. Grenader, E. J. Rhodes, C. P. McGuire, R. M.
Langridge, A. Nicol, and A. E. Hatem (2016), Evolution and progressive geomorphic
manifestation of surface faulting: A comparison of the Wairau and Awatere faults, South
155
Island, New Zealand, Geology, 44(8), doi:10.1130/G38188Y.1.
Zinke, R., J. F. Dolan, E. J. Rhodes, R. Van Dissen, and C. P. McGuire (2017), Highly Variable
Latest Pleistocene-Holocene Incremental Slip Rates on the Awatere Fault at Saxton River,
South Island, New Zealand, Revealed by Lidar Mapping and Luminescence Dating,
Geophys. Res. Lett., 44(22), 11,301-11,310, doi:10.1002/2017GL075048.
Zinke, R., J. F. Dolan, E. J. Rhodes, R. Van Dissen, C. P. McGuire, A. E. Hatem, N. D. Brown,
and R. M. Langridge (2018), Multi-millennial incremental slip rate variability of the
Clarence fault at the Tophouse Road site, Marlborough fault system, New Zealand,
Geophys. Res. Lett., doi:10.1029/2018GL080688.
Zinke, R., J. Hollingsworth, J. F. Dolan, and R. Van Dissen (2019), 3D surface deformation in
the 2016 M W 7.8 Kaikōura, New Zealand earthquake from optical image correlation:
Implications for strain localization and long-term evolution of the Pacific-Australian plate
boundary , Geochemistry, Geophys. Geosystems, 1609–1628, doi:10.1029/2018gc007951.
156
Chapter 2 Figures
Figure 1
157
Figure 2
158
Table 1
Labe
l on
Figu
re 3
Site
Garlo
ck
Segme
nt
Minim
um
Rate
(mm/yr
)
Preferr
ed Rate
(mm/yr
)
1
Maxim
um
Rate
(mm/yr
)
Sense of
displacem
ent
Offset
Feature
Dating
Method
Citatio
n
a;
CW
Clark
Wash
wester
n
5.3 6.1-7.8
*
10.7 sinstral
channel
incised
into
alluvial
fan
14
C
McGill
et al.,
2009
b;
KL
Koehn
Lake
central 4.5
#
5.3
#
6.1
#
sinistral
gravel
bar
14
C
Clark
and
Lajoie,
1974
c; SR
Summit
Range 1
central 2.8 5.3 6.3 sinistral
alluvial
fan
10
Be
Ganev
et al.,
2012
d;
SR
Summit
Range 2
central 4.8 5.1 5.4 sinistral
channel
incised
into
alluvial
fan
inferred
climatic
event at
end of
younger
Dryas
(~11.5 ka)
Ganev
et al.,
2012
e; SR
Summit
Range 3
central 5.4 6.6 7.8 sinistral
channel
incised
into
alluvial
fan
inferred
climatic
event at
onset of
summer
monsoon
rainfall
pattern
(~8-10 ka)
Ganev
et al.,
2012
f; SR
Summit
Range 4
central 5.4 5.5 5.6 sinistral
alluvial
fan
post-IR 50-
IRSL 225
Dolan
et al.,
2015
g; SL
Searles
Lake
central 4 5.0-7.0 9 sinistral
abandon
ed lake
shorelin
e
14
C
McGill
and
Sieh,
1993
h;
QM
Quail
Mounta
ins
central 1.7 2.7 4.6 sinistral
multiple
risers
observe
d in
lidar
relative
surface
characteris
tics
determined
from
regional
mapping
Crane,
2014
159
i;
AM
Avawat
z
Mounta
ins
eastern 0.5
^
1 2.5^ sinistral
multiple
risers
observe
d in
lidar
relative
surface
characteris
tics
determined
from
regional
mapping
Crane,
2014
j
Sierra
Nevada
Frontal
fault
N/A 0.1 0.15 0.2
horizontal
dextral-
extension
(slip
vector
~0°)
alluvial
fan and
rockslid
e
10
Be
Le et
al.,
2006
k
Panami
nt
Valley
fault
N/A 1.6 2.1 2.6
horizontal
dextral-
extension
(slip
vector
~325°)
debris
flow
levee
10
Be
Hoffm
an,
2009
l
Souther
n Death
Valley
fault
N/A 0.8 1.0 1.2
horizontal
dextral-
extension
(slip
vector
323°)
Alluvial
fan
10
Be and
OSL
Franke
l et al.,
2015
1
mean rate if preferred rate not reported by original authors
*
preferred range expanded by Hatem and Dolan to reflect broad peak of slip rate PDF; see McGill et al., [2009],
figure 10C for PDF
#
ages recalibrated by Ganev et al., [2012], which reduced slip rate from originally reported 7 mm/yr from Clark and
Lajoie, [1974]
^ error ranges estimated from Crane [2014], Figure 19
160
Figure 3
161
Figure 4
162
Figure 5
163
Chapter 3 Figures
Figure 1
164
Figure 2
165
Figure 3
166
Figure 4
167
Figure 5
168
Table 1
169
170
Table 2
171
Figure 6
172
Table 3
Fault Event Minimum age (CE) Maximum age (CE) Preferred sequence
Alpine
1
A1 1717
2
-- S2
A2 1549 1594 S3
A3 1388 1407 Is1
A4 1008 1213 S4
A5 915 961 Is3
A6 592 646 S5
A7 370 416 S6
Hope (Hurunui)
3
MF1 1888 -- S1
MF2 1652 1840 S2
MF3 1630 1424 S3
Hope (Hurunui)
4
HS1 1888 -- S1
HS2 1818 1840 S2
HS3 1233 1735 S3
HS4 821 1100 S4
HS5 439 587 S5
HS6 375 428 S6
Hope (Conway)
5
GB1 1730 1840 S1
GB2 1657 1797 S2
GB3 1495 1611 S3
GB4 1230 1277 Is2
GB5 476 1240 S4
GB6 476 1240 S5
eastern Kekerengu
6
EK0 2016 -- N/A
EK1 1701 1840 S1
EK2 1422 1594 S3
EK3 701 1047 S4
EK4 224 857 S5
Wairarapa
7
CC1 1855 -- S1
CC2 1030 1150 S4
1
Howarth et al., 2012; 2014; 2016; 2018
2
Wells et al., 1999
3
Langridge et al., 2013
4
Khajavi et al., 2016
5
This study
6
Little et al., 2018
7
Little et al., 2009
173
Figure 7
174
Chapter 4 Figures
Figure 1
175
Figure 2
176
Figure 3
177
Figure 4
178
Figure 5
179
Figure 6
180
Figure 7
181
Figure 8
182
Figure 9
183
Figure 10
184
Figure 11
185
Figure 12
186
Figure 13
187
Figure 14
188
Figure 15
189
Figure 16
190
Figure 17
191
Chapter 5 Figures
Figure 1
192
Figure 2
193
Figure 3
194
Figure 4
195
Figure 5
196
Figure 6
197
Chapter 6 Figures
Figure 1
198
Figure 2
199
Figure 3
200
Figure 4
201
Figure 5
202
APPENDICES
Appendix A: A 2000-year paleoearthquake record along the Conway segment of the Hope
fault: Implications for patterns of earthquake occurrence in northern South Island and
southern North Island, New Zealand
In Figure S1, we provide an interpreted lidar hillshade and contour (1 m) map to illustrate
the origin of the shutter ridge. The topographic high at GBE, which is interpreted to be primarily
a shutter ridge, and secondarily a coseismically uplifted scarp. Here, we indicate contours of
specific intervals (Figure S1A). For reference, the southern end of GBE trench is located just
north of the northeastern edge of the light blue 296 m closed contour. We also plotted this 296 m
contour occurring east of the GBE site, across the south-flowing, steeply incised drainage.
The 296 m contour (and others) are distinctly offset by about 38 m (Figure S1B).
Restoring the GBE ridge by 38 m forces this ridge to intrude into the active south-flowing
drainage, indicating that the extremely steep, south-east facing incision must be younger than 38
m of displacement along the Conway segment. Restoring the GBE shutter ridge by a minimum
of 38 m places the south end of our trench at the base of the alluvium/colluvium coated bedrock
slope north of the trench, which provides a source for the colluvial wedges coseismically
deposited.
A larger offset, ≥200 m, is recorded by the displacement of the thalweg this incising
stream (Figure S1C). This offset places the GBE ridge well within the present-day drainage,
immediately south of the steeply incised south-east facing channel wall. This restoration
highlights that the origin of the GBE ridge is likely the nose of the bedrock ridge which the
channel has subsequently incised.
In Figure S2, we plot high-resolution lidar data courtesy of Open Topography at the
GBW site. We provide an oblique view of the trench site to highlight the steep, landslide-prone
slope north of the trenches, and also to show evidence of minor-to-no channelization at the site.
Contour intervals are 50 cm, and coloring represents relative elevation above sea level.
In Figure S3, we present the full trench length photomosaics of both the east and west
wall of GBE. These mosaics were made using Agisoft Photoscan. Scale below east wall mosaic
applies to both mosaics. On the west wall mosaic, orange spray paint numbers denote vertical
string markers, and green spray paint numbers denote horizontal string markers. Note the
203
curvature of the east wall mosaic is due to shoveling out trench spoils that were compressed into
the trench wall during digging to prevent the backhoe track from sinking into the marsh.
In Figure S4, we present the full west wall of GBW T-1 photomosaic. Pink spray paint
numbers represent vertical string markers, and green spray paint numbers represent horizontal
string markers.
In Figure S5, we present the full exposure of the SE wall of GBW T-2. Orange numbers
represent inferred-landslides (L units). Blue numbers represent paleosols (P units).
In addition to the previously discussed radiocarbon samples, we collected an additional
eleven samples from the north end of the GBE trench (Figure S6). These “north marsh” samples
were collected with the aim to provide a high resolution age profile throughout the marsh. Such
an age profile could aid in providing limiting ages on scarp colluviation/earthquake event
horizons. Such an age profile would be useful to provide this age control if the deposits were in
direct, stratigraphic contact with the earthquake event horizons. However, the north marsh age
profile is separated from the “south marsh” and scarp deposits by a large wood deposit. The
wood mass likely acted as a depositional divide between the main marsh (north marsh) and a
narrower secondary marsh (south marsh) adjacent to the scarp. Because of this physical
disruption in sedimentation from north to south, we cannot use the north marsh ages to constrain
events preserved in the scarp. Where possible to correlate stratigraphy across the tree, the north
marsh ages are older than the south marsh ages. The material sampled from the north marsh
included seeds and individual plant leaves, providing a nearly annual signal for radiocarbon
determination. This observation leads us to believe that sedimentation was slower in the north
marsh compared to the south marsh, potentially due to near scarp south-ward warping of the
marsh, creating a deeper depo-center in the south marsh.
In Figure 6C of the main text, we plot these 11 north marsh samples as age PDFs in an
OxCal model, and have them labeled in place on SI Figure 3 on a photomosaic of the east wall of
the GBE excavation. For reference, we also provide samples included in the composite
GBE/GBW age model from the east wall in the scarp (yellow) and south marsh (green).
For some comparisons, we include calibrated radiocarbon age PDFs in Figure S6, which
shows the GBE east wall. In the top left plot, we show that samples SF-6 and -48 from the north
marsh are >400 years older than sample SF-40 in the south marsh. We also plot results from SF-
204
15 and -16 of the scarp deposits, which are stratigraphically above SF-40. SF-15 and -16 are
younger than SF-40, as would be expected by stratigraphic order. Because SF-6 and -48 are at
nearly the same stratigraphic level as SF-15 and -16, yet are nearly 800 years older that SF-15
and -16, we cannot correlate ages of seemingly similar stratigraphic horizons from the north
marsh, across the tree, to the south marsh and therefore to the scarp.
While the north marsh ages collected provided a calibrate age profile in perfect
stratigraphic order (Figure 6C main text; Table 1 main text), we cannot include these results in
our final age model. We prefer including ages that are in direct stratigraphic contact with other
dated layers, including the south marsh samples.
In Figure S7, we present, with other data discussed in the main text, an alternative
method for arriving at a combined age for GBE and GBW paleoearthquakes. This method takes
the average of 10,000 randomly selected points from a given GBE and GBW probability density
function, plots the results as a histogram with 15 bins, and then fits a kernel density estimation to
the histogram. This results of this alternative method are plotted with the GBE probability
density function, the GBW probability density function, and the OxCal combine GBE and GBW
age for events 1-3 (the events which have pre- and post-date information from the GBW
excavations).
In Table S1, we present abbreviated unit descriptions of all layers logged in Figure 3,
with additional event related information for each unit, in Table 1. For additional unit
information, we direct the reader to the main text.
205
Figure S1
206
Figure S2
207
Figure S3
208
Figure S4
209
Figure S5
210
Figure S6
211
Figure S7
212
Appendix B: Towards quantification of the three-dimensional deformation field associated
with the 1952 Mw 7.3 Kern County earthquake along the White Wolf fault, southern
California, USA
The following is a report generated by Agisoft to summarize parameters and other technical
information used when building the dense point cloud of post-event imagery presented in Figure
2 of Chapter 6 of this thesis.
213
214
215
216
217
218
219
220
221
222
223
224
225
Abstract (if available)
Abstract
The spatial and temporal patterns of earthquake occurrence, despite theoretical predictions, remains enigmatic. Quantification of these patterns has implications for fault evolution processes, particularly variable fault strength at depth, as well as for probabilistic seismic hazard assessments. Many methods are available to constrain the variability in earthquake behavior over many time and length scales. Regional tectonic reconstructions can illuminate loading patterns within complex fault systems, such as the understanding the role of the sinistral Garlock fault in the largely dextral southern California Pacific-North American plate boundary. Results from this study uncover direct contributions from the Eastern California Shear Zone north and south of the Garlock, as well as from the southern San Andreas. The results of this study (Chapter 2) imply that the Garlock fault aids in stabilization of the current plate boundary configuration of southern California, and plays a major role in modulating strain accommodation between various faults in the region. Smaller scale, site specific studies to obtain primary data are needed to make compilations to address fault system behavior through time. Paleoseismic studies show the temporal complexity of earthquake occurrence at a given site, and can be synthesized with other paleoseismic records within the same fault system to assess the presence of potential earthquake sequences versus isolated events, such as using the temporally irregular paleoearthquake record of the Conway segment of the Hope fault to better understand the Pacific-Australian plate boundary of the South Island of New Zealand (Chapter 3). In addition to paleoseismic records, incremental slip rate records document the covariance of incremental, discrete displacement along a fault coupled with the elapsed time between the ages of offset features. Incremental slip rates are presented at two sites along the Conway segment of the Hope fault, in the west (Chapter 4) and in the east (Chapter 5), showing variations in fault slip rate along strike of up to 200-300% over centennial and millennial time scales. Finally, on the shortest time scale, coseismic displacements can reveal the amount of off-fault deformation compared to on-fault, discrete slip. Such measurements are achieved using optical image correlations methods, which I have improved upon by developing a novel workflow to obtain the three-dimensional deformation field of the 1952 Mw 7.3 Kern County earthquake along the White Wolf fault, southern California (Chapter 6). Together, these studies add to a growing, global dataset describing how faults work over wide spatial and temporal skills. This thesis shows that, no matter the temporal or spatial scale of observation, earthquake behavior is complex. The results of these studies beg the question of why is earthquake occurrence in space and time irregular? Answers, or hints of answers, likely like below kilometers below the surface, far below the study area of the works presented in this thesis.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Characterizing fault behavior and earthquake surface expression on timescales of single events to multiple millennia
PDF
Constancy of strain release rates along the North Anatolian fault
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
Towards an understanding of fault-system mechanics: from single earthquakes on isolated faults to millenial-scale collective plate-boundary fault-system behavior
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
Quantifying ground deformation of large magnitude earthquakes using optical imgaging systems
PDF
Fault slip rates, constancy of seismic strain release, and landscape evolution in the eastern California shear zone
PDF
Extreme events as drivers of landscape evolution in active mountain belts
PDF
Microseismicity, fault structure, & the seismic cycle: insights from laboratory stick-slip experiments
PDF
Multi-scale imaging of the fault zone velocity structure: double-difference tomography, inversion of fault zone headwaves, and fault zone sensitivity kernels
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
Multi-scale imaging and monitoring of crustal and fault zone structures in southern California
PDF
High-resolution imaging and monitoring of fault zones and shallow structures: case studies in southern California and on Mars
PDF
Stress-strain characterization of complex seismic sources
Asset Metadata
Creator
Hatem, Alexandra Elise
(author)
Core Title
Observations of temporal and spatial patterns of strain accommodation and earthquake occurrence along strike-slip faults of New Zealand and southern California, USA
School
College of Letters, Arts and Sciences
Degree
Doctor of Philosophy
Degree Program
Geological Sciences
Publication Date
07/25/2019
Defense Date
05/07/2019
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
earthquake,faulting,image correlation,OAI-PMH Harvest,paleoseismology,plate boundary evolution,remote sensing,slip rates,tectonics
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Dolan, James Francis (
committee chair
), Nutt, Steven (
committee chair
), West, A. Joshua (
committee chair
)
Creator Email
ahatem@usc.edu,alexhatem1@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c89-192082
Unique identifier
UC11663054
Identifier
etd-HatemAlexa-7619.pdf (filename),usctheses-c89-192082 (legacy record id)
Legacy Identifier
etd-HatemAlexa-7619.pdf
Dmrecord
192082
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Hatem, Alexandra Elise
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
faulting
image correlation
paleoseismology
plate boundary evolution
remote sensing
slip rates
tectonics