SlideShare a Scribd company logo
1 of 18
Download to read offline
1	
Large earthquake production sensitivity to downdip limit of seismogenic zone determination
using subduction zone 2D computer modeling
Austin Butler
3/9/16
Introduction
The Tonga Trench, located roughly 2,000 km north-northeast of New Zealand, is a
convergent plate boundary where the Pacific plate is subducting beneath the Tonga microplate.
At its highest rate along the trench, the Pacific plate is subducting at a rate of 24 cm/yr, making it
the fastest subduction on the planet (Wright, Bloomer, MacLeod, Taylor, & Goodlife, 2000).
With such fast convergence rates, the Tonga Trench is a region of high seismicity, making it an
important and interesting region of study for the field of seismology.
Subduction zones are the cause of many of the largest earthquakes that we see. Some
convergent plate boundaries, such as the one between the Nazca and South American Plates in
western South America, occur at a continent-ocean boundary and cause earthquakes that produce
direct devastating effects. In some cases, convergent plate boundaries, such as the one in South
America, cause large earthquakes below the seafloor that then produce extremely destructive
tsunamis when a large amounts of displaced water collide with a landmass. In such cases, much
damage is done as a result of these megathrust earthquakes that greatly impacts the lives of many
around the globe. Because of this, the study of the generation of these earthquakes is very
valuable to the general population in addition to geoscientists.
A subduction zone fault model constructed by Colella, Dieterich, Richards-Dinger, and
Rubin (2012) suggests that the type of seismic activity in a subduction zone can be attributed to
2	
its depth on the fault plane (Figure 1). Colella et al. uses the Cascadia subduction zone to
distinguish between three zones at varying depths: the seismogenic zone, the region where the
most and largest seismic events occur, lying roughly between 5 and 25 km deep, the transition
zone lying between 25 and 40 km deep, and the continuous creep zone at depths greater than 40
km. Other similar models of megathrust faults by Liu (2013) and Lay, Kanamori, Ammon,
Koper, Hutko, Ye, Yue, & Rushing (2012) have been made that also differentiate between
different regions (Figures 2 & 3). Despite Tonga’s impressive convergence rates and frequent
seismic activity, it fails to produce great earthquakes comparable to those seen in South America,
as much of the stress is released aseismically (Beavan et al., 2010). One explanation for this is
that the downdip limit of Tonga’s seismogenic zone may not be deep enough to support
earthquakes of that magnitude. The goal of this project will be to determine if the downdip limit
of the seismogenic zone is an important control in the occurrence of large earthquakes (> 8 Mw).
To do this, I will create a model of the subduction zone megathrust based on Mohr-Coulomb
theory assuming a brittle elastic medium to look at failure propagation along the fault plane due
to changes in stress.
Tectonic Influences and Details
The driving tectonic forces that cause the subduction of tectonic plates come from
occurrences such as the formation of new oceanic lithosphere at divergent plate boundaries and
flow of the upper mantle. These forces drive plates together resulting in one of the plates at a
boundary to subduct, or to move past and beneath the other plate. The slow tectonic forces put
constant stress on the plates that are interacting. The interface between the two plates at a
convergent plate boundary has a finite strength, or resistance to these stresses. Once the stress
3	
has reached a critical level in a location on the interface, that location between the plates is no
longer able to support the stress, and slip occurs as the subducting plate slides past the adjacent
plate as an earthquake. This initial location of slip between plates is called the nucleation of the
earthquake. Depending on the conditions of the fault plane surrounding the nucleation, the slip
may continue, and the fault may rupture over great distances resulting in a large earthquake. The
other possibility is that a critical level of stress in the locations adjacent to the nucleation is not
reached and the rupture terminates, which is seen as a small earthquake.
The model I will produce will represent both the geometry and the physics of a
subduction zone with inspiration from the Tonga Trench. Since the seismogenic zone is limited
by depth, the dip angle of the fault would directly affect the size of the seismogenic zone, with
steeper downdip angles resulting in a smaller area of the seismogenic zone, and shallower
downdip angles yielding a larger seismogenic zone. This depth limitation of the seismogenic
zone in subduction zones is generally thermally controlled. The colder lithosphere is able to
deform brittlely, while the deeper, warmer asthenosphere supports ductile deformation. While
ruptures may propagate into the asthenosphere, this transition is generally the lower boundary for
the nucleation of earthquakes, and therefore the downdip limit of the seismogenic zone.
The dimensions of the seismogenic zone on the fault interface have an effect on the
rupture propagation of large earthquakes. After the nucleation of an earthquake, if the conditions
allow, the rupture may propagate in all directions away from the epicenter maintaining a roughly
1:1 aspect ratio. When the ruptures reach the boundaries of the seismogenic zone or regions
where brittle deformation is not supported, such as the deeper transition zone or the boundaries
of the fault, propagation cannot continue and terminates. In the case of megathrusts, the aspect
ratio of the seismogenic zone is much higher with an along-strike length much longer than the
4	
along-dip length. This allows for the rupture to continue propagating along-strike even though
rupture has stopped vertically. In such scenarios, larger earthquakes with magnitudes greater
than 8.5 are capable of being produced as the area of the rupture can increase horizontally.
Laboratory tests have been done to study the nucleation of earthquakes and what
characterizes the setting where they occur. Reches (1999) experimented with the triaxial loading
of both intact rock samples as well as the interaction of multiple intact samples to study both
rupture and friction, respectively, in the initiation of earthquakes. After considering instances of
both slip on pre-existing faults and rupture of intact rocks, it was concluded that it is sufficient to
assume that the rock is intact, and that the strength of the rock at different locations is must be
considered. This is in part due to the fact that under the heat and pressure conditions of the
locations of earthquake nucleation, fault zones may heal due to the re-cementation of fault
gouge. This information can be applied to my model in that it may not be necessary to
distinguish the modeled fault interface as either a pre-existing fault or completely intact rock, but
rather assume that it is intact rock with variable strengths, as this allows for the possibility that
either scenario may be the case. Another important finding was that it is not the nucleation and
growth of a single rupture that causes an earthquake, but rather the interaction of multiple
ruptures in close proximity of each other. The P wave of the main shock often followed two
smaller events, indicating that the first two triggered the main one (Figure 4). This phenomenon
can be described as the “cascade model” (Reches, 1999). In order to take this into account, my
model may require the stress threshold to be exceeded by multiple locations within a
predetermined maximum distance from each other so that a slip may cascade and result in a large
earthquake. On the same topic, a study conducted by Ohnaka (2003) had different intentions.
Ohnaka tries to solve the problem of scale-dependence of the rupture of earthquakes while taking
5	
into consideration the fact that rupture during an earthquake consists of both slip failure on pre-
existing faults as well as the fracture of intact rock. The result is a scaling law that “…enables
one to provide a consistent, unified comprehension for small-scale frictional slip failure and
shear fracture in the laboratory, and large-scale earthquake rupture in the field” (Ohnaka, 2003).
The idea that asperities on the subducting plate can create “locked patches” and effect the
distribution of stress on a megathrust fault interface has been explored. Konca et al. (2008)
studied the sequence of Sumatra-Andaman megathrust earthquakes in 2007 using GPS geodetic
data along the fault to analyze strain accumulation and determine the relationship between the
locked patches where coupling is high and the magnitude 8.4 and 7.9 earthquakes in 2007
(Figure 5). The inspiration for this study came from the observations that places where coupling
is low, only moderate earthquakes occur, and that places where there is high coupling, much
larger earthquakes are produced. The idea is that subduction zones where asperities lock the
plates together are able to accumulate more stress between seismic events and release more at
once than if friction between the plates was low where stress could be released more frequently
and in smaller quantities.
The use of Byerlee’s Law may prove to be helpful in setting a yield stress that will
determine when locations on the fault interface slip. Looking at friction between rocks, Byerlee
experimentally tested the shear stress required for slip to occur for a given normal stress on many
different rock types (Byerlee, 1978). His findings led to two very useful equations describing
just this scenario. In each case, the shear stress necessary for slip to occur increases linearly with
normal stress, regardless of rock type. For normal stresses less than 2 kbar, this relationship is
described as 𝜏 = 0.85𝜎!, and for normal stresses greater than 2 kbar, described as 𝜏 = 0.5 +
0.6𝜎!, where τ is shear stress and σn is normal stress (Byerlee, 1978). Since the shallow end of
6	
the seismogenic zone lies approximately 5 km deep, it is sufficient to assume lithostatic stresses
in this environment (Stein & Wysession, 2003). At 5 km depth, the lithostatic pressure is
roughly 4.5 kbar, and the second of Byerlee’s equations can be used to determine the shear stress
where failure occurs. Byerlee notes, however, that if fault gouge is present between the two
surfaces, friction is much lower than if there were no gouge. On the other hand, asperities being
subducted, such as seamounts, can cause very high levels of friction between the plates.
Hypothesis
I predict that having a greater seismogenic zone downdip limit will enhance the
probability of large earthquakes (> 8 Mw) by allowing failure around locked patches. I will be
testing this hypothesis by systematically controlling the downdip limit of the seismogenic zone
in my 2D computer model of the megathrust fault interface. Upon analysis of my results, a
statistically significant increase in pre-defined cascading events would confirm my hypothesis,
where a statistically insignificant increase or a decrease in cascading events would falsify it.
Methods and Data
To solve the physics of the effects of stress in a brittle elastic medium, I will be
implementing the finite differences method for the differential equations of stress interaction,
taking a numerical approach. The model will consist of a rectangular grid of elements, each
representing a small area on the fault plane. Elements will experience different levels normal
and shear stresses as the “plate subducts”, and failure will occur at an element once shear stress
has overcome its force of friction. Failure and displacement of an element will then directly
affect the stresses of elements adjacent to it, which may or may not induce failure in those. A
7	
large earthquake would be shown as slip cascading in which failure spreads rapidly due to a
chain reaction until the strength of an element is enough to stop slip from propagating further.
All modeling will be done using Matlab.
It is important to determine the relationship between stress and strain at points (nodes) in
the continuous elastic medium I will be modeling. Assuming both isotropy and homogeneity at
the fault interface, the constitutive equation I will use for a given node is a 2-dimensional
variation of Hooke’s law,
𝜎!" = 𝜆𝑒!! 𝛿!" + 2𝜇𝑒!",
where σij is the ijth
component of the stress tensor; eij is the ijth
component of the strain tensor; ekk
is the dilatation or the trace of the strain tensor; δij is the Kronecker delta or the ijth
component of
the identity matrix; and λ and µ are the Lamé constants, the latter of which is the shear modulus.
In terms of bulk modulus, Κ, the same equation takes the form 𝜎!" = 𝛫𝑒!! 𝛿!" + 2𝜇 𝑒!" −
!!!!!"
!
(Stein & Wysession, 2003). According to this equation, the stress and strain can be calculated
and recorded for each element of the interface. Each run of the simulation will begin with some
amount of stress being applied to a group of elements. Each element may have different material
properties and therefore may deform at different times and in different directions. A yield stress
will be calculated, and when the stress at a node exceeds this value, “slip” will occur at that
node, resulting in a dramatic and sudden decrease in stress as well as displacement of that node.
It is this displacement of the node that will quickly apply new stresses to the neighboring nodes
and will result in a higher probability that the slip will cascade and continue to propagate.
One control that I will be varying is the strength of the subducting plate, where I will be
implementing a Monte Carlo method that randomly varies strength numerous times to yield a
sufficient amount of numerical data. The strength of the plate and the interface will be
8	
characterized by the elastic moduli at each node, K and µ, as well as a coefficient of friction.
Also referred to as incompressibility and rigidity as well as bulk modulus and shear modulus,
respectively, K and µ will determine the deformability of the plate at each node. Elements that
are more incompressible and rigid will result in higher values of stress from the same amount of
strain experienced by elements that are more compressible and less rigid. Inversely, the same
amount of stress applied to an element with high values of K and µ will be strained much less
than elements with lower values of K and µ. Upon deformation due to stresses, the stress will be
checked after each iteration to see if it has exceeded the yield stress and if it has overcome the
coefficient of friction. The coefficient of friction at each node is a free parameter and may be
used to simulate areas of locking or areas of low friction. High coefficients of friction can
represent asperities on the subducting plate such as seamounts, rugged topography, or anything
else that may resist slip in a location. Low coefficients of friction could represent regions that
are clay-rich or have any other slip-promoting characteristics. As the relationship between stress
and strain is linear in an isotropic and homogeneous medium, Κ and µ are the constants at each
node that will convert the strain tensor to the stress tensor. By utilizing a Mohr’s circle, the yield
stress can be calculated for each node. When stress at that node exceeds the yield stress, slip
would occur and put new stress and strain on its neighboring nodes whose stress must be
checked as well, and the process repeats itself. If upon slip at a location the strength of the
neighboring locations is increased or decreased, it will be described as rate-weakening or rate-
strengthening, respectively (Figure 2). Rate-weakening, as demonstrated in the seismogenic
zone, depends upon the speed of slip, and results in a weaker fault interface as slip speed
increases (Colella, Dieterich, Richards-Dinger, & Rubin, 2012).
9	
A deeper downdip limit of the seismogenic zone could affect the production of large
earthquakes in several ways. First, a seismogenic zone with a longer along-dip dimension has a
larger area than a seismogenic zone with a shorter along-dip dimension, assuming equal along-
strike length. A larger area means more locations that can slip and more chances for nucleation
of earthquakes. Megathrust faults such as those in Japan, South America, and Tonga have more
than enough length horizontally, or along-strike, to produce earthquakes of enormous
magnitudes, and therefore is not likely the limiting factor in large earthquake production in
subduction zones. This is why I will be less concerned about the length of the horizontal, or
along-strike, dimension of the seismogenic zone. The aspect ratio that I will choose for my
model will be large enough that the along-strike length of the fault will not limit the potential
size of the an earthquake that can be produced. The limiting factor in the potential size of a
megathrust earthquake, rather, is determined by the strength of the fault plane that the rupture is
propagating through, which leads to another reason large earthquake production is dependent on
the along-dip dimension of the seismogenic zone. If the horizontally-propagating rupture has
more pathways in which it can circumvent high-strength locations in the fault plane, it has a
higher chance of passing around one of these strong locations and continuing to propagate
further.
The moment magnitude of an earthquake is directly dependent on the log of the seismic
moment of an earthquake. The seismic moment (M0) with dimensions of energy in dyn-cm can
be calculated using the equation
𝑀! = 𝜇𝐴𝐷,
10	
where µ equals the shear modulus or rigidity of the rock, a physical property of the material with
a unit of pressure; A is the rupture area along the fault plane; and D is the displacement of one of
the plates relative to the other (Stein & Wysession, 2003). Using seismic moment, the
dimensionless moment magnitude of an earthquake is calculated by the equation
𝑀! =
2
3
log!" 𝑀! − 10.73
(Stein & Wysession, 2003). With the horizontal dimension of seismogenic zones of large
subduction zones on the order of hundreds of kilometers long and the vertical dimension only
tens of kilometers, it is clear that in order for a massive earthquake to take place, the rupture
must take advantage of the along-strike length of the seismogenic zone. Avoiding locations that
would stop the propagation of a rupture is crucial for this to happen, and is more likely to occur
with deeper seismogenic zones.
Once the simulation is run many times and ample data have been obtained, the statistical
significance of the results must be determined in order for the hypothesis to be either confirmed,
falsified, or found inconclusive. To do this, I will determine the normalized covariance between
seismogenic zone depth (independent variable) and production of large cascading events
(dependent variable) using Pearson’s correlation coefficient, 𝑟 =
!
!!!
(!!!!)
!!
(!!!!)
!!
!
!!! , where xi
and yi are the seismogenic zone depth and the quantity or frequency of large events produced for
each data point, respectively; n is the number of data points; 𝑥 and 𝑦 are the algebraic means of
the independent and dependent variables, respectively; and sx and sy are the standard deviation of
the independent and dependent variables, respectively. Once the covariance is determined, its
significance can be determined by using a resampling method such as jackknifing or
11	
bootstrapping. This will provide an objective result to the data and will aid in the conclusion of
the project.
Discussion
Upon completion of my project, the results must be analyzed in order to come to a
conclusion on my hypothesis. As this is a computer-modeling project, many iterations may be
done in a small amount of time to test many different scenarios and different variables. Contrary
to other projects relying on physical experiments, changes to the model can be made without
entirely starting the model-building process over. With this being said, trial and error can lead to
the production of many different models and will allow them to eventually evolve into a
physically very accurate one. The processes being replicated in the model show an example of
deterministic chaos, similar to the results of the slider block experiment (Turcotte, 1997). As
some computer models may come to the same result given the same initial conditions,
deterministic chaos implies that a result cannot be predicted, even if the initial conditions are
known, although the result may be known within limits (Figure 6). This requires that the model
must be run many times and the results of all the iterations looked at at once. This may allow
determination of the probability of an event occurring given certain initial conditions.
In order to test my hypothesis, I must model subduction zones with a variety of aspect
ratios and areas. This will be testing the effect of the downdip limit or the along-dip length of
the seismogenic zone on generating great earthquakes. The number or frequency of cascading
events and large earthquakes for each scenario must be determined and compared. In order for
my hypothesis to be validated, there must be a significantly larger amount of cascade events in
model runs corresponding to deeper seismogenic downdip limits than in runs corresponding to
12	
shallower downdip limits. If my hypothesis is proven to be correct, this could potentially explain
the relative absence of great earthquakes seen on the Tonga megathrust despite its high
convergence rates and frequency of seismicity. If no significant relationship is seen between
seismogenic zone downdip limit and large earthquakes, it could rule out downdip limit as a
controlling factor in the production of large earthquakes, and a different explanation must be
proposed to explain Tonga’s lack of great earthquakes.
Timeline
The following is an approximation of what will need to be finished at what time in order
to stay on track to graduate at the beginning of the Summer 2016 semester:
March
• Construct 2D model of subduction zone
• Begin writing of thesis
April
• Finish model and thesis
• Prepare presentation for thesis defense
Late April/Early May
• Defend thesis
May
• Make any changes to thesis and/or model for reevaluation if necessary
• Submit corrected thesis to graduate school
13	
Figures
Figure 1 - A megathrust fault model based of off the Cascadia subduction zone differentiating
between 3 distinct regions of slip (Colella et al., 2012)
Figure 2 - A cross-section of a subduction zone showing locations of velocity- (or rate)
weakening and velocity-strengthening slip (Liu, 2013)
14	
A
B
Figure 3 –
A.) A to-scale cross-section of the subduction zone off the eastern coast of Japan differentiating
between 4 distinct regions of slip
B.) A more detailed look at the slip on the fault interface of the 4 regions shown in 3A.
(Lay et al., 2012)
15	
A
B
Figure 4 –
A.) Velocity seismogram schematic where the P-wave of the main shock followed a separate
initial phase of multiple smaller events (left), and the slip history of the 3 events (right)
B.) A series of steps showing the cooperation of nearby cracks leading to the growth of a fault in
intact rock
(Reches, 1999)
16	
Figure 5 – Locations of strong coupling (red regions) on the Sunda megathrust due to subducted
asperities creating locked patches correspond to locations of large earthquakes (Konca et al.,
2008)
Figure 6 – Results of an asymmetrical two-block slider block model showing chaotic behavior
(Turcotte, 1997)
17	
Works Cited
Beavan, J., Wang, X., Holden, C., Wilson, K., Power, W., Prasetya, G., … Kautoke, R. (2010).
Near-simultaneous great earthquakes at Tongan megathrust and outer rise in September
2009. Nature, 466(7309), 959–963. http://doi.org/10.1038/nature09292
Byerlee, J. (1978). Friction of rocks. Pure and Applied Geophysics PAGEOPH, 116(4-5), 615–
626. http://doi.org/10.1007/BF00876528
Colella, H. V., Dieterich, J. H., Richards-Dinger, K., & Rubin, A. M. (2012). Complex
characteristics of slow slip events in subduction zones reproduced in multi-cycle
simulations. Geophysical Research Letters, 39(20), n/a–n/a.
http://doi.org/10.1029/2012GL053276
Konca, A. O., Avouac, J.-P., Sladen, A., Meltzner, A. J., Sieh, K., Fang, P., … Helmberger, D.
V. (2008). Partial rupture of a locked patch of the Sumatra megathrust during the 2007
earthquake sequence. Nature, 456(7222), 631–635. http://doi.org/10.1038/nature07572
Lay, T., Kanamori, H., Ammon, C. J., Koper, K. D., Hutko, A. R., Ye, L., … Rushing, T. M.
(2012). Depth-varying rupture properties of subduction zone megathrust faults. Journal of
Geophysical Research: Solid Earth, 117(4), 1–21. http://doi.org/10.1029/2011JB009133
Liu, Y. (2013). Numerical simulations on megathrust rupture stabilized under strong dilatancy
strengthening in slow slip region. Geophysical Research Letters, 40(7), 1311–1316.
http://doi.org/10.1002/grl.50298
Ohnaka, M. (2003). A constitutive scaling law and a unified comprehension for frictional slip
failure, shear fracture of intact rock, and earthquake rupture. Journal of Geophysical
Research, 108(B2), 1–21. http://doi.org/10.1029/2000JB000123
Reches, Z. (1999). Mechanisms of slip nucleation during earthquakes. Earth and Planetary
18	
Science Letters, 170(4), 475–486. http://doi.org/10.1016/S0012-821X(99)00122-3
Stein, S., & Wysession, M. (2003). An Introduction to Seismology, Earthquakes, and Earth
Structure (1st ed.). Blackwell Publishing.
Turcotte, D. (1997). Fractals and Chaos in Geology and Geophysics (2nd editio). Cambridge
University Press.
Wright, D. J., Bloomer, S. H., MacLeod, C. J., Taylor, B., & Goodlife, a. M. (2000). Bathymetry
of the Tonga Trench and Forearc: A map series. Marine Geophysical Researches, 21(5),
489–511. http://doi.org/10.1023/A:1026514914220

More Related Content

What's hot

Jack Oughton - Impact Calculator Report.doc
Jack Oughton - Impact Calculator Report.docJack Oughton - Impact Calculator Report.doc
Jack Oughton - Impact Calculator Report.docJack Oughton
 
Laboratory Manual in Physical Geology 11th Edition American Solutions Manual
Laboratory Manual in Physical Geology 11th Edition American Solutions ManualLaboratory Manual in Physical Geology 11th Edition American Solutions Manual
Laboratory Manual in Physical Geology 11th Edition American Solutions Manuallycegoka
 
Disaster Management
Disaster ManagementDisaster Management
Disaster Managementbhavypoladia
 
Basic of seismic engineering
Basic of seismic engineeringBasic of seismic engineering
Basic of seismic engineeringmayankparekh98
 
GeophyicsPapergel110
GeophyicsPapergel110GeophyicsPapergel110
GeophyicsPapergel110Cody McMurdie
 
Introduction to earthquake engineering by Engr. Basharat Ullah
Introduction to earthquake engineering by Engr. Basharat UllahIntroduction to earthquake engineering by Engr. Basharat Ullah
Introduction to earthquake engineering by Engr. Basharat Ullahbasharat ullah
 
Gravity and Magnetic Mapping
Gravity and  Magnetic MappingGravity and  Magnetic Mapping
Gravity and Magnetic MappingBrenton Garside
 
9 magnetic - introduction
9  magnetic - introduction9  magnetic - introduction
9 magnetic - introductionEmre Demirhan
 
Similarities and differences between gravity and magnetic
Similarities and differences between gravity and magneticSimilarities and differences between gravity and magnetic
Similarities and differences between gravity and magneticAkhtar Hussain
 
EARTHQUAKE ENGINEERING 1
EARTHQUAKE ENGINEERING 1EARTHQUAKE ENGINEERING 1
EARTHQUAKE ENGINEERING 1Edz Gapuz
 
Gravity & magnetic methods in geology
Gravity & magnetic methods in geologyGravity & magnetic methods in geology
Gravity & magnetic methods in geologyMd Asif Hasan
 
Introduction to earthquake
Introduction to earthquakeIntroduction to earthquake
Introduction to earthquakeArvinder Singh
 

What's hot (20)

Jack Oughton - Impact Calculator Report.doc
Jack Oughton - Impact Calculator Report.docJack Oughton - Impact Calculator Report.doc
Jack Oughton - Impact Calculator Report.doc
 
Laboratory Manual in Physical Geology 11th Edition American Solutions Manual
Laboratory Manual in Physical Geology 11th Edition American Solutions ManualLaboratory Manual in Physical Geology 11th Edition American Solutions Manual
Laboratory Manual in Physical Geology 11th Edition American Solutions Manual
 
Basic seismology
Basic seismologyBasic seismology
Basic seismology
 
Presentation
PresentationPresentation
Presentation
 
Impact cratering lab
Impact cratering labImpact cratering lab
Impact cratering lab
 
01 chapter:Earthquake
01 chapter:Earthquake01 chapter:Earthquake
01 chapter:Earthquake
 
Seismic
SeismicSeismic
Seismic
 
Disaster Management
Disaster ManagementDisaster Management
Disaster Management
 
Basic of seismic engineering
Basic of seismic engineeringBasic of seismic engineering
Basic of seismic engineering
 
Earthquakes
EarthquakesEarthquakes
Earthquakes
 
GeophyicsPapergel110
GeophyicsPapergel110GeophyicsPapergel110
GeophyicsPapergel110
 
Task Earthquake
Task EarthquakeTask Earthquake
Task Earthquake
 
Introduction to earthquake engineering by Engr. Basharat Ullah
Introduction to earthquake engineering by Engr. Basharat UllahIntroduction to earthquake engineering by Engr. Basharat Ullah
Introduction to earthquake engineering by Engr. Basharat Ullah
 
Gravity and Magnetic Mapping
Gravity and  Magnetic MappingGravity and  Magnetic Mapping
Gravity and Magnetic Mapping
 
9 magnetic - introduction
9  magnetic - introduction9  magnetic - introduction
9 magnetic - introduction
 
Similarities and differences between gravity and magnetic
Similarities and differences between gravity and magneticSimilarities and differences between gravity and magnetic
Similarities and differences between gravity and magnetic
 
EARTHQUAKE ENGINEERING 1
EARTHQUAKE ENGINEERING 1EARTHQUAKE ENGINEERING 1
EARTHQUAKE ENGINEERING 1
 
Gravity & magnetic methods in geology
Gravity & magnetic methods in geologyGravity & magnetic methods in geology
Gravity & magnetic methods in geology
 
Introduction to earthquake
Introduction to earthquakeIntroduction to earthquake
Introduction to earthquake
 
Foundation on Layered Soil under Torsional Harmonic Vibration using Cone model
Foundation on Layered Soil under Torsional Harmonic Vibration using Cone modelFoundation on Layered Soil under Torsional Harmonic Vibration using Cone model
Foundation on Layered Soil under Torsional Harmonic Vibration using Cone model
 

Viewers also liked

JODIE M ROSE - Resume Long 05.2016
JODIE M ROSE - Resume Long 05.2016JODIE M ROSE - Resume Long 05.2016
JODIE M ROSE - Resume Long 05.2016Jodie Rose
 
新生訓練04 認識線上教室01
新生訓練04 認識線上教室01新生訓練04 認識線上教室01
新生訓練04 認識線上教室01正信 賴
 
Golden eagle decline may 10 2016
Golden eagle decline may 10 2016Golden eagle decline may 10 2016
Golden eagle decline may 10 2016Mueller Darryl
 
Automatic translation gone wrong
Automatic translation gone wrongAutomatic translation gone wrong
Automatic translation gone wrongGiovanna Mussini
 
Beneath our boots final 21042016
Beneath our boots final 21042016Beneath our boots final 21042016
Beneath our boots final 21042016Derrick McClure
 
Act 4.1 velasco_torres_informemariana
Act 4.1 velasco_torres_informemarianaAct 4.1 velasco_torres_informemariana
Act 4.1 velasco_torres_informemarianaLIVIA VELASCO TORRES
 
Sand hill project_may10
Sand hill project_may10Sand hill project_may10
Sand hill project_may10Mueller Darryl
 
Dura Vermeer Divisie Infra - Fietsportfolio - oktober 2015
Dura Vermeer Divisie Infra - Fietsportfolio - oktober 2015Dura Vermeer Divisie Infra - Fietsportfolio - oktober 2015
Dura Vermeer Divisie Infra - Fietsportfolio - oktober 2015Dianne Huiberts
 
Reciclaje de pet catherine garzón bravo
Reciclaje de pet   catherine garzón bravoReciclaje de pet   catherine garzón bravo
Reciclaje de pet catherine garzón bravocatherine garzon
 
Eric Kabukuru's defense
Eric Kabukuru's defenseEric Kabukuru's defense
Eric Kabukuru's defenseEric Kabukuru
 
Abc derecho minero
Abc derecho mineroAbc derecho minero
Abc derecho minerolayfu_2016
 

Viewers also liked (20)

JODIE M ROSE - Resume Long 05.2016
JODIE M ROSE - Resume Long 05.2016JODIE M ROSE - Resume Long 05.2016
JODIE M ROSE - Resume Long 05.2016
 
新生訓練04 認識線上教室01
新生訓練04 認識線上教室01新生訓練04 認識線上教室01
新生訓練04 認識線上教室01
 
Perfect Aspect
Perfect AspectPerfect Aspect
Perfect Aspect
 
Golden eagle decline may 10 2016
Golden eagle decline may 10 2016Golden eagle decline may 10 2016
Golden eagle decline may 10 2016
 
CV DANIELE Ferdinando
CV DANIELE FerdinandoCV DANIELE Ferdinando
CV DANIELE Ferdinando
 
Automatic translation gone wrong
Automatic translation gone wrongAutomatic translation gone wrong
Automatic translation gone wrong
 
Beneath our boots final 21042016
Beneath our boots final 21042016Beneath our boots final 21042016
Beneath our boots final 21042016
 
Alia khan1
Alia khan1Alia khan1
Alia khan1
 
Act 4.1 velasco_torres_informemariana
Act 4.1 velasco_torres_informemarianaAct 4.1 velasco_torres_informemariana
Act 4.1 velasco_torres_informemariana
 
Sand hill project_may10
Sand hill project_may10Sand hill project_may10
Sand hill project_may10
 
DisneyCase
DisneyCaseDisneyCase
DisneyCase
 
Dura Vermeer Divisie Infra - Fietsportfolio - oktober 2015
Dura Vermeer Divisie Infra - Fietsportfolio - oktober 2015Dura Vermeer Divisie Infra - Fietsportfolio - oktober 2015
Dura Vermeer Divisie Infra - Fietsportfolio - oktober 2015
 
Z-Scope*7 EN
Z-Scope*7 ENZ-Scope*7 EN
Z-Scope*7 EN
 
Yordi peña cardona
Yordi peña cardonaYordi peña cardona
Yordi peña cardona
 
Netflix SMP Report
Netflix SMP ReportNetflix SMP Report
Netflix SMP Report
 
Бизнес Центр "СИТИ Центр"
Бизнес Центр "СИТИ Центр"Бизнес Центр "СИТИ Центр"
Бизнес Центр "СИТИ Центр"
 
Reciclaje de pet catherine garzón bravo
Reciclaje de pet   catherine garzón bravoReciclaje de pet   catherine garzón bravo
Reciclaje de pet catherine garzón bravo
 
Gabinete y memorias extraibles
Gabinete y memorias extraiblesGabinete y memorias extraibles
Gabinete y memorias extraibles
 
Eric Kabukuru's defense
Eric Kabukuru's defenseEric Kabukuru's defense
Eric Kabukuru's defense
 
Abc derecho minero
Abc derecho mineroAbc derecho minero
Abc derecho minero
 

Similar to Large earthquake production sensitivity to downdip limit of seismogenic zone determination using subduction zone 2D computer modeling

Earthquake an explanation text
Earthquake an explanation textEarthquake an explanation text
Earthquake an explanation textsman 2 mataram
 
Earthquake horrible
Earthquake  horribleEarthquake  horrible
Earthquake horriblemehmet669
 
GEOTECHNICAL_EARTHQUAKE_ENGINEERING (1).pptx
GEOTECHNICAL_EARTHQUAKE_ENGINEERING (1).pptxGEOTECHNICAL_EARTHQUAKE_ENGINEERING (1).pptx
GEOTECHNICAL_EARTHQUAKE_ENGINEERING (1).pptxRajanDas20
 
Earthquake disaster prevention in thailand sent 13-5-2013
Earthquake disaster prevention in thailand sent 13-5-2013Earthquake disaster prevention in thailand sent 13-5-2013
Earthquake disaster prevention in thailand sent 13-5-2013Tanakrom Pangam
 
Forsyth what is_the_lithosphere
Forsyth what is_the_lithosphereForsyth what is_the_lithosphere
Forsyth what is_the_lithospherequinterofelipe010
 
The Scaling of Geological Faults
The Scaling of Geological FaultsThe Scaling of Geological Faults
The Scaling of Geological FaultsAli Osman Öncel
 
Annals of Limnology and Oceanography
Annals of Limnology and OceanographyAnnals of Limnology and Oceanography
Annals of Limnology and Oceanographypeertechzpublication
 
Annals of Limnology and Oceanography
Annals of Limnology and OceanographyAnnals of Limnology and Oceanography
Annals of Limnology and Oceanographypeertechzpublication
 
M6.0 2004 Parkfield Earthquake : Seismic Attenuation
M6.0 2004 Parkfield Earthquake : Seismic AttenuationM6.0 2004 Parkfield Earthquake : Seismic Attenuation
M6.0 2004 Parkfield Earthquake : Seismic AttenuationAli Osman Öncel
 
Earthquakes
EarthquakesEarthquakes
EarthquakesVibhansh
 
Vulcanic flooding on_the_moon
Vulcanic flooding on_the_moonVulcanic flooding on_the_moon
Vulcanic flooding on_the_moonSérgio Sacani
 
A RESEARCH BASED STUDY ON EARTHQUAKES (case study included)
A RESEARCH BASED STUDY ON EARTHQUAKES (case study included)A RESEARCH BASED STUDY ON EARTHQUAKES (case study included)
A RESEARCH BASED STUDY ON EARTHQUAKES (case study included)ishuvishy
 
Öncel Akademi: İstatistiksel Sismoloji
Öncel Akademi: İstatistiksel SismolojiÖncel Akademi: İstatistiksel Sismoloji
Öncel Akademi: İstatistiksel SismolojiAli Osman Öncel
 

Similar to Large earthquake production sensitivity to downdip limit of seismogenic zone determination using subduction zone 2D computer modeling (20)

Connecting earthquakes and faults
Connecting earthquakes and faultsConnecting earthquakes and faults
Connecting earthquakes and faults
 
Earthquake an explanation text
Earthquake an explanation textEarthquake an explanation text
Earthquake an explanation text
 
Relationship Between Fault and Earthquake in sumatera
Relationship Between Fault and Earthquake in sumateraRelationship Between Fault and Earthquake in sumatera
Relationship Between Fault and Earthquake in sumatera
 
Earthquakes
EarthquakesEarthquakes
Earthquakes
 
Earthquake horrible
Earthquake  horribleEarthquake  horrible
Earthquake horrible
 
GEOTECHNICAL_EARTHQUAKE_ENGINEERING (1).pptx
GEOTECHNICAL_EARTHQUAKE_ENGINEERING (1).pptxGEOTECHNICAL_EARTHQUAKE_ENGINEERING (1).pptx
GEOTECHNICAL_EARTHQUAKE_ENGINEERING (1).pptx
 
Earthquake disaster prevention in thailand sent 13-5-2013
Earthquake disaster prevention in thailand sent 13-5-2013Earthquake disaster prevention in thailand sent 13-5-2013
Earthquake disaster prevention in thailand sent 13-5-2013
 
Iotsunami
IotsunamiIotsunami
Iotsunami
 
Forsyth what is_the_lithosphere
Forsyth what is_the_lithosphereForsyth what is_the_lithosphere
Forsyth what is_the_lithosphere
 
The Scaling of Geological Faults
The Scaling of Geological FaultsThe Scaling of Geological Faults
The Scaling of Geological Faults
 
Eartquake report
Eartquake reportEartquake report
Eartquake report
 
Annals of Limnology and Oceanography
Annals of Limnology and OceanographyAnnals of Limnology and Oceanography
Annals of Limnology and Oceanography
 
Annals of Limnology and Oceanography
Annals of Limnology and OceanographyAnnals of Limnology and Oceanography
Annals of Limnology and Oceanography
 
M6.0 2004 Parkfield Earthquake : Seismic Attenuation
M6.0 2004 Parkfield Earthquake : Seismic AttenuationM6.0 2004 Parkfield Earthquake : Seismic Attenuation
M6.0 2004 Parkfield Earthquake : Seismic Attenuation
 
Earth quakes
Earth quakesEarth quakes
Earth quakes
 
Iotsunami
IotsunamiIotsunami
Iotsunami
 
Earthquakes
EarthquakesEarthquakes
Earthquakes
 
Vulcanic flooding on_the_moon
Vulcanic flooding on_the_moonVulcanic flooding on_the_moon
Vulcanic flooding on_the_moon
 
A RESEARCH BASED STUDY ON EARTHQUAKES (case study included)
A RESEARCH BASED STUDY ON EARTHQUAKES (case study included)A RESEARCH BASED STUDY ON EARTHQUAKES (case study included)
A RESEARCH BASED STUDY ON EARTHQUAKES (case study included)
 
Öncel Akademi: İstatistiksel Sismoloji
Öncel Akademi: İstatistiksel SismolojiÖncel Akademi: İstatistiksel Sismoloji
Öncel Akademi: İstatistiksel Sismoloji
 

Large earthquake production sensitivity to downdip limit of seismogenic zone determination using subduction zone 2D computer modeling

  • 1. 1 Large earthquake production sensitivity to downdip limit of seismogenic zone determination using subduction zone 2D computer modeling Austin Butler 3/9/16 Introduction The Tonga Trench, located roughly 2,000 km north-northeast of New Zealand, is a convergent plate boundary where the Pacific plate is subducting beneath the Tonga microplate. At its highest rate along the trench, the Pacific plate is subducting at a rate of 24 cm/yr, making it the fastest subduction on the planet (Wright, Bloomer, MacLeod, Taylor, & Goodlife, 2000). With such fast convergence rates, the Tonga Trench is a region of high seismicity, making it an important and interesting region of study for the field of seismology. Subduction zones are the cause of many of the largest earthquakes that we see. Some convergent plate boundaries, such as the one between the Nazca and South American Plates in western South America, occur at a continent-ocean boundary and cause earthquakes that produce direct devastating effects. In some cases, convergent plate boundaries, such as the one in South America, cause large earthquakes below the seafloor that then produce extremely destructive tsunamis when a large amounts of displaced water collide with a landmass. In such cases, much damage is done as a result of these megathrust earthquakes that greatly impacts the lives of many around the globe. Because of this, the study of the generation of these earthquakes is very valuable to the general population in addition to geoscientists. A subduction zone fault model constructed by Colella, Dieterich, Richards-Dinger, and Rubin (2012) suggests that the type of seismic activity in a subduction zone can be attributed to
  • 2. 2 its depth on the fault plane (Figure 1). Colella et al. uses the Cascadia subduction zone to distinguish between three zones at varying depths: the seismogenic zone, the region where the most and largest seismic events occur, lying roughly between 5 and 25 km deep, the transition zone lying between 25 and 40 km deep, and the continuous creep zone at depths greater than 40 km. Other similar models of megathrust faults by Liu (2013) and Lay, Kanamori, Ammon, Koper, Hutko, Ye, Yue, & Rushing (2012) have been made that also differentiate between different regions (Figures 2 & 3). Despite Tonga’s impressive convergence rates and frequent seismic activity, it fails to produce great earthquakes comparable to those seen in South America, as much of the stress is released aseismically (Beavan et al., 2010). One explanation for this is that the downdip limit of Tonga’s seismogenic zone may not be deep enough to support earthquakes of that magnitude. The goal of this project will be to determine if the downdip limit of the seismogenic zone is an important control in the occurrence of large earthquakes (> 8 Mw). To do this, I will create a model of the subduction zone megathrust based on Mohr-Coulomb theory assuming a brittle elastic medium to look at failure propagation along the fault plane due to changes in stress. Tectonic Influences and Details The driving tectonic forces that cause the subduction of tectonic plates come from occurrences such as the formation of new oceanic lithosphere at divergent plate boundaries and flow of the upper mantle. These forces drive plates together resulting in one of the plates at a boundary to subduct, or to move past and beneath the other plate. The slow tectonic forces put constant stress on the plates that are interacting. The interface between the two plates at a convergent plate boundary has a finite strength, or resistance to these stresses. Once the stress
  • 3. 3 has reached a critical level in a location on the interface, that location between the plates is no longer able to support the stress, and slip occurs as the subducting plate slides past the adjacent plate as an earthquake. This initial location of slip between plates is called the nucleation of the earthquake. Depending on the conditions of the fault plane surrounding the nucleation, the slip may continue, and the fault may rupture over great distances resulting in a large earthquake. The other possibility is that a critical level of stress in the locations adjacent to the nucleation is not reached and the rupture terminates, which is seen as a small earthquake. The model I will produce will represent both the geometry and the physics of a subduction zone with inspiration from the Tonga Trench. Since the seismogenic zone is limited by depth, the dip angle of the fault would directly affect the size of the seismogenic zone, with steeper downdip angles resulting in a smaller area of the seismogenic zone, and shallower downdip angles yielding a larger seismogenic zone. This depth limitation of the seismogenic zone in subduction zones is generally thermally controlled. The colder lithosphere is able to deform brittlely, while the deeper, warmer asthenosphere supports ductile deformation. While ruptures may propagate into the asthenosphere, this transition is generally the lower boundary for the nucleation of earthquakes, and therefore the downdip limit of the seismogenic zone. The dimensions of the seismogenic zone on the fault interface have an effect on the rupture propagation of large earthquakes. After the nucleation of an earthquake, if the conditions allow, the rupture may propagate in all directions away from the epicenter maintaining a roughly 1:1 aspect ratio. When the ruptures reach the boundaries of the seismogenic zone or regions where brittle deformation is not supported, such as the deeper transition zone or the boundaries of the fault, propagation cannot continue and terminates. In the case of megathrusts, the aspect ratio of the seismogenic zone is much higher with an along-strike length much longer than the
  • 4. 4 along-dip length. This allows for the rupture to continue propagating along-strike even though rupture has stopped vertically. In such scenarios, larger earthquakes with magnitudes greater than 8.5 are capable of being produced as the area of the rupture can increase horizontally. Laboratory tests have been done to study the nucleation of earthquakes and what characterizes the setting where they occur. Reches (1999) experimented with the triaxial loading of both intact rock samples as well as the interaction of multiple intact samples to study both rupture and friction, respectively, in the initiation of earthquakes. After considering instances of both slip on pre-existing faults and rupture of intact rocks, it was concluded that it is sufficient to assume that the rock is intact, and that the strength of the rock at different locations is must be considered. This is in part due to the fact that under the heat and pressure conditions of the locations of earthquake nucleation, fault zones may heal due to the re-cementation of fault gouge. This information can be applied to my model in that it may not be necessary to distinguish the modeled fault interface as either a pre-existing fault or completely intact rock, but rather assume that it is intact rock with variable strengths, as this allows for the possibility that either scenario may be the case. Another important finding was that it is not the nucleation and growth of a single rupture that causes an earthquake, but rather the interaction of multiple ruptures in close proximity of each other. The P wave of the main shock often followed two smaller events, indicating that the first two triggered the main one (Figure 4). This phenomenon can be described as the “cascade model” (Reches, 1999). In order to take this into account, my model may require the stress threshold to be exceeded by multiple locations within a predetermined maximum distance from each other so that a slip may cascade and result in a large earthquake. On the same topic, a study conducted by Ohnaka (2003) had different intentions. Ohnaka tries to solve the problem of scale-dependence of the rupture of earthquakes while taking
  • 5. 5 into consideration the fact that rupture during an earthquake consists of both slip failure on pre- existing faults as well as the fracture of intact rock. The result is a scaling law that “…enables one to provide a consistent, unified comprehension for small-scale frictional slip failure and shear fracture in the laboratory, and large-scale earthquake rupture in the field” (Ohnaka, 2003). The idea that asperities on the subducting plate can create “locked patches” and effect the distribution of stress on a megathrust fault interface has been explored. Konca et al. (2008) studied the sequence of Sumatra-Andaman megathrust earthquakes in 2007 using GPS geodetic data along the fault to analyze strain accumulation and determine the relationship between the locked patches where coupling is high and the magnitude 8.4 and 7.9 earthquakes in 2007 (Figure 5). The inspiration for this study came from the observations that places where coupling is low, only moderate earthquakes occur, and that places where there is high coupling, much larger earthquakes are produced. The idea is that subduction zones where asperities lock the plates together are able to accumulate more stress between seismic events and release more at once than if friction between the plates was low where stress could be released more frequently and in smaller quantities. The use of Byerlee’s Law may prove to be helpful in setting a yield stress that will determine when locations on the fault interface slip. Looking at friction between rocks, Byerlee experimentally tested the shear stress required for slip to occur for a given normal stress on many different rock types (Byerlee, 1978). His findings led to two very useful equations describing just this scenario. In each case, the shear stress necessary for slip to occur increases linearly with normal stress, regardless of rock type. For normal stresses less than 2 kbar, this relationship is described as 𝜏 = 0.85𝜎!, and for normal stresses greater than 2 kbar, described as 𝜏 = 0.5 + 0.6𝜎!, where τ is shear stress and σn is normal stress (Byerlee, 1978). Since the shallow end of
  • 6. 6 the seismogenic zone lies approximately 5 km deep, it is sufficient to assume lithostatic stresses in this environment (Stein & Wysession, 2003). At 5 km depth, the lithostatic pressure is roughly 4.5 kbar, and the second of Byerlee’s equations can be used to determine the shear stress where failure occurs. Byerlee notes, however, that if fault gouge is present between the two surfaces, friction is much lower than if there were no gouge. On the other hand, asperities being subducted, such as seamounts, can cause very high levels of friction between the plates. Hypothesis I predict that having a greater seismogenic zone downdip limit will enhance the probability of large earthquakes (> 8 Mw) by allowing failure around locked patches. I will be testing this hypothesis by systematically controlling the downdip limit of the seismogenic zone in my 2D computer model of the megathrust fault interface. Upon analysis of my results, a statistically significant increase in pre-defined cascading events would confirm my hypothesis, where a statistically insignificant increase or a decrease in cascading events would falsify it. Methods and Data To solve the physics of the effects of stress in a brittle elastic medium, I will be implementing the finite differences method for the differential equations of stress interaction, taking a numerical approach. The model will consist of a rectangular grid of elements, each representing a small area on the fault plane. Elements will experience different levels normal and shear stresses as the “plate subducts”, and failure will occur at an element once shear stress has overcome its force of friction. Failure and displacement of an element will then directly affect the stresses of elements adjacent to it, which may or may not induce failure in those. A
  • 7. 7 large earthquake would be shown as slip cascading in which failure spreads rapidly due to a chain reaction until the strength of an element is enough to stop slip from propagating further. All modeling will be done using Matlab. It is important to determine the relationship between stress and strain at points (nodes) in the continuous elastic medium I will be modeling. Assuming both isotropy and homogeneity at the fault interface, the constitutive equation I will use for a given node is a 2-dimensional variation of Hooke’s law, 𝜎!" = 𝜆𝑒!! 𝛿!" + 2𝜇𝑒!", where σij is the ijth component of the stress tensor; eij is the ijth component of the strain tensor; ekk is the dilatation or the trace of the strain tensor; δij is the Kronecker delta or the ijth component of the identity matrix; and λ and µ are the Lamé constants, the latter of which is the shear modulus. In terms of bulk modulus, Κ, the same equation takes the form 𝜎!" = 𝛫𝑒!! 𝛿!" + 2𝜇 𝑒!" − !!!!!" ! (Stein & Wysession, 2003). According to this equation, the stress and strain can be calculated and recorded for each element of the interface. Each run of the simulation will begin with some amount of stress being applied to a group of elements. Each element may have different material properties and therefore may deform at different times and in different directions. A yield stress will be calculated, and when the stress at a node exceeds this value, “slip” will occur at that node, resulting in a dramatic and sudden decrease in stress as well as displacement of that node. It is this displacement of the node that will quickly apply new stresses to the neighboring nodes and will result in a higher probability that the slip will cascade and continue to propagate. One control that I will be varying is the strength of the subducting plate, where I will be implementing a Monte Carlo method that randomly varies strength numerous times to yield a sufficient amount of numerical data. The strength of the plate and the interface will be
  • 8. 8 characterized by the elastic moduli at each node, K and µ, as well as a coefficient of friction. Also referred to as incompressibility and rigidity as well as bulk modulus and shear modulus, respectively, K and µ will determine the deformability of the plate at each node. Elements that are more incompressible and rigid will result in higher values of stress from the same amount of strain experienced by elements that are more compressible and less rigid. Inversely, the same amount of stress applied to an element with high values of K and µ will be strained much less than elements with lower values of K and µ. Upon deformation due to stresses, the stress will be checked after each iteration to see if it has exceeded the yield stress and if it has overcome the coefficient of friction. The coefficient of friction at each node is a free parameter and may be used to simulate areas of locking or areas of low friction. High coefficients of friction can represent asperities on the subducting plate such as seamounts, rugged topography, or anything else that may resist slip in a location. Low coefficients of friction could represent regions that are clay-rich or have any other slip-promoting characteristics. As the relationship between stress and strain is linear in an isotropic and homogeneous medium, Κ and µ are the constants at each node that will convert the strain tensor to the stress tensor. By utilizing a Mohr’s circle, the yield stress can be calculated for each node. When stress at that node exceeds the yield stress, slip would occur and put new stress and strain on its neighboring nodes whose stress must be checked as well, and the process repeats itself. If upon slip at a location the strength of the neighboring locations is increased or decreased, it will be described as rate-weakening or rate- strengthening, respectively (Figure 2). Rate-weakening, as demonstrated in the seismogenic zone, depends upon the speed of slip, and results in a weaker fault interface as slip speed increases (Colella, Dieterich, Richards-Dinger, & Rubin, 2012).
  • 9. 9 A deeper downdip limit of the seismogenic zone could affect the production of large earthquakes in several ways. First, a seismogenic zone with a longer along-dip dimension has a larger area than a seismogenic zone with a shorter along-dip dimension, assuming equal along- strike length. A larger area means more locations that can slip and more chances for nucleation of earthquakes. Megathrust faults such as those in Japan, South America, and Tonga have more than enough length horizontally, or along-strike, to produce earthquakes of enormous magnitudes, and therefore is not likely the limiting factor in large earthquake production in subduction zones. This is why I will be less concerned about the length of the horizontal, or along-strike, dimension of the seismogenic zone. The aspect ratio that I will choose for my model will be large enough that the along-strike length of the fault will not limit the potential size of the an earthquake that can be produced. The limiting factor in the potential size of a megathrust earthquake, rather, is determined by the strength of the fault plane that the rupture is propagating through, which leads to another reason large earthquake production is dependent on the along-dip dimension of the seismogenic zone. If the horizontally-propagating rupture has more pathways in which it can circumvent high-strength locations in the fault plane, it has a higher chance of passing around one of these strong locations and continuing to propagate further. The moment magnitude of an earthquake is directly dependent on the log of the seismic moment of an earthquake. The seismic moment (M0) with dimensions of energy in dyn-cm can be calculated using the equation 𝑀! = 𝜇𝐴𝐷,
  • 10. 10 where µ equals the shear modulus or rigidity of the rock, a physical property of the material with a unit of pressure; A is the rupture area along the fault plane; and D is the displacement of one of the plates relative to the other (Stein & Wysession, 2003). Using seismic moment, the dimensionless moment magnitude of an earthquake is calculated by the equation 𝑀! = 2 3 log!" 𝑀! − 10.73 (Stein & Wysession, 2003). With the horizontal dimension of seismogenic zones of large subduction zones on the order of hundreds of kilometers long and the vertical dimension only tens of kilometers, it is clear that in order for a massive earthquake to take place, the rupture must take advantage of the along-strike length of the seismogenic zone. Avoiding locations that would stop the propagation of a rupture is crucial for this to happen, and is more likely to occur with deeper seismogenic zones. Once the simulation is run many times and ample data have been obtained, the statistical significance of the results must be determined in order for the hypothesis to be either confirmed, falsified, or found inconclusive. To do this, I will determine the normalized covariance between seismogenic zone depth (independent variable) and production of large cascading events (dependent variable) using Pearson’s correlation coefficient, 𝑟 = ! !!! (!!!!) !! (!!!!) !! ! !!! , where xi and yi are the seismogenic zone depth and the quantity or frequency of large events produced for each data point, respectively; n is the number of data points; 𝑥 and 𝑦 are the algebraic means of the independent and dependent variables, respectively; and sx and sy are the standard deviation of the independent and dependent variables, respectively. Once the covariance is determined, its significance can be determined by using a resampling method such as jackknifing or
  • 11. 11 bootstrapping. This will provide an objective result to the data and will aid in the conclusion of the project. Discussion Upon completion of my project, the results must be analyzed in order to come to a conclusion on my hypothesis. As this is a computer-modeling project, many iterations may be done in a small amount of time to test many different scenarios and different variables. Contrary to other projects relying on physical experiments, changes to the model can be made without entirely starting the model-building process over. With this being said, trial and error can lead to the production of many different models and will allow them to eventually evolve into a physically very accurate one. The processes being replicated in the model show an example of deterministic chaos, similar to the results of the slider block experiment (Turcotte, 1997). As some computer models may come to the same result given the same initial conditions, deterministic chaos implies that a result cannot be predicted, even if the initial conditions are known, although the result may be known within limits (Figure 6). This requires that the model must be run many times and the results of all the iterations looked at at once. This may allow determination of the probability of an event occurring given certain initial conditions. In order to test my hypothesis, I must model subduction zones with a variety of aspect ratios and areas. This will be testing the effect of the downdip limit or the along-dip length of the seismogenic zone on generating great earthquakes. The number or frequency of cascading events and large earthquakes for each scenario must be determined and compared. In order for my hypothesis to be validated, there must be a significantly larger amount of cascade events in model runs corresponding to deeper seismogenic downdip limits than in runs corresponding to
  • 12. 12 shallower downdip limits. If my hypothesis is proven to be correct, this could potentially explain the relative absence of great earthquakes seen on the Tonga megathrust despite its high convergence rates and frequency of seismicity. If no significant relationship is seen between seismogenic zone downdip limit and large earthquakes, it could rule out downdip limit as a controlling factor in the production of large earthquakes, and a different explanation must be proposed to explain Tonga’s lack of great earthquakes. Timeline The following is an approximation of what will need to be finished at what time in order to stay on track to graduate at the beginning of the Summer 2016 semester: March • Construct 2D model of subduction zone • Begin writing of thesis April • Finish model and thesis • Prepare presentation for thesis defense Late April/Early May • Defend thesis May • Make any changes to thesis and/or model for reevaluation if necessary • Submit corrected thesis to graduate school
  • 13. 13 Figures Figure 1 - A megathrust fault model based of off the Cascadia subduction zone differentiating between 3 distinct regions of slip (Colella et al., 2012) Figure 2 - A cross-section of a subduction zone showing locations of velocity- (or rate) weakening and velocity-strengthening slip (Liu, 2013)
  • 14. 14 A B Figure 3 – A.) A to-scale cross-section of the subduction zone off the eastern coast of Japan differentiating between 4 distinct regions of slip B.) A more detailed look at the slip on the fault interface of the 4 regions shown in 3A. (Lay et al., 2012)
  • 15. 15 A B Figure 4 – A.) Velocity seismogram schematic where the P-wave of the main shock followed a separate initial phase of multiple smaller events (left), and the slip history of the 3 events (right) B.) A series of steps showing the cooperation of nearby cracks leading to the growth of a fault in intact rock (Reches, 1999)
  • 16. 16 Figure 5 – Locations of strong coupling (red regions) on the Sunda megathrust due to subducted asperities creating locked patches correspond to locations of large earthquakes (Konca et al., 2008) Figure 6 – Results of an asymmetrical two-block slider block model showing chaotic behavior (Turcotte, 1997)
  • 17. 17 Works Cited Beavan, J., Wang, X., Holden, C., Wilson, K., Power, W., Prasetya, G., … Kautoke, R. (2010). Near-simultaneous great earthquakes at Tongan megathrust and outer rise in September 2009. Nature, 466(7309), 959–963. http://doi.org/10.1038/nature09292 Byerlee, J. (1978). Friction of rocks. Pure and Applied Geophysics PAGEOPH, 116(4-5), 615– 626. http://doi.org/10.1007/BF00876528 Colella, H. V., Dieterich, J. H., Richards-Dinger, K., & Rubin, A. M. (2012). Complex characteristics of slow slip events in subduction zones reproduced in multi-cycle simulations. Geophysical Research Letters, 39(20), n/a–n/a. http://doi.org/10.1029/2012GL053276 Konca, A. O., Avouac, J.-P., Sladen, A., Meltzner, A. J., Sieh, K., Fang, P., … Helmberger, D. V. (2008). Partial rupture of a locked patch of the Sumatra megathrust during the 2007 earthquake sequence. Nature, 456(7222), 631–635. http://doi.org/10.1038/nature07572 Lay, T., Kanamori, H., Ammon, C. J., Koper, K. D., Hutko, A. R., Ye, L., … Rushing, T. M. (2012). Depth-varying rupture properties of subduction zone megathrust faults. Journal of Geophysical Research: Solid Earth, 117(4), 1–21. http://doi.org/10.1029/2011JB009133 Liu, Y. (2013). Numerical simulations on megathrust rupture stabilized under strong dilatancy strengthening in slow slip region. Geophysical Research Letters, 40(7), 1311–1316. http://doi.org/10.1002/grl.50298 Ohnaka, M. (2003). A constitutive scaling law and a unified comprehension for frictional slip failure, shear fracture of intact rock, and earthquake rupture. Journal of Geophysical Research, 108(B2), 1–21. http://doi.org/10.1029/2000JB000123 Reches, Z. (1999). Mechanisms of slip nucleation during earthquakes. Earth and Planetary
  • 18. 18 Science Letters, 170(4), 475–486. http://doi.org/10.1016/S0012-821X(99)00122-3 Stein, S., & Wysession, M. (2003). An Introduction to Seismology, Earthquakes, and Earth Structure (1st ed.). Blackwell Publishing. Turcotte, D. (1997). Fractals and Chaos in Geology and Geophysics (2nd editio). Cambridge University Press. Wright, D. J., Bloomer, S. H., MacLeod, C. J., Taylor, B., & Goodlife, a. M. (2000). Bathymetry of the Tonga Trench and Forearc: A map series. Marine Geophysical Researches, 21(5), 489–511. http://doi.org/10.1023/A:1026514914220