Depth Dependent Rupture Properties
in Circum-Pacific
Subduction
Zones
Susan L. Bilek and Thorne Lay
Institute of Tectonics and Earth Sciences Dept.,
University of California, Santa Cruz
Depth dependence of the source rupture duration of interplate thrust
earthquakesis examinedfor sevensubductionzonesaround the Pacific to explore variations in faulting properties. Multi-station deconvolutionsof teleseismicP wavesfor moderate size earthquakesyield estimates of the source
time function and centroid depth for each event. Analysis of 17 to 75 earthquakesin eachregionrevealsa consistenttrend of decreasingsourceduration
(inferredfrom the sourcetime functions,after correctionfor differences
in
total energyrelease)with increasing
depth. Rupturedurationpatternsvary
somewhat betweensubductionzonesas well as along strike within a given
zone, and the data have large scatter, implying significantvariation in rupture processesalong the interplate megathrusts,but the depth dependence
appearsto be robust. The rupture duration variationsprompt consideration
of two end-membermodels:1) depth-dependent
rupturevelocityis caused
by variations of rigidity of materials in the fault zone, while static stress
drop is constant,and 2) staticstressdrop varieswith depthwhile material
propertiesand rupture velocity are constant. For the first model, the volumetrically averagedrigidity of the fault zone must increasewith depth in
eachregionby a factor of 5 betweendepthsof 5 to 20 km. If rupture velocity is constant,the stressdrop must increaseby an order of magnitudeover
the same depth range. This systematicvariation in rupture behavior with
depth may reflect spatial variationsin the amount, compactionand porosity
of sedimentin the fault zone, topographyon the subductingplate, phase
transitionsin the fault zone materials,thermal structureof the megathrust,
and varying presenceof fluids in the fault zone. Such physicalvariations
appear to control the physicsof rupture propagation, leading to intrinsic
dependenceof rupture velocityon materialsand fluidswithin the fault zone.
INTRODUCTION
faults [Pachecoet al., 1993]. Variationsin earthquake
rupture complexity and total energy releasehave been
Approximately90% of global seismicenergyrelease
related to grossproperties of subductionzone geomeoccurs in subduction zone earthquakes, primarily intry and kinematics, with the underlying premisethat
volvingthrusting motionson the interplate megathrust
earthquake faulting varies with the megathrust environment [Ruff and Kanamori, 1980; Lay et al., 1982;
Kanamori, 1986; Ruff, 1992]. While someintriguing
GeoComplexity
andthePhysics
of Earthquakes
general
associationshave become evident, we are far
Geophysical
Monograph120
from
a
detailed
understandingof what controlsfaulting
Copyright2000 by theAmericanGeophysical
Union
165
166
RUPTURE
PROPERTIES
IN SUBDUCTION
ZONES
in subductionzones. It is becomingincreasinglywell
documentedthat frictionalpropertiesof the megathrust
are verycomplex,and the entirespectrumof convergent
motionsmustbe considered.For example,someregions
experiencelarge amountsof slip occurringover periods
as long as one year after large thrust events. This has
beenobservedin Japan,whereHeki et al. [1997]used
geodeticmeasurementsto detect postseismicslip equal
to the amountof coseismic
slip in the year followingthe
December24, 1994 Sanriku-okievent. Similarly,signif-
with increasingdepth accompaniedby enhancedscatter in duration for intermediate depth events. Bos et
icant postseismicslip has been observedfor the 1992
parison of the rupture and radiation characteristicsof
intermediate and deep earthquakes,submitted, J. Geo-
Sanriku-okieventin Japan[Kawasakiet al., 1995].
Several subduction zones have also experienceddestructive tsunami earthquakes,which producea larger
tsunami than expected given their seismicmagnitude
al. [1998]and Houstonet al. [1998]stackbroadband
recordsfor events from 100 to 650 km deep, finding
a weak trend of decreasingsourceduration with increasingdepth that can be completely accountedfor
by the expected increase in shear wave velocity with
depth. Houstonet al. [1998]find evidenceof asymmerry and complexity in the stackedtime functionsfor
eventsfrom 350-550km depth. CampusandDas [Com-
phys. Res.] do not detect any unusualdepth depen-
dencefor intermediate and deep focuseventsfor events
in Fiji and Japan. Overall, no dramatic depth dependence
of rupture propertieshas been revealedwithin the
[Kanamori,1972].Theseeventscommonly
releaseseisintraplate
environment of intermediate and deep focus
mic wavesthat are depletedin high frequencyenergy,
earthquakes.
apparently as a result of having unusuallyslow rupMany studiesof earthquake rupture have been made
ture propagation[Pelayoand Wiens,1990;Kanamori
and Kikuchi, 1993; $atake, 1994; Tanioka and $atake, for individual or suites of events on the megathrust
1996; Johnsonand $atake, 1997; Ihrnldet al., 1998]. zones,and some have focusedon depth dependentsysThe distinctive seismicwave spectra of theseeventshas tematics. Tichelaarand Ruff [1991, 1993]examined
even served as a basis for near reM-time
detection of
variations in maximum depth of interplate thrust events
tsunami earthquakesinvolving calculation of spectral to assesscontrols on seismic coupling. Zhang and
energyratios [Newmanand Okal, 1998;Shapiroet al.,
1998].The observationthat tsunamieventsruptureat
Schwartz[1992]analyzedthe depthdistributionof moment
release in different
zones to assess variations
in
very shallow depths where weak sedimentsare likely to
be present in the fault zone has prompted speculation
thermal structure and stress. EkstrSm and Engdahl
that rupture in low rigidity materialscausesslowrupture velocitiesand resultsin a larger faulting displacement for a given seismicmoment. Thesefactorscan explain the spectral characteristicsand enhancedtsunami
excitationof tsunamiearthquakes[e.g.,Kanamori and
Kikuchi,1993].If this is correct,onemightexpectshal-
quakesin the central Aleutian Islands to examinevari-
lower thrust events in the interplate thrust zone to have
[1989]determinedsourceparametersfor many earth-
ationsin stressdistribution,and Taniokaet al. [1996]
consideredlateral variations in earthquake rupture and
slabmorphologyalongthe Japantrench. Bilek andLay
[1998,1999a]focusonearthquake
rupturedurationvariationsfor shalloweventsin the Japan and Middle Amer-
ica subductionzones,findingthat the shallowestevents
have anomalouslylong sourcedurations in both subductionzones.Bilek and Lay [1999b]showevidencefor
similar behaviorin other regions,and proposeseveral
possiblemechanismsto accountfor this observation,involvingphysicalattributesof the subductionzonesuch
as roughnessof the subductingplate, amountand type
of sedimentbeing subducted,thermal structure, and
fluid processesin the fault zone. Earthquake behavthrust zone.
ior in the interplate seismogeniczone can thus guide
There have been severalinvestigationsof earthquake interpretationsof the thermal and petrologicalstrucrupture duration variations with depth in subduction ture of subduction
zones[e.g.,Peacock,1993;Hyndman
zones, primarily focused on intraplate events within
et al., 1997; Oleskevichet al., 1999]. This study exthe subductedslab. Vidaleand Houston[1993]stacked tends the examination of depth dependentvariationsof
short period P waveformsto estimate sourceduration
earthquakerupture duration in sevensubductionzones:
of 160 intermediateto deepevents(depth > 100 km). Japan, Kuriles, Alaska-AleutianIslands, Mexico, MidThey found evidencefor a decreaseof sourceduration
dle America, Peru, and Chile.
longer rupture durations than deeper events, and relatively long duration ruptures should be observedin
regionswhere significantsedimentis subductingcompared to regionswhere no sedimentis present. It is not
clear what to expect in terms of complexity of rupture,
other than the possibilitythat the shallowenvironment
is intrinsically more heterogeneous
in material properties and stressthan the deeper portion of the interplate
BILEK
120'
150'
180'
-150'
-120'
-90'
-60'
• II
..•,
........
30'
.
LAY
1'67
time function that representsthe time history of seis-
60'
'.i!i•?i•ii:•i•ii.i!ii??i..
-.?:... '7""'•:'21'::5i!i
.............
ii'ii
.............'60'
30'
AND
"•
mic moment releasefrom the source[Kanamori and
Ruff, 1983; Ruff, 1989; Tichelaar and Ruff, 1991; Ruff
and Miller, 1994]. The deconvolution
methodis based
on computingsynthetic P wave Green's functionsfor
eacheventusingthe bestdoublecoupleof the Harvard
CMT solution for a model with a water layer over a
O'ß..•,"•x•
•'•:•:•:•5•.:•
................... • '
-
'
e•
O'
uniformhalf spacewith a P wavevelocityof 6.0 km/s.
We deconvolvethe P waves by Green's functions generated for a rangeof 15-25point sourcedepths,obtaining sourcefunctionsat eachtrial depth. The depth at
-30' ............
• •...
-30'
_60.•
' ' l' I
-60
ø
120'
150'
180'
-150 ø'
which the deconvolution
minimizes
the misfit between
the data and syntheticseismograms
is preferred.While
-120'
-•'
-60'
simultaneous
inversion for a revised focal mechanism
may reducesomeuncertainties,it is not viable for most
of
our eventsgiven the limitations of the availableP
Figure 1. Map showingsubductionzonesstudied. Boxes
enclose the areas of event locations.
wavedata. In general,tradeoffsbetweensourcemechanism and either source depth or source time function
are not too severefor shallow thrusting events,and the
METHODOLOGY
CMT solutions are probably fairly robust for most of
The subduction zones around the Pacific that we anour events. The general processingsequenceis shown
alyze (Figure 1) are selectedprimarilyon the basisof in Figure 2 for an event in the Aleutian Islandsregion.
abundant shallow interplate seismicity. Our focus is The stations used are well-distributed azimuthally from
on frictional and faulting processesalong the megath- the source,which ensuresa range of wave shapesand
Green'sfunctionsthat reducesthe severe
rust zones,and severalcriteria are used to select events corresponding
trade-offs between depth and source function. From
located on the main thrust fault. Earthquakes are initially selected
basedon' (1) closeproximityto the main the deconvolutionprocedure, performed for 23 depths
thrustzoneof interest,(2) havinga faultingmechanism in Figure 2, we determinean optimal depth of 31 km
(from the HarvardCentroidMomentTensor(CMT)
catalog)with strike,dip, and rake consistent
with un-
based on the minimum
in the misfit
recordings
are usedfor mostevents).Thesecriteriaare
when the true source radiation
curve.
The corre-
spondingsourcefunctionhasa simpletrapezoidalshape
derthrustingof the subductingplate (typically,events with a duration of 7 s, followedby somelow amplitude
have a strike within 20-300 of the local strike of the
oscillations. The latter oscillationsare highly variable,
trench, a dip • 30-35ø, and a rake of 90ø+30ø), (3) and representinstabilities causedby accumulatinginhavinga momentmagnitude(Mw) of 5.0-7.5,and (4) accuraciesin the Green's functions with lapse time into
availability of at least 4 good quality broadbandtele- the signal. This eventand the associatedP wavesignals
seismicP wave recordingsthat are well distributed az- are well characterizedby the strong trapezoidal pulse,
imuthallyfrom the source(between7 and 15 P wave althoughthere is intrinsically somesubjectivity as to
was finished.
Figures 3-9 show final sourcetime functionsfor all
events in the 7 subduction zones. The panel for each
[e.g., Zhangand Schwartz,1992; Tichelaarand Ruff, event shows the source time function for the optimal
sourcedepth and our preferred depth and sourcedu1991,1993].
For the selectedevents, we obtain all available tele- ration. In some cases, the depth determination is roseismicvertical componentbroadband recordingsfrom bust, with a singledistinct minimumin the misfit curve.
the IRIS data center with time windowsappropriate for However, in other cases,there are double or multiple
the direct P phaseand associated
depth phases(pP, minima in the misfit curves,or a range in depths with
sP) neededfor accuratesourcedepth determination. relatively uniform misfit. For thesecases,we examine
Ground displacementtracesare obtainedby deconvolv- the sourcetime function producedfor each depth, and
ing the instrumentresponse,and the P waveonsetsare choosethe depth which yields the simplest time funcmanuallypickedon eachrecord. A multistation decon- tion, with most of the moment releasedearly in the
similar to those used by other authors in earlier studies of depth dependenceof subduction zone properties
volution
method
is then used to determine
the source
signal[Christensen
andRuff, 1985].
168 RUPTURE
PROPERTIES
IN SUBDUCTION
ZONES
58 ø
56 ø
54 ø
52 ø
ae
50 ø
170 ø
180 ø
-170 ø
O.4OO
_150 ø
_160 ø
ALE
10.1
0.375'
FFC 54.6
0.350'
ANMO
78.8
0.325
CTAO
214.1
DAV
246.8
0.300
CHTO
0.275
'' '1'5'' '2•5'' '3)5'' '45
Depth (km)
be
BJT 282.4
HIA
292.0
AAK
Best
Depth 31km
Error
0.279
276.6
BRVK
ANTO
LVZ
310.1
320.1
336.6
346.1
BFO 355.6
n
de
7s
Figure 2. Example of data processingfor M•=6.49 event in the Aleutian Islandsregion. (a) Focal
mechanismof the event taken from the Harvard CMT catalog. The event was chosenbecauseof its
underthrustingmechanism(strike 262ø, dip 24ø, rake 114ø),closeproximityto the trench,and moderate
magnitude.(b) Error or misfitas a functionof depth for the deconvolutions.
We performthe deconvolution for 23 different depths to minimize the misfit between the data and synthetics. Minimum occurs at
31 km depth. (c) Sourcetime functionfor deconvolution
at 31 km depth. On left is sourcetime function,
on right is sourcetime function with misfit boundsin gray dashedlines. Sourceduration is measured
from the first large peak of the source time function; the black bar indicates the measuredduration of
7 s. (d) Data (solidblacklines)and syntheticseismograms
(dashed)shownwith the stationcodeand
azimuth
from
the event.
The point-sourcetime functionsin Figures3 to 9 further illustrate the difficultiesintrinsic to measuringthe
time interval of sourceenergy release. As in Figure 2,
the deconvolutionssometimesshow significant amplitude oscillations after the main pulse of moment rate,
or significantnegativeovershootfollowingthe primary
pulse. For the largest events, unparameterizedspatial
finitenessmay contribute to these features, but their
sensitivityto sourcedepth suggeststhat inaccuracyof
the later portions of the Green'sfunctionsis the primary
BILEK
89/01/22
9 km
89/11/01
2 ]• 12
km
89/11/04
II
90/0'•/10
91/11/26
28
km• 17
kmI
14km
4s
•
93/0•25
32km
/
94/1•28
169
24 km
Jl
92/07/12
14km
92/07/13
50
km 47
km 41
km
ls
9•07/18
9•07/18
9•07/18
7 km
12km
18
93/03/25
93/09/11
93/10/28
30km
•lkm
2s
LAY
km
• 89/10/26
I 789/10/27
• 89/10/29
•, 89/10/29
9s
9•07/16
AND
3s
!
•km
•
ls
ls
9•07/25
•
13km
9•07/29
•
9•/0•/08
[
5km
5 km
9•/08/1•
/
33km
1
94/1•28
95/01/01
95/01/06
95/01/06
95/01/11
6km I 6km• 9km • 4km • 37km 43km
5s
95/01/21
95/11/23
95/0•23
95/1 •30
95/04/25
95/05/25
ls
3s
ls
95/07/08.
95/09/26
95/1 •30
Figure 3. Sourcetime functionsfor Japanevents. Each eventpanellists the eventdate, the best
sourcetime function,and the preferreddepth and duration. Duration is measuredfrom the sourcetime
functions, with each horizontal step of 1 s.
culprit. This is not unexpectedgivenpossibleerrorsin
the focal mechanismand the very simplevelocity model
assumedin computingthe Green'sfunctions.Until realistic near-surfaceand wedgevelocitystructureis known
in eachregionand fully three-dimensionalGreen'sfunctions can be computed for such structures, it will be
hard to improvethe accuracyof the time function estimates. The time function complexity makes it difficult
170
RUPTURE
PROPERTIES
I:
IN SUBDUCTION
ZONES
90/07/06
91/12/07
34km
17km
91/12/13
]• 91/12/13
91/12/13
I 91/12/19
20km
9km
15km
9km
I 3s 2•
7••ff•
]
•/•
92/07/12
31km
91/12/22
24km
92/02/02
25km
2s
94/01/24
ls
94/01/26
s
]
94/05/18
! los
92/07/14
16km
93/03/22
26km
93/09/16
36km
4s
ls
4s
94/08/18
94/08/18
94/08/20
7s
ls
30km
tli7km
] 35km
I 7km13km
35km
I
94/08/20
I! 94/08/28
94/10/25
]• 40km
95/01/12 30km
95/02/14
26km
16km
•f['[ 94/10/09
32km
14km
ls
•
95/02/21
95/03/11 L• 95/04/17
30km
43km
34km
27km
2s
95/04/18
'1• 31km
f
26
29
km
3s
I
16km
95/04/19 •. 95/04/28
30
km
3s
8km
5s
'ix,-,
30km
9,107
8km
95/12/07 95/12/08
[r• 95/12/10] 96/01/06 96/01/31 96/02/04
30km
2s
16
km
2s
[1 13
km
II 36km
18
km
3s
I• 2s
37km
Figure 4. Sourcetimefunctions
for Kitrileevents,in sameformatasFigure3.
to definethe cessation
of coherentenergyreleasefrom
the source,and in somecasesthere may be isolated
secondary
pulsesof sourceradiation. However,every
caseshowndoeshavea primary pulseof momentrate
in the early part of the signalthat accounts
wellfor the
predominant features of the teleseismic P waves. We
proceedby measuringthe durationof the primarypositive pulseof the sourcetime function,whichusually
sufficesto characterizethe energyreleasehistory. We
attempt to be consistentin measuringthe width of this
BILEK
AND
96102/14
96/03/09
96/03/30
96/04/06
96/05/07
36km
33km
9km
36km
18km
44km
96/10/01
45kl•
96/10/28
30km
97/02/28
31km
97/10/08
14km
7km
97/10/09
97/10/11
14km
]• 96/02/07
LAY
171
2s
98/01/20
Figure 4. Continued
pulse,definingthe end as when the pulsereturns to a
baseline level. We incorporate error bars on the duration estimatesto subjectivelyreflect the difficultiesin
measuringthe time function duration.
There are severaltrade-offsin determining the source
duration and depth parameters. Sourcetime function
duration and depth have particularly strong trade-offs
Holdingthe focalmechanismconstantis anothersource
of possibleerror in the depth determinations,but tests
of the effects of focal mechanism uncertainty indicate
that the depth estimatesare not very sensitiveto small
focalmechanismchanges.The error barsthat we assign
to the depth estimatesreflectthesesourcesof error. The
point-sourceassumptionexplicit in our deconvolutions
for individualstations[e.g.,Christensen
andRuff,1985]; clearly resultsin someaveragingof spatial finitenessefhowever,by performingthe deconvolution
for manywell fects that may bias our sourceduration estimates. Esdistributed stations, the ability to separate these pa- timation of source radiation duration from the source
rameters is greatly improved. Additionally, there are time functions involvesan approximation that there is
trade-offs between velocity model and source depth. a simple rupture processthat yields negligibledirectivThe choiceof a half spacevelocityof 6.0 km/s undoubt- ity effect. As most of our events are of moderate size,
edly biasesour depth estimates.Tests usinga different and the deconvolution processemphasizeswave periods
averagevelocityshowthat a 10%changein averageve- longer than 1 or 2 s, the directivity effectsin teleseislocity leadsto a corresponding
8-12% changein source mic P waves are likely to be below the resolution of
depth(approximately
3-4 km), similarto the resultsof our measurementsin almost all cases. The good fit of
Tichelaarand Ruff [1991].Giventhe likelihoodof low the point sourcesyntheticsto the observationsgivesdivelocitiesin the sedimentarywedge above the thrust rect support for this assertion. The residual waveform
plane,we probablyoverestimate
true depthsby several mismatch error varies somewhat, but not in a fashion
kilometers,particularly for the shallowestevents. Our simply linked to the shape of the sourcetime functions
depth estimatesstill tend to be shallowerthan thosein or the event moment (see Figure 2). As a first-order
the CMT catalog(Figure10), partly dueto the higher approximation, we take the duration of the main pulse
averagecrustal and upper mantle velocityof the Pre- of the source time function as an estimate of the ac-
liminaryReferenceEarth Model (PREM) [Dziewonski tual rupture duration. We correct the rupture duraand Anderson,1981]structureusedin the CMT inver- tion estimatesfor the effect of varying seismicmoment
sion. There are also differencescausedby the practice (Mo), as it has beenempiricallyshownthat the duraof fixingCMT depthsat either15km or 33 km for shal- tion is proportionalto the cuberoot of Mo [Kanamori
low earthquakeswhenthe depth resolutionis not good. and Anderson, 1975; Houston et al., 1998; Campus and
172 RUPTURE
PROPERTIES
IN SUBDUCTION
ZONES
89/09/20Jl 89/10/07•1 90/01/08 |
5km
•l•l. 15km 11110km
•
•
90/03/12II
91/05/30
91/07/20
18km I1 30km
32km
-u•j•_
•v
9•W•V1
• • •'•
91108105
27km •
91/11/26]
38km
l
9•0•05
29 km
•.
•U•
9•03/26 I
5 km
/
3•
9•06/03 /
17km
•
9•06/24
17km
15km r• 21km ] 16km
10km
36km II• 7km
•9•09/30
n9•09/30
• 9•09/30
• 9•10/01
9•11/10
.. 9•11/11
/•
I
5•
6•
9•11/24
93/01/23I 93/04/16
14km
36km
93/05/28
93/0•02
3s
94/04/05
/
28km
93/08/19
•
93/05/15
23
km
93/05/17
•/ 39
93/05/25
33
km
km
93/10/29
38km
• 15km
ls
94/07/29
3s
94/08/05
•1•
93/11/12
93/1•03
1 Skm Okm
94/08/05
94/09/28
94/10/10
10km7km 29km 9km• 8km• 20km
95/01/16
95/03/14
95/04/23
95/05/23
95/06/07
95/07/08
8km• 36
km
•10
km 17
km 29
km 7km
13s
I s
2s
11 s
Figure 5. Source time functions for Alaska-Aleutian Islands events,in same format as Figure 3.
Das, 1999]. Each durationestimateis dividedby the
cube root of the associated Mo from the Harvard CMT
catalog,normalizedto a momentmagnitude(Mw) 6.0
event. The CMT moments are used for scalingrather
than the moments
determined
from the deconvolution
process,as we feel the CMT moments are more stable
estimatesbecausethey are derived from large numbers
of long-perioddata. Our deconvolutionmomentstend
to underpredict the values found by Harvard, in part
due to the velocity model used, and possibly due to
BILEK
95/10/30
96/02/15
96/03/22
96/03/26
96/03/28
I
AND
LAY
173
96/03/28
2s
96/03/28
96/03/30
96/03/30
96/04/21
8kin
7kin
7kin
• 32kin ]
35kin
9•04/23
96/04/26
96/06/08
96/0•11
96/06/11
9•06/12
9•06/16
]
16kin
96/07/10
• 31kin
• 19kin
• 17kin17kin38kin
L 34kin
3s
96/08/19
96/1•08
O,m
•
•.
ls
3s
97/03/26
97/03/27
97/04/02
•o•
•o•
1 ,,,m I.
2s 1• u
1• '•
•o•o•
•m
ls
97/01/19
3s
!
•o•o•
•o•o
'•m
,•m
ls
97/1•17
97/1•26
16km
33
km
ls
•o•o•
I
•m
1• •m
•
,•m
2s
Figure 5. Continued.
omissionof later low amplitude energyrelease.Scaling any dependenceon event size. Our data are alsofree of
by the cuberoot of momentis widelypracticed,and al- any systematicrelationship between sourcedepth and
lowsour resultsto be easily comparedwith other stud- seismicmoment(Figure11b), whichis importantin asies [Ekstr6mand Engdahl,1989; Vidale and Houston, sessingthe results.
1993;Bos et al., 1998;Houstonet al., 1998],after takRESULTS
ing into accountthe differences
in the referencemoment
usedin the variousstudies. Figure 11a showsthe obFigure 11c showsnormalized duration estimates as
servedrelationshipbetweenour durationestimatesand
Mo. The unscaleddurations show a clear increasewith a function of source depth for the entire dataset, repincreasingMo, while the scaleddurationsare free of resenting 354 events. We include 68 events from a
174
RUPTURE
PROPERTIES
IN SUBDUCTION
92/06/07
14 km
3s
ZONES
92/06/07
7 km
5s
93/03/31
20 km
3s
93/05/15
20 km
2s
93/09/30
13 km
93/10/24
22 km
8s
4s
93/11/13
7 km
5s
95/09/14
22 km
10s
95/10/06
17 km
95/10/12
19 km
3s
95/10/30
16 km
3s
96/02/25
11 km
96/02/25
6 km
96/02/26
8 km
5s
4s
96/03/19
12 km
3s
96/03/27
17 km
2s
96/07/15
19 km
6s
96/07/18
20 km
2s
97/07/19
9 km
12s
97/1 2/16
7 km
7s
98/02/03
24 km
ls
3s
13s
Figure 6. Sourcetime functions for Mexico events,in same format as Figure 3.
study of maximum couplingdepth in subductionzones depth. With typical error bars being +1.5 s, the long
[TichelaarandRuff, 1991;TichelaarandRuff, 1993], duration estimates appear to be real anomalies.
Figure 12 showsthe same data, subdividedby geowhich provideddepth and duration estimatesby the
samegeneralprocedureusedhere. Thoseduration es- graphicregion and with error bars on depth and dutimates are alsoscaledusingcube root of moment scal- ration values. The trend of decreasingsourceduration
ing. The trend of decreasingduration with increasing with increasingdepth is apparent for each region, aldepth is very clear, even in this compositeplot which though somesubductionzoneshave strongerpatterns
combines data from subduction zones with different rethan others. For instance,data for the Alaska-Aleutian
gionalcharacteristics.
A sourcedurationof about 3 s is Islandsregionshow a very dramatic decreasein duratypical of an Mw = 6.0 event, but scaleddurationsas tion down to a depth of 15-20 km, then a flatter dislongas 18 s are foundfor eventsshallowerthan 15 km tribution at depths deeper than 20 km. The Kuriles,
BILEK
91/06/21
92/02/21
AND
LAY
175
92/04/05
_90/03/25
90/04/28
91/03/16
• 21km
I 6km
i 4s
15km
h•11km4 7s
10 km
92/05/30
92/08/10
92/08/11
92/09/05
93/05/2C
93/07/16
il 22km 10km8km•, 13km
•1 20km
l 19km
3s 4s•
93/07/21
93/09/03
.
10krn
22 km
'•1 11km
93/09/03
•
93/09/10
!
7 km
I
93/09/10
.
7km
It 13km
2SI1•1••1•
93/09/10
I1•
• 7
93/09/14
93/09/19
93/09/11
• 93/09/13
I•
km
13
km q 93/09/30
13
kmII. 94/03/12
14
km
20 km
_
94/05/01
Ii
17km
96/04/01
,'
24km
96/08/27
95/01/20
95/09/06
5
96/03/03
•
.
96/03/03
25km •h 16km
I 16
km
•Lkm
I[• I1•
96/08/28
96/09/04
3• 3it
• •
96/12/17
•
94/06/29
•
96/11/17
96/12/14
, 3•••.•r•.
97/12/22
10km
Figure 7. Sourcetime functions for Central America events,in same format as Figure 3.
Mexico and Peru showsimilar abrupt changesat depths
shallower than 25 km, but these three regions have
fewer events than the Alaska-Aleutian Islands region.
The South American zones have slightly shorter duration plateaus at depths greater than 20 km than
do Japan and the Aleutians, which may reflect differencesin crustal structure overlying the subduction
zones. Central America displays the weakest variation
with depth, with events lessthan 10 krn deep showing
some increase in duration.
The scatter
in duration
is
176
RUPTURE
PROPERTIES
IN SUBDUCTION
ZONES
90/09/02
17 km
3s
91/07/01
12 km
9s
92/05/16
40 km
4s
92/07/17
39 km
ls
94/1 2/14
21 km
2s
96/08/05
9 km
96/11/12
24 km
7s
7s
96/11/13
11 km
96/11/14
9 km
7s
5s
96/12/16
6 km
6s
97/08/15
16 km
4s
97/O2/O9
19 km
3s
96/01/19
32 km
2s
96/11/13
22 km
3s
96/11/15
12 km
4s
97/06/3O
25 km
ls
98/08/04
23 km
7s
Figure 8. Sourcetime functionsfor Peruevents,in sameformatas Figure3.
comparablebetweenregionsfor depthsgreaterthan 10
km, and indicatesthe intrinsic variability in earthquake
rupture processesand errors in duration estimation.
Given that our depth estimates differ from both the
CMT values and earthquake bulletin values, we seek
to confirm that the events are truly on the megathrust, and not intraplate ruptures. Plate i shows our
depth and duration data plotted as a function of distance perpendicularto the regional trench axis. This
distancewasmeasuredusingbathymetry mapsto track
the trench axis. Eventslocatedwithin the dashedgray
boxes are consideredlikely to lie on the main plate interface, which may itself involvemultiple thrust planes,
within a 4-5-10 km region. Those eventswhich lie outsidethe box may either be mislocated(laterallyor in
depth), are intraplateearthquakeswith thrust mechanismssimilar to underthrustingat the plate boundary,
or arethrusteventsoccurringin the accretionary
wedge
abovethe seismogenic
zone.The eventsdefinedipping
plate interfacesin eachregion,with someof the scatter
attributableto changesin plate geometry.The distributionsare much tighter than providedby either the
CMT or earthquakebulletinlocations,soour depthestimation processgenerallyappearsto have succeeded.
In regionsthat exhibit a lot of scatter and where there
is significantalong-strikegeometrychange,we binned
the data to ensurethat we include only eventsthat are
confidentlyidentified as interplate thrust events. The
data still show a decrease in source duration
with in-
creasingdepth alongthe plate interfacefor eachregion,
BILEK
90/08/02
92/05/22
92/06/21
92/11/04
93/03/19
AND
LAY
177
93/07/11
] 9kmI 26km33km
I 27km
• 9km[•49km
ls
21
km ]I
34 km
6
95/05/28
95/07/30 I
47 km
3s
0
6s
94/09/1294/09/17
19 km
29
94/12/10
km
38 km
95/08/01 I
95/08/02
95/08/02
95/08/02
38
km•L6kmI 21
km• 20
km 22
km 19
km
95/08/03
!1] 20km I
95/11/30
28 km
95/10/03
95/10/16
95/10/31
95/11/01
22km
23km
ls
18km •
2s
20km I• 5km
95/11/21
96/04/19
44 km
96/07/03 ]
I] 6 km
•
97/03/09
38 km
97/04/13
24 km
97/05/25
13km
••/•97/07/06
[ 97/07/06
t 97/07/24
I 97/07/25
•,97/07/27
97/08/18
21km
I
97/11/03 I
41km
7 km
98/01/12 I
!
15km
7 km
98/07/29
98/09/03
46km
0•1 7 km
[
14km
27km
Figure 9. Sourcetime functionsfor C!file events,in sameformat as Figure 3.
mainly becausevery few of the shallowestevents have
been excluded.
ALASKA-ALEUTIAN
ISLANDS
DATASET
The strongpattern and large numberof data in Figure 12 for the Alaska-AleutianIslandsregionprompted
a more detailed look at the seismic parameters for
eventsin this area. There have been many great earthquakes as well as tsunami earthquakes in this region,
which emphasizesthe importance of understandingthe
nature of the seismogeniczone here. Figure 13a shows
the locations of the events analyzed for the AlaskaAleutian Islands region;a total of 74 earthquakesfrom
1989 to 1997. We consider the patterns in the data in
both depth and along-strikebins.
178 RUPTURE
PROPERTIES
IN SUBDUCTION
ZONES
7O
30 km has the shortest
6O
age sourcefunction for the shalloweventssuggestsevidenceof greater rupture complexity as well, although
the subeventfeatures are highly variable from event to
•50
duration
of 7.0 s.
The
aver-
event.
We alsoconsideralong-strikevariationsin the source
•
time
40
• •o
2O
ß
10
20
30
of the Alaska-Aleutian
tions for the shallowest
%
lO
0
functions
40
Islands
events.
The along-strikebins are based on distinct tectonic
blocksas describedby Geist et al. [1988]. Initially we
consideredall events in each spatial bin, but this was
problematic due to the distinct shapeof the time func-
ß
50
60
70
Our Depths(km)
Figure 10. Comparisonof depths as listed by the Harvard
CMT catalog and depths determined in this study. Our
depths tend to be more shallow than the Harvard CMT
depths, and our depths do not show any bias towards certain
depths, such as the large grouping of Harvard CMT depths
of 15 km and 33 km.
Figure 13b showsthe averagesourcetime functions
alongthe entire Alaska-Aleutianarc for four depth bins
of 0-10km (20 events),10-20km (26 events),20-30km
(10events),and30-40km (18events).Boththe seismic
moment rate amplitudes and the sourcedurations were
scaledto removethe effectsof varying seismicmoment,
events.
The
shallowest
events
in eachregiondisplayqualitatively similar broadening,
but the total numbersare too small to averagejust the
shallowestevents. For this reason,we stack only events
deeperthan 10 km in the along-strikebinsshownin Figure 13a usingthe same procedureas describedabove.
Figure 13cshowsthe averagesourcetime functionsfor
eachsubregion,plotted on samescaleto facilitate comparison between regions. Four of the six bins look
remarkably similar, but events in the Unimak Block
and ShumaginBlock differ somewhatfrom the others.
There may still be some depth effect in these stacks,
as events in the Unimak Block have mainly 10-20 km
depthswhile eventsin the ShumaginBlock havemainly
30-40 km depths. The causeof these variationsis unclear at this point, but lateral variationsalongthe arc
are certainly lesspronouncedthan the commondepth
variation.
Our data for the Alaska-Aleutian
Islands events do
suggestvariationsin maximumcouplingdepthfor this
region. Tichelaarand Ruff [1993]examined11 events
in the Aleutian Islands region and 5 eventsnear the
Alaskan peninsulato determinean averagemaximum
the reference
Mw 6.0 event(Mo 1.16x1025
dyne-cm). couplingdepth for the Aleutiansof 35-41 km and 37The moment rate amplitudeswere divided by the event
41 km for Alaska. Our larger data set supportsthese
M2/3inordertomaintain
therelationship
between
mo- averagesof maximumcouplingdepth, althoughit apment and area under the time function. Individual time
pearsthat there is significantvariationin the coupling
depth alongstrikeof the trench,with somesectionsof
functionswere set to 0 prior to the beginningof the large
the seismogenic
zonehavingmuchdeepercouplingthan
pulseof energyin the sourcetime function, and 0 after
othersections.Tichelaarand Ruff [1993]mentionthat
our pick of the termination point of the faulting radiathe easternmostportion of the 1957earthquakerupture
tion. The stackedsourcefunctionsfor eachdepth range
have been scaled so that the area beneath each curve
zone may have shallowercoupling; this region correis equalto 1.16x1025
dyne-cm,to facilitatecomparison. spondsapproximatelyto the Unimak Blockbin, where
The most obvious feature is the dramatic difference in
we see earthquakesonly down to 20 km depth. We
alsoobservethe possibilityof shallowercouplingaround
the shape of the sourcetime function for the 0-10 km
longitude1800within the Rat and DelarofBlockbins,
bin, as this visually demonstratesthe generaltrend of
with deepercouplingfoundon eitherside. We are only
decreasingsourceduration with increasingdepth. The
time function for the 0-10 km bin has a duration of
considering
eventswith Mw 5-7.7,soour resultson coupling depth are appropriatefor that magnituderange.
16.4 s while the time function for events greater than
followingthe procedureof Houstonet al. [1998],prior
to binning the eventsand averagingthe time functions.
The time scalesweredividedby ••oo, normalizedby
o
BILEK
AND
LAY
DistanceFrom Trench(km)
0
o
25
50
75
':"•
. -
100 125 150 175 200 225
'""
•
'l ....... d'
' 'w'"'-
.•,•:e•'• .•
0
0
25
50
75
,ill[
Japan 10
40
40
•
50
50
60
0
I
25
2
3 4
50
75
...............
5
6
7
8 9 1011
.....
;....,,.,
•.,, •,•.._, ,, •_
..-'-lO
- -
Aleutiang
20
30
40
40
50
50
ß
0
100 125 150 175 200 225
25
l
Illlllll
III
6 8 10 12 14 16 18
50
75
100 125 150 175 200 225
;,.
•..,
,..
.,,,..,
.._,.
Mexico
2 3 4 5 6 7 8 9 1011 1213
25
•7•-• •-
.............
'
1o
4
Illtill
0
75
__
60
0
50
I ]•l,I
ß
• -" •'•.n-•
n.
1 2 3 4 5 6 7 8 9 101112131415
25
Ill[
Kuriles
...............
10
30
2
0
0
20
o
I Iw
ß
Alaska-
Inllin
,wIII
•
0
100 125 150 175 200 225
.
III
6O
II
-1 0
Ill.
-
n
&
100 125 150 175 200 225
I !.
-
20
30
•
•
i
10
50
- :-
75
100 125 !50
.,,• ........-
175 200 225
, .....
.;......, •:::•: . •:
..•.•
20
30
30
40
50
50
60
60
l
012345678910
0
25
50
01234567891011
75
100 125 150 175 200 225
o
1o
20
.•,.
30
40
5o
6o
0 1 2 3 4 5 6 7 8 9 101112
Plate 1. Normalized source duration and source depth as a function of distance from the respective
trench The color bar below indicates source duration for each symbol. Error bars show range in depth
estimatesfrom the range in misfit from the deconvolutionprocessing.The dashedboxes enclosethose
events that define the plate interface, allowing for some scatter due to variation in the along-strike
subductionzonegeometry.Eventswhichlie outsidethe box are likely either accretionarywedgeor intra-
plate events(gray symbolsin Figure 12). Trianglesrepresenteventsdeterminedby Tichelaarand Ruff
[1991,1993].
1'79
180 RUPTURE
PROPERTIES
IN SUBDUCTION
ZONES
lOO
le+28
le+27
•1e+26
• le+25
•
le+24
bo
ao
0.1
ß
le+23
i
ß
1e+•4
i
ß
1e+•
i
ß
1e+•
le+23
ß
i
1e+•?
le+28
ß 10I ß 20I ß 30I ß 4 •) ß 50I
0
Seismicmoment(dyne-cm)
i
Depth(km)
Figure 11. Relationship of sourceduration and depth with
18
seismicmoment. (a) Plot of sourcedurationas a function
ß
of moment.
16
• 14
tion
ß ß ßß
8
• 46
2
0
moments
are taken
from
the Harvard
rLg"u•dg"J
ß
ß
i
.
is measured
from
the source
time
function
in I s in-
tervals. Solid circles represent normalized source duration
ß
as a function
•**-.•t*
. ß*ß o*ø o-.
etr..ø..o_-•k
10
eO Iß
'f'
--
ß
20
i
30
i
40
i
50
ß
60
As data accumulate, it may prove viable to map out
lateral variations in faulting complexity and coupling
depth in greater detail, but the presentresults indicate
subtle variations
increase
in source duration
trend in this plot, with moments scattered over our depth
-1'ß ' '• ulo•o.%eoc.
'
The
(b) Seismicmomentasa functionof eventdepth. We seeno
** -
*.. '
i
of moment.
with increasing moment has been removed from the data.
Depth•m)
that rather
Event
CMT catalog. Open squaresrepresentraw durations, which
show an expected increase with increasingmoment. Linear
pattern of symbols results from the fact that source dura-
will be found.
range. (c) Compositeplot of normalizedsourcedurationas
a function of depth. There is an overall trend of decreasing
sourceduration with increasing depth, and wide scatter is
likely due to the composite nature of the plot.
sedimentsat very shallowdepths. Rigidity is a key material property that can be related to source duration
through its direct influenceon shear wave velocity. For
a simple unilateral rupture model, sourceduration is in-
verselyrelatedto rupture velocity. The rupture velocity
DISCUSSION
Further mapping out of lateral variationsin behavior
of earthquake ruptures within each zone and between
zonesremains a desirablegoal, but for the remainder of
this study we focus on the common depth-dependence
apparent in Figure 11c. Seismologicalanalysis reveals
only grossattributes of the sourceenergy release,and
there is substantialnon-uniquenesswhen interpreting
sourcetime function characteristicsin terms of dynamical behavior. We will consider several simple endmember possibilitiesto frame the problem. One possible explanation for the observed variations in source
duration is systematicvariation with depth of rigidity of
the material in the seismogeniczone. This is an extension of the notion that tsunami earthquakes have slow
rupture velocitiesbecausethe slip occursin low rigidity
(V•) is empiricallyfoundto be approximately
equalto
80%4-10%of the shearwavevelocity(/3)[Scholz,
1990],
/3-•-P
(1)
where/• is the rigidity and p is the material density. If
we assumea constant static stress drop model for scaling, variations in the duration can be associatedwith
variations in rigidity. The constant stressdrop model
appearsvalidfor a rangeof earthquakemagnitudes
[e.g.,
Abercrombie,
1995], althoughwe do expectscatterto
result from variations in stressdrop. Thus we can use
our source duration
measurements
to estimate
volume-
averaged(overthe rangeof largestrainaccumulation)
rigidity variationswith depth in the seismogenic
zone.
In order to calculate the rigidity, we use a constant
densityof 2.7 g/cm3. It is clearthat the densitywill
BILEK
.•,
15
10
10
I
ø
o
•
0
LAY
10
20
+i'
30
40
o
50
60
0
10
20
30
40
Aleutians
50
60
Mexico
..o 15
•
•o
o
........
o
20
i
10
....
!
20
....
i
30
....
i
40
....
i
,
,
,
50
....
i
•
60
0
10
20
30
40
50
60
....
Central
Peru
America
15
10
• •o
o
10
o
20
....
!
....
20'
•
....
;0
i
....
40
!
....
50'
i
0
60
,
0
10
20
+.+,,
30
40
50
60
Depth (km)
....
Chile
o
o
10
181
Japan
''•...................
Kuriles
• 15
•
AND
20
30
40
50
6o
Depth (km)
Figure 12. Normalizedsourcedurationas a functionof sourcedepthfor eachof the analyzedregions.
Open squaresindicateeventsanalyzedin a similarfashionby Tichelaarand Ruff [1991, 1993]. All of
the 354 event sourcedurationshave been scaledusingthe cube root of seismicmoment, normalizedto a
M,o=6.0 event. Gray symbolsindicate thosethrust eventssatisfyingour initial criteria, but have depth
estimatessuggestingthat they did not occur at the plate interface.
182
RUPTURE
PROPERTIES
IN SUBDUCTION
ZONES
60 ø
o O-lOkm
[
10-20
km[
20-30km I
-•3o-4o
kmI
55 ø
Peninsula
Rat Block
Delarof
in Block
Block
Andreanof
50 ø
170 ø
175 ø
!
[
180 ø
-175 ø
He
Block
!
_170 o
_160 o
_165 o
_150 ø
_155 ø
4'le+241
3.1e+24-•
/'
2'1e+24
1/'
1.0e+24
• _•
.
5.2e+24•
Delarof
Block
4'le+24
1
2'1e+241
/
z
0
5
10
15
20
Duration (s)
5.2e+24
•
Un•Bl•k
4.1e+24
•
Figure 13. (a) Detailed map of Alaska-AleutianIslands
study area.
Symbols show NEIC location of events with
symbolshapedenotingdepth of event (seelegend). Boxes
indicatedistincttectonicblocksfrom Geistet a/.[1988].(b)
Average normalized time functions for the depth bins 0-10
km (sohdblack), 10-20km (dashedgray), 20-30km (dashed
black), and 30-40km (solidgr•y). (•) Alongstrikebinsof
sourcetime functionsfor blocksindicatedin (a). Average
sourcetime functionsweregeneratedin sameway as in (b),
however only events with depths greater than 10 km were
included
in these bins.
'
'
' Alaska •msula
5.2e+24
•
•.le+•l
•.,e+•l /
•.le+•l /
,.•2•
•
•o
'
•5
C.
•o
T•e
•5
16.o
•2.5
(s)
change somewhat with depth, but we use a constant event,with a 3.0-3.5 km/s rupture velocity.This choice
density for simplicity. The other parameter needed to of a rupture dimensionaffectsthe baselineof the rigidity
estimate rigidity is a source dimension, as the source values, but not any systematic depth variations. It is
not presently possible to independently determine the
duration (v) equals
rupture dimensionsfor all of our eventsusingdirectivity
source dimension
analysis becauseof their small magnitude. Since it is
-
We assume a uniform
.
source dimension
(2)
of 10 km for
unlikelythat they all rupture precisely(scaled)10 km
dimensions,we expect significant scatter to remain.
Figure 14 showsthe seismogeniczone rigidity esti-
our moment-scaled rupture durations. This dimension
mates obtained
is c9nsistentwith the 3-4 s duration for a typical Mw 6.0
ered in this study. There is an order of magnitude
for the seven subduction
zones consid-
BILEK
1o5
....
i
ß
•
ß
....
I
....
AND
LAY
183
Studies of tsunami earthquakes have inferred compa-
I
RigklJy(kbars)
Averagerigidily(kbars)
Rigidity
(g•bars)TR dataset
rable (factor of 5 to 10) reductionsof rigidity for the
seismogeniczone necessaryto reducerupture velocities
1oa
to about i km/s and to accountfor the enhancedslip
neededto satisfy tsunami excitation given the observed
seismicmoments[I(anamori, 1972;Pelayoand Wiens,
1992; I(anamori and Kikuchi, 1993; $atake, 1994; Hein-
rich et al., 1998].Thus, thesecalculations
suggestthat
100
10
0
60
Figure 14. Plot of estimatedrigidity variationsalongthe
megathrust for the entire dataset. Small black circles indicate rigidity valuescalculatedfrom our measuredsourcedurations, trianglesindicate rigidity estimatesfor the 68 events
in the Tichelaar and Ruff [1991, 1993] datasets,and open
circlesare averagerigidity valuesover a depth interval. The
solid black line indicates the rigidity valuesestimated from
PREM
shear wave velocities and densities.
scatter in the data at all depths, but we find a general trend of increasingrigidity with increasingdepth
in the rangeof 5-50 km. Rigidity valuesestimatedfrom
P REM are includedin the figure to provide a reference
earth comparisonfor a layeredcrust/mantleintraplate
oceanicenvironment. Our averagevaluesare very similar to PREM at depths of 20-40 km, but are lower than
P REM by a factor of as much as 5 at depths shallower
than 20 km. Given the likely contributions to error in
the rigidity estimatesfrom the factorsdescribedabove,
only the average trend should be considered,not the
full range of values.
One concern inherent
in these estimates
is the use of
Harvard CMT momentsin scalingthe sourcedurations.
These momentsare calculatedusing the PREM structure for seismicwave excitation. If the rigidity variations in Figure 14 are correct, the excitation should be
recomputedfor a correspondingdecreaseof rigidity near
the surface.Overestimatingthe rigidity in the modeling
may have yielded seismic moments that are too small
for shallow events, which would underestimate the correctionsfor moment scaling. A quantitative correction
for this effectis very difficult, both becausethe rigidity
is not independently known, and the excitation should
be computed for a realistic three-dimensionalmodel to
obtain
unbiased
moments.
We believe
that
the overall
effect of suchcorrectionfor moment scalingwould be a
slightly reducedrangeof rigidity values,but the general
trend would be preserved.
all subduction zones may have shallow regions with
low rigidity properties that could enable large tsunami
earthquakes to take place. Whether they do or not
is likely to be a consequenceof the slip history of the
deeperportionsof the seismogenic
zone, along with the
myriad factors that control the transition from stable
sliding to stick-slip instabilities.
If the rupture velocity is decreased because of low
shear velocities, a fairly thick zone of reduced rigidity must be present to affect the volumetric strain releaseduring rupture. The rigidity estimatesin Figure
14 would then be averagedover the seismogenic
zones,
and local properties right at the plate interface may
vary even more. The fact that our rigidity estimatesdo
not displayany seismicmomentdependence
(especially
whenvery largetsunamieventresultsare included),indicatesthat the scaleof the regionof low rigidity may be
substantial. Various factors could produce a distributed
zoneof low rigidity material near the fault zone. A thick
zone of sediments
both
above
and within
the seismo-
geniczoneis a likely candidatefor significantreduction
of rigidity relative to hard rock values. Lower rigidities
can also be associated
with
increased
water
content
in
the material, which is abetted by having high porosity
sediments.
Increased
water
content
can also be the re-
sult of phase transitions such as smectite to illite, and
even basalt to eclogite, which can releasewater from
the hydrous phase.
Other factors may be important. Previous studies
have examined the depth dependence of moment re-
lease[Zhangand Schwartz,1992], maximumcoupling
depth [Tichelaarand Ruff, 1991;1993],stressdrop and
sourcedurationsfor deep earthquakes[e.g., EkstrSm
and Engdahl, 1989; Vidale and Houston, 1993; Bos et
al., 1998;Houstonet al., 1998],and lateral variations
in earthquakeoccurrence
[Taniokaet al., 1997].These
authors have invoked a number of possiblefactors for
causingthe variability in earthquake rupture, such as
amount and types of sediment being subducted, thermal state, hydrologiceffects,and changesin subducting
plate roughness.Systematicsin any of thesefactorsis
also a plausible causefor the depth dependenceof the
sourceduration and rigidity. However,with the current
184
RUPTURE
PROPERTIES
IN SUBDUCTION
ZONES
Figure 16 showscorresponding
estimatesof the static
stressdrop as a function of depth for eachof the studied
regions. For eachcalculation,we usethe Harvard CMT
..........
•_•••'"'"'"•verridingplate
•
•o•
Subdueting
plate •
catalogmomentsand a constant/• - 3.5 km/s. The
shearwave velocity will likely vary with depth because
of rigidity variationswith depth, but we use a constant
value for simplicity for these calculations. This model
predictsthat stressdropincreaseswith increasingdepth
ranging over an order of magnitude. This is certainly
plausible,however,a study of dynamic stressdrop for
recentlarge earthquakesshowsthat the stressdrop estimated from sourcetime functions is basicallyconstant
compaction
•/
'"'•
'"•L.•
•o
'74•water
ex•lled
'"
o,..o
• • 'ph•e
tr•sitio•
OOo *:'
_
defo•tion
of segments
for the eventsoncethe changein shearvelocity(and
thereforerigidity variations)with depth and regionis
takeninto account[Ruff, 1998]. Effortsto improveour
ability to estimate rupture dimensionsfor moderatesize
events are essential if we are to resolve the trade-off
Figure 15. Generalizedschematicof a subductionzone,
with a detailedblow up of the seismogenic
zoneinterface,
modifiedfrom Bilek and Lay [1999b].
data set, it is difficult to determine which is the most
important causeof the variations. Figure 15 showsa
schematicof the seismogenic
zoneindicatingsomeof
the possibleprocesses
likely involvedin changingthe
material rigidity.
The model of rigidity variationsis non-unique. We
assumedthat the scaled rupture area of the eventsis
constant with volumetric material properties controlling the variations in duration. Another end member
modelinvolvesa constantrupture velocitywith a varying rupture area for each event. Such a model implies
static stressdrop variations with depth. We lack independentestimatesof the fault areafor eachevent,which
is neededto resolvethe trade-offbetweenrupture area
be-
tween fault area and rupture velocity variations.
The discussionabove invokesvery simple notions of
earthquakemechanics,either with rupture velocity being controlledby volumetrically averagedrigidity and
shear velocity variations, or with variable static stress
drop resulting from depth-dependentchangesin frictional behavior. It is likely that such effectscould be
coupled,with low rigidity regionstendingto havelarger
fault areas,so that the true explanationlies in between.
However, perhaps the most plausible model is one in
which the micromechanicalpropertiesof the fault zone
influencethe macroscopicearthquake rupture. In particular, porous sedimentsat shallow depths in the seismogeniczonemay be fluid saturated,with fluidsplaying
a critical role in earthquake slip. Such a model is con-
sideredby Kanamoriand Heaton[1999],whoespecially
and rupturevelocity[Vidale and Houston,1993]. Instead we relate static stressdrop to seismicmoment
throughits relationshipto fault displacementand area.
For a circular crack model, static stressdrop is
1000
7rrMo
Art-- 16ra
(3)
where r is the radius of the circular fault area and Mo
is the seismicmoment[KanamoriandAnderson,1975].
If we make a further assumptionthat the rupture velocity approximatelyequalsthe shear velocity,we can
substitute the product of the shear velocity/5 and the
measuredsourcedurationr for the fault radius,leaving
for the static stressdrop.
10
20
30
40
50
60
Depth(km)
Figure 16. Plot of stressdrop variationsalongthe megath-
7rrMo
A•r- 16/•ar
a
0
(4)
rust with depth. Small black circlesindicate stressdrop estimates calculatedfrom our sourcedurationsand triangles
indicate stressdrop estimates calculated from the Tichelaar
and Ruff [1991,1993]datasets.
BILEK
emphasizethe microscaleinteractions of frictional heating and fluid pressurizationas key factorsin the macroscopicbehavior of faulting. Essentially,it may not be
necessaryto have large volumesof low rigidity material if the very presenceof fluid rich sedimentsin the
seismogeniczone can directly reduce rupture velocity
intrinsically. Such models need further elaboration, but
our observationsprovide key targets for explanation.
AND LAY
185
properties with depth in the Japan subductionzone, Science, 281, 1175-1178, 1998.
Bilek, S.L., and Lay, T., Comparisonof depth dependent
fault zone properties in the Japan trench and Middle
America trench, Pure Appl. Geophys., 15•, 433-456,
1999a.
Bilek, S.L., and Lay, T., Rigidity variationswith depth along
interplate megathrust faults in subduction zones, Nature,
•00, 443-446, 1999b.
Bos, A.G., Nolet, G., Rubin, A., Houston, H., and Vidale,
J.E., Duration of deep earthquakesdetermined by stackCONCLUSIONS
ing of Global SeismographNetwork seismograms,J. Geophys. Res., 103, 21059-21065, 1998.
Earthquake source time functions for a large num- Christensen, D.H., and Ruff, L.J., Analysis of the tradeber of events in seven circum-Pacific
zones indicate
off between hypocentral depth and sourcetime function,
Bull. Seisin. Soc. Am., 75, 1637-1656, 1985.
that moment-normalized source rupture duration deDziewonski, A.M., and Anderson, D.L., Preliminary refercreaseswith increasing depth along the seismogenic
ence earth model, Phys. Earth Planet. Inter., 25, 297-
zone. There are minor regionaldifferencesin details of
this relationshipthat may be related to the tectonic environment, but the depth dependenceappears to be robust. The Alaska-Aleutian Islandsregion has the most
completedataset, and showsdramatic differencesin the
shapeof the sourcetime functionswith depth, with minor changesin time function shape along strike of the
trench, indicating that the depth dependenceis more
than any along-strike variations for this region. Two
end member models of earthquake rupture processare
consideredto explain the observations,one with varying volumetrically-averaged
sourcezonerigidity and the
other with variable static stress drop. For the first
model, we find that rigidity increasesby a factor of 5
over the depth range of 5 to 20 km. Low rigidity of the
shallowportion of the seismogenic
zonemay the result
of sediments, high porosity, and weakly consolidated
materials, all of which diminish in volume with depth.
The variablestress•lrop modelpredictsan orderof magnitude increasein stressdrop with increasingdepth. It
is possiblethat the two scenariosare coupled, with material property variationsinfluencingstressdrop, but it
may alsobe true that our observationsreflect microscale
influences
of sediments
and fluids in the fault
zone on
rupture propagation.
Acknowledgments. We made extensive use of the Har-
vard CMT catalog.GMT software[WesselandSmith,1991]
was used for figure preparation. All data were obtained
from the IRIS DMS. This work was supported by NSF EAR
9418643.
This
is contribution
number
407 of the Institute
of Tectonics, University of California, Santa Cruz.
REFERENCES
Abercrombie,R.E., Earthquakesourcescalingrelationships
from-1 to 5 ML using seismogramsrecordedat 2.5 km
depth, J. Geophys.Res., 100, 24015-24036, 1995.
Bilek, S.L., and Lay, T., Variation of interplate fault zone
356, 1981.
EkstrSm, G., and Engdahl, E.R., Earthquake sourceparameters and stressdistribution in the Adak Island region of
the central Aleutian Islands, J. Geophys.Res., 9•, 1549915519, 1989.
Geist, E.L., Childs, J.R., and Scholl, D.W., The origin of
summit basins of the Aleutian ridge: Implications for
block rotation of an arc massif, Tectonics, 7, 327-341,
1988.
Heinrich, P., Schindele,F., Guibourg, S., and Ihmld, P.M.,
Modeling of the February 1996 Peruvian tsunami, Geophys. Res. Lett., 25, 2687-2690, 1998.
Heki, K., Miyazaki, S., and Tsuji, H., Silent fault slip following an interplate thrust earthquake at the Japan trench,
Nature, 386, 595-598, 1997.
Houston, H., Benz, H.M., and Vidale, J.E., Time functions
of deep earthquakes from broadband and short period
stacks, J. Geophys. Res., 103, 29895-29913, 1998.
Hyndman, R.D., Yamano, M., and Oleskevich, D.A., The
seismogeniczone of subduction thrust faults, The Island
Arc, 6, 244-260, 1997.
Ihml•, P.M., Gomez, J.-M., Heinrich, P., and Guibourg,
S., The 1996Peru tsunamigenic
earthqu•tke:
Broadband
source process, Geophys. Res. Lett., 25, 2691-2694, 1998.
Johnson, J.M., and Satake, K., Estimation of seismic moment and slip distribution of the April 1, 1946, Aleutian
tsunami earthquake, J. Geophys. Res., 102, 11765-11779,
1997.
Kanamori, H., Mechanism of tsunami earthquakes, Phys.
Earth Planet. Int., 6, 246-259, 1972.
Kanamori, H., Rupture processof subduction-zone earthquakes, Ann. Rev. Earth Planet. Sci., 1•, 293-322, 1986.
Kanamori, H., and Anderson, D.L., Theoretical basis of
some empirical relations in seismology,Bull. Seisin. Soc.
Am., 65, 1073-1095, 1975.
Kanamori, H., and Heaton, T.H., Microscopic and macroscopicphysicsof earthquakes, this volume, 1999.
Kanamori, H., and Kikuchi, M., The 1992 Nicaragua earthquake: A slow tsunami earthquake associatedwith subducted sediments,Nature, $61, 714-716, 1993.
Kawasaki, I., Asai, Y., Tamura, Y., Sagiya, T., Mikami,
N., Okada, Y., Sakata, M., and Kasahara, M., The 1992
Sanriku-Oki, Japan, ultra-slow earthquake, J. Phys.
Earth, J3, 105-116, 1995.
186 RUPTURE
PROPERTIES
IN SUBDUCTION
ZONES
Lay, T., Kanamori, H., and Ruff, L., The asperity model and
the nature of large subduction zone earthquakes, Earthquake Pred. Res., 1, 3-71, 1982.
Newman, A.V., and Okal, E.A., Teleseismicestimates of ra-
diatedseismicenergy:The E/Mo discriminantfor tsunami
earthquakes, J. Geophys. Res., 103, 26885-26898, 1998.
Oleskevich,D.A., Hyndman, R.D., and Wang, K., The updip and downdip limits to great subduction earthquakes:
Thermal and structural models of Cascadia, south Alaska,
SW Japan, and Chile, J. Geophys. Res., 10•, 14965-
684-694, 1993.
Pelayo, A.M., and Wiens, D.A., The November 20, 1960
Peru tsunami earthquake: Source mechanism of a slow
earthquake, Geophys. Res. Left., 17, 661-664, 1990.
Pelayo, A.M., and Wiens, D.A., Tsunami earthquakes: Slow
thrust-faulting events in the accretionary wedge, J. Geophys. Res., 97, 15321-15337, 1992.
Ruff, L.J., Multi-trace deconvolution with unknown trace
Omnilinear
inversion
of P and $ waves for
source time functions, Geophys. Res. Left., 16, 10431046, 1989a.
Ruff, L.J., Do trench sediments affect great earthquake
occurrence
202-230, 1983.
Ruff, L.J., and Miller, A.D., Rupture processof large earthquakes in the northern Mexico gubduction zone, Pure
Appl. Geophys., 1•œ, 101-171, 1994.
Satake, K., Mechanismof the 1992 Nicaragua tsunami earthquake, Geophys. Res. Left., œ1,2519-2522, 1994.
Scholz, C., The mechanics of earthquakes and faulting, 449
pp., Cambridge University Press, Cambridge, U.K., 1990.
Shapiro, N.M., $ingh, $.K., and Pacheco,J., A fast and sim-
ple diagnosticmethod for identifying tsunamigenicearth-
14991, 1999.
Pacheco, J.F., Sykes, L.R., and Scholz, C.H., Nature of seismic couplingalong simple plate boundariesof the subduction type, J. Geophys. Res., 98, 14133-14159, 1993.
Peacock, S.M., Metamorphism, dehydration, and the importance of the blueschist-eclogitetransition in subducting oceanic crust, Geologic. Soc. of Amer. Bull. 105,
scale factors:
period diffracted P-waves, Phys. Earth Planet. Int., 31,
in subduction
zones?
In:
L.
Ruff
and
H.
Kanamori (Editors), $ubductionZones,II., Pure Appl.
Geophys., 1œ9,263-282, 1989b.
Ruff, L.J., Asperity distributions and large earthquake occurrence in subduction zones, Tectonophys., œ11, 61-83,
quakes, Geophys. Res. Left., œ5,3911-3914, 1998.
Tanioka, Y., and Satake, K., Fault parameters of the 1896
Sanriku tsunami earthquake estimated from tsunami numerical modeling, Geophys. Res. Lett., œ3, 1549-1552,
1996.
Tanioka, Y., Ruff, L., and Satake, K., What controls the
lateral variation of large earthquake occurrencealong the
Japan trench?, Island Arc, 6, 261-266, 1996.
Tichelaar, B.W., and Ruff, L.J., Seismiccouplingalong the
Chilean subduction zone, J. Geophys. Res., 96, 1199712022, 1991.
Tichelaar, B.W., and Ruff, L.J., Depth of seismiccoupling
along subductionzones,•J. Geophys. Res., 98, 2017-2037,
1993.
Vidale, J.E., and Houston, H., The depth dependence of
earthquake duration and implications for rupture mechanics, Nature, 35, 45-47, 1993.
Wessel,P., and Smith, W.H.F., Free softwarehelpsmap and
display data, EOS, 7œ,441-445, 1991.
Zhang, Z., and Schwartz, S.Y., Depth distribution of moment releasein underthrustingearthquakesat subduction
zones, J. Geophys. Res., 97, 537-544, 1992.
1992.
Ruff, L.J., Quasi-dynamic stress drop of recent large earthquakes, EOS Trans., A G U, 79, 657, 1998.
Ruff, L.J., and Kanamori, H., Seismicityand the subduction
process, Phys. Earth Planet. Inter., œ3,240-252, 1980.
Ruff, L.J., and Kanamori, H., The rupture processand asperity distribution of three great earthquakesfrom long
S.L. Bilek and T. Lay, Earth SciencesDept., Earth and
Marine SciencesBldg. A232, University of California, Santa
Cruz, Santa Cruz, CA 95064) (e-mail: [email protected];
[email protected])