University of California, Santa Cruz - Tsunamigenic Major and …thorne/TL.pdfs/LRL_Tsunamigenic... · 2016-07-21 · T Tsunamigenic Major and Great Earthquakes (2004–2013): Source

T

Tsunamigenic Major and GreatEarthquakes (2004–2013): SourceProcesses Inverted from Seismic,Geodetic, and Sea-Level Data

S. Lorito1, F. Romano1 and T. Lay21Roma 1, Sez. Sismologia e Tettonofisica, IstitutoNazionale di Geofisica e Vulcanologia, Roma,Italy2Department of Earth and Planetary Sciences,University of California Santa Cruz, Santa Cruz,CA, USA

Glossary

Earthquake Location and FaultParameters

Dip The angle between the Earth’shorizontal surface and theearthquake fault

Epicenter The position (longitude,latitude) of the hypocenterprojected on the Earth’s surface

EarthquakeFault

The fracture within a rockvolume that separates two rockmasses where there is evidenceof a relative displacement (slip)between them during anearthquake; the surface wherethe slip occurred is termed thefault surface; if the fault

surface is inclined, the rockmasses above and below thefault are termed hanging walland footwall, respectively

Hypocenter The position (longitude,latitude, depth) of theearthquake nucleation point

Rake (or SlipDirection)

The angle of the relativedisplacement between thehanging wall and the footwallof a fault during an earthquakemeasured in the fault surfacerelative to the horizontal strikedirection

Strike The angle, measured relative tonorth, of the intersection(trace) of the earthquake faultwith the Earth’s surface

Faulting Style

Normal Fault Fault where the hanging wallmoves down with respect to thefootwall

Reverse Fault Fault where the hanging wallmoves up with respect tothe footwall; if the dip angle is<= 45�, it is termed a thrustfault

Strike-SlipFault

Fault (often steeply dipping)where the two blocks separatedby the fault surface movehorizontally along the fault tracerelative to each other; if the

# Springer Science+Business Media New York 2015R.A. Meyers (ed.), Encyclopedia of Complexity and Systems Science,DOI 10.1007/978-3-642-27737-5_641-1

motion of the block across thefault trace from an observermoves toward left, themechanism is termed leftlateral, otherwise themechanism is termed rightlateral

TranspressiveFault

Reverse fault with an obliquerake component

Kinematic Rupture Parametersand Inversion

Inversion(or InverseProblem)

A procedure that uses a set ofdirect or indirectobservations to retrieve theparameters of a model that isconsistent with theobservations; this is acommon approach (manytechniques are useddepending on the problem tobe solved) in seismology toretrieve earthquakekinematic parameters suchas the seismic moment, slipdistribution, and rupturevelocity

Rupture Area Portion of the fault surfacethat slips during anearthquake

Rupture Velocity Velocity at which the rupturefront propagates over thefault surface during anearthquake; it may be veryirregular spatially

Slip Distribution Pattern that describes theearthquake rupture in termsof timing and amount of slipalong the fault surface

Earthquake Size

MomentMagnitude

Measure of earthquakemagnitude computed usingthe seismic moment, M0,estimated from variousmethods

Mw ¼ 23log M0ð Þ � 10:73,

�

withM0 in dyne � cmÞRigidity(or ShearModulus)

Shear stress to shear strainratio of a material; thiscoefficient is used to measurethe stiffness of a material anddescribe its response toapplied shear stress; insubduction zoneenvironments, the rigidityvaries with depth, with lowervalues in the shallowsedimentary part of themegathrust

Seismic Moment A scalar measure of theearthquake size based on therupture area (A), the averageslip (D) on the fault surface,and the rigidity (m) of themedium surrounding thefault; the relationship amongthese parameters is

M0 ¼ A � D � mSeismic Cycle

CoseismicPhase

The interval in which theaccumulated stress on the faultis suddenly released causing anearthquake

InterseismicCoupling

The capability of accumulatingstress on a fault in the period thatseparates two large seismicevents (interseismic phase);strongly coupled (or locked)faults accumulate stress at therate at which the plates aremoving relative to each otherand, in principle, are more proneto host large earthquakes

PostseismicPhase

The interval in which additionalslip or viscoelastic relaxationoccurs after the mainshock aseffect of the stress and strainredistribution

Seismic Cycle The three phases involving theprocess of stress and strainbuildup (interseismic), release

2 Tsunamigenic Major and Great Earthquakes (2004–2013)

(coseismic), and redistribution(postseismic) that occurs alongthe fault surface before, during,and after an earthquake,respectively; each phase occurswith a very different time scale(many years for interseismic,seconds for coseismic, fromminutes to months forpostseismic)

Strain Is the deformation of a rockmass subjected to tectonicstresses

Stress Is the force per unit area appliedon a surface (e.g., a faultsurface) within a volume

Subduction

Megathrust An extremely large thrust fault,formed along a subduction zoneplate boundary, hosting very largeearthquakes

SubductionZone

Convergence zone of two tectonicplates where one plate movesbeneath the other; the contactsurface between the twoconvergent plates is termed thesubduction interface ormegathrust; subduction zones arethe most seismically activeenvironments on Earth and wheremost of the largest earthquakesoccur

Trench The deep bathymetric troughposition where the subductingplate begins to descend beneaththe overriding plate

Wedge The shallow portion of theoverriding plate characterized bythe accumulation of sedimentarymaterial due to the collision withthe subducting plate

Tsunami

Run-up The maximum topographicelevation reached by the tsunamiduring the inundation

TsunamiAmplitude

The difference, computed on atsunami waveform, between thepeak of the tsunami wave and theundisturbed sea-level value

TsunamiHeight

The difference, computed on atsunami waveform, betweensuccessive crest and troughvalues of a tsunami wave

TsunamiPeriod

The time between twosuccessive tsunami waveextrema (generally consideringthe crest of the wave) that is theamount of time taken by atsunami wave to complete acycle

TsunamiWavelength

The distance between twoconsecutive tsunami waves(generally considering the crestof the wave)

Definition of the Subject

The decade 2004–2013 was characterized by anunusually large rate (1.7 per year) of great earth-quakes (MW �8) (Lay 2015). The majorityoccurred on subduction zone megathrusts. Signif-icant tsunamis were generated by most of them, aswell as by several major (MW 7.0–7.9) earth-quakes at very shallow depth on megathrusts, orby great intraplate normal faulting events nearsubduction zones.

Two of the great events, the 2004 Sumatra-Andaman MW 9.2 and the 2011 Tohoku MW 9.0earthquakes are, respectively, the third and thefourth largest earthquakes seismologically recorded(since 1900) and the largest events of the last50 years (http://earthquake.usgs.gov/earthquakes/world/10_largest_world.php). These events werelarger than had been anticipated for their sourceregions and produced two of the most damagingtsunamis in modern times. It is now widelyaccepted that great tsunamigenic earthquakes mayoccur in most subduction zones worldwide,although very infrequently in some regions (e.g.,Stein and Okal 2011; Kagan and Jackson 2013).

The occurrence of many important seismicevents during a time of greatly expanded global

Tsunamigenic Major and Great Earthquakes (2004–2013) 3

http://earthquake.usgs.gov/earthquakes/world/10_largest_world.php


seismological, geodetic, and tsunami recordingsystems has provided the scientific communitywith an unprecedented set of geophysical obser-vations. Data availability has prompted the devel-opment of new data inversion techniques forreconstruction and understanding of the space-time fault sliding process during the earthquake,which, in turn, provides a better understanding ofthe resulting tsunamigenic potential.

We address most of the major and greattsunamigenic earthquakes that occurred over thedecade. The objective of this entry is to discuss thegeneral features of source models – as retrieved byinversion of seismic, geodetic, and tsunamidata – and of the observed tsunamis. We highlightthe kinematic complexity of the rupture processesin subduction zones and the relation between faultslip distribution, bathymetry, and coastal featureson one hand and the observed tsunami impacts onthe other hand.

Introduction

The seismological instrumental era begun in theearly twentieth century (http://earthquake.usgs.gov/learn/topics/seismology/history/history_seis.php), asdid systematic production of global seismicevent catalogs (http://www.isc.ac.uk/iscgem/).Self-recording tide gauges (http://noc.ac.uk/science-technology/climate-sea-level/sea-level/tides/tide-gauges/tide-gauge-instrumentation) have been record-ing tsunamis since the mid-nineteenth century,whereas the tide-gauge catalog seems to be globallycomplete since the mid-twentieth century (Geist andParsons 2011; https://www.ngdc.noaa.gov/hazard/tide.shtml).

The largest megathrust earthquakes seismolo-gically recorded prior to the decade 2004–2013were the 1952 MW 9.0 Kamchatka, 1964 MW 9.2Alaska, and 1960 MW 9.5 Chile earthquakes(Kanamori 1977; http://earthquake.usgs.gov/earthquakes/world/10_largest_world.php), all ofwhich generated transoceanic tsunamis in thePacific. The cumulative global seismic momentrelease of those three events exceeds that of allgreat events of the recent decade, but the rate ofgreat earthquakes was higher from 2004 to 2013

than previously observed (Lay 2015). Most ofthese great earthquakes occurred near the Pacificand Indian Plate boundary margins where previ-ous great earthquakes have occurred, but �40 %ruptured within lithospheric plates rather thandirectly on plate boundaries, and the location ofthe latter events could not be directly anticipatedbased on plate motions. Even for the plate bound-ary events, some of the ruptures had surprisingspatial extent and features of their slip distribu-tion, as well as nature of the tsunamis theygenerated.

The overall size of the 2004 MW 9.2 Sumatra-Andaman 2004 megathrust earthquake had notbeen anticipated for the region along on theSunda Trench from Northern Sumatra to theAndaman Islands. This was due in part to thelack of knowledge of its subsequently discoveredmiddle-ages predecessors (Sieh et al. 2015 andreferences therein) and in part because of widelyaccepted assumptions about the seismic potentialof the highly oblique convergence of oceanic lith-osphere along this plate boundary. The rupturelength (>1300 km) and duration (�8 min) wereunprecedented. This earthquake rupture spreadacross the rupture zones of several prior major-to-great events (1847, 1881, 1941), without beingconstrained by any lateral segmentation of themegathrust. The very long rupture duration neces-sitated improvements of seismological inversionmethods for finite source models (see, e.g.,Shearer and B€urgmann 2010). For example, thefirst reliable estimate of moment magnitude wasmade long after the earthquake (e.g., Parket al. 2005), and it took several months to con-verge on reliable finite-fault models from seismol-ogy (e.g., Ammon et al. 2005) and even longer foranalysis of the measurements from campaigngeodesy (e.g., Vigny et al. 2005).

This earthquake generated a huge tsunami thatexpanded throughout the Indian Ocean and intothe Pacific and Atlantic Oceans (Titov et al. 2005).The catastrophic loss of life (�228,000 people)triggered global efforts aimed at providing tsu-nami warning systems for more regions than thePacific. This includes the Caribbean and Mediter-ranean, where tsunamigenic events have a lowoccurrence rate and the hazard from infrequent


http://earthquake.usgs.gov/learn/topics/seismology/history/history_seis.php

http://earthquake.usgs.gov/learn/topics/seismology/history/history_seis.php

http://www.isc.ac.uk/iscgem/

http://noc.ac.uk/science-technology/climate-sea-level/sea-level/tides/tide-gauges/tide-gauge-instrumentation



https://www.ngdc.noaa.gov/hazard/tide.shtml




catastrophic events was underappreciated prior tothe 2004 Sumatra event.

The December 2004 event was the first in asequence of earthquakes and tsunamis to strikealong the Sunda Trench from 2004 to 2010 thatwe describe in some detail in this entry and whichincluded not only “ordinary” megathrust earth-quakes but also two “tsunami earthquakes”.Within the Indo-Australian plate seaward of the2004 rupture, the largest ever recorded strike-slipearthquake (MW 8.7) also occurred in 2012, gen-erating a small tsunami.

Similar to the 2004 Sumatra-Andaman earth-quake, the Tohoku 2011 MW 9.0 earthquake waslarger than had been anticipated in the region,mostly because of incomplete information aboutprior earthquakes and incorrect assumptionsconcerning fault segmentation and the maximumpossible magnitude (Stein and Okal 2011). Therupture length of significant slip spanned only�350 km along strike, much shorter than the2004 Sumatra earthquake. Both events havequite extreme values compared to the expectedmean value for earthquake scaling laws (e.g.,Strasser et al. 2010). How large the variability ofrupture size can be is also appreciated consideringthat the 2010 Maule earthquake (MW 8.8,corresponding to a two to three smaller seismicmoment) ruptured a fault length of �500 km.

The 2011 Tohoku event involved a very largeamount (>�50 m) of slip reaching all the way tothe trench, which strongly increased the tsunamiexcitation. It is commonly thought that the shal-low sedimentary wedge cannot accumulate signif-icant elastic strain and the shallow fault frictionshould be slip velocity strengthening, favoringaseismic creep rather than stick-slip earthquakefailure (e.g., Wang and Kinoshita 2013), so thelarge coseismic slip at shallow depth was surpris-ing. Dynamic rupture processes appear to haveplayed a role in this large slip at shallow depth.

An important context for anticipating futurestrong earthquakes is the identification of seismicgaps (e.g., Nishenko 1991), i.e., regions wherelarge documented earthquakes occurred in thepast and the time since the last large rupture is asignificant fraction of the average recurrenceinterval (which can be inferred from frequency

of past events or from the strain accumulationrate). Based on the elastic rebound mechanismof stress/strain accumulation during theinterseismic period that results from frictionalresistance to motion between two moving platesand the sudden stress/strain release during subse-quent earthquake sliding, the earthquake occur-rence probability on a fault may (simplistically)tend to increase with time since the last earth-quake. Over the past several decades, most greatsubduction zone megathrust earthquakes have infact occurred or initiated within previously iden-tified seismic gaps or regions that at least hadunknown prior large event history (e.g., Lay2015). This may reflect the relative simplicity ofthe seismic cycle involved in subduction zoneplate convergence. However, the timing of futureevents is not well-constrained, and seismic gapsare often not robustly identified due to limitedknowledge of prior events. Additionally, the rup-ture areas have not always conformed to identifiedgaps, likely due to the nonlinear cascade involvedin large earthquake ruptures; earthquake rupturesare affected by local scale frictional and geomet-rical heterogeneity, and dynamic rupture propaga-tion may occasionally overcome “barriers” ofrelatively low stress such as slip velocity-strengthening regions, as shown by numericalmodels (e.g., Kaneko et al. 2010). Some eventsonly partially ruptured seismic gaps (e.g., the2001 Peru (MW 8.4) event ruptured only 60 % ofthe 1868 rupture zone in southeastern Peru, andthe 2014 Iquique, Chile (MW 8.0), event rupturedonly 20 % of the 1877 rupture zone in northernChile); others ruptured the megathrust wellbeyond identified gaps (e.g., the Maule MW 8.8and the 2004 Sumatra MW 9.2 earthquakes); yetothers rupture the whole megathrust, as the 2011Tohoku MW 9.0 earthquake did, breaking acrossinferred regions of lateral down-dip-up-dip seg-mentation based on previous earthquake history(e.g., Stein et al. 2012). Together with the partialre-rupture in the Mentawai gap of the 1797 Suma-tra rupture zone by the large events in 2007, theseprovide examples of variable megathrust failureindicating that rupture zone-based segmentation isnot a reliable guide to future earthquake size. Theresulting variability of large earthquake ruptures


has prompted questions as to the statistical valueof the gap model (e.g., Jackson and Kagan 2011).

Subduction zones also occasionally host theabovementioned “tsunami earthquakes,”addressed elsewhere in this encyclopedia (Poletand Kanamori 2009). These are generally major(MW <8) and very shallow earthquakes that gen-erate a tsunami larger than expected for their mag-nitude. Observationally, they feature large slowslip and relatively long duration and weak groundmotion. For this reason they rarely trigger self-evacuation of coastal populations based on therecognition of strong shaking. Their rupture pro-cess is difficult to model and their frequency ofoccurrence is difficult to assess; thus, they areoften overlooked in tsunami hazard assessments(e.g., González et al. 2009; UNISDR 2015). Wehere discuss two tsunami earthquakes thatoccurred in the recent decade along the SundaTrench, the 2006 Java and 2010 Mentawai earth-quakes. Another case challenging the commonexpectation of aseismic or slow-slip behavior ofthe accretionary wedge that we discuss is theHaida Gwaii (2012) earthquake.

Other earthquakes that occurred in the decade2004–2013 demonstrated additional complexityof the earthquake rupture process and tsunamigeneration. For example, earthquake doubletsdemonstrate triggering interactions between tworegional fault systems and increased overall tsu-nami hazard, while events located in complicatedtectonic environments, such as deformed slabs orjunctions of three plates, also illustrate the diffi-culty of anticipating tsunami excitation. Under-standing the complexity of ruptures iscomplicated by the nonuniqueness of inversionfor the earthquake slip distribution and challeng-ing model uncertainty quantification, and thisenhances the value of analyzing seismic, geodetic,and tsunami data collectively to better constrainthe solutions.

The next section gives a brief discussion ofsome important concepts regarding tsunami gen-eration by earthquakes and their impact on localand distant coastlines (“Tsunamis and the SeismicSource”). Then we remark on inversion for theearthquake source by seismic, geodetic, and tsu-nami data (“Inversion for the Seismic Source”).

The rest of the entry is a summary of selectedtsunamigenic major-to-great earthquakes thatoccurred in the decade 2004–2013. It beginswith “Megathrust Events”, including those in“The 2004–2010 Sunda Trench Sequence”(“Sumatra-Andaman (2004)”, “Nias (2005)”,“Bengkulu (2007)”), “Peru (2007)”, “Maule(2010)”, “Tohoku (2011)”, and “Santa CruzIslands (2013)”. Then we describe the “GreatSubduction Earthquakes Doublets” involvingpaired outer-rise and megathrust earthquakes: the“Kurils (2006, 2007)” and the “Samoa (2009)”doublets. “Strike-slip Events” section follows,describing the “Sumatra (2012)” earthquakes.Large “Tsunami Earthquakes” section deals with“Java (2006)” and “Mentawai (2010)” events.Some “Special Cases” of tsunamigenic earth-quakes complete this overview (“Solomon Islands(2007)”, “Haida Gwaii (2012)”).

Tsunamis and the Seismic Source

Tsunamis are water gravity waves resulting fromseafloor deformation induced by submarine orcoastal earthquakes (tsunamigenic earthquakes).Less frequently, tsunamis are generated by sub-marine landslides, volcanic eruptions, atmo-spheric phenomena, or meteorite impacts. Themajority of tsunamigenic earthquakes occuralong subduction zones as “ordinary” interplatethrust earthquakes (rupturing on the shallow dip-ping plate boundary interface); less frequentlythey show a “tsunami earthquake” character. Sig-nificant tsunamis are sometimes generated byintraplate normal faulting earthquakes (e.g.,outer-rise events). Also oceanic strike-slip earth-quakes can be tsunamigenic in the presence ofsteep bathymetric slopes. Magnitude and featuresof earthquake-generated tsunamis depend on sev-eral factors.

Great megathrust earthquakes located forexample on the Pacific “ring of fire” commonlygenerate tsunamis that propagate to long dis-tances. To first order, the tsunami amplitudes andmean run-up recorded at long distances scale withearthquake magnitude (Okal 1988; Geist 2009);nevertheless, tsunami impact can be


heterogeneous depending on source orientationand bathymetry features. The tsunami generationprocess by complex earthquake slip distributionshas been discussed within this encyclopedia (e.g.,Geist and Oglesby 2014 and references therein).A major control factor is the amount and distribu-tion of vertical seafloor coseismic displacement,particularly in the near field, which depends notonly on the earthquake magnitude but also on thefocal depth and mechanism of the earthquake andon the slip distribution. The complex relationbetween slip distribution and tsunami features inthe near field may depend also on the relativeposition and orientation of the source and thecoastline, the water depth above the source zone,the steepness of the initial tsunami field, and theperiods of the tsunami compared to the size of thecontinental shelf and to the features and the com-plexity of the coastline where the inundationoccurs.

Inversion for the Seismic Source

The earthquake source is commonly investigatedby inversions of geophysical data in order toretrieve the kinematic parameters of the earth-quake rupture, such as the space-time slip distri-bution, the rise time for slip at a point on the fault,and the rupture expansion velocity. Generally,some assumptions are needed to limitnonuniqueness of the problem, such as a prioriinformation on the fault geometry or positivity ofthe slip vector, and regularization of the solution(e.g., Das 2011).

Inversions of tsunami data for the earthquakekinematic parameters (Satake 2009, this encyclo-pedia, and references therein) are usuallyperformed by linear least-squares method orglobal search techniques (e.g., Piatanesi andLorito 2007). In case of a tsunamigenic eventwhose causative source is thought to be partiallyor totally non-seismic (e.g., a submarine land-slide) or if fault geometry is not enoughconstrained, tsunami travel-time inversion issometimes used to determine the location andextent of the tsunami source (e.g., Layet al. 2005a; Neetu et al. 2005; Ishii et al., 2012),

or tsunami and other geophysical data are invertedto retrieve the initial sea surface displacementinstead of the kinematic earthquake source param-eters (e.g., Hossen et al. 2015 and referencestherein).

More generally, different inversion techniquesfor the seismic source can be used, depending onboth the kind and complexity of the problem thathas to be solved, such as linear least-squaresmethod (e.g., Hartzell and Heaton 1983; Satake2009; Yue and Lay 2011), Bayesian inversion(e.g., Minson et al. 2013), or global searchmethods (e.g., simulated annealing: Rothman1986; Piatanesi and Lorito 2007; Romanoet al. 2014).

The data used for inversion depend on theavailable instruments that recorded the earth-quake. For example, global seismic data are reg-ularly used for investigating all moderate to greatearthquakes; strong-motion and geodetic data(e.g., GPS, land leveling) are very useful if theseismic source is relatively close to an instrumen-tal network. Recently, differential interferogramsfrom InSAR, potentially giving a wider image ofthe inland displacement, and high-rate GPS data,complementing strong-motion seismic data atlower frequencies, are being successfully used ininversions. Tsunami data measured by tidegauges, bottom pressure sensors, and morerecently GPS buoys can be used if the earthquakegenerated a tsunami large enough to be instrumen-tally recorded (see Satake 2009; this encyclope-dia); occasionally, also satellite altimetry hasdetected the tsunami waves in the open ocean.

The accuracy of the retrieved seismic ruptureimage depends on the data used, and each observ-able has its specific resolving power for differentrupture features (e.g., Lay 2015). For example,seismic waves are quite sensitive to faulting ori-entation and the time function of the source, whiletsunami waves are primarily sensitive to total faultslip and long-wavelength slip features.Teleseismic data also provide reliable seismicmoment estimation, but, compared to tsunamidata, they may have a lower resolving power forthe shallow offshore slip; static geodetic data canconstrain the seismic moment, but they onlyresolve local slip. An increasingly used strategy


is to perform joint inversions of different kinds ofdata, in order to exploit their possibly complemen-tary resolving power and/or sensitivity to differentaspects of the seismic rupture (e.g., Romanoet al. 2012; Yue et al. 2015). Relative weights tobe assigned to the different datasets are difficult toassess; synthetic checkerboard-like slip modelsare commonly employed for assessing the resolu-tion offered by individual datasets and for cali-brating their relative weights for the jointinversion (Fig. 1). However, there are still severalchallenges to be faced: models that satisfactorilymatch one data type may fail to match others,uncertainty in model parameterization, andnoncomparable accuracy in some cases of theforward modeling techniques. Also, the widerange of fault deformation time scales that isnow being recognized and the intrinsic variationsin sensitivity to time scales of the source defor-mation need to be taken into account; for example,deformation (or even slumps) too slow for seismicwave excitation may still excite tsunami, whileless tsunamigenic deep ruptures may provokestrong ground motion; so it may be difficult toforce consistency with a single representation ofthe source. Robust assessment of the earthquakekinematic model uncertainties remains an openresearch problem.

Megathrust Events

The 2004–2010 Sunda Trench SequenceThe Sunda Trench in the Indian Ocean (Fig. 2)marks the partitioned oblique convergence of theIndo-Australian Plate under the Burma silvermicroplate along the Andaman and NicobarIslands and under the Sunda Plate along Indone-sia. The Sunda Trench forms a curved arc trendingroughly NW-SE, extending from the AndamanIslands, along Sumatra and Java, to the LesserSunda Islands. Between 13�N and 7�S, thetrench-parallel component of the relative platemotion is accommodated by the Andaman-Nicobar back-arc system and the Great Sumatranfault. From 2004 to 2010 the Sunda megathrustruptured over much of its length in a sequence oftsunamigenic earthquakes, including three great

megathrust earthquakes in 2004 (MW 9.2), 2005(MW 8.6), and 2007 (MW 8.4) and four MW

7.6–7.9 earthquakes (2006, 2007, 2009, 2010),two of which are classified as “tsunami earth-quakes.” Additionally, two great strike-slip earth-quakes (MW 8.7, 8.2) occurred within the Indo-Australian Plate seaward of the trench in 2012.A comparable spatiotemporal clustering of greatearthquakes occurred along the Alaska-Aleutiansarc from 1938 to 1965 (Lay 2015) including the1964 MW �9.2 Alaska earthquake and four greatearthquakes in 1938, 1946 (a tsunami earth-quake), 1957, and 1965.

The MW 9.2 Sumatra-Andaman earthquake isthe first giant earthquake to be observed by themodern global network of broadband seismome-ters and GPS campaign networks (Subaryaet al. 2006; Ammon et al. 2010). No prior eventhad been historically recorded along northwesternSumatra where the 2004 earthquake had thelargest slip.

The Sumatra-Andaman earthquake caused thedeadliest tsunami of modern times. Together withthe tsunami generated by the 21 July 365 CreteM8+ earthquake, it is likely one of the two dead-liest tsunamis ever. The coast of northern Sumatrasuffered the greatest losses, but all margins of theBay of Bengal including Thailand, Sri Lanka,India, and Myanmar were hit within a few hours;5–10 h later the Maldives and Somalia experi-enced damaging tsunami. Very little warningwas given even far from the event, as there wasno operational tsunami warning system in theIndian Ocean. In response to this tragic event,IOC/UNESCO received a mandate from the inter-national community to coordinate the establish-ment of regional tsunami early warning systemsglobally, which are now largely operational(http://www.ioc-tsunami.org).

The adjacent segment of the Sunda Trenchimmediately southeast of the 2004 event ruptured3 months later in the March 2005 MW 8.6 Niasearthquake. The rupture zone of the 2005 eventoverlaps the deeper part of the estimated sourcezone of a great 1861 earthquake (Natawidjajaet al. 2006). The shallow portion of the 1861rupture may have failed in the 1907 tsunami earth-quake (Kanamori et al. 2010). Due to the


http://www.ioc-tsunami.org/

05

1015

20Sl

ip(m

)

134˚

136˚

138˚

140˚

142˚

144˚

146˚

32˚

34˚

36˚

38˚

40˚

42˚

010

0km

Targ

et

Long

itud

e (°

E)

Latitude (°N)

010

0

km

GPS

ons

hore

010

0km

GPS

ons

hore

+ of

fsho

re g

eod

etic

010

0km

Tsun

ami

010

0km

Join

t

Tsun

amigen

icMajor

andGreat

Earthqua

kes(200

4–20

13):So

urce

Proc

esses

Inve

rted

from

Seismic,G

eodetic,a

ndSe

a-Le

velD

ata,

Fig.1

Checkerbo

ardtest

forindividu

alandjointinversions.In

acheckerboard

test,aspecified

pattern

offinal

faultslip

isprescribed,andcorrespo

ndingpredictio

nsaremadeforgeod

etic,seismic,

andtsun

amiwaves

generatedby

that

pattern.The

leftpanelshow

sacheckerboard

pattern

ofun

itandzero

slipov

erafaultoffshoreof

Japan.The

predictio

ns(usuallywith

someno

iseinjected)aremadeforeither

ady

namicruptureexpansionmod

elor

astatic

mod

el,atlocations

correspo

ndingto

actualdataob

servations

foreach

wavetype.T

hecompu

tedsign

alsarethen

separately

orjointly

inverted

totryto

retrieve

theinpu

tmod

el,with

theintrinsiclim

itatio

nsof

each

data

type

andrecordingconfi

guratio

nlim

iting

therecovery

ofthetargetcheckerboard

pattern.In

principle,combining

GPS

andoffsho

regeod

eticandtsun

amid

ataallowsthebestreconstructio

nof

thecomplete

checkerboard

target

mod

el;ho

wever,this

kind

oftest

does

notaccoun

tfordata

mod

elingun

certainties(From

Rom

anoetal.2

012)


characteristics and location of the Nias earthquakeslip distribution, discussed in the next section, theensuing tsunami was weak relative to the earth-quake magnitude (Geist et al. 2006), demonstrat-ing how difficult it can be to forecast tsunami

impact based only on hypocenter and magnitudeof an earthquake.

Southeast of the 2005 rupture zone, a mostlylocked section of the megathrust extends about200 km along the Mentawai Islands

Tsunamigenic Major and Great Earthquakes(2004–2013): Source Processes Inverted from Seis-mic, Geodetic, and Sea-Level Data, Fig. 2 Tectonicplates and earthquakes along the Sunda Trench (From

Lay 2015) (Reprinted from Earth and Planetary ScienceLetters, Vol. 409, Lay T., The surge of great earthquakesfrom 2004 to 2014, pp.133-146, Copyright 2015, withpermission from Elsevier)


approximately to Enggano Island, where the lastgreat earthquake occurred in 1797 (Philibosianet al. 2014). The subsequent 2006–2010 sequencestraddles this region.

A major MW 7.8 tsunami earthquake rupturedfar offshore of Java in 2006, producing a damag-ing tsunami; then in September 2007, the greatMW 8.4 Bengkulu earthquake struck in the regionof a great 1833 earthquake, which overlapped thesoutheastern portion of the 1797 event. This eventwas followed 12 h 39 min later by an MW 7.9aftershock, which ruptured the down-dip portionof the overlapping zones of the 1797 and 1833ruptures. The two 2007 events generated modesttsunamis. When a large MW 7.6 81 km deepintraslab earthquake occurred in September 2009below Padang, the city was heavily damaged bythe strong shaking, and people fled to high groundin fear of a possible tsunami. Due to the focaldepth and location of the 2009 earthquake, thetsunami was very small and only instrumentallyrecorded. The up-dip portion of the overlappingzones of the 1797 and 1833 ruptures was then

struck by a tsunami earthquake in 2010, locatedseaward of the Pagai Islands and up-dip from thetwo 2007 earthquake source zones. There is a�400 km long section of the Sunda Trenchbetween the 2005 and 2007 rupture zones, the“Mentawai gap” (e.g., McCloskey et al. 2010),with remaining potential for large-to-greatmegathrust earthquakes and tsunamis accordingto the estimated moment deficit (Koncaet al. 2008; Fig. 3).

Sumatra-Andaman (2004)The giant Sumatra-Andaman megathrust earth-quake (MW 9.2) nucleated �80 km off the north-western coast of Sumatra and propagatednorthward for �1300 km (Fig. 4), causing a dev-astating tsunami in the Indian Ocean.

Different teams (Choi et al. 2006 and refer-ences therein) conducted post-tsunami surveyson Indian Ocean coasts, collecting more than1500 run-up measurements. The highest run-upoccurred in the Aceh region (northern Sumatra),near the maximum slip featured by most of the

{

Nias2005

1833

1797

2007

2007

0–6° –4° –2° 0° 4°2°

2

4

8

10

6

Mw = 8.4

Mw = 7.9

Mw = 8.4 plus Mw = 7.9

Sumatra–Andaman2004

Latitude

Mom

ent p

er h

alf d

egre

e of

latit

ude

(1021

N m

)

Tsunamigenic Major and Great Earthquakes(2004–2013): Source Processes Inverted from Seis-mic, Geodetic, and Sea-Level Data, Fig. 3 The purpleline represents the accumulated moment deficit since thelast great rupture in the Mentawai gap (From Koncaet al. 2008) (Reprinted by permission from MacmillanPublishers Ltd: [Nature], Konca A O, Avouac J P,

Sladen A, Meltzner A J, Sieh K, Fang P, Li Z, Galetzka J,Genrich J, Chlieh M, Natawidjaja D H, Bock Y, FieldingE J, Ji C, Helmberger D V, Partial rupture of a locked patchof the Sumatra megathrust during the 2007 earthquakesequence, Nature, 456:631–35, doi:10.1038/nature07572,Copyright 2008)


finite-fault models. The tsunami penetrated morethan 2 km inland, and run-up values were as highas �50 m at some places (s). Aceh province hadthe highest death toll (160,000+ victims of a totalof �228,000). The impact was extensive in manycountries: Indonesia (average run-up �10 m),Thailand (average run-up �5 m), Sri Lanka(average run-up �5 m), and India (averagerun-up �4 m), and significant waves reachedacross the Indian ocean striking, for example, thecoasts of Somalia (run-up�10 m in some places),Bangladesh, Malaysia, Maldives, and SeychellesIslands.

The 2004 tsunami propagated well beyond theIndian Ocean (Fig. 5), and for the first time atsunami was measured by worldwide tide gauges.Numerical simulations revealed how bathymetricwaveguides trapped and directed tsunami energy;for example, the Southwest Indian andMid-Atlantic ridges guided the tsunami towardthe Atlantic Ocean, whereas Southeast Indianand Pacific-Antarctic ridges guided it toward thePacific Ocean (Titov et al. 2005).

The tsunami waves at tide gauges located in thefar field (e.g., Callao in Peru, �19,000 km from

the epicenter, or Halifax in Nova Scotia,�24,000 km) were sometimes higher than thosemeasured in the near field (e.g., Cocos Island), dueto amplification effect during propagation on thecontinental shelf, enhanced by local resonances(e.g., Thomson et al. 2007). Commonly, in thefar-field tsunami signals, the maximum waveheights arrive several hours after the leadingwave due to multiple reflections and coastal reso-nances (see also discussion of the 2006 MW 8.3Kuril earthquake); this is a very important factorwhen evaluating an “all clear” announcementafter issuing a tsunami alert message.

Many studies were performed as inversiontechniques were progressively updated to handlethe long rupture duration (e.g., Ammonet al. 2005; Gahalaut et al. 2006; Banerjeeet al. 2007; Chlieh et al. 2007; Piatanesi andLorito 2007; Rhie et al. 2007; and many others).These studies used different kinds of geophysicaldata (teleseismic, tsunami, GPS, satellite altime-try) and different inversion techniques, whicheach offer variable resolution of the source pro-cess (see discussions in Menke et al. 2006;Shearer and B€urgmann 2010; Poisson

70˚ 75˚ 80˚ 85˚ 90˚ 95˚ 100˚ 105˚

0˚

5˚

10˚

15˚

20˚

Longitude

Latit

ude

Sumatra

Thailand

Andaman Is.

India

Sri Lanka

Indian OceanMaldives

Myanmar

Malaysia

0

5

10

15

20

25

30

Runu

p

m

2004 Sumatra rupture

Tsunamigenic Major and Great Earthquakes (2004–2013): Source Processes Inverted from Seismic, Geodetic,and Sea-Level Data, Fig. 4 Extent of the 2004 Sumatra-Andaman rupture and run-up distribution in the Indian Ocean


et al. 2011; Lay 2015). Notably, sea-level anoma-lies up to �80 cm were measured in the openIndian ocean by multiple-satellite altimetry passesof Jason-1 and TOPEX/Poseidon satellites andwere used for the first time in source inversions(Fujii and Satake 2007; Pietrzak et al. 2007;Hoechner et al. 2008; Lorito et al. 2010).

The source models are still not fully in agree-ment about the total seismic moment (estimates ofMW range from 9.2 to 9.3) or the slip location andamount (estimates of maximum slip range from11 to 35 m, although this is usually a poorlyresolved value for slip inversions). While themodels also include a wide range of rupture veloc-ities (from �1 to �4 km/s), several models indi-cate that the rupture progressively slowed downnorthward, and slow slip in the northernmost seg-ment may have occurred, as suggested by geo-detic observations (Bilham 2005; Layet al. 2005b). Guided by different analyses, othersargue against this possibility (Ishii et al. 2007;Shearer and B€urgmann 2010). The very largelocal run-up values remain underpredicted by

most of the published slip distributions.Coseismic activation of a splay fault on the west-ern edge of Banda Aceh has been proposed as apossible explanation (Plafker et al. 2006), andseismic evidence for disrupted oceanic crust inthe epicentral region has been reported (Singhet al. 2008).

Nias (2005)Three months later, the great megathrust Niasearthquake (28 March 2005, MW 8.6) occurredbetween the Nias and Simeulue Islands,�200 km southeast from the great 2004 earth-quake epicenter. Several studies suggest that thisevent was triggered by the stress increase resultingfrom the adjacent 2004 Sumatra earthquake(Nalbant et al. 2005), with a possible poroelasticcomponent (Hughes et al. 2010).

The 2005 tsunami mainly struck the coasts ofSimeulue Island (run-up up to �4 m; Borreroet al. 2012), Nias, Banyak, and Batu Islandsabove the rupture area. Tsunami waves up to�1.2 m (peak to trough) were recorded by the

Tsunamigenic Major and Great Earthquakes(2004–2013): Source Processes Inverted from Seis-mic, Geodetic, and Sea-Level Data, Fig. 5 Globalreach of the 2004 tsunami indicated by peak wave heightpredictions (Titov et al. 2005 (Reprinted with permissionfrom AAAS). (Readers may view, browse, and/or down-load material for temporary copying purposes only,

provided these uses are for noncommercial personal pur-poses. Except as provided by law, this material may not befurther reproduced, distributed, transmitted, modified,adapted, performed, displayed, published, or sold inwhole or in part, without prior written permission fromthe publisher)


Sibolga tide gauge, in the near field of the source,whereas only centimetric waves were measured inthe far field by several tide gauges located alongthe coasts of Oman, Sri Lanka, Mauritius, Mal-dives, and Seychelles Islands.

Source models from teleseismic, GPS, andcoral displacement inversions (e.g., Briggset al. 2006; Hsu et al. 2006, 2011; Koncaet al. 2007) inferred two patches of slip (up to�12 m) mainly located under Nias and southernSimeulue Islands (Fig. 6). In addition, analysis ofpostseismic deformation highlighted significantaseismic afterslip, equivalent to an MW 8.2event, during the 11 months after the mainshock(Hsu et al. 2006), both up-dip and down-dip of themain coseismic rupture area (Fig. 6).

Despite the great magnitude, the tsunami wasmoderate, due to the depth of the slip distributionand the vertical displacement field lying almostentirely under the islands or relatively shallowwater (e.g., Geist et al. 2006). This is noteworthycompared to the tsunamigenic excitation by earth-quakes with smaller magnitude (e.g., tsunamiearthquakes or the shallow 2007 Solomon Islandrupture). Maximum run-up was observed alongthe southeastern tip of Simeulue Island locatedfacing the northern patch of the inferred slipdistribution.

Bengkulu (2007)A great MW 8.4 interplate earthquake occurred on12 September 2007, with an epicenter located�130 km offshore from the city of Bengkulu. Itwas followed 12 hr 39 min later by an MW 7.9aftershock to the northwest. Source models of theMW 8.4 earthquake obtained by inverting, sepa-rately or jointly, geophysical datasets includingteleseismic, GPS, coral uplifts, InSAR, and tsu-nami data (e.g., Lorito et al. 2008; Koncaet al. 2008; Gusman et al. 2010) agree to firstorder on northwestward rupture expansion fromthe epicenter toward South Pagai Island, withmaximum slip in the range of �7–12 m close toa previously strongly coupled region of theinterface.

The tsunami following the 2007 earthquakewas moderate, likely because most of the ruptureoccurred relatively deep along the megathrust.

The run-up measurements collected along thecoast of the Bengkulu province are <4 m(Borrero et al. 2009), and the closest tide gauge(Padang) measured a tsunami wave of�1m (peakto trough).

Peru (2007)On 15 August 2007, an MW 8.0 earthquakeoccurred offshore of central Peru �60 km westfrom the city of Pisco. The hypocenter waslocated at a depth of �40 km on the subductioninterface between the Nazca and South AmericanPlates. The earthquake generated a tsunami thatstruck the Peruvian coast from Lima (�12�S) to50 km south of the Paracas Peninsula (14.5�S).The run-up measurements range from 1 to 10 m(Fritz et al. 2008). Three victims were reported inthe area of maximum inundation at Lagunilla(Ioualalen et al. 2013). The moderate size of therupture and its depth limited the regional tsunamiexcitation for this event. The 2007 Peru earth-quake generated centimetric-to-decimetric tsu-nami amplitudes measured by tide gaugesinstalled in French Polynesia, California, Mexico,Chile, New Zealand, Hawaii, and Japan.

Both teleseismic (Lay et al. 2010b) and InSAR(Sladen et al. 2010) data have been used to invertfor finite-fault rupture models of the 2007 Peruearthquake. These models (Fig. 7) both have twopatches of slip, one near the epicenter and theother located in the southwest off the ParacasPeninsula. The moment rate function (time historyof moment release) features a first pulse within afew seconds after the rupture began and a secondlarger pulse �60 s later (Lay et al. 2010b). Thisprolonged source process could conceivably beassociated with a slow rupture velocity(<1.5 km/s), which would be unusual for a rup-ture confined in the deeper part of the megathrust.There is no clear evidence from geodetic data forslow or aseismic slip on the megathrust in thisregion. Thus, it appears likely that the �60 s hia-tus in the seismic radiation is due to discretefailure of an initial smaller sub-event (MW 7.8)that dynamically triggered the second largersub-event (MW 8.0) (Lay et al. 2010b). This typeof compound failure also appears to have occurredduring the 2001 Peru MW 8.4 earthquake (Lay


Fig. 6 (continued)


et al. 2010b; Pritchard et al. 2007), which initiatedwith an MW 7.5 rupture that appears to havedynamically triggered a larger rupture about120 km away that grew into an MW 8.4 event.

The 2007 Peru event, together with the adja-cent 1940 (MW 8.0) and 1974 (MW 8.1) earth-quakes, forms a patchwork collectively spanningthe estimated source zone of the 1687 M8.5+earthquake (Biggs et al. 2009). Similarly, on theColombia-Ecuador margin, the rupture zonecorresponding to the 1906 earthquake (MS 8.8,Kanamori and McNally 1982) re-ruptured inthree smaller separate events (1942, MS 7.9;1958, MS 7.7; 1979, MW 8.2).

While the entire stretch of Peruvian coast fromnorth of Lagunilla to Pisco runs broadside of therupture zone, the largest run-ups were measuredsouth of the Paracas Peninsula (Fig. 8). Indeed,the southern side of the peninsula was most

affected by the tsunami, with run-ups from 5 to7 m measured in Lagunilla and the maximumtsunami run-up of 10 m measured in PlayaYumaque. Lagunilla experienced the maximuminundation distance, as the tsunami penetrated2 km inland on flat terrain. This pattern is likelyan effect of the peninsula extending obliquely tothe NNWabove the maximum slip area, exposingits southernmost side to the leading wave, muchmore than the northernmost stretch of coast. Onthe north side of the peninsula, the measuredrun-ups were only 3 m on average. Additionally,the islands located west and north of the peninsulamight have attenuated the tsunami impact north ofParacas Peninsula (Fritz et al. 2008; Ioualalenet al. 2013). Run-ups up to �5 m were also mea-sured south of Lagunilla, within the Bahia de laIndependencia, which is located obliquely to thetsunami source. These southern surges may have

��

Tsunamigenic Major and Great Earthquakes(2004–2013): Source Processes Inverted from Seis-mic, Geodetic, and Sea-Level Data, Fig. 6 Coseismic(top panel, from Hsu et al. 2011) and postseismic (bottompanel, from Hsu et al. 2006) (Reprinted with permissionfrom AAAS) (Readers may view, browse, and/or down-load material for temporary copying purposes only,

provided these uses are for noncommercial personal pur-poses. Except as provided by law, this material may not befurther reproduced, distributed, transmitted, modified,adapted, performed, displayed, published, or sold inwhole or in part, without prior written permission fromthe publisher). panel, from Hsu et al. 2006) slip of the Nias2005 earthquake

400

400

400

400

600

800

78°W 77°W 76°W 75°W

15°S

14°S

13°S

76 km

32 km

8 km

2 km

6 º

20 º

30 º

78˚W 77˚W 76˚W 75˚W

15˚S

14˚S

13˚S

0 50

km

a. InSAR

0

200

400

600

800

Slip [cm]

Tsunamigenic Major and Great Earthquakes(2004–2013): Source Processes Inverted from Seis-mic, Geodetic, and Sea-Level Data, Fig. 7 Slip distri-butions estimated for the 2007 Peru earthquake (left panel

from Sladen et al. 2010; right panel from Lay et al. 2010b)peak slip for the first event (red tones) is 3.83m; for thesecond event (gray tones) is 9.14


been generated by edge waves, which are wavestrapped by refraction to propagate along the coast.

Maule (2010)On 27 February 2010, a great megathrust earth-quake (MW 8.8) occurred off the Maule region,central Chile, on the subduction zone interfacebetween the converging Nazca and South Ameri-can Plates. An ensuing tsunami struck the Chileancoast, extending �300 km north and �300 kmsouth of the epicenter, taking almost 200 lives.More than 400 run-up measurements were col-lected with values ranging from 3 m up to�30 m (Fritz et al. 2011b), distributed �600 kmalong the coast between 33�S and 39�S. ManyChilean tide gauges recorded wave amplitudesbetween 1 and 2 m or more.

Significant tsunami waves were recordedaround the Pacific Ocean, which is typical formegathrust earthquakes of this size. For example,the wave amplitude reached �1.8 m at Hiva OaIsland of the Marquesas archipelago, French

Polynesia, and �1 m at several places in NewZealand, at Hokkaido in Japan, at the KurilIslands, Russia, and at several locations in bothCalifornia and Hawaii in the United States. Theoverall magnitude of the tsunami might have beenreduced by the fact that a significant fraction of thecoseismic displacement occurred inland (mostlysubsidence) and not on the seafloor. Locally, sub-sidence enhances the tsunami impact.

The seismic and tsunami source of the Mauleearthquake has been investigated by a number ofdifferent groups (e.g., Delouis et al. 2010; Layet al. 2010a; Lorito et al. 2011; Morenoet al. 2012; Koper et al. 2012; Lin et al. 2013),often using joint inversion of different geophysi-cal datasets, including static data from GPS,InSAR, and land leveling, high-rate GPS, strong-motion, teleseismic body and surface wave data,and deep water and tide-gauge tsunami data.

To first order, all of the earthquake sourcemodels for the 2010 event feature a bilateral rup-ture extending north and south of the epicenter,

−78˚ −77˚ −76˚

−14˚

−13˚

−12˚

Longitude

Latit

ude

Pisco

Paracas

NazcaPlate

SouthAmerica

0

2

4

6

8

10

Runu

p

m

Tsunamigenic Major and Great Earthquakes(2004–2013): Source Processes Inverted from Seis-mic, Geodetic, and Sea-Level Data, Fig. 8 Run-updistribution (left panel) and simulated tsunami maximumwave heights for the 2007 Pisco earthquake (right panelfrom Ioualalen et al. 2013) (Reprinted with kind

permission from Springer Science+Business Media: Pureand Applied Geophysics, Tsunami Modeling to ValidateSlip Models of the 2007 Mw8.0 Pisco Earthquake, CentralPeru, vol.170, 2013, pp.433-451, Ioualalen M,Perfettini H, Condo S Y, Jimenez C, Tavera H, Figure 3)


with nonuniform coseismic slip along strike from�38.5 to �34�S, a main patch of slip locatedoffshore north of the epicenter with �15–20 mof slip and a secondary patch to the south withlower slip (<�10 m). Slip up-dip of the hypocen-ter, in the region between the two large-slippatches, is small in most models. Most inversionsdo not have slip extending to the trench, although

onshore geodetic data cannot constrain this well,and inversions of seismic data give variable place-ment of slip along depth.

Yue et al. (2014) performed a joint inversion oftsunami, geodetic, and seismic data, concludingthat the rupture likely propagated up-dip to thetrench along the two large-slip patches north andsouth of the epicenter (Fig. 9). Large outer-rise

1928

1928 1928

1928

1960 1960

19601960

1835

1835

1835

1835

LongitudeLongitude

Latit

ude

Latit

ude

a b

c d

Tsunamigenic Major and Great Earthquakes(2004–2013): Source Processes Inverted from Seis-mic, Geodetic, and Sea-Level Data,Fig. 9 Comparison of different slip distributions for the

2010 Maule earthquake (Modified after Yue et al. 2014);dashed white arrow indicates the estimated 1835 earth-quake rupture area, namely, the Darwin gap (Loritoet al. 2011)


normal faulting aftershock seismicity is concen-trated in the Nazca Plate seaward of these slippatches, as commonly observed seaward oflarge-slip regions in other shallow megathrustruptures (2006 Java and 2010 Mentawai tsunamiearthquakes; 2011 Tohoku and 2012 Haida Gwaiiearthquakes). Resolution of the shallowmegathrust slip was provided by the improvedmodeling of deep water tsunami signals, withpropagation phase corrections accounting fortwo types of inaccuracy: neglect of seafloor elas-ticity and of density variations with depth in theocean. These factors are particularly important forpropagation to distant stations, as phase errorsprogressively accumulate, resulting in severalminute too-early simulated arrival times withrespect to observations. Without correction,these travel-time errors tend to shift inferred slipcloser to the coast to delay the seaward arrivals.Other factors, such as imperfect bathymetrymodels, neglect of tsunami dispersion, andgeopotential variations induced by the tsunamiitself (e.g., Tsai et al. 2013; Watada et al. 2014)can also limit the precise determination of tsunamisource area.

The 2010 Maule earthquake occurred in aregion that had previously been identified as aseismic gap (Madariaga et al. 2010) betweenlarge ruptures in 1960 to the south and 1928 tothe north. This portion of the plate boundary hadlast ruptured in an 1835 (M �8.5) earthquake, anevent that was experienced by Charles Darwinduring his voyage on the Beagle. The “Darwingap” (Lorito et al. 2011) was shown by GPSobservations to be a region of slip deficit relativeto plate convergence over the prior decade andinferred to have strong plate coupling (Morenoet al. 2010). The 2010 event nucleated within thegap, but the large-slip patches are not locatedwithin the presumed 1835 rupture area, with thenorthern patch overlapping the estimated rupturezone of the 1928 MW 8.0 earthquake (Becket al. 1998). Yue et al. (2014) note that manylarge aftershocks are located on the megathrustup-dip of the hypocenter, so there may be a mixof stable and unstable friction in the central regionof the gap.

As reported by Fritz et al. (2011b), the distri-bution of run-up measurements shows a broadrange of values (3–30 m, Fig. 10), which arespatially well correlated with the along-strike-slip distribution. However, the maximum value(�30 m) in Constitución, located on the coastnear the main slip patch, involved a local effectof the wave surging up a coastal bluff. Therun-ups decay both northward and southward butstill show relatively high values (5–15 m) alongthe large-slip zone. On the other hand, the stretchof coast near the epicenter is characterized byrelatively low run-up values, corresponding tothe central region of low slip found in mostpublished rupture models.

The inundation pattern did not involve largeinundation distances, mainly due to the generalsteepness of the Chilean coast. The largest inun-dation distances were observed in Constitución(500 m) along the side of the city adjacent to theMaule River. Significant run-up values (�20 m)were observed along the coast of Mocha Island,located 30 km offshore, directly exposed to wavestraveling from the southern large-slip patch. Sim-ilarly, large run-up values are recorded in Tirúa, acity located on the Chilean coast east of MochaIsland.

In the near field of the coseismic displacement,destructive late edge wave’s arrival occurred some3 h after the earthquake at Talcahuano harbor. Thisphenomenon is due to refraction and trapping ofthe energy along the continental shelf and slope,and both linear and nonlinear resonances of edgewave modes may amplify the tsunami. Edge waveexcitation is a complex phenomenon related tofine slip distribution details (Geist 2009, 2012)and to coastal complexity (Yamazaki and Cheung2011); thus, it is difficult to evaluate more pre-cisely. Edge wave resonance likely also causedarrivals from multiple directions observed northof the epicenter as modeled by numerical simula-tions by Yamazaki and Cheung (2011). Signifi-cant edge wave excitation was observed alsofollowing the 2006 Kurils and 2009 Samoa earth-quakes. One lesson from these events is that evenif accurate modeling of tsunami edge waves maybe difficult with conventional models (e.g., Liuet al. 1998), using too simplified earthquake slip


−80˚ −78˚ −76˚ −74˚ −72˚ −70˚

−38˚

−36˚

−34˚

Longitude

Latit

ude

Nazca Plate

SouthAmerica

Plate

Constitucion

Juan Fernandez

Tirua

0

5

10

15

20

25

30

Runu

p

m

–74° –72°

–38°

–36°

–34°

Mw 8.8

Santiago

Valparaíso

Constitución

ConcepciónIsla Santa Maria

Isla Mocha

–1 0 1 2 3 4 5Surface Deformation

(m)

–120° –100° –80°

–40°

20°

0°

Easter Isl.

Rob. Crusoe Isl.

LimaLagunilla

Isla Mocha

Valparaiso

Talcahuano

0 0.01 0.05 0.1 0.2 0.5 1 1.5 2 3 5 10Wave Height

(m)

Tsunamigenic Major and Great Earthquakes(2004–2013): Source Processes Inverted from Seis-mic, Geodetic, and Sea-Level Data, Fig. 10 Run-updistribution (top panel), maximum tsunami wave amplitudesover the South Pacific Ocean (bottom-left panel), and seafloordeformation field (bottom-right panel) for the 2010 Maule

earthquake (bottom-left and bottom-right panels from Fritzet al. 2011b) (Reprinted with kind permission from SpringerScience+Business Media: Pure Appl Geophys, Field Surveyof the 27 February 2010 Chile Tsunami, 168, 2011b, 1989-2010, Fritz H M, Petroff C M, Catalán, Cienfuegos R,Winckler P, Kalligeris N, Weiss R, Figure 10)


and tsunami modeling may lead to severe under-estimation for tsunami hazard assessments.

Finally, very high run-ups were reported alongthe coasts of islands comprising the JuánFernández Archipelago, located more than650 km off Valparaiso, in the perpendicular direc-tion from the middle of the main slip patch. Thisdirection is within the main lobe of the tsunamiradiation pattern toward the Pacific Ocean,according to the simulation of Fritzet al. (2011b). Robinson Crusoe Island, impacted�50 min after the earthquake origin time, firstexperienced a relatively slow-moving floodfollowed by a violent wave that destroyed mostof the houses and local facilities. The maximumrun-up measured was �18 m, and the maximuminundation distance was �300 m.

Tohoku (2011)On 11 March 2011, a great earthquake (MW 9.0)occurred offshore of the Tohoku region (Japan) onthe subduction interface between the Okhotsk andunderthrusting Pacific Plates. This is the largestrecorded earthquake to strike Japan. It generated ahuge tsunami that devastated the eastern coast ofHonshu, killing at least 15,890 people and causingwidespread destruction. It also precipitated a melt-down of the Fukushima Daiichi reactor, which,being classified as Level 7 on the IAEA eventscale, is one of the two worst nuclear plantdisasters ever.

Researchers from more than 60 universitiesand institutes participated in the post-tsunami sur-vey that began only 2 days after the earthquake.They collected more than 5300 run-ups (Fig. 11)and flow depth observations along the easterncoast of Japan (Mori et al. 2011). A gap in mea-surements exists around the Fukushima area,which could not be surveyed due to the conse-quences of the nuclear accident. The run-upvalues range up to �40 m along the coasts ofIwate prefecture (northwest from the epicenter).Sustained run-ups from 10 to 40 m occurred alonga 300+ km stretch of coast; run-ups were generally�2–3 m along the southern Honshu (Chiba) andHokkaido coasts.

Comparable run-up heights along Iwate wereobserved in the 1896 Meiji and 1933 Sanriku-oki

earthquakes (Fig. 11). Recorded earthquakes inthe past few centuries off of central Honshu wereless than magnitude �8.2, much smaller than the2011 event, and that had guided earthquake andtsunami hazard assessments, with the potential fora much bigger rupture with very large tsunamialong Sendai not receiving detailed consideration.Some recent work on the 869 earthquake hadindicated run-up and inundation of the SendaiPlain that was to be matched by the 2011 event,but this had not yet penetrated into tsunami miti-gation measures. Combined with other too sim-plistic assumptions, this reliance on recordedtsunami heights led, for example, to the deploy-ment of 5.7 m tsunami walls near Fukushima(based on modern peak recorded tsunami alongthat stretch of coast that was caused by the distant1960 Chile earthquake), which proved futile whenthe �15 m wave generated by the 2011 eventarrived (see discussion in Synolakis and Kanoğlu2015).

The extensive Japanese network of sea-levelsensors recorded unprecedented tsunami observa-tions. Many Japanese tide gauges measured tsu-nami amplitudes >10 m (up to �18 m atOnagawa; Satake et al. 2013a); however, there isuncertainty regarding maximum values, as severaltide-gauge facilities in the Iwate, Miyagi, andFukushima prefectures were washed away ordamaged during the tsunami inundation. Addi-tionally, the tsunami was measured by severaloffshore GPS buoys, peaking up to 6 m alongIwate prefecture. Oceanward, deep water tsunamisensors recorded the first tsunami amplitudes>1m.Wave amplitude exceeded 5m at the sensorlocated �100 km from the coast north of theepicenter and �2 m at a DART buoy �450 kmeast of the Japan Trench. The tsunami propagatedover the Pacific Ocean with wave amplitudesrecorded by tide gauges up to �1 m in Papua(Indonesia) and New Zealand, up to �2 m inHawaii, up to �1.5 m at Nuku Hiva in FrenchPolynesia, up to �2.3 m in the Kuril Islands, andup to �2.5 m in Chile and California(Crescent City).

The very dense Japanese geophysical networksalso recorded an unprecedented amount of high-quality seismic and geodetic ground motion data


in the near field of the source, and there is a superbglobal dataset of seismic observations, enablingmany studies of this earthquake. Most of the morethan 50 retrieved slip and rupture history models(e.g., Yokota et al. 2011; Lay et al. 2011b; Yue andLay 2013; Satake et al. 2013a; Romanoet al. 2014; Bletery et al. 2014, and many others,Fig. 12) share two first-order features: an overallspatially concentrated rupture, with a large-slipzone extending <200 km along strike at shallowdepths with very large slip values (>50 m innumerous models) and extending <400 kmalong strike at greater depth with smaller slip.The entire 200 km wide expanse along dip of themegathrust appears to have ruptured, includingregions of moderate size ruptures in the down-

dip portion that failed in the last century. Modelsdiffer in whether and where significant slipextended to the trench.

Interpretation of the concentrated nature of thelarge-slip zone has been explored using seismicvelocity anomalies around the fault plane. The slipin the central portion of the megathrust, wheregreat earthquake ruptures typically occur(“Domain B” of Lay et al. 2012), correspondswell with a near-source volume characterized byhigh seismic wave velocity (Zhao et al. 2011) andhigh pre-event coupling (Loveless and Meade2010). This region is confined between twolow-velocity, low-coupling zones along strike;the latter may have acted as low strain or slip

Longitude

Latitud

e

(1896)

Tsunamigenic Major and Great Earthquakes (2004–2013): Source Processes Inverted from Seismic, Geodetic,and Sea-Level Data, Fig. 11 Run-up distribution of the 2011 Tohoku earthquake (From Mori et al. 2011)


velocity-strengthening barriers (Ye et al. 2012;Romano et al. 2014).

Extension of the 2011 rupture to the shallowestpart of the megathrust, with peak slip (60–80 m)twice as large as that on the central megathrust,runs contrary to the common perception that theshallow subduction interface mainly experiencesaseismic slip up-dip of the “seismic front,” a tran-sition from seismic to aseismic behavior at a depthof �10 km on the megathrust. However, theresults of the JFAST drilling project whichretrieved core from the Tohoku megathrust nearthe trench (Fulton et al. 2013 and referencestherein) suggest that, under conditions of highslip rate (e.g., Faulkner et al. 2011), the abundantsmectite-rich clays on the megathrust may haveexhibited slip velocity-weakening frictionalbehavior, with the huge slip possibly being alsoaugmented by dynamic normal stress variationdue to the reflection of seismic waves off theseafloor (Oglesby et al. 1998; Kozdon andDunham 2013). Having large slip extend veryclose to the trench may have contributed to thehuge intraplate extensional faulting sequence sea-ward of the trench within the Pacific Plate. Overone thousand normal faulting events occurredover a widespread area extending hundreds ofkilometers from the trench (e.g., Lay et al. 2011a).

Most seismic and geodetic rupture models donot show significant shallow slip north of�39�N;however, the offshore tsunami buoy measure-ments off Iwate are compatible with little slipfurther north (e.g., Romano et al. 2012; Bleteryet al. 2014). Accordingly, several models usingtsunami data alone (Satake et al. 2013a) or incombination with static GPS offsets (Romanoet al. 2014) in slip model inversions have inferredlarger, delayed, or slow slip in the shallowsediment-rich megathrust (Satake et al. 2013a;Kubo and Kakehi 2013) extending to at least39.5�N. This would overlap the estimated sourcearea of the 1896 Meiji tsunami earthquake(Kanamori 1972), reinforcing the notion that the2011 rupture was a combination of regularmegathrust rupture and shallow tsunami earth-quake. The northern extent of slip is importantfor evaluating whether the 1896 zone could pro-duce another large tsunami earthquake. Some

high-frequency-depleted seismic radiation fromthis region has been suggested (Yagi et al. 2012;Maercklin et al. 2012), although it is not evident inall back-projection studies, and any late momentrelease within the seismogenic passband shouldhave been readily apparent in the rupture centroidtime and long-period slip inversions.

Some authors (e.g., Tappin et al. 2014) argueinstead that reconciling seismic and tsunamiinversions with the large run-ups observed in theSanriku region suggests the occurrence of a sub-marine landslide, whose scar would extend sev-eral tens of km along the strike direction, possiblycorresponding to a feature mapped on the land-ward slope instead of near the trench axis (see alsoSatake and Fujii 2014). Quantification of the�40 m run-up along the Sanriku Coast remainslimited due to the lack of very high-resolutiontopo-bathymetric models, so the north-easternextent of the 2011 rupture remains an open issue.

The large trench-reaching slip enhanced thehuge tsunamigenic nature of this earthquake,with devastating effects and huge run-ups alongmost of the eastern coast of Honshu, broadside ofthe tsunami source. It is nevertheless interesting todecompose the slip distribution (Satakeet al. 2013a) into its deep and shallow slip com-ponents (Fig. 13), in order to highlight a somehowcounterintuitive feature, contradicting the com-monly used simplification that the shallower theearthquake slip, the most damaging the tsunamiwill be. The relationship between source featuresand inundation particularly in the near field ismore complex (e.g., Geist 2009). The Tohokutsunami penetrated up to �5 km inland along theSendai and Ishinomaki plains. Simulating sepa-rately the tsunamis generated by the shallow andthe deep slip shows that the latter generated aninitial water displacement with larger wavelength(and thus longer tsunami periods) which can aloneaccount for the observed inundation distanceabove the plains; on the other hand, the shallowslip generates higher and steeper tsunami waveswith more dissipative inland character achievingshorter inundation distances. This dualtsunamigenic slip behavior can be also observedin the tsunami waveforms recorded by the oceanbottom pressure sensors near the epicenter and at


9 March ‘11M7.4

1896 Meiji Eq.

JFASTSite C0019HF

Fukushima

Ibaraki

42°

40°

38°

36°

34°

136˚ 140˚

36˚

40˚

144˚

Syn

Obs

5 m hr-GPS

25 m OB-GPS

−10000 −5000 0

m10 20 30 40 50

m

138˚ 140˚ 142˚ 144˚ 146˚

0 100

km

05

101520253035404550

Slip(m)

Longitude

0 4 8 12 16 20 24 28 32 36 40Slip (m)

36˚N

38˚N

40˚N

100 200

km

140˚E 142˚E 144˚E

0

Miyagi MMiyagi M

TM−1TM−1TM−2TM−2Iwate MIwate M

Miyagi NMiyagi N

Iwate NIwate N

FukushimaFukushima

Iwate SIwate S

Miyagi M

TM−1TM−2Iwate M

Miyagi N

Iwate N

Fukushima

Iwate S

Fig. 12 (continued)


some GPS buoys located 20 km off the Iwate andMiyagi coasts, where both the shorter- and thelonger-period components are clearly visible.Conversely, the largest run-ups have beenobserved in the near-oblique regime of the tsu-nami source, where the morphology of theSanriku Coast contributed to the amplification ofthe tsunami waves.

Santa Cruz Islands (2013)On 6 February 2013, a large underthrusting earth-quake (MW 8.0) occurred close to the Santa CruzIslands (part of the Solomon Islands) on the plateboundary interface between the Australian andPacific Plates. This is a particularly complexmegathrust section, where plate geometry is notwell known (Fig. 14). The hypocenter was located76 kmwest of Lata, the main city of Nendo Island,at the northern end of the Vanuatu subductionzone where a transpressive transform faultextends westward to the Solomon Islands subduc-tion zone. The 90� change in plate boundary strikeproduces a concentration of intraplate activitywith normal faulting in the subducting AustralianPlate and compressional and strike-slip faulting inthe Pacific Plate. There had not been a prior major(M>7) interplate earthquake in the vicinity of the2013 event dating back to 1900.

This earthquake, the largest event to occurworldwide in 2013, generated a tsunami thatstruck the coasts of Nendo Island with maximumobserved tsunami run-up>11 m (Fritz et al. 2014)along the southwestern side of the island. Thetsunami was recorded by tide gauges located inthe Santa Cruz, Solomon, and Fiji Islands, withthe maximum wave height (peak to trough) beingrecorded in Lata (>�2 m); several NOAA DARTbuoys located in the southwestern Pacific Oceanspanning a wide range of azimuths recorded thistsunami too.

There are two published studies of the earth-quake source, using teleseismic body wave

inversion optimized by iterative forward model-ing of deep-sea tsunami recordings (Layet al. 2013a) and teleseismic body and surfacewaves (Hayes et al. 2014) and one of the tsunamisource, from the inversion of DART and tide-gauge tsunami signals (Romano et al. 2015).

These models differ (Fig. 15) regarding slipamplitude, direction (rake), and distribution, as aresult of using distinct datasets and fault modelgeometries. This highlights the nonuniqueness ofrupture models and their dependence on datachoice and modeling assumptions.

The models of Hayes et al. (2014) and Romanoet al. (2015) are essentially end members, as thefirst one is derived from ground shaking recordedat teleseismic distances, while the second one isderived by tsunami signals alone, although itimposes rupture velocity inferred from seismicmodeling by Lay et al. (2013a). Intrinsically,these data types provide different sensitivity tothe source process, and the models may not nec-essarily reproduce data that are not used in theinversion. The model of Hayes et al. (2014), char-acterized by moderate (<3.5 m) and quite deepslip, significantly underpredicts the observed tsu-nami amplitudes. Based on the analysis of after-shock locations and mechanisms, Hayeset al. (2014) speculated that there may have beensome aseismic slip on the southeastern margin oftheir slip zone over hours or days; if this occurredfast enough (<�15 min), it could have contrib-uted to the tsunami generation modeled byRomano et al. (2015). The tsunami source modelof Romano et al. (2015) will almost certainly notprovide a good fit to the seismic data even thoughit assumed a reasonable rupture expansion veloc-ity, as it places slip away from the hypocentralarea where the models of seismic data place strongslip. However, it is possible to account for thetsunami generation and seismic waves with aself-consistent model, as shown by Layet al. (2013a).

��

Tsunamigenic Major and Great Earthquakes(2004–2013): Source Processes Inverted from Seis-mic, Geodetic, and Sea-Level Data, Fig. 12 Slip dis-tribution models of the 2011 Tohoku earthquake (top-left

panel, from Yue and Lay 2013; top-right panel, fromSatake et al. 2013a; bottom panel, from Romanoet al. 2014)


The model of Lay et al. (2013a) should, inprinciple, offer the most complete image of theearthquake and its effects in terms of long distanceshaking and tsunami generation, being deter-mined using both kinds of data. Conversely, ifaseismic tsunami generation occurred, it cannotbe accounted for by forward modeling of thetsunami using the results of teleseismic inversion.A planar rectangular fault model is used, shifted

shallower by �10 km from that used by Hayeset al. (2014). Romano et al. (2015) use a 3D faultsurface that may reduce uncertainty in tsunamimodeling. However, the choice of model structureis very uncertain. Lay et al. (2013a) have reportedrupture extending to the seafloor in the deepestpart of the trench; the other models extend�10 km further seaward, but without rapid

a b

Tsunamigenic Major and Great Earthquakes(2004–2013): Source Processes Inverted from Seis-mic, Geodetic, and Sea-Level Data, Fig. 13 Inundation

computed from deep (light gray) and shallow (dark gray)slip compared to inundation observed (black line) after the2011 Tohoku earthquake (From Satake et al. 2013a)


shallowing of the plate boundary. It is not clearwhat choices are best.

The low rupture velocity used in the models ofLay et al. (2013a), Hayes et al. (2014), andRomano et al. (2015) and the large slip (>10 m)at relatively shallow depth, extending up to thetrench in some cases, are features that are typicalof tsunami earthquakes. However, Layet al. (2013a) find that the moment-scaled radiatedenergy is higher than typical of end-member tsu-nami earthquakes, suggesting that the event is acomposite of typical interplate thrusting accom-panied by some tsunami earthquake-likeshallow slip.

The maximum observed run-up (�11 m) onthe southwestern side of Nendo is�1 km from thetown of Malemgeulu and locates northeast oflarge coseismic slip (>10 m) regions in themodels of Lay et al. (2013a) and Romanoet al. (2015). This is likely a localized result ofthe direct arrival of the main tsunami waves gen-erated by this patch with the coast lying broadside

to the coseismic slip, particularly if the slip isdominantly in the thrust direction. Two additionalrelatively high run-ups were observed on thenorthwestern side of Nendo, 3 m along a stretchof coast facing the Pacific Ocean and 1.5 m nearthe city of Lata. The former could be due to therefraction of the tsunami waves traveling aroundthe island from the southern slip patch, whereasthe latter could be due to resonance effects withinGraciosa Bay. An alternative hypothesis is that themain tsunami features are instead due to the hypo-central thrust slip patch in the model of Layet al. (2013a), which they show is moretsunamigenic than the up-dip oblique-slip patchto the south.

Great Subduction Earthquake Doublets

When a large earthquake is soon followed byanother nearby event with comparable size, thepair of seismic events is termed a doublet (e.g.,

160˚ 162˚ 164˚ 166˚ 168˚−14˚

−12˚

−10˚

−8˚

0 50 100

km

9.4 cm/yr9.4 cm/yr

Longitude

Latit

ude

Pacific Plate

Australia Plate

Solomon Trench

New

Hebrides Trench

−600

−500

−400

−300

−200

−100

0

SlabDepth (km)Mw 8.0

February 06, 2013

Tsunamigenic Major and Great Earthquakes(2004–2013): Source Processes Inverted from Seis-mic, Geodetic, and Sea-Level Data, Fig. 14 Historicalseismicity from 1976 and subduction zone features in the2013 Santa Cruz Islands earthquake area, Romano et al.

2015 (With kind permission from Springer Science+Busi-ness Media: Nat Hazard Earth Syst Sci, Source of the6 February 2013 Mw 8.0 Santa Cruz Islands tsunami,15, 2015, 1371–1379, Romano F, Molinari I, Lorito S,Piatanesi A.)


0 50

km

9.4 cm/yr

Longitude

02468

10

12Slip (m)

Nendo

Lata

10

20

30

40

50

164° 165° 166°

–12°

–11°

–10°

–9°

0

1

2

3

4

Slip (m)

Dept

h (k

m)

(a)

–10°

–11°

–12°

–164° –165° –166°

Tsunamigenic Major and Great Earthquakes(2004–2013): Source Processes Inverted from Seis-mic, Geodetic, and Sea-Level Data, Fig. 15 Slip dis-tributions of the 2013 Santa Cruz earthquake (top-leftpanel, from Lay et al. 2013a (Reprinted fromTectonophysics, Vol. 608, Lay T, Ye L, Kanamori H,Yamazaki Y, Cheung K F, Ammon C J, The February6, 2013 Mw 8.0 Santa Cruz Islands earthquake and tsu-nami, pp.1109-1121, Copyright 2013, with permission

from Elsevier); top-right panel, from Hayes et al. 2014(Reprinted from Earth and Planetary Science Letters, Vol.388, Hayes G P, Furlong K P, Benz H M, Herman H W,Triggered aseismic slip adjacent to the 6 February 2013Mw8.0 Santa Cruz Islands megathrust earthquake,pp. 265-272, Copyright 2014, with permission fromElsevier); bottom panel, modified after Romanoet al. 2015)


Lay and Kanamori 1980). The second event in adoublet is essentially an unusually large after-shock, deviating from the common tendency forthe largest aftershock to be about a magnitude unitsmaller than the first event. Two great earthquakedoublets occurred during the decade 2004–2013involving unusual pairs of MW 8 earthquakesincluding strong outer-rise tsunamigenesis; thefirst one struck the central Kuril Islands region,with an MW 8.3 plate boundary thrust earthquakeon 15 November 2006 being followed by an MW

8.1 outer-rise extensional earthquake on 13 Janu-ary 2007; the second doublet was on 9 September2009 at the northern end of the Tonga subductionzone and involved an initial MW 8.1 extensionalouter-rise earthquake followed within a minute byinterplate thrust faulting with MW 8.0. The recentKurils and Samoa-Tonga events involve two ofthe larger recorded outer-rise normal faultingevents. The 2004 (MW 9.2) and 2005 (MW 8.6)Sumatra great thrust earthquakes and the 2012(MW 8.7, 8.2) Indo-Australia intraplate strike-slip events are other recent examples of eventswith unusually large aftershocks.

Great megathrust events are commonlyfollowed by normal faulting aftershocks in theouter rise (2011 Tohoku, 2010 Maule, and 2012Haida Gwaii). The 2006–2007 Kuril doublet is thefirst documented case where the outer-rise-triggered event was also a great earthquake (the1896 and 1933 Sanriku earthquakes could beanother case, although the time separation of37 years is longer than typically associated withdoublets). The Samoa-Tonga doublet appears tobe the first documented case for the oppositebehavior, with megathrust faulting being dynam-ically triggered by an outer-rise “mainshock.”

Faulting interactions associated with doubletsare important to consider for earthquake and tsu-nami hazard assessment after large events,although they tend to have low probability.There was high concern about the possibility ofa 1933 Sanriku-like normal faulting event sea-ward of the 2011 Tohoku thrust zone and theassociated danger of a second large tsunamiimpacting the damaged coast soon after the mainrupture, but fortunately to date the extensionalfaulting occurred diffusely in many events on

separate faults. Only a few previous documentedcases of damaging tsunamis generated by outer-rise earthquakes exist; notably the 1933 Sanrikuand the 1977 Sumba events (Kanamori 1971;Lynnes and Lay 1988).

Due to their relative infrequency, large outer-rise faulting is generally not considered inpre-calculated tsunami scenarios used for opera-tional forecasting, and such events represent achallenge for tsunami warning operations. Retro-spectively, both the Kurils and Samoa-Tongaevents would have been a challenge even fortsunami forecast based on fast moment tensorsdetermination. In the first case, this is because ofseismological ambiguity between the two conju-gate steep focal planes, either southeast or north-west dipping, with strike direction ranging fromsubparallel to slightly oblique to the trench direc-tion (Ammon et al. 2008; Lay et al. 2009). How-ever, strain and accumulation release scenariosincluding earthquakes from 1963 to 2009 andthe interseismic period seem to slightly favor anorthwest dipping plane consistent with tensionalstress induced by bending and downward pull ofthe subducting plate (Lay et al. 2009; Lay 2015).In the second case, the overall complexity of thesequence of earthquakes that occurred almostsimultaneously and very close in space was hardto resolve (Beavan et al. 2010; Lay et al. 2010c;Satake 2010). Only very detailed analysis led tounraveling the complexity of the event that hadinitially been classified as a single earthquakewith a complex moment tensor (Lay et al. 2010c).

Kurils (2006, 2007)The great interplate MW 8.3 Kuril Islands earth-quake on 15 November 2006 occurred �100 kmsoutheast from the Simushir Island, of the Kurilarchipelago. The hypocenter is located on theshallow part of the subduction interface betweenthe Pacific and Okhotsk Plates. The earthquakenucleated between the rupture areas of the 1952Kamchatka MW 9.0 and the 1963 Kuril IslandsMW 8.5 earthquakes; in principle, this regioncould be considered a seismic gap. However, incontrast to some other cases (e.g., the Darwin gapin Chile), the lack of large recorded historicalearthquakes and the presence of a large forearc


basin unlike anywhere else along the arc (e.g., Lay2015) made the seismic potential assessment ofthis area particularly uncertain. On 13 January2007, a great MW 8.1 normal faulting earthquakeoccurred in the outer-rise region of the 2006 event.

Both earthquakes generated tsunamis strikingmostly uninhabited islands in the central area ofthe Kuril archipelago. The run-up measurements,collected during a post-tsunami survey conducteda few months after the 2007 earthquake,highlighted an average run-up value of �10 mestimated over a distance of �200 km(MacInnes et al. 2009), with a maximum observedrun-up of 22 m. Furthermore, the tsunami inunda-tion distance varies from �20 to �500 m. Sincethe survey was conducted after the second of twoevents, it is difficult to clearly distinguish thecausative source of the run-ups. It is likely thatthese observations are related to the first of the twoevents. MacInnes et al. (2009) argue that theobservations are associated with the 2006 earth-quake because at the time of the 2007 earthquake,Landsat images showed a snow accumulation inthe Kuril Islands; under this condition, the tsu-nami inundation would not necessarily erode thecoast making it difficult to clearly identify therun-up heights from debris deposits.

Both tsunamis propagated over the PacificOcean reaching Japan, New Zealand, FrenchPolynesia, Hawaii, Chile, and the WesternUSA. Larger tsunami signals were recorded bythe distant tide gauges for the 2006 earthquakethan for the 2007 event, with the DART buoysdocumenting centimetric tsunami signals. Forboth earthquakes, the remote tsunami energywas mainly directed southeast (Rabinovichet al. 2008).

Interactions with submarine ridges and numer-ous seamounts located along the tsunami pathwere documented as sources of secondary tsu-nami arrivals larger than the direct wave for the2006 tsunami (Kowalik et al. 2008). This can beclearly seen in the signals recorded by DARTbuoys D46404 and D46409, where late peaksappear, respectively, 2 and 3 h after the onset ofthe tsunami record. Accordingly, the initial waveat Kahului (Maui) of 50 cm amplitude wasfollowed 2 h later by a wave of 76 cm amplitude,

and similar patterns were observed elsewhere. Anadditional local amplification effect of note isalong the Mendocino Escarpment, whichentrapped, redirected, and amplified the tsunamiwave heading to Crescent City (California). Thesmall initial tsunami of �20 cm was barelynoticed there, whereas the highest wave of�88 cm (�1.7 m peak to trough) was recorded2–3 h later, resulting from a combination of dis-tant and local amplifications. Similar anomalies atCrescent City previously occurred for tsunamisfrom the great 1960 Chile and 1964 Alaskaearthquakes.

The earthquake and tsunami source of theseevents have been investigated by inversion ofteleseismic (e.g., Ammon et al. 2008; Layet al. 2009) and tsunami (Fujii and Satake 2008;Baba et al. 2009) data. All models for the 2006event show that most of the coseismic slipoccurred in the shallowest part of megathrustwith the rupture propagating northeastward fromthe hypocenter. The models inferred fromteleseismic data have greater slip (>13 m,Fig. 16) than the model obtained by tsunami datainversion (�8 m). The rupture zone inferred byboth models of the 2007 event is very narrow, andthe slip distributions and northeastward rupturepropagation are similar.

The run-up distribution associated with the2006 event (Fig. 16) broadside to the tsunamisource is fairly homogenous, with an averagevalue of �10 m; some particularly large values(�20 m) have been observed on Matua andSimushir Islands, likely due to the coastal plainbeaches that are backed by cliff or steep slopes.This morphological feature is common to most ofKuril Islands, and it is the cause of a limitedinundation distance (�500 m); in addition, thepresence of the islands is likely to have limitedthe tsunami penetration into and propagationwithin the Sea of Okhotsk.

Rabinovich et al. (2008) compared theobserved and modeled tsunamis for both the2006 and 2007 events. They demonstrated thatthe 2006 thrust earthquake resulted in largeredge waves than the 2007 outer-rise earthquake,as the former source was closer to the coast/shelfand aligned with it, consistently with the finding


of Geist (2012) that this is a factor enhancing edgewave excitation. Edge waves can be also gener-ated in the far field and in insular environments.They were detected after the 2006 earthquake alsoin Hawaii, where they were responsible for thehighest surges (e.g., Munger and Cheung 2008).

Samoa (2009)The great composite event of Samoa-Tongaoccurred on 29 September 2009 in the southwest-ern Pacific Ocean near the Tonga Trench. Theinitial event epicenter was at a depth of �20 kmand located �160 km southwest from the SamoaIslands. A normal faulting outer-rise event of MW

8.1 occurred first, followed by two thrustsub-events on the nearby plate boundarymegathrust located �50 km south of the normalfaulting.

The earthquake generated a tsunami that struckthe nearby coasts of Samoa, American Samoa,and northern Tonga Islands. The run-ups alongthe coasts of Samoa vary from a few meters upto �15 m, whereas they reach more than 20 malong the coasts of American Samoa and northernTonga Islands (e.g., Okal et al. 2010). Run-ups upto�10 m and�15 m occurred along the southernside of the Sava’i and Upolu islands, respectively;the highest number of fatalities (�150 people)was reported in Upolu. Maximum inundation dis-tance was 500 m along the Salani River in UpoluIsland and 200 m in Taga (Sava’i Island). Thetsunami had a significant impact on both thenorth and south side of Tutuila Island (AmericanSamoa), with �18 m in Poloa (on the western tipof the island) and 8 m in Pago Pago, the capital ofthe archipelago. The tsunami also propagated overthe Pacific Ocean with decimetric wave ampli-tudes reaching the coasts of Chile, New Zealand,French Polynesia, Mexico, Peru, Hawaii, and Cal-ifornia. In the near field, the largest tsunami waveswere recorded by tide gauges located in Apia(Samoa, amplitude �0.8 m) and Pago Pago(American Samoa, amplitude �2.7 m).

Beavan et al. (2010) (Fig. 17) embraced thehypothesis of an earthquakes doublet similar tothe 2006–2007 Kuril, where the outer-rise eventhas been triggered in by an interplate rupture,based on the analysis of GPS and tsunami data.

Eastward offset of a campaign GPS station on anorthern Tonga Island is opposite to the expecta-tion of only outer-rise normal faulting, andBeavan et al. (2010) postulated a thrust faultingevent of about magnitude 8. Additionally, at someDART buoys, the signals are well reproducedonly by considering a composite source, whereasnormal faulting is sufficient for reproducing thetide-gauge signal in Pago Pago in the near field(Fritz et al. 2011a). Since a thrust event was notimmediately identified from the seismic signals,basing on tsunami and geodetic evidence, a slowthrust event was hypothesized. This interplatethrust earthquake could have then promoted nor-mal faulting in the outer rise, thus making such asequence of earthquakes similar to the 2006–2007Kuril.

In-depth investigation of seismic wavesrecorded at teleseismic distances (Layet al. 2010c, Fig. 17), supported by the back-projection analysis, demonstrated that thrustfaulting actually followed the normal faulting.While there may have been some initial slow-slip thrust event that was not seismogenic, themoment of the triggered thrust faulting is suffi-cient to account for the GPS signal analyzed byBeavan et al. (2010), so there is no independentevidence for any slow-slip component to theprocess.

Samoa and American Samoa coasts were sig-nificantly struck by the tsunami (Okal et al. 2010;Fritz et al. 2011a, Fig. 18). In terms of source andpropagation effects, tsunami simulations showthat canyon-like bathymetric features offshore ofPago Pago could have driven the tsunami wavestoward the city where the geometry of the harborfurther amplified the tsunami, which penetrated500 m inland (Okal et al. 2010).

The islands of the Tonga archipelago mostimpacted by the tsunami were Tafahi andNiuatoputapu. The largest run-ups (up to�23 m) were measured on Tafahi, which is char-acterized by very steep coasts. The presence ofcoral reefs extending �2 km offshore reduced thetsunami impact on Niuatoputapu, where the max-imum run-up reached�10 m; however, due to theflat coastal topography of Tafahi Island, therewere larger inundation distances (�1 km). As for


the Kuril 2006 event, edge waves have beendocumented and modeled, due to resonance withthe shelf and embayment at Tutuila Islands, andthey were likely responsible for the severest prop-erty damage and casualties that occurred duringthe 2009 tsunami (Roeber et al. 2010).

Strike-Slip Events

Sumatra (2012)On 11 April 2012, anMW 8.7 earthquake occurredoffshore of Sumatra, some 100–200 km southwestof the Sunda Trench, with epicenter �300 kmsouth from that of the 2004 Sumatra earthquake.It was followed 2 h later by an MW 8.2 event withan epicenter �200 km south. An MW 8.7 intra-plate strike-slip earthquake had not been recordedbefore.

Due to its strike-slip character, involving lim-ited vertical seafloor displacement, this earth-quake generated a relatively small tsunami. Thelargest wave amplitudes were recorded by the tide

gauges located in Indonesia (Meulaboh, �1 m;Sabang �0.4 m); the tsunami also reached thecoasts of Thailand (Ko Taphao, �0.1 m) andpropagated over the Indian Ocean toward theMaldives Islands (Hanimaadhoo, �0.3 m), Mau-ritius Islands (Rodrigues,�0.2 m), the SeychellesIslands (Pointe la Rue, �0.1 m), and La ReunionIsland (Sainte Marie, �0.2 m).

Earthquakes outside of subduction zones, orwith unexpected mechanisms, are always a par-ticular challenge for tsunami warning procedures.The Sumatra 2012 tsunami was initiallyoverpredicted at several places, e.g., in Thailandand the Nicobar Islands, due to the assumption ofa worst-case thrust event scenario and the earlylarge magnitude estimates, before it was recog-nized that the event was a strike-slip earthquake.The warning was partially or totally adjusted byseveral warning centers as soon as focal mecha-nisms or tsunami measurements (e.g., at theDART buoys 23401 and STB01 midway betweenSumatra and Sri Lanka) became available. Falsealarms can always occur, and in a situation of

154°E

48°N

152°E150°E

46°NUrup

RasshuaUshishir

Matua

Shiashkotan

KetoiSimushir

Onekotan

2006 rupture area

2015

105

SCALE (m

)

Spaseniya Bay

Yuzhni Bay

Ryponkicha & Yankicha

Rasshua

Ainu BaySouth Bay

Sarychevo Toporkov

25

Dushnaya Bay

Steep, rockyCoastal plain

Inundation limitProfile locationRunup measurement15

8.68.6 8

6.1 7.3

9.610.49.9

7.8

7.87.9

5.74.9

6.45.7

7.1

5.77.312.920.8

20.218.3

17.118.1

1714

13.6

Ainu Bay South Bay

Southern Sarychevo

0 1km 2

Tsunamigenic Major and Great Earthquakes(2004–2013): Source Processes Inverted from Seis-mic, Geodetic, and Sea-Level Data, Fig. 16 Left panel,slip distributions of the 2006, 2007 Kuril earthquakes(Modified after Ammon et al. 2008) (Adapted by permis-sion from Macmillan Publishers Ltd: [Nature], AmmonC J, Kanamori H, Lay T, A great earthquake doublet andseismic stress transfer cycles in the Central Kuril Islands,Nature, 451:561–565, doi:10.1038/nature06521,

Copyright 2008); right panel, run-up distribution of the2006 Kuril earthquake (From MacInnes et al. 2009)(Reprinted with kind permission from Springer Science+Business Media: Pure and Applied Geophysics, FieldSurvey and Geological Effects of the 15 November 2006Kuril Tsunami in the Middle Kuril Islands, vol.166, 2009,pp.9-36, MacInnes B T, Pinegina T K, Bourgeois J,Razhigaeva N G, Kaistrenko V M, KravchunovskayaE A, Figure 4)


large uncertainty immediately following a seismicevent, erring on the cautious side may be consid-ered better than an underprediction or a missingalarm. However, the rate of false alarms should belimited as much as possible since they underminethe credibility of warning systems which maylimit the reaction of the population to the nextwarning, and avoiding unnecessary evacuationscan save a lot of money.

It was initially speculated that the causativefault of the MW 8.7 earthquake was aligned withthe system of roughly NS trending crustal fossil

transform faults off the Sunda Trench, within thefossil fabric in the Wharton Basin, a diffuse zoneof deformation at the boundary between theIndian and the Australian Plates, both convergingunder the Sunda Trench (Delescluse et al. 2012;Fig. 19).

Subsequent detailed source studies, mainlyusing teleseismic data, revealed a very complexmultiple fault rupture process adding up to an MW

8.7 event. It involved a complex system of orthog-onal faults (Fig. 19), both with right-lateralWNW-ESE and left-lateral NNE-SSW strike-slip

Pacific plate

Australian plate(Tonga block)

Tonga

Outer rise

Tsunami sources

Pull force

Trench

SamoaDisplacement detected by GPS

L1B2B1

L2

Beavan et al .1

Lay et al .2

B1 is slow and not detectedby seismic data

L2 buried in seismic data

Observed seismograms

Time

Slip rate

Slip rate

a

b

B1 B2

L1 L2

Time

Ground displacement

Tsunamigenic Major and Great Earthquakes(2004–2013): Source Processes Inverted from Seis-mic, Geodetic, and Sea-Level Data,Fig. 17 Interpretations of the Samoa 2009 doublet

earthquakes (From Satake 2010) (Reprinted by permissionfrom Macmillan Publishers Ltd: [Nature], Satake K, Dou-ble trouble at Tonga, Nature, 466:931-932, doi:10.1038/466931a, Copyright 2010)


mechanisms (e.g., Yue et al. 2012;Wei et al. 2013;Hill et al. 2015). The maximum slip of�37 mwasobserved on a fault centered around the hypocen-ter; the overall rupture zone, taking into accountthe entire fault system, spans �300 km andextends in depth to 50–60 km. Using regionalseismic data (Wei et al. 2013), a bilateral rupturealong a NNE-SSW trending fault with maximumslip of �8 m has been inferred for the MW 8.2aftershock.

The complex mechanism revealed by thesource analyses is compatible with areactivation/stress buildup process of this systemof faults due to the elastic stress transfer following

the 2004 Sumatra and 2005 Nias events and sub-sequent viscoelastic relaxation of the astheno-spheric mantle occurred during the �7 years inbetween the megathrust events and the intraplatesequence. The occurrence of strike-slip faultingrather than normal faulting as observed offshoreof the Kuril Islands in 2007 and Samoa in 2009 isaccounted for by the principle compressionalstress lying in the Indo-Australian Plate due tothe collision of India with Eurasia toward thenorthwest direction. Therefore, on a relativelyshort time scale, these events highlighted theinterplay between megathrust events at the sub-duction interface and intraplate deformation

0

2

m

–17

–16

–15

–14

–13

–174 –173 –172 –171 –170

Savai’i

Upolu

Tutuila

Niuatoputapu

(a)

Tsunamigenic Major and Great Earthquakes(2004–2013): Source Processes Inverted from Seis-mic, Geodetic, and Sea-Level Data, Fig. 18 Run-updistribution (left panels, from Fritz et al. 2011a (Reprintedfrom Earth-Science Reviews, Vol. 107, Fritz H M, BorreroJ C, Synolakis C E, Okal E A, Weiss R, Titov V V, Jaffe

B E, Foteinis S, Lynett P J, Chan I, Liu P L F, Insights onthe 2009 South Pacific tsunami in Samoa and Tonga fromfield surveys and numerical simulations, pp. 66-75, Copy-right 2011, with permission from Elsevier); right panel,from Okal et al. 2010)


within the incoming plates; on the longer timescale, the detaching process of Indian and Austra-lian Plates is driven by their overall differentialmotion (Delescluse et al. 2012).

Tsunami Earthquakes

During the decade from 2004 to 2013, there weretwo major earthquakes along the Sunda Trenchthat are classified as “tsunami earthquakes” andcaused deadly and particularly damaging tsu-namis along the coasts of Java and Sumatra.These are the 2006 Java and 2010 Mentawaiearthquakes, both with MW around 7.8. Priorevents in this class have struck near Java in 1994and near Sumatra in 1917.

Distinctive features of tsunami earthquakes arerelatively long source durations, depleted short-period seismic source spectra, slow rupture veloc-ities (�1.0–1.5 km/s), and deficiency of seismicmoment-scaled radiated energy in comparison toother earthquakes (e.g., Polet and Kanamori 2009;this encyclopedia). Their mechanical propertiesare particularly intriguing as they do not involvelarge ruptures that begin deeper on the megathrustand then propagate to shallow depth like the 2011Tohoku event. They occur entirely within theshallow and less rigid part of the megathrust,beneath the toe of sedimentary wedges. Many ofthe distinctive attributes of tsunami earthquakescan be qualitatively explained by the presence oflow rigidity materials (e.g., wet, low seismicvelocity sediments) that have greater slip for thesame seismic moment than for high rigidity rockslike those deeper on the megathrust or in thesurrounding crust (e.g., Geist and Bilek 2001).Large slip at shallow depth produces large dis-placement of the seafloor and efficient tsunamiexcitation near the deep trench (Geist 2009).Their occurrence in the vicinity of steep seafloordip at the trench can also increase the net verticalmotion of the seafloor resulting from horizontaldisplacement of the slope (Tanioka and Satake1996).

The strong tsunami excitation for tsunamiearthquakes results in greater coastal impact, par-ticularly in the near field and at regional distances.

Relatively weak ground shaking due to the deple-tion of short-period energy from the source canexacerbate the tsunami impact because people arenot prompted to self-evacuate. It is of utmostimportance to stress, during any tsunami pre-paredness campaign, that the duration of the shak-ing (even if it is weak) should be considered bycoastal populations among the possible naturalwarnings of an imminent large tsunami.A sustained duration of about 1 min or more ofshaking means: run to high ground!

Java (2006)On 17 July 2006 an MW 7.8 earthquake occurredoff the south coast of Java, at shallow depth on thesubduction interface where the Australian Plateunderthrusts the Sunda Plate approximately inthe north-northeast direction across the SundaTrench. This event generated a tsunami that killedmore than 600 people in Java.

Very weak ground shaking preceded the largetsunami arrival. Permisan experienced 21 m peakrun-up along a 1–2 km long stretch of coast, withan inundation distance of �1 km. This was muchlarger than the average run-up value of 5–7 malong �250–300 km of the southern Java coast(Fritz et al. 2007). This tsunami was clearlyrecorded at tide gauges around the southeasternIndian Ocean. Some inundation occurred, close tohigh tide, on the Western Australian coast,peaking at �7 m on a steep cliff within ShelterBay. However, tsunami effects were compara-tively larger in the near field than in the far fieldwith respect to those of the tsunamis generated bylarger magnitude earthquakes.

The slip distribution of the 2006 event has beenstudied using body and surface wave (Ammonet al., 2006, Fig. 20) and tsunami (Fujii and Satake2006) data, respectively. Both studies infer a200–250 km rupture extent expanding unilaterallysoutheastward from the epicenter, with slip mostlyconfined to the shallowest part of the megathrust,up-dip of the 15 km hypocenter. The low esti-mated rupture velocity (1–1.5 km/s), very longduration (�190 s), and compound spectral fea-tures of the source radiation, with high frequen-cies originating from localized regions, and long-period waves excited by the entire rupture, along


–4°

–2°

0°2°4°6°8°10°

12°

14°

16°

1 cm

yr–1

(rel

. to

Indi

a)

10.0

× 1

0–9 y

r–1

1 cm

yr–1

(rel

. to

Sun

da)

Mw 8

.6

86°

88°

90°

92°

94°

96°

98°

100°

Long

itude

CIB

CIB

NYR NYR

WB

WB

CIB

NYR

WB

Mw 8

.28.2

Mw 8

.2

Latitude

89°

90°

91°

92°

93°

94°

1°2°3°4°

bc

d

e

fa

g

Long

itud

e

Latitude

102030

T 0 = 3

0 s

2468

5101520

Max

imum

slip

= 1

2 m

Mw =

7.8

4

246810

5 5

T 0 = 0

s

T 0 = 7

0 s

Max

imum

slip

= 1

1 m

Max

imum

slip

= 3

7 m

Max

imum

slip

= 2

6 m

Mw =

7.9

4

Mw =

8.3

0

Mw =

8.5

4

e

E E

E E

EE

SSNN

50250

Dist

ance

alo

ng st

rike

(km

)0

5010

0

= 1

6°,

= 8

0°,

= –

10°

= 1

09°,

= 8

0°,

= 1

80°

= 2

86°,

= 7

5°,

= 1

70°

= 1

11°,

= 7

4°,

= 1

80°

50250 500 500

= 1

06°,

= 8

5°,

= 1

85°

Distance along dip (km)

bbbaaa WW

1010101010W

20202030 3030

20202010 1010

111

(m)

cccd

(m)

111333555777

1 11333

5 55

T 0 = 6

0 s

(m)

WWW101015152020 101520

T 0 = 1

10 s

(m)

WWWfff2255 25

Tsun

amigen

icMajor

andGreat

Earthqua

kes(200

4–20

13):So

urce

Proc

esses

Inve

rted

from

Seismic,G

eodetic,a

ndSe

a-Le

velData,

Fig.1

9Tectonicfram

e-workandslipdistribu

tionof

the20

12greatintraplateearthq

uakes;(leftpanel,fromWei

etal.201

3;middlepanel,mod

ified

afterDelescluseetal.201

2(A

dapted

byperm

ission

from

Macmillan

Pub

lishers

Ltd:[N

ature],DelescluseM,C

hamot-Roo

keN,C

attin

R,

Fleito

utL,Trubienko

O,Vigny

C,April20

12intra-oceanicseismicity

offSum

atra

boosted

bythe

Banda-A

ceh

megathrust,

Nature,

490:24

0-24

4,do

i:10.10

38/

nature1152

0,Cop

yright

2012

);righ

tpanels,mod

ified

afterYue

etal.20

12(A

dapted

byperm

ission

from

Macmillan

Pub

lishersLtd:[Nature],Y

ueH,L

ayT,

Kop

erKD,E

néchelonandorthog

onalfaultrup

turesof

the11

April20

12greatintraplateearthq

uakes,

Nature,49

0:24

5-24

9,do

i:10.10

38/nature114

92,C

opyright

2012

)


with the large shallow megathrust slip and largetsunami excitation, identify the 2006 Java event asa tsunami earthquake.

Significant differences exist in the inferred slipon the fault; the teleseismic inversion estimatestwo to three times larger localized slip valuesrelative to the coarser model used in the tsunamiinversion. This appears to be largely due to thelow, 10 GPa rigidity value, typical of sedimentarymaterials, used to convert seismic moment to slipfor the teleseismic inversion. The actual rigidity isnot well known, and large variations are commonin shallow subduction zones. On the other hand,the tsunami recordings used for inversion aremostly obtained from tide gauges located atregional and far-field distances, and consequentlythe source slip is “averaged” over 50 � 50 km2

patches, because only near-field tsunami data aresensitive to finer scale slip heterogeneity.

The length of the Java coast hit by the tsunamiinundation is consistent with the along-strikeextent of the inferred source, to which it liesbroadside. The tsunami earthquake nature of theJava event, with large shallow slip, accounts forthe damaging tsunami, whose consequences, mer-cifully, were likely limited by the fact that itoccurred on Monday afternoon, when there werevery few people on the beaches.

The run-up “outlier” observed at Permisan ledto the suggestion of a possible additionalnon-seismic source for this tsunami, namely, asubmarine landslide (Fritz et al. 2007; Fig. 20).However, a combination of the source featuresand the tsunami propagation toward land offers aconvincing alternative explanation. The convexshape of the subduction as seen from the Javacoast may have led to a strong focusing effect,without ruling out additional focusing and ampli-fication effects due to the diffractive behavior ofthe bathymetric features which are not known indetail (Kanoğlu et al. 2013 and referencestherein).

Mentawai (2010)About 4 years after and �1000 km northwestalong the Sunda Trench from the Java 2006event, another large (MW 7.8) underthrusting tsu-nami earthquake occurred on 25 October 2010

offshore from the southeastern Mentawai Islands.Including the 1994 Java tsunami earthquakewhich occurred �600 km to the east of the 2006Java event, this was the third tsunami earthquakein 16 years along the same megathrust. The hypo-center of the 2010 Mentawai event was located�40 km offshore southwest from the South PagaiIsland, just up-dip of the rupture zone of the12 September 2007 Bengkulu MW 8.4.

The 2010 earthquake generated a large tsunamithat struck the west coasts of the MentawaiIslands, with the strongest effects on North andSouth Pagai Islands (Hill et al. 2012). In particu-lar, the run-up observations vary over a range of2–17 mwith peak values along the mid-west coastof South Pagai Island. The death toll followingthis tsunami exceeded 500. A tsunami warningwas issued that did not reach the majority of thepopulation due to the absence of sufficient infra-structure for alert dissemination. Again, self-evacuation was limited: the earthquake producedweak ground shaking and was perceived as beingmuch weaker than the 2007 Bengkulu earthquake,which generated a relatively modest tsunami, orthe 2009 intermediate depth Padang earthquake,which only generated small anomalies recordedby the Padang tide gauge.

The tsunami has been measured offshore bybuoys (DART and a nearby GPS buoy whichrecorded the tsunami from 10 min after the earth-quake) and tide gauges spread around the IndianOcean. Most recorded relatively weak, but dis-tinct, centimetric and decimetric tsunami signalsthat have been used by different groups for inver-sions or for calibration of source models wereobtained by teleseismic inversions.

Several source models have been proposed forthe 2010 Mentawai earthquake. Even thoughretrieved using different kinds of geophysicaldata, they all share a common pattern of slipmostly confined along the shallowest part of themegathrust up-dip of the epicenter (Newmanet al. 2011; Lay et al. 2011c; Hill et al. 2012;Satake et al. 2013b; Yue et al. 2015). The mostrecent and best constrained model (Yueet al. 2015), obtained by joint inversion of tsu-nami, regional high-rate GPS, and teleseismicdata, indicates that the rupture propagated up-dip


and northward from the epicenter and that most oralmost all the slip was released in a very narrowpatch close to the trench, elongated along thestrike direction, with an almost purely thrust slipdirection.

The large (up to 15 m, Fig. 21) shallow slip canaccount for the run-up observations on the Pagai

Islands (Hill et al. 2012; Satake et al. 2013b). Forexample, the tsunami penetrated up to �400 minland on Muntei (North Pagai Island) and up to�300 on Maonai (South Pagai Island), likely dueto the U-shaped morphology of these bays. Inaddition, the maximum run-up measurement(�17 m) for this event was observed on Sibigau,

Tsunamigenic Major and Great Earthquakes(2004–2013): Source Processes Inverted from Seis-mic, Geodetic, and Sea-Level Data, Fig. 20 Slip (top

panel, from Ammon et al. 2006) and run-up distribution(bottom panel, from Fritz et al. 2007) of the 2006 Javaearthquake


located �5 km off South Pagai and closer to thetsunami source; on this island, the tsunami pene-trated up to 600 m inland. Conversely, the PagaiIslands acted as a natural shield greatly mitigatingthe tsunami impact on cities like Padang on theSumatran coast to the northeast.

The limited impact of the Java and Mentawaitsunami earthquakes in the far field, with respectto other events induced by earthquakes of largermagnitudes, demonstrates once more that localtsunamis are greatly influenced by the features ofthe slip distribution, while distant tsunamis arecontrolled to first order by the earthquake magni-tude (e.g., Okal 1988; Geist 2009).

Special Cases

Solomon Islands (2007)The Solomon Islands-New Britain subductionzones have hosted numerous large earthquakes,often occurring in triggered doublets or multiplets(e.g., Lay and Kanamori 1980), and some eventshave been tsunamigenic (e.g., Geist and Parsons2005) with impact on the adjacent islands. Thecentral Solomon Islands subduction zone is thesite of a triple junction, with the Australian andSolomon Sea Plates underthrusting the PacificPlate. There is slight divergence of the twosubducting plates, and they are separated by aridge-transform system with very young(<3.5 Ma) lithosphere entering the trench. Histor-ical seismicity near the triple junction has beenlow, and there is a small slabless window at depth,so even intermediate depth seismicity is low in thevicinity of the triple junction. Conventional ideasabout subduction of very young, thin lithospherehad suggested that seismic coupling might be lowin the region, accounting for the lack of any largehistoric earthquakes, but this was proved incorrectwhen a major rupture stretched right across thetriple junction in 2007 (Furlong et al. 2009).

On 1 April 2007, a great interplate earthquakeoccurred off the New Georgia Group of islands(Solomon Archipelago) with magnitude MW 8.1.The hypocenter is shallow on the subductioninterface between the Australian and PacificPlates, close to the triple junction (Fig. 22), but

the rupture extended northwest, crossing over thetriple junction, with large slip also occurring onthe interface between the Solomon Sea and PacificPlates. This surprised many researchers.

This earthquake generated a tsunami thatstruck most of the islands inWestern and Choiseulprovinces of the Solomon Archipelago. In gen-eral, when a typical megathrust earthquake occursvery close (a few tens of km) to the coast, then the“lead time” between detection of the shaking andthe tsunami impact can be very short. Luckily, inthis case, the message “run to high ground after anearthquake” had been passed on to younger gen-erations by the survivors of prior tsunamis, andthis triggered an effective self-evacuation that,despite massive tsunami impact (more than 6000damaged or destroyed houses), limited the deathtoll to less than 60 fatalities (Fritz and Kalligeris2008).

Several source models of the Solomon Islandsearthquake have been published. For example,based on inversion of teleseismic body waves(Furlong et al. 2009, Fig. 22) or coral uplift andcoastal subsidence data (Chen et al. 2009). Fur-long et al. (2009) inverted global seismic bodywaves, finding a unilateral rupture that propagatednorthwest from the epicenter at a relatively highrupture velocity of 2.5 km/s. Their model featurestwo shallow slip patches, straddling the junction.The rake angle (�90�) of the northernmost patch(maximum slip �7 m) is consistent with the con-vergence direction north of the junction, whereasthe southernmost patch (maximum slip �4 m) isconsistent with the oblique convergence direction(rake �50�) of the Australian Plate south of thejunction. This suggests that the event involvedtwo plates underthrusting a third plate in differentdirections during a single earthquake, consistentwith the tectonic setting.

The pattern of slip found by the analysis ofgeodetic data is to first order consistent with theteleseismic model but features larger slip values(up to 30 m locally) over a narrower zone at veryshallow depth; this difference might be due topostseismic uplift being included in the static off-set data used, as the campaign GPS sites werereoccupied 1 or more weeks after the earthquake.Also, the geodetic data have diminished


resolution of slip as distance from the coastincreases. There is also uncertainty in the valueof rigidity used to convert seismic moment intoslip in seismic inversions. Nevertheless, this is thefirst well-documented case of a great event rup-turing through a triple junction. Together withother cases, such as the 2002 MW 7.9 Denaliearthquake, in which a complex rupture involvedthree different faults and both a thrust and a strike-slip mechanism (Eberhart-Phillips et al. 2003),this earthquake is a further caveat against assump-tions of strong segmentation based on the pres-ence of structural barriers in hazard analyses(Taylor et al. 2008), as discussed for severalevents analyzed in this entry.

The post-tsunami surveys also measured tsu-nami run-up (Fig. 22). The largest impactoccurred on the Simbo, Ghizo, and RanonggaIslands which are located within the verticalcoseismic displacement field; the maximumrun-up (�12 m) was measured on Simbo Islandand is consistent with the shallow slip near thetrench (Chen et al. 2009), whereas �7 m havebeen measured on Ranongga Island and �6 mon Ghizo Island. Run-ups of �5 m are reportedalong the coasts of Choiseul Island located in frontof the northernmost and largest patch of slip. Thepresence of coral reefs and the straits that separatethe New Georgia islands is likely to have

attenuated the tsunami impact on these islands(Fritz and Kalligeris 2008).

Haida Gwaii (2012)On 28 October 2012, a major (MW 7.8) shallowthrust earthquake occurred along a transpressiveboundary of predominantly right-lateral strike-slip motion between the Pacific and North Amer-ican Plates. The thrusting represents slippartitioning in a restraining bend of the QueenCharlotte Fault, which has pure strike-slip motionfurther to the NNW. There is uncertainty in theprecise amount of convergence over the past6 Ma, during which the relative plate motionbetween the Pacific and North American Plateshas changed slightly, but it is likely that the PacificPlate underthrusts to at least the eastern side of thesouthern island of Haida Gwaii (Moresby Island)(Fig. 23).

The earthquake hypocenter was located closeto the Pacific coast beneath western MoresbyIsland, in the Haida Gwaii archipelago, Canada.The accompanying tsunami struck primarily onthe nearby western coast of the island, whichshielded the coast of British Columbia. During apost-tsunami survey on the island, Leonard andBednarski (2015) collected a number of run-upobservations ranging from�3 to 13 m, peaking inthe Davidson inlet in the near field of the tsunamisource. The maximum inundation distances are

99˚ 100˚

−3˚

−2˚

Sanding

Lighthouse

SibarubaruPagai−Selatan

SilauLimoksua

SiumangSouth−LibautNorth−Libaut

BakeSibigauPitojetsabeu

Malakopa

Siduamata

SiopasabeuSabeugukgung

MacaroniSmall−Island

Tumalei

Katiet

Kandui

Simakakak

100 km

0 2 4 6 8 10 12 14 16 18

Runup height

Tsunami height

Tsunami height (m)

Tsunamigenic Major and Great Earthquakes(2004–2013): Source Processes Inverted from Seis-mic, Geodetic, and Sea-Level Data, Fig. 21 Slip (left

panel, from Yue et al. 2015) and run-up distribution (rightpanel, from Hill et al. 2012) of the 2010 Mentawaiearthquake


Fig. 22 (continued)


relatively short due to steep coastal morphology;the tsunami penetrated the Pacific coast up to�60 m, with maximum inundation distanceobserved in the northernmost Otard Bay, some130 km northwest from the epicenter.

The tsunami was also detected by several tidegauges, as documented by the NGDC/WDS data-base and by Fine et al. (2015). It was recorded byseveral tide gauges located along the coast ofBritish Columbia, in most cases with very smallheights (<20 cm). The highest tsunami waves(�0.5 m peak to trough) were recorded by thetide gauges in Henslung Cove on Langara Island(a small island north of Graham Island), and inWinter Harbour on the northwestern coast of Van-couver Island (Fine et al. 2015).

The tsunami also propagated over the northernPacific Ocean; decimetric maximum heights wererecorded by coastal tide gauges in the Hawaiianarchipelago (at Kahului, the tide gauge recorded�1.5 m from peak to trough); centimetric heightswere measured elsewhere. Tsunami signals ofseveral centimeters recorded by the DARTbuoys located off the Alaska, Oregon, Washing-ton, and Canada coasts were used for the inferenceof the tsunami source, as described below. Addi-tional tsunami observations arise from the oceanbottom pressure sensors of the NEPTUNE-Canada cabled observatory (Fine et al. 2015).Overall, in the far field, the Haida Gwaii earth-quake generated modest tsunami waves. In termsof near-field effects, this was the largest tsunamiever recorded on Moresby Island.

Teleseismic data inversion, iteratively opti-mized by forward modeling of the tsunami signalat the surrounding DART buoys (Lay et al. 2013b,Fig. 24), indicates that the rupture started approx-imately beneath the trace of the Queens Charlottefault offshore of Moresby Island, but thecoseismic slip release occurred essentially sea-ward of the hypocenter. The rupture propagatedbilaterally along the trench direction for a total

extent of �150 km, extending seaward (up-dip)to the toe of the sedimentary wedge. GPS obser-vations indicate that the model of Layet al. (2013b) has rupture extending too fardown-dip beneath the coast, and a revised modelwith all slip locating offshore is preferred(Nykolaishen et al. 2015). As a result, it appearsthat the slip was concentrated in a narrow band�30 km wide on the subduction interfaceextending to the trench and localized entirelyunder the sedimentary wedge seaward of theQueen Charlotte Fault offshore of Moresby Island(Fig. 24).

As already noted for the 2006 Java and the2010 Mentawai “tsunami earthquakes,” it isoften assumed that the sedimentary wedge nearthe toe and the subduction interface beneath canonly slip and deform aseismically. Conversely,this earthquake indicates that the interface belowthe Queen Charlotte terrace accumulates com-pressional strain perpendicular to the plate bound-ary and experiences stick-slip faulting.A surprising attribute of the faulting is that therupture velocity inferred from seismic modeling isnot particularly low (Lay et al. 2013b, find arupture expansion velocity of 2.3 km/s), despitethe shallow depth and sedimentary wedge contactzone. This is similar to the 2007 Solomon Islandsearthquake, which ruptured the very shallowmegathrust with comparable rupture expansionvelocity (2.5 km/s). The moment-scaled radiatedenergy estimated by Lay et al. (2013b) is 1.28 �10�5, which is also higher than typical tsunamiearthquakes, and the source spectrum is onlyslightly depleted in short-period spectral ampli-tudes. Thus, the Haida Gwaii earthquake is not atsunami earthquake per se; this may be associatedwith the specific sedimentary materials, largelyderived from erosion of the ancient metamorphicterrane uplifted in the islands, in contrast to deepabyssal muds on the plate interface in othertrenches.

��

Tsunamigenic Major and Great Earthquakes(2004–2013): Source Processes Inverted from Seis-mic, Geodetic, and Sea-Level Data, Fig. 22 Triplejunction and slip distribution (top panel, Lay 2015)(Reprinted from Earth and Planetary Science Letters, Vol.

409, Lay T., The surge of great earthquakes from 2004 to2014, pp. 133-146, Copyright 2015, with permission fromElsevier) and run-up distribution (bottom panel, from Fritzet al. 2008) of the 2007 Solomon earthquake


To first order, the largest run-up observations(Fig. 24) are spatially correlated with the tsunamisource resulting from the teleseismic inversion(Leonard and Bednarski 2015). They lie mainlybroadside to the inferred coseismic slip, withvalues ranging from �3 to 13 m. In particular,large run-up values (�5 m), although not thelargest, were observed in front of the broadest(along dip direction) slip patch, which also hadthe peak slip. They occurred inside Goski Bay,near the maximum (simulated) tsunami ampli-tudes, offshore the Pacific coast. The relativelyflat area inside Goski Bay also features the max-imum observed inundation distances of �50 m.Notably, despite having occurred along a regionof lower slip and smaller offshore tsunami ampli-tudes, the absolute run-up maxima (�8–13 m)were observed northernmost in the DavidsonInlet, with a shorter inundation distance than atother places (�35 m).

The observations indicate a complex relation-ship between the main source features, i.e., sizeand location of the slip patches, and the features ofthe generated tsunami, i.e., amplitude and period,which in turn determine the impact along thecoastline. First, the slip pattern determines theamplitude and the steepness of the initial tsunamiwave elevation, for example, the narrower the slipdistribution, the shorter the tsunami wavelengthsat the source and the shorter the period of theresulting waves. Second, the offshore tsunamiamplitudes tend to peak near the broadest slipregion, not necessarily in correspondence withthe steepest and highest initial tsunami wavefield. Third, an important role is played by theinteraction of the tsunami period with the coastalfeatures, with, for example, funneling and reso-nant phenomena causing higher run-up values forthe short-period waves within the fjord-likeDavidson Inlet (Leonard and Bednarski 2015),and longer inundation distances with run-upheights limited by frictional dissipation withinGoski Bay for the longer-period tsunami compo-nents. This situation is similar to that alreadydiscussed for the Tohoku 2011 tsunami, with acorrespondence between the tsunami periods gen-erated in different source regions and the differentinundation features observed in the Sendai plain

and in the bays of the northernmost Iwateprefecture.

Another interesting feature is the verysustained inundation observed at Gilbert Bay,peaking at �9 m and reaching to �50 m fromthe coast, occurring in the near-field obliqueregime with respect to the source. This is likelycaused by the late arrival of trapped edge waves,which traveled along the coast of Moresby Island.

Future Directions

The decade from 2004 to 2013 is striking both forthe number of major tsunamigenic events and thegreat damage and loss of life produced by tsu-nami. It is also notable for the vastly expandedseismic, geodetic, and tsunami recordings avail-able to analyze the source processes, tsunamiexcitation, and complexity of the events. Thereare still major limitations in instrument coverage,and not all events have comparable recordings,but instrument deployments are steadily increas-ing, and operational earthquake and tsunamiwarning capabilities are advancing in manyregions.

Progress in quantification of the nature of greatearthquake ruptures has been dramatic, but therecent events also exhibit great complexity inindividual ruptures and in triggering interactionsthat present many challenges. Compound faultingand doublet events are very challenging for warn-ing systems, as are shallow “tsunami earth-quakes.” The analyses of multiple datasets havedemonstrated the complementary sensitivity tosource processes provided by seismic, geodetic,and tsunami observation and the value of jointanalysis when possible. Performing such compre-hensive analyses in near real time to augmenttsunami warning systems is technically possiblebut remains very challenging logistically.

Improved assessment of strain accumulationdistributions, using both onshore and offshoregeodetic methods, is very desirable for identifyingregions of likely future large-slip events and formaking progress on bounding likely ruptureextent beyond the limited performance of classicseismic gap model approaches. Tsunami hazard


assessments need to consider not only ordinaryinterplate megathrust faulting but also “extremes”in which shallow slip is enhanced and even “tsu-nami earthquakes” or intraplate rupture

possibilities in many regions. Nonetheless, prob-abilities for these features are difficult to assess.Great caution is moreover needed regardingassumptions on earthquake segmentation, and

Tsunamigenic Major and Great Earthquakes(2004–2013): Source Processes Inverted from Seis-mic, Geodetic, and Sea-Level Data, Fig. 23 Slip dis-tribution and tectonic context/model of the 2012 HaidaGwaii earthquake (from Lay et al. 2013b) (Reprintedfrom Earth and Planetary Science Letters, Vol.

375, Lay T, Ye L, Kanamori H, Yamazaki Y, CheungK F, Kwong K, Koper K D, Mw 7.8 Haida Gwaii under-thrusting earthquake and tsunami: Slip partitioning alongthe Queen Charlotte Fault transpressional plate boundary,pp. 57-70, Copyright 2013, with permission from Elsevier)


50 k

m

Mor

esby

Isla

ndGra

ham

Isla

nd

Que

enC

harlo

tte

Lang

ara

Isla

nd

Kun

ghit

Isla

nd

N

Afte

rsho

ck

zone

Engl

efie

ldB

ay

Tasu

Soun

d

Gow

gaia

Bay

53°N

132°

W

133°

W13

4°W54°N

54°N

Infe

rred

runu

p >

4 m

Infe

rred

runu

p <

3-3.

5 m

Mea

sure

d ru

nup

2 m

46

810

12 Run

up (m

abo

ve s

tate

of t

ide)

Flow

dep

th (m

abo

ve g

roun

d)R

unup

at e

xpos

ed s

ites

Bon

anza

Bea

ch

Sea

l Inl

et

Mitc

hell

Inle

tN

ewco

mbe

Inle

t

Gos

ki B

ay

Sta

ki B

ay

Ros

e In

let

Lous

coon

e In

let

Ota

rd B

ay

Sau

nder

s Is

land

Dav

idso

n In

let

Blu

e H

eron

Bay

Kw

oon

Cov

eS

unda

y In

let

Poc

ket I

nlet

Ta'd

asl

Gilb

ert B

ay

Hen

slun

gC

ove

tide

gaug

e

Hea

d of

Ren

nell

Sou

ndG

udal

Bay

Mik

e In

let

Puf

fin C

ove

Hus

band

Har

bour

Tsun

amigen

icMajor

andGreat

Earthqua

kes(200

4–20

13):So

urce

Proc

esses

Inve

rted

from

Seismic,G

eodetic,andSe

a-Le

velD

ata,Fig.24

Maxim

umtsunam

iwaveam

plitu

des(leftpanel,from

Lay

etal.2013b)(Reprinted

from

EarthandPlanetary

Science

Letters,V

ol.375,L

ayT,YeL,K

anam

oriH

,Yam

azakiY

,CheungKF,KwongK,

Koper

KD,Mw

7.8

Haida

Gwaiiunderthrustin

gearthquake

and

tsunam

i:Slip

partitioningalongtheQueen

CharlotteFaulttranspressionalplateboundary,p

p.57-70,

Copyright

2013,with

perm

ission

from

Elsevier)

andrun-up

distributio

nof

the2012

Haida

Gwaiiearthquake

(right

panel,from

Fineet

al.2015)(Reprinted

with

kind

perm

ission

from

SpringerScience+BusinessMedia:PureandAppliedGeophysics,

Observatio

nsandnumerical

modelingof

the2012

Haida

Gwaiitsunam

ioffthecoast

ofBritishColum

bia,vol.172,2015,pp.699-718,FineIV,CherniawskyJY,Thomson

RE,R

abinovichAB,K

rassovskiM

V,F

igure14)


the possibility that an M9 earthquake may occurvirtually on any subduction zone should not beoverlooked.

Acknowledgments The authors are grateful to Eric L.Geist for his constructive comments, to the Section EditorsEric L. Geist and W. H. K. Lee, and to the Senior EditorAndrew Spencer. We acknowledge large use of sea-leveland tsunami data from the following websites: https://www.ngdc.noaa.gov/hazard/tide.shtml; http://nctr.pmel.noaa.gov/Dart/; http://www.ioc-sealevelmonitoring.org;and of earthquake parametric data and information fromhttp://earthquake.usgs.gov/contactus/golden/neic.php andother USGS web pages. Global seismic recordings for theanalyses discussed here were primarily obtained from thedata centers of the Federation of Digital Seismic Networksand the Incorporated Research Institutions for Seismology.We also made large use of the International Tsunami Infor-mation Center website (http://itic.ioc-unesco.org). Somefigures were produced with Generic Mapping Tools(GMT, https://www.soest.hawaii.edu/gmt/). We are grate-ful to the numerous colleagues who sent us high-resolutionversions of all the figures reproduced in this paper (withpermission from publisher when requested). S. Lorito's andF. Romano's work on earthquakes and tsunamis issupported by Italian Flagship project RITMARE; by pro-ject ASTARTE (Assessment, Strategy And Risk Reductionfor Tsunamis in Europe) FP7-ENV2013 6.4-3, grant603839; by the agreement between Italian DPC andINGV, annex B2 (2015). T. Lay's work on earthquakes issupported by NSF grant EAR1245717.

Bibliography

Ammon CJ, Ji C, Thio HK, Robinson D, Ni S,Hjorleifsdottir V, Kanamori H, Lay T, Das S,Helmberger D, Ichinose G, Polet J, Wald D (2005)Rupture process of the 2004 Sumatra-Andaman earth-quake. Science 308:1133–1139. doi:10.1126/science.1112260

Ammon CJ, Kanamori H, Lay T, Velasco AA (2006) The17 July 2006 Java tsunami earthquake. Geophys ResLett 33:L24308. doi:10.1029/2006GL028005

Ammon CJ, Kanamori H, Lay T (2008) A great earthquakedoublet and seismic stress transfer cycles in the CentralKuril Islands. Nature 451:561–565. doi:10.1038/nature06521

Ammon CJ, Lay T, Simpson DW (2010) Great earthquakesand global seismic networks. Seismol Res Lett81:965–971. doi:10.1785/gssrl.81.6.965

Baba T, Cummins PR, Thio HK, Tsushima H (2009) Val-idation and joint inversion of teleseismic waveformsfor earthquake source models using deep ocean bottompressure records: a case study of the 2006 Kurilmegathrust earthquake. Pure Appl Geophys166:55–76. doi:10.1007/s00024-008-0438-1

Banerjee P, Pollitz FF, Nagarajan B, B€urgmann R (2007)Coseismic slip distributions of the 26 December 2004Sumatra-Andaman and 28 March 2005 Nias earth-quakes from GPS static offsets. Bull Seismol Soc Am97:S86–S102. doi:10.1785/0120050609

Beavan J, Wang X, Holden C, Wilson K, Power W,Prasetya G, Bevis M, Kautoke R (2010) Near-simultaneous great earthquakes at Tongan megathrustand outer rise in September 2009. Nature 466:959–963.doi:10.1038/nature09292

Beck S, Barrientos S, Kausel E, Reyes M (1998) Sourcecharacteristics of historic earthquakes along the centralChile subduction zone. J South Am Earth Sci11:115–129

Biggs J, Robinson DP, Dixon TH (2009) The 2007 Pisco,Peru, earthquake (M 8.0): seismology and geodesy.Geophys J Int 176:657–669. doi:10.1111/j.1365-246X.2008.03990.x

Bilham R (2005) A flying start, then a slow slip. Science308:1126–1127. doi:10.1126/science.1113363

Bletery Q, Sladen A, Delouis B, Vallée M, Nocquet JM,Rolland L, Jiang J (2014) A detailed source model forthe Mw 9.0 Tohoku-Oki earthquake reconciling geod-esy, seismology, and tsunami records. J Geophys Res119:7636–7653. doi:10.1002/2014JB011261

Borrero JC, Weiss R, Okal EA, Hidayat R, ArcasSuranto D, Titov VV (2009) The tsunami of 2007 Sep-tember 12, Bengkulu province, Sumatra, Indonesia:post-tsunami field survey and numerical modelling.Geophys J Int 178:180–194. doi:10.1111/j.1365-246X.2008.04058.x

Borrero JC, McAdoo B, Jaffe B, Dengler L,Gelfenbaum G, Higman B, Hidayat R, Moore A,Kongko W, Lukijanto PR, Prasetya G, Titov V,Yulianto E (2012) Field survey of the March 28, 2005Nias-Simeulue earthquake and tsunami. Pure ApplGeophys 168:1075–1088. doi:10.1007/s00024-010-0218-6

Briggs RW, Sieh K, Meltzner AJ, Natawidjaja D,Galetzka J, Suwargadi B, Hsu YJ, Simons M,Hananto N, Suprihanto I, Prayudi D, Avouac JP,Prawirodirdjo L, Bock Y (2006) Deformation and slipalong the Sunda megathrust in the great 2005 Nias-Simeulue earthquake. Science 311:1897–1901.doi:10.1126/science.1122602

Chen T, Newman AV, Feng L, Fritz HM (2009) Slip dis-tribution from the 1 April 2007 Solomon Islands earth-quake: a unique image of the near-trench rupture.Geophys Res Lett 36:L16307. doi:10.1029/2009GL039496

Chlieh M, Avouac JP, Hjorleifsdottir V, Song TR, Ji C,Sieh K, Sladen A, Hebert H, Prawirodirdjo L, Bock Y,Galetzka J (2007) Coseismic slip and afterslip of thegreat (Mw 9.15) Sumatra-Andaman earthquake of2004. Bull Seismol Soc Am 97:S152–S173.doi:10.1785/0120050631

Choi BH, Hong SJ, Pelinovsky E (2006) Distribution ofrunup heights of the December 26, 2004 tsunami in the




http://nctr.pmel.noaa.gov/Dart/

http://nctr.pmel.noaa.gov/Dart/

http://www.ioc-sealevelmonitoring.org

http://earthquake.usgs.gov/contactus/golden/neic.php

http://itic.ioc-unesco.org

https://www.soest.hawaii.edu/gmt/

Indian Ocean. Geophys Res Lett 33:L13601.doi:10.1029/2006GL025867

Das S (2011) Earthquake rupture: inverse problem. In:Gupta HK (ed) Encyclopedia of solid earth geophysics.Springer, Netherlands. doi:10.1007/978-90-481-8702-7

Delescluse M, Chamot-Rooke N, Cattin R, Fleitout L,Trubienko O, Vigny C (2012) April 2012 intra-oceanicseismicity off Sumatra boosted by the Banda-Acehmegathrust. Nature 490:240–244. doi:10.1038/nature11520

Delouis B, Nocquet JM, Vallée M (2010) Slip distributionof the February 27, 2010 Mw = 8.8 Maule earthquake,central Chile, from static and high-rate GPS, InSAR,and broadband teleseismic data. Geophys Res Lett 37:L17305. doi:10.1029/2010GL043899

Eberhart-Phillips D, Haeussler PJ, Freymueller JT, FrankelAD, Rubin CM, Craw P, Ratchkovski NA,Anderson G, Carver GA, Crone AJ et al (2003) The2002 Denali fault earthquake, Alaska: a large magni-tude, slip-partitioned event. Science 300:1113–1118.doi:10.1126/science.1082703

Faulkner DR, Mitchell TM, Behnsen J, Hirose T,Shimamoto T (2011) Stuck in the mud? Earthquakenucleation and propagation through accretionaryforearcs. Geophys Res Lett 38:L18303. doi:10.1029/2011GL048552

Fine IV, Cherniawsky JY, Thomson RE, Rabinovich AB,Krassovski MV (2015) Observations and numericalmodeling of the 2012 Haida Gwaii tsunami off thecoast of British Columbia. Pure Appl Geophys172:699–718. doi:10.1007/s00024-014-1012-7

Fritz HM, Kalligeris N (2008) Ancestral heritage savestribes during 1 April 2007 Solomon Islands tsunami.Geophys Res Lett 35:L01607. doi:10.1029/2007GL031654

Fritz HM, Kongko W, Moore A, McAdoo B, Goff J,Harbitz C, Uslu B, Kalligeris N, Suteja D, Kalsum K,Titov V, Gusman A, Latief H, Santoso E, Sujoko S,Djulkarnaen D, Sunendar H, Synolakis C (2007)Extreme runup from the 17 July 2006 Java tsunami.Geophys Res Lett 34:L12602. doi:10.1029/2007GL029404

Fritz HM, Kalligeris N, Borrero JC, Broncano P, OrtegaE (2008) The 15 August 2007 Peru tsunami runupobservations and modeling. Geophys Res Lett 35:L10604. doi:10.1029/2008GL033494

Fritz HM, Borrero JC, Synolakis CE, Okal EA, Weiss R,Titov VV, Jaffe BE, Foteinis S, Lynett PJ, Chan I, LiuPLF (2011a) Insights on the 2009 South Pacific tsu-nami in Samoa and Tonga from field surveys andnumerical simulations. Earth Sci Rev 107:66–75.doi:10.1016/j.earscirev.2011.03.004

Fritz HM, Petroff CM, Catalán CR, Winckler P,Kalligeris N, Weiss R (2011b) Field survey of the27 February 2010 Chile tsunami. Pure Appl Geophys168:1989–2010. doi:10.1007/s00024-011-0283-5

Fritz H M, Papantoniou A, Biukoto L, Gilly A, WeiY (2014) The Solomon Islands Tsunami of 6 February2013 in the Santa Cruz Islands: Field Survey and

Modeling, EGU General Assembly 2014, held27 April - 2 May, in Vienna, Austria, id.15777

Fujii Y, Satake K (2006) Source of the July 2006West Javatsunami estimated from tide gauge records. GeophysRes Lett 33:L24317. doi:10.1029/2006GL028049

Fujii Y, Satake K (2007) Tsunami source of the 2004Sumatra-Andaman earthquake inferred from tidegauge and satellite data. Bull Seismol Soc Am 97:S192–S207. doi:10.1785/0120050613

Fujii Y, Satake K (2008) Tsunami sources of the November2006 and January 2007 Great Kuril earthquakes. BullSeismol Soc Am 98:1559–1571. doi:10.1785/0120070221

Fulton PM, Brodsky EE, Kano Y, Mori J, Chester F,Ishikawa T, Harris RN, Lin W, Eguchi N, Toczko S,Expedition 343, 343T, and KR13-08 Scientists(2013) Low coseismic friction on the Tohoku-Okifault determined from temperature measurements. Sci-ence 342:1214–1217. doi:10.1126/science.1243641

Furlong KP, Lay T, Ammon CJ (2009) A great earthquakerupture across a rapidly evolving three-plate boundary.Science 324:226–229. doi:10.1126/science.1167476

Gahalaut VK, Nagarajan B, Catherine JK, Kumar S (2006)Constraints on 2004 Sumatra-Andaman earthquakerupture from GPS measurements in Andaman-Nicobarislands. Earth Planet Sci Lett 242:365–374.doi:10.1016/j.epsl.2005.11.051

Geist EL (2009) Phenomenology of tsunamis: statisticalproperties from generation to runup. Adv Geophys51:107–169. doi:10.1016/S0065-2687(09)05108-5

Geist EL (2012) Near-field tsunami edge waves and com-plex earthquake rupture. Pure Appl Geophys170:1475–1491

Geist EL, Bilek SL (2001) Effect of depth-dependent shearmodulus on tsunami generation along subductionzones. Geophys Res Lett 28:1315–1318

Geist EL, Oglesby DD (2014) Tsunamis: stochastic modelsof occurrence and generation mechanisms. In: RobertA. Meyers (ed.) Encyclopedia of complexity and sys-tems science, Springer, New York. doi:10.1007/978-3-642-27737-5_595-1

Geist EL, Parsons T (2005) Triggering of tsunamigenicaftershocks from large strike-slip earthquakes: analysisof the November 2000 New Ireland earthquakesequence. Geochem Geophys Geosyst 6:Q10005.doi:10.1029/2005GC000935

Geist EL, Parsons T (2011) Assessing historical ratechanges in global tsunami occurrence. Geophys J Int187:497–509. doi:10.1111/j.1365-246X.2011.05160.x

Geist EL, Bilek SL, Arcas D, Titov VV (2006) Differencesin tsunami generation between the December 26, 2004and March 28, 2005 Sumatra earthquakes. EarthPlanets Space 58:185–193. doi:10.1186/BF03353377

González FI et al (2009) Probabilistic tsunami hazardassessment at Sea-side, Oregon, for near- and far-fieldseismic sources. J Geophys Res 114:C11023.doi:10.1029/2008JC005132

Gusman AR, Tanioka Y, Kobayashi T, Latief H, PandoeW (2010) Slip distribution of the 2007 Bengkulu


earthquake inferred from tsunami waveforms andInSAR data. J Geophys Res 115:B12316.doi:10.1029/2010JB007565

Hartzell SH, Heaton TH (1983) Inversion of strong groundmotion and teleseismic waveform data for the faultrupture history of the 1979 Imperial Valley, California,earthquake. Bull Seismol Soc Am 73:1553–1583

Hayes GP, Furlong KP, Benz HM, Herman HW(2014) Triggered aseismic slip adjacent to the 6 Febru-ary 2013 Mw8.0 Santa Cruz Islands megathrust earth-quake. Earth Planet Sci Lett 388:265–272.doi:10.1016/j.epsl.2013.11.010

Hill EM, Borrero JC, Huang Z, Qiu Q, Banerjee P,Natawidjaja DH, Elosegui P, Fritz HM, SuwargadiBW, Pranantyo IR, Li LL, Macpherson KA,Skanavis V, Synolakis CE, Sieh K (2012) The 2010Mw 7.8Mentawai earthquake: very shallow source of arare tsunami earthquake determined from tsunami fieldsurvey and near-field GPS data. J Geophys Res 117:B06402. doi:10.1029/2012JB009159

Hill EM, Yue H, Barbot S, Lay T, Tapponnier P,Hermawan I, Hubbard J, Banerjee P, Feng L,Natawidjaja D, Sieh K (2015) The 2012 Mw 8.6 Whar-ton Basin sequence: a cascade of great earthquakesgenerated by near-orthogonal, young, oceanic-mantlefaults. J Geophys Res 120:1–25. doi:10.1002/2014JB011703

Hoechner A, Babeyko AY, Sobolev SV (2008) EnhancedGPS inversion technique applied to the 2004 Sumatraearthquake and tsunami. Geophys Res Lett 35:L08310.doi:10.1029/2007GL033133

Hossen MJ, Cummins PR, Dettmer J, Baba T (2015) Tsu-nami waveform inversion for sea surface displacementfollowing the 2011 Tohoku earthquake: importance ofdispersion and source kinematics. J Geophys Res SolidEarth 120:1–22. doi:10.1002/2015JB011942

Hsu YJ, Simons M, Avouac JP, Galetzka J, Sieh K,Chlieh M, Natawidjaja D, Prawirodirdjo L, BockY (2006) Frictional afterslip following the 2005 Nias-Simeulue earthquake, Sumatra. Science312:1921–1925. doi:10.1126/science.1126960

Hsu YJ, Simons M, Williams C, Casarotti E (2011) Three-dimensional FEM derived elastic Green’s functions forthe coseismic deformation of the 2005 Mw 8.7 Nias-Simeulue, Sumatra earthquake. Geochem GeophysGeosyst 12:1–19. doi:10.1029/2011GC003553

Hughes K, Masterlark T, Mooney WD (2010) Poroelasticstress-triggering of the 2005 M8.7 Nias earthquake bythe 2004 M9.2 Sumatra-Andaman earthquake. EarthPlanet Sci Lett 293:289–299. doi:10.1016/j.epsl.2010.02.043

Ioualalen M, Perfettini H, Condo SY, Jimenez C, TaveraH (2013) Tsunami modeling to validate slip models ofthe 2007 Mw8.0 Pisco earthquake, Central Peru. PureAppl Geophys 170:433–451. doi:10.1007/s00024-012-0608-z

Ishii M, Shearer PM, Houston H, Vidale JE(2007) Teleseismic Pwave imaging of the 26December2004 Sumatra-Andaman and 28 March 2005 Sumatra

earthquake ruptures using the Hi-net array. J GeophysRes 112:B11307. doi:10.1029/2006JB004700

Ishii M, Kiser E, Geist EL (2012) Mw 8.6 Sumatran earth-quake of 11 April 2012: rare seaward expression ofoblique subduction. Geology 41:319–322.doi:10.1130/G33783.1

Jackson DD, Kagan YY (2011) Characteristic earthquakesand seismic gaps. In: Harsh K. Gupta (ed.) Encyclope-dia of solid earth geophysics

Kagan YY, Jackson DD (2013) Tohoku earthquake: asurprise? Bull Seismol Soc Am 103:1181–1194.doi:10.1785/0120120110

Kanamori H (1971) Seismological evidence for a litho-spheric normal faulting – the Sanriku earthquake of1933. Phys Earth Planet In 4:289–300

Kanamori H (1972) Mechanism of tsunami earthquakes.Phys Earth Planet Inter 6:346–359. doi:10.1016/0031-9201(72)90058-1

Kanamori H (1977) The energy release of great earth-quakes. J Geophys Res 82:2981–2987

Kanamori H, McNally KC (1982) Variable rupture modeof the subduction zone along the Ecuador-Colombiacoast. Bull Seismol Soc Am 72:1241–1253

Kanamori H, Rivera L, Lee WHK (2010) Historicalseismograms for unraveling a mysterious earthquake:the 1907 Sumatra earthquake. Geophys J Int183:358–374. doi:10.1111/j.1365-246X.2010.04731.x

Kaneko Y, Avouac JP, Lapusta N (2010) Towards inferringearthquake patterns from geodetic observations ofinterseismic coupling. Nat Geosci 3:363–369.doi:10.1038/NGEO843

Kanoğlu U, Titov VV, Aydın B, Moore C, Stefanakis TS,Zhou H, Spillane M, Synolakis CE (2013) Focusing oflong waves with finite crest over constant depth. ProcR Soc A 469:1–20. doi:10.1098/rspa.2013.0015

Konca AO, Hjorleifsdottir V, Song TRA, Avouac JP,Helmberger DV, Ji C, Sieh K, Briggs R, MeltznerA (2007) Rupture kinematics of the 2005 Mw 8.6Nias-Simeulue earthquake from the joint inversion ofseismic and geodetic data. Bull Seismol Soc Am 97:S307–S322. doi:10.1785/0120050632

Konca AO, Avouac JP, Sladen A, Meltzner AJ, Sieh K,Fang P, Li Z, Galetzka J, Genrich J, Chlieh M,Natawidjaja DH, BockY, Fielding EJ, Ji C, HelmbergerDV (2008) Partial rupture of a locked patch of theSumatra megathrust during the 2007 earthquakesequence. Nature 456:631–635. doi:10.1038/nature07572

Koper KD, Hutko AR, Lay T, Sufri O (2012) Imagingshort-period seismic radiation from the 27 February2010 Chile (Mw 8.8) earthquake by back-projectionof P, PP, and PKIKP waves. J Geophys Res 117:B02308. doi:10.1029/2011JB008576

Kowalik Z, Horrillo J, Knight W, Logan T (2008) KurilIslands tsunami of November 2006: 1. Impact at Cres-cent City by distant scattering. J Geophys Res 113:C01020. doi:10.1029/2007JC004402

Kozdon JE, Dunham EM (2013) Rupture to the trench:dynamic rupture simulations of the 11 March 2011


Tohoku earthquake. Bull Seismol Soc Am103:1275–1289. doi:10.1785/0120120136

Kubo H, Kakehi Y (2013) Source process of the 2011Tohoku earthquake estimated from the joint inversionof teleseismic body waves and geodetic data includingseafloor observation data: source model with enhancedreliability by using objectively determined inversionsettings. Bull Seismol Soc Am 103:1195–1220.doi:10.1785/0120120113

Lay T (2015) The surge of great earthquakes from 2004 to2014. Earth Planet Sci Lett Invited Front Pap409:133–146. doi:10.1016/j.epsl.2014.10.047

Lay T, Kanamori H (1980) Earthquake doublets in theSolomon Islands. Phys Earth Planet In 21:283–304

Lay T, Kanamori H, Ammon CJ, Nettles M, Ward SN,Aster RC, Beck SL, Bilek SL, Brudzinski MR,Butler R, DeShon HR, Ekström G, Satake K, SipkinS (2005a) The Great Sumatra-Andaman earthquake ofthe 26 December 2004. Science 308:1127–1133.doi:10.1126/science.1112250

Lay T, Kanamori H, Ammon CJ, Nettles M, Ward SN,Aster RC, Beck SL, Bilek SL, Brudzinski MR,Butler R, DeShon HR, Ekström G, Satake K, SipkinS (2005b) Response to comment on “The GreatSumatra-Andaman Earthquake of the 26 December2004”. Science 310:1431. doi:10.1126/science.1119662

Lay T, Kanamori H, Ammon CJ, Hutko AR, Furlong K,Rivera L (2009) The 2006–2007 Kuril Islands greatearthquake sequence. J Geophys Res 114:B113208.doi:10.1029/2008JB006280

Lay T, Ammon CJ, Kanamori H, Koper KD, Sufri O,Hutko AR (2010a) Teleseismic inversion for ruptureprocess of the 27 February 2010 Chile (Mw 8.8) earth-quake. Geophys Res Lett 37:L13301. doi:10.1029/2010GL043379

Lay T, Ammon CJ, Hutko AR, Kanamori H (2010b)Effects of kinematic constraints on teleseismic finite-source rupture inversions: Great Peruvian earthquakesof 23 June 2001 and 15 August 2007. Bull Seismol SocAm 100:969–994. doi:10.1785/0120090274

Lay T, Ammon CJ, Kanamori H, Rivera L, Koper KD,Hutko AR (2010c) The 2009 Samoa–Tonga greatearthquake triggered doublet. Nature 466:964–968.doi:10.1038/nature09214

Lay T, Ammon CJ, Kanamori H, Kim MJ, Xue L (2011a)Outer trench-slope faulting and the 2011Mw 9.0 off thePacific coast of Tohoku earthquake. Earth Planets Space63:713–718. doi:10.5047/eps.2011.05.006,713-718

Lay T, Ammon CJ, Kanamori H, Xue L, Kim MJ (2011b)Possible large near-trench slip during the 2011 Mw 9.0off the Pacific coast of Tohoku earthquake. EarthPlanets Space 63:687–692. doi:10.5047/eps.2011.05.033

Lay T, Ammon CJ, Kanamori H, Yamazaki Y, Cheung KF,Hutko AR (2011c) The 25 October 2010 Mentawaitsunami earthquake (Mw 7.8) and the tsunami hazardpresented by shallow megathrust ruptures. GeophysRes Lett 38:L06302. doi:10.1029/2010GL046552

Lay T, Kanamori H, Ammon CJ, Koper KD, Hutko AR,Ye L, Yue H, Rushing TM (2012) Depth-varying rup-ture properties of subduction zone megathrust faults.J Geophys Res 117:B04311. doi:10.1029/2011JB009133

Lay T, Ye L, Kanamori H, Yamazaki Y, Cheung KF,Ammon CJ (2013a) The February 6, 2013 Mw 8.0Santa Cruz Islands earthquake and tsunami.Tectonophysics 608:1109–1121. doi:10.1016/j.tecto.2013.07.001

Lay T, Ye L, Kanamori H, Yamazaki Y, Cheung KF,Kwong K, Koper KD (2013b) Mw 7.8 Haida Gwaiiunderthrusting earthquake and tsunami: slippartitioning along the Queen Charlotte Faulttranspressional plate boundary. Earth Planet Sci Lett375:57–70. doi:10.1016/j.epsl.2013.05.005

Leonard LJ, Bednarski JM (2015) The preservation poten-tial of coastal coseismic and tsunami evidence observedfollowing the 2012 Mw 7.8 Haida Gwaii Thrust earth-quake. Bull Seismol Soc Am 105:1280–1289.doi:10.1785/0120140193

Lin YN, Sladen A, Ortega-Culaciati F, Simons M, AvouacJP, Fielding EJ, Brooks BA, Bevis M, Genrich J,Rietbrock A, Vigny C, Smalley R, Socquet A (2013)Coseismic and postseismic slip associated with the2010 Maule earthquake, Chile: characterizing theArauco Peninsula barrier effect. J Geophys Res118:3142–3159. doi:10.1002/jgrb.50207

Liu PLF, Yeh H, Lin P, Chang KT, Cho YS (1998) Gener-ation and evolution of edge-wave packets. Phys Fluids10:1635–1657

Lorito S, Romano F, Piatanesi A, Boschi E (2008) Sourceprocess of the September 12, 2007, Mw 8.4 southernSumatra earthquake from tsunami tide gauge recordinversion. Geophys Res Lett 35:L02310. doi:10.1029/2007GL032661

Lorito S, Piatanesi A, Cannelli V, Romano F, MeliniD (2010) Kinematics and source zone properties ofthe 2004 Sumatra-Andaman earthquake and tsunami:nonlinear joint inversion of tide-gauge, satellite altim-etry, and GPS data. J Geophys Res 115:B02304.doi:10.1029/2008JB005974

Lorito S, Romano F, Atzori S, Tong X, Avallone A,McCloskey J, Cocco M, Boschi E, Piatanesi A (2011)Limited overlap between the seismic gap and coseismicslip of the great 2010 Chile earthquake. Nat Geosci4:173–177. doi:10.1038/ngeo1073

Loveless JP, Meade BJ (2010) Geodetic imaging of platemotions, slip rates, and partitioning of deformation inJapan. J Geophys Res 115:B02410. doi:10.1029/2008JB006248

Lynnes CS, Lay T (1988) Source process of the great 1977Sumba earthquake. J Geophys Res 93:13407–13420

MacInnes BT, Pinegina TK, Bourgeois J, Razhigaeva NG,Kaistrenko VM, Kravchunovskaya EA (2009) Fieldsurvey and geological effects of the 15 November2006 Kuril Tsunami in the Middle Kuril Islands. PureAppl Geophys 166:9–36. doi:10.1007/s00024-008-0428-3


Madariaga R, Métois M, Vigny C, Campos J (2010) Cen-tral Chile finally breaks. Science 328:181–182.doi:10.1126/science.1189197

Maercklin N, Festa G, Colombelli S, Zollo A (2012) Twinruptures grew to build up the giant 2011 Tohoku, Japan,earthquake. Sci Rep 709:1–6. doi:10.1038/srep00709

McCloskey J, Lange D, Tilmann F, Nalbant SS, Bell AF,Natawidjaja DH, Rietbrock A (2010) The September2009 Padang earthquake. Nat Geosci 3:70–71.doi:10.1038/ngeo753

Menke W, Abend H, Bach D, Newman K, Levin V (2006)Review of the source characteristics of the GreatSumatra-Andaman islands earthquake of 2004. SurvGeophys 27:603–613. doi:10.1007/s10712-006-9013-4

Minson SE, SimonsM, Beck JL (2013) Bayesian inversionfor finite fault earthquake source models I – theory andalgorithm. Geophys J Int 194:1701–1726. doi:10.1093/gji/ggt180

Moreno M, Rosenau M, Oncken O (2010) 2010 Mauleearthquake slip correlates with pre-seismic locking ofAndean subduction zone. Nature 467:198–204.doi:10.1038/nature09349

Moreno M, Melnick D, Rosenau M, Baez J, Klotz J,Oncken O, Tassara A, Chen J, Bataille K, Bevis M,Socquet A, Bolte J, Vigny C, Brooks B, Ryder I,Grund V, Smalley B, Carrizo D, Bartsch M, HaseH (2012) Toward understanding tectonic control onthe Mw 8.8 2010 Maule Chile earthquake. Earth PlanetSci Lett 321–322:152–165. doi:10.1016/j.epsl.2012.01.006

Mori N, Takahashi T, Yasuda T, Yanagisawa H (2011)Survey of the 2011 Tohoku earthquake tsunami inun-dation and run-up. Geophys Res Lett 38:L00G14.doi:10.1029/2011GL049210

Munger S, Cheung KF (2008) Resonance in Hawaii watersfrom the 2006 Kuril Islands tsunami. Geophys Res Lett35:L07605. doi:10.1029/2007GL032843

Nalbant SS, Steacy S, Sieh K, Natawidjaja D, McCloskeyJ (2005) Earthquake risk on the Sunda trench. Nature435:756–757. doi:10.1038/nature435756a

Natawidjaja DH, Sieh K, Chlieh M, Galetzka J, SuwargadiBW, Cheng H, Edwards RL, Avouac JP, Ward SN(2006) Source parameters of the great Sumatranmegathrust earthquakes of 1797 and 1833 inferredfrom coral microatolls. J Geophys Res 111:B06403.doi:10.1029/2005JB004025

Neetu S, Suresh I, Shankar R, Shankar D, Shenoi SSC,Shetye SR, Sundar D, Nagarajan B (2005) Comment on“The Great Sumatra-Andaman Earthquake of the26 December 2004”. Science 310:1431. doi:10.1126/science.1118950

Newman AV, Hayes G, Wei Y, Convers J (2011) The25 October 2010 Mentawai tsunami earthquake, fromreal-time discriminants, finite-fault rupture, and tsu-nami excitation. Geophys Res Lett 38:L05302.doi:10.1029/2010GL046498

Nishenko SP (1991) Circum-Pacific seismic potential:1989–1999. Pure Appl Geophys 135:169–259

Nykolaishen L, Dragert H, Wang K, James TS, SchmidtM (2015) GPS observations of crustal deformationassociated with the 2012 Mw 7.8 Haida Gwaii earth-quake. Bull Seismol Soc Am 105:1241–1252.doi:10.1785/0120140177

OglesbyDD,Archuleta RJ, Nielsen SB (1998) Earthquakeson dipping faults: the effects of broken symmetry.Science 280:1055–1059. doi:10.1126/science.280.5366.1055

Okal EA (1988) Seismic parameters controlling Far-fieldtsunami amplitudes: a review. Nat Hazards 1:67–96

Okal EA, Fritz HM, Synolakis CE, Borrero JC, Weiss R,Lynett PJ, Titov VV, Foteinis S, Jaffe BE, Liu PLF,Chan I (2010) Field survey of the Samoa tsunami of29 September 2009. Seismol Res Lett 81:577–591.doi:10.1785/gssrl.81.4.577

Park J, Song T-R A, Tromp J, Okal E, Stein S, Roult G,Clevede E, Laske G, Kanamori H, Davis P, Berger J,Braitenberg C, Van Camp M, Lei X, Sun H, Xu H,Rosat S (2005) Earth’s free oscillations excited by the26 December 2004 Sumatra-Andaman earthquake. Sci-ence 308(5725):1139–1144. doi:10.1126/science.1112305

Philibosian B, Sieh K, Avouac JP, Natawidjaja DH, ChiangHW, Wu CC, Perfettini H, Shen CC, Daryono MR,Suwargadi BW (2014) Rupture and variable couplingbehavior of the Mentawai segment of the Sundamegathrust during the supercycle culmination of 1797to 1833. J Geophys Res 119:7258–7287. doi:10.1002/2014JB011200

Piatanesi A, Lorito S (2007) Rupture process of the 2004Sumatra-Andaman earthquake from tsunami waveforminversion. Bull Seismol Soc Am 97:S223–S231.doi:10.1785/0120050627

Pietrzak J, Socquet A, HamD, SimonsW, Vigny C, LabeurRJ, Schrama E, Stelling G, Vatvani D (2007) Definingthe source region of the Indian Ocean tsunami fromGPS, altimeters, tide gauges and tsunami models. EarthPlanet Sci Lett 261:49–64. doi:10.1016/j.epsl.2007.06.002

Plafker G, Nishenko S, Cluff L, Syahrial M (2006) Thecataclysmic 2004 tsunami on NWSumatra – preliminary evidence for a near-field second-ary source along the western Aceh basin. Seismol ResLett 77:231

Poisson B, Oliveros C, Pedreros R (2011) Is there a bestsource model of the Sumatra 2004 earthquake for simu-lating the consecutive tsunami? Geophys J Int185:1365–1378. doi:10.1111/j.1365-246X.2011.05009.x

Polet J, Kanamori H (2009) Tsunami earthquakes. In:Meyers A (ed) Encyclopedia of complexity and sys-tems science. Springer, New York. doi:10.1007/978-0-387-30440-3_567

Pritchard ME, Norabuena EO, Ji C, Boroschek R,Comte D, SimonsM, Dixon TH, Rosen PA (2007) Geo-detic, teleseismic, and strong motion constraints on slipfrom recent southern Peru subduction zone earth-quakes. J Geophys Res 112:B03307. doi:10.1029/2006JB004294


Rabinovich AB, Lobkovsky LI, Fine IV, Thomson RE,Ivelskaya TN, Kulikov EA (2008) Near-source obser-vations and modeling of the Kuril Islands tsunamis of15 November 2006 and 13 January 2007. Adv Geosci14:105–116

Rhie J, Dreger D, BurgmannR, Romanowicz B (2007) Slipof the 2004 Sumatra–Andaman earthquake from jointinversion of long-period global seismic waveforms andGPS static offsets. Bull Seismol Soc Am 97:S115–S127. doi:10.1785/0120050620

Roeber V, Yamazaki Y, Cheung KF (2010) Resonance andimpact of the 2009 Samoa tsunami around Tutuila,American Samoa. Geophys Res Lett 37:L21604.doi:10.1029/2010GL044419

Romano F, Piatanesi A, Lorito S, D’Agostino N, Hirata K,Atzori S, Yamazaki Y, Cocco M (2012) Clues fromjoint inversion of tsunami and geodetic data of the2011 Tohoku-Oki earthquake. Sci Rep 2:385.doi:10.1038/srep00385

Romano F, Trasatti E, Lorito S, Piromallo C, Piatanesi A,Ito Y, Zhao D, Hirata K, Lanucara P, Cocco M (2014)Structural control on the Tohoku, earthquake ruptureprocess investigated, by 3D FEM, tsunami and geo-detic data. Sci Rep 5631:1–11. doi:10.1038/srep05631

Romano F, Molinari I, Lorito S, Piatanesi A (2015) Sourceof the 6 February 2013 Mw 8.0 Santa Cruz Islandstsunami. Nat Hazard Earth Syst Sci 15:1371–1379.doi:10.5194/nhess-15-1371-2015

Rothman D (1986) Automatic estimation of large residualstatics corrections. Geophysics 51:332–346.doi:10.1190/1.1442092

Satake K (2009) Tsunamis, inverse problem of. In: MeyersA (ed) Encyclopedia of complexity and systems sci-ence. Springer, New York

Satake K (2010) Double trouble at Tonga. Nature466:931–932. doi:10.1038/466931a

Satake K, Fujii Y (2014) Review: source models of the2011 Tohoku earthquake and long-term forecast oflarge earthquakes. J Dis Res 9:272–280

Satake K, Fujii Y, Harada T, Namegaya Y (2013a) Timeand space distribution of coseismic slip of the 2011Tohoku earthquake as inferred from tsunami waveformdata. Bull Seismol Soc Am 103:1473–1492.doi:10.1785/0120120122

Satake K, Nishimura Y, Putra PS, Gusman AR,Sunendar H, Fujii Y, Tanioka Y, Latief H, YuliantoE (2013b) Tsunami source of the 2010 Mentawai,Indonesia earthquake inferred from tsunami field sur-vey and waveform modeling. Pure Appl Geophys170:1567–1582. doi:10.1007/s00024-012-0536-y

Shearer P, B€urgmann R (2010) Lessons learned from the2004 Sumatra-Andaman megathrust rupture. AnnuRev Earth Planet Sci 38:103–131. doi:10.1146/annurev-earth-040809-152537

Sieh K, McKinnon EE, Pilarczyk JE, Chiang HW,Horton B, Rubin CM, Shen CC, Ismail N, Vane CH,Feener RM (2015) Penultimate predecessors of the2004 Indian Ocean tsunami in Aceh, Sumatra: strati-graphic, archeological, and historical evidence.

J Geophys Res 120:308–325. doi:10.1002/2014JB011538

Singh SC, Carton H, Tapponnier P, Hananto ND, ChauhanAPS, Hartoyo D, Bayly M, Moeljopranoto S,Bunting T, Christie P, Lubis H, Martin J (2008) Seismicevidence for broken oceanic crust in the 2004 Sumatraearthquake epicentral region. Nat Geosci 1:777–781.doi:10.1038/ngeo336

Sladen A, Tavera H, Simons M, Avouac JP, Konca AO,Perfettini H, Audin L, Fielding EJ, Ortega F,Cavagnoud R (2010) Source model of the 2007 Mw8.0 Pisco, Peru earthquake – implications forseismogenic behavior of subduction megathrusts.J Geophys Res 115:B02405. doi:10.1029/2009JB006429

Stein S, Okal EA (2011) The size of the 2011 Tohokuearthquake need not have been a surprise. Eos92:227–228

Stein S, Geller RJ, Liu M (2012) Why earthquake hazardmaps often fail and what do about it. Tectonophysics562–563:1–25. doi:10.1016/j.tecto.2012.06.047

Strasser FO, ArangoMC, Bommer JJ (2010) Scaling of thesource dimensions of interface and intraslabsubduction-zone earthquakes with moment magnitude.Seismol Res Lett 81:941–950

Subarya C, ChliehM, Prawirodirdjo L, Avouac JP, Bock Y,Sieh K, Meltzner AJ, Natawidjaja DH, McCaffreyR (2006) Plate-boundary deformation associated withthe great Sumatra-Andaman earthquake. Nature440:46–51. doi:10.1038/nature04522

Synolakis C, Kanoglu U (2015) The Fukushima accidentwas preventable. Phil Trans R Soc A 373:2053

Tanioka Y, Satake K (1996) Tsunami generation by hori-zontal displacement of ocean bottom. Geophys ResLett 23:861–886

Tappin DR, Grilli ST, Harris JC, Geller RJ, Masterlark T,Kirby JT, Shi F, Ma G, Thingbaijam KKS, Mai PM(2014) Did a submarine landslide contribute to the 2011Tohoku tsunami?Mar Geol 357:344–361. doi:10.1016/j.margeo.2014.09.043

Taylor FW, Briggs RW, Frohlich C, Brown A,Hornbach M, Papabatu AK, Meltzner AJ, BillyD (2008) Rupture across arc segment and plate bound-aries in the 1 April 2007 Solomons earthquake. NatGeosci 1:253–257. doi:10.1038/ngeo159

Thomson RE, Rabinovich AB, Krassovski MV(2007) Double jeopardy: concurrent arrival of the2004 Sumatra tsunami and storm-generated waves onthe Atlantic coast of the United States and Canada.Geophys Res Lett 34:L15607. doi:10.1029/2007GL030685

Titov V, Rabinovich AB, Mofjeld HO, Thomson RE,González FI (2005) The global reach of the 26 Decem-ber 2004 Sumatra Tsunami. Science 309:2045–2048.doi:10.1126/science/1114576

Tsai VC, Ampuero JP, Kanamori H, Stevenson DJ(2013) Estimating the effect of Earth elasticity andvariable water density on tsunami speeds. GeophysRes Lett 40:492–496. doi:10.1002/grl.50147


UNISDR (2015) Making development sustainable: thefuture of disaster risk management. Global assessmentreport on disaster risk reduction. United Nations Officefor Disaster Risk Reduction (UNISDR), Geneva

Vigny C, Simons WJF, Abu S, Bamphenyu R, SatirapodC et al (2005) Insight into the 2004 Sumatra-Andamanearthquake from GPS measurements in southeast Asia.Nature 436:201–206

Wang K, Kinoshita M (2013) Dangers of being thin andweak. Science 342:2013. doi:10.1126/science.1246518

Watada S, Kusumoto S, Satake K (2014) Traveltime delayand initial phase reversal of distant tsunamis coupledwith the self-gravitating elastic Earth. J Geophys Res119:4287–4310. doi:10.1002/2013JB010841

Wei S, Helmberger D, Avouac JP (2013) Modeling the2012 Wharton basin earthquakes off-Sumatra: com-plete lithospheric failure. J Geophys Res118:3592–3609. doi:10.1002/jgrb.50267

Yagi Y, Nakao A, Kasahara A (2012) Smooth and rapid slipnear the Japan Trench during the 2011 Tohoku-Okiearthquake revealed by a hybrid back-projectionmethod. Earth Planet Sci Lett 355–356:94–101.doi:10.1016/j.epsl.2012.08.018

Yamazaki Y, Cheung KF (2011) Shelf resonance andimpact of near-field tsunami generated by the 2010Chile earthquake. Geophys Res Lett 38:L12605.doi:10.1029/2011GL047508

Ye L, Lay T, Kanamori H (2012) The Sanriku-Okilow-seismicity region on the northern margin of thegreat 2011 Tohoku-Oki earthquake rupture.J Geophys Res 117:B02305. doi:10.1029/2011JB008847

Yokota Y, Koketsu K, Fujii Y, Satake K, Sakai S,Shinohara M, Kanazawa T (2011) Joint inversion of

strong motion, teleseismic, geodetic, and tsunamidatasets for the rupture process of the 2011 Tohokuearthquake. Geophys Res Lett 38:L00G21.doi:10.1029/2011GL050098

Yue H, Lay T (2011) Inversion of high-rate (1 sps) GPSdata for rupture process of the 11 March 2011 Tohokuearthquake (Mw 9.1). Geophys. Res. Lett. 38:L00G09.doi:10.1029/2011GL048700

Yue H, Lay T (2013) Source rupture models for theMw 9.02011 Tohoku earthquake from joint inversions of high-rate geodetic and seismic data. Bull Seismol Soc Am103:1242–1255. doi:10.1785/0120120119

Yue H, Lay T, Koper KD (2012) En échelon and orthogo-nal fault ruptures of the 11 April 2012 great intraplateearthquakes. Nature 490:245–249. doi:10.1038/nature11492

Yue H, Lay T, Rivera L, An C, Vigny C, Tong X, Báez SotoJC (2014) Localized fault slip to the trench in the 2010Maule, Chile Mw = 8.8 earthquake from joint inver-sion of high-rate GPS, teleseismic body waves, InSAR,campaign GPS, and tsunami observations. J GeophysRes 119:1–19. doi:10.1002/2014JB011340

YueH, Lay T, Li L, Yamazaki Y, Cheung KF, Rivera L, HillEM, Sieh K, KongkoW,Muhari A (2015) Validation oflinearity assumptions for using tsunami waveforms injoint inversion of kinematic rupture models: applica-tion to the 2010Mentawai Mw 7.8 tsunami earthquake.J Geophys Res 120:1728–1747. doi:10.1002/2014JB011721

Zhao D, Huang Z, Umino N, Hasegawa A, KanamoriH (2011) Structural heterogeneity in the megathrustzone and mechanism of the 2011 Tohoku-oki earth-quake (Mw 9.0). Geophys Res Lett 38:L17308.doi:10.1029/2011GL048408


Documents

University of California, Santa Cruz - Tsunamigenic Major and …thorne/TL.pdfs/LRL_Tsunamigenic... · 2016-07-21 · T Tsunamigenic Major and Great Earthquakes (2004–2013): Source