Highly variable coastal deformation in the 2016 MW7.8 Kaikōura earthquake reflects rupture complexity along a transpressional plate boundary

(1)

Citation for this paper:

Clark, K.J., Nissen, E.K., Howarth, J.D., Hamling, I.J., Mountjoy, J.J., Ries, W.F., …

Strong, D.T. (2017). Highly variable coastal deformation in the 2016 M

W

7.8 Kaikōura earthquake reflects rupture complexity along a transpressional plate

boundary. Earth and Planetary Science Letters, 474(September), 334-344.

http://dx.doi.org/10.1016/j.epsl.2017.06.048

_____________________________________________________________

Highly variable coastal deformation in the 2016 M

W

7.8 Kaikōura earthquake reflects

rupture complexity along a transpressional plate boundary

K.J. Clark, E.K. Nissen, J.D. Howarth, I.J. Hamling, J.J. Mountjoy, W.F. Ries, K.

Jones, S. Goldstien, U.A. Cochran, P. Villamor, S. Hreinsdóttir, N.J. Litchfield, C.

Mueller, K.R. Berryman, D.T. Strong

September 2017

©2017 The Authors. Published by Elsevier B.V. This is an open access article under

the CC BY-NC-ND license (

http://creativecommons.org/licenses/by-nc-nd/4.0/

).

This article was originally published at:

http://dx.doi.org/10.1016/j.epsl.2017.06.048

(2)

Contents lists available atScienceDirect

Earth

and

Planetary

Science

Letters

www.elsevier.com/locate/epsl

Highly

variable

coastal

deformation

in

the

2016

M

_W

7.8 Kaik ¯oura

earthquake

reﬂects

rupture

complexity

along

a

transpressional

plate

boundary

K.J. Clark

a

,

∗

,

E.K. Nissen

b

,

J.D. Howarth

a

,

1

,

I.J. Hamling

a

,

J.J. Mountjoy

c

,

W.F. Ries

a

,

K. Jones

a

,

S. Goldstien

d

,

U.A. Cochran

a

,

P. Villamor

a

,

S. Hreinsdóttir

a

,

N.J. Litchﬁeld

a

,

C. Mueller

a

,

K.R. Berryman

a

,

D.T. Strong

a

a_GNS_Science,_PO_Box₃₀_368,_Lower_Hutt_5040,_New_Zealand

b_School_of_Earth_and_Ocean_Sciences,_University_of_Victoria,₃₈₀₀_Finnerty_Road,_Victoria_B.C.,_V8P_5C2,_Canada c_National_Institute_of_Water_and_Atmospheric_Research,_Private_Bag_14901,_Wellington_6241,_New_Zealand d_School_of_Biological_Sciences,_University_of_Canterbury,_Private_Bag_4800,_{Christchurch,}_New_Zealand

a

r

t

i

c

l

e

i

n

f

o

a

b

s

t

r

a

c

t

Articlehistory:

Received1April2017

Receivedinrevisedform25June2017 Accepted27June2017

Availableonline20July2017 Editor:A.Yin

Keywords:

coastaldeformation multi-faultrupture Kaik ¯ouraearthquake lidardifferencing plateboundary

Coseismiccoastaldeformationisoftenusedtounderstandsliponoffshorefaultsinlargeearthquakesbut inthe2016MW7.8 Kaik ¯ouraearthquakemultiplefaultsrupturedacrossandsub-paralleltothecoastline.

Along_∼110kmofcoastline,arichdatasetofcoastaldeformationcomprisingairbornelidardifferencing, field surveyingand satellitegeodesy revealshighlyvariable verticaldisplacements,rangingfrom −2.5 to 6.5m. Theseinformarefined slipmodel forthe Kaik ¯ouraearthquake whichincorporateschanges tothesliponoffshorefaultsandinclusionofanoffshorereversecrustalfaultthataccountsforbroad, low-amplitudeupliftcenteredonKaik ¯ouraPeninsula.Theexceptionaldetailaffordedbydifferentiallidar andthehighvariabilityincoastaldeformationcombinetoformthehighest-resolutionandmostcomplex record ofcoseismic coastal deformationyetdocumented. Thisshould promptreassessmentof coastal paleoseismicrecordsthatmaynothaveconsideredmulti-faultrupturesandhighcomplexitydeformation fields.

1. Introduction

Sudden coastal uplift and its direct association with earth-quakes was ﬁrst documented by FitzRoy and Darwin in south-central Chile in 1835 (FitzRoy, 1839) and since the 1960s, mea-surements of coseismic coastal uplift and subsidence have been usedtounderstandthelocationandgeometryofthefaultsource, slipdistributionandsegmentationofmanyearthquakes(e.g.Awata etal., 2008; Briggsetal., 2006; Hayesetal., 2010; Plafker,1969; Taylor et al., 2008). Observations of coseismic coastal uplift also underpin the use of uplifted coastal geomorphology for deter-mining the timing and magnitude of paleoearthquakes on off-shore faults (e.g. Berryman et al., 2011; Rockwell et al., 2016; Shawet al., 2008). Paleoseismic applications ofcoastal uplift are typically predicatedon simple assumptions of single planar fault

*

Correspondingauthor.

E-mailaddress:K.Clark@gns.cri.nz(K.J. Clark).

1 _Now_at:_School_of_Geography,_Environment_and_Earth_Sciences,_Victoria

Univer-sityofWellington,POBox600,Wellington6140,NewZealand.

rupturesanddolittletoaccountforrupturecomplexity.Well doc-umented records of coastal uplift during earthquakes, especially complexruptures,provideanimportantcontextforevaluatingthe assumptionsthatunderpincoastalpaleoseismicstudies.

Extensive coastal uplift was one of the more apparent im-pacts of the 2016 MW7

.

8 Kaik ¯oura earthquake, which is one of themostcomplexearthquakeseverobservedwithinstrumentation (Hamling et al., 2017) (Figs. 1 and 2). There remain ambiguities in theslipmodelforthis eventthat are not yetresolved despite it being well captured by seismological and geodetic instrumen-tation, and satellite interferometric synthetic aperture radar (In-SAR)measurements ofgrounddeformation (Hamling etal., 2017; Kaiseretal.,2017).Ruptureofanunderlyinglow-anglethrustfault or the subduction interface is generally required to fit teleseis-mic observations(Duputel andRivera, 2017; Hollingsworthetal., 2017), andmodeling of two sizable regions of seafloor deforma-tion (one consistent witha subduction interface source) produce the best-fit to tsunami tide gauge data (Bai et al., 2017). Yet, geodetic and InSARmeasurements, consistent withfield observa-tions ofextensive and large (upto 12 m) fault surface ruptures,

http://dx.doi.org/10.1016/j.epsl.2017.06.048

(3)

Fig. 1. Tectonicsetting,surfacerupturesandupliftedcoastlineofthe2016MW7.8 Kaik ¯ouraearthquake.(a)PlatetectonicsettingoftheKaik ¯ouraearthquake,MFS:

Marlbor-oughfaultsystem,platemotionratesfromWallaceetal. (2012).(b)FaultsurfacerupturesoftheKaik ¯ouraearthquake(Litchfieldetal.,2017) andactivefaultsoftheregion (onshorefaultsfromLangridgeetal.,2016,offshorefaultscourtesyofNIWA,KPF:KaikouraPeninsulafaultasinferredbyBarrell,2015);onlyfaultrupturesnearthecoast andoffshorearelabeled.(2c)and(2d)denotesthelocationofthephotosshowninFig. 2cand2d.(c)ThecoastaltraceofthewesternstrandofthePapateafault(redarrow) atWaipapaBay.Whitetrianglesshowthelocationoffieldmeasurementsofcoastaluplift.Photoviewdirectionissouthward,photocredit:SteveLawson.(Forinterpretation ofthereferencestocolorinthisfigurelegend,thereaderisreferredtothewebversionofthisarticle.)

Fig. 2. ExamplesofcoastalupliftduetotheMw7.8Kaik ¯ouraearthquake.(a)UpliftedcoastalplatformbetweenthetwostrandsofthePapateafault(seelocationinFig. 3c),

thebrownalgaeformerlyinhabitedthesubtidalzone,verticaldisplacementherewas4.4±0.25 m.(b)Detailofthedecayingholdfastofbullkelp(Durvillaeaantarctica).

(c) UpliftedcoastlinenearGooseBay,herethedistinctivehorizontalwhitebandofbleachedcorallinealgaemarkstheformermeanlowwatermark;verticaldisplacement herewas1.5±0.27 m.PhotolocationismarkedinFig. 1b.(d)CoastalupliftnorthoftheWaimaRiver,verticaldisplacementherewas2.9±0.34 m.Photolocationismarked inFig. 1b.

indicate crustal faulting dominated the seismic moment release (Hamlingetal.,2017).Resolvingtheinvolvementofthesubduction interfaceandthe extentofoffshore faultrupturein theKaikoura earthquakehasimplicationsforfutureseismicandtsunamihazard onthesouthern Hikurangiplateboundary,andforunderstanding earthquakehazardsatanalogoustranspressionalplateboundaries globally.

Inthisstudywe presenta high-resolution recordofcoseismic coastal deformation that contributesto the understandingof slip on faults that cross the coastline, provides insights into offshore fault ruptures, and offers context for re-evaluating paleoseismic studies using coastal deformation as a primary dataset. We col-lected post-earthquakeﬁeld measurements ofcoastaluplift using displaced low-tidal biota, and mapped coseismic vertical change

(4)

usingpre- and post-earthquakeairborne lidar surveysthat cover 70% of the deformed coastline. Post-earthquake mapping of the freshrupture scarp along the offshore Needlesfault reveals sub-marine coseismic vertical offsets, which can be integrated with onshoreverticalcoastaldeformationtoconstrainslip. Lidar differ-encing, along with closely-spaced ﬁeld surveys, capture unprece-denteddetailofthecoastaldeformationassociatedwithacomplex earthquakeonatranspressionalplateboundary.

2. Background

The 14 November 2016 Mw7.8 Kaik ¯oura earthquake occurred in thenorth-eastern South Island of New Zealand atthe bound-ary between the Australian and Pacific tectonic plates (Fig. 1a). Thenorth-easternMarlboroughregionrepresentsatransitionzone from oblique continental collision in the south along the Alpine fault to subduction in the north along the Hikurangi subduction zone(Wallaceetal.,2012).TheMarlboroughFaultSystemconsists ofa seriesof dextralstrike-slipfaults that accommodate at least 80%ofAustralia/PacificPlatemotioninthenorthernSouthIsland, andit ispossiblethat theymaylinkatdepthwiththefar south-ern partofthe Hikurangisubduction zone (Wallace et al., 2012). Immediatelytothe southinNorth Canterburythefaults are pre-dominantlyreverseandtranspressional,withmuchlowersliprates (Litchfield etal., 2014) (Fig. 1b). Pleistocene marineterraces pro-videlong-termupliftratesrangingfrom

∼

1.1mm/yrto0.5mm/yr alongtheKaik ¯ouracoastline(Otaetal.,1996) whileHolocene ma-rine terraces, although observed along the same coastline, have only been documented at Kaik ¯oura Peninsula where there have beenatleast3uplifteventsinthepast3000years(Barrell,2015). TheKaik ¯ouraearthquakeinitiated nearWaiauinNorth Canter-buryat12:03amlocaltime atadepthof15kmwithanoblique thrust mechanism; therupture propagated south–west to north– eastover

∼

2 minandterminatedoffshoreinCookStrait(Kaiseret al.,2017) (Fig. 1a).Meter-scalesurfaceruptureoccurredonatleast fourteenfaults,including

>

10 mdextraldisplacementsonthe Kek-erengufault(Litchﬁeldetal.,2017),andatsunamiwithrunupof upto7mwasgenerated(Poweretal.,2017).

Inhistoric andrecentlarge earthquakes, coseismiccoastal de-formation hasbeen measured by a variety of methods, withthe diversity oftechniques, accuracy andspatial density of measure-ments generallyincreasingover time. Acommonfeature ofmost studiesofcoseismiccoastalupliftistheuseofdisplacedintertidal biozones to measure uplift. In tropical regions, coral microatolls areparticularlysensitiverecordersoftide levels(e.g.Briggsetal., 2006; Meltzneretal., 2006; Tayloret al.,2008) whilein temper-ate regions coralline algae, macroalgae and sessile molluscs are used (e.g. Awata et al., 2008; Bodin andKlinger, 1986; Jaramillo et al., 2017; Melnick et al., 2012a; Plafker, 1969). GPS observa-tionsare asourceofprecisepointmeasurements ofuplift(Konca etal.,2007;Subaryaetal.,2006)whilesatelliteimageryandInSAR can provide more complete spatial coverage (Hayes et al., 2010; Meltzner et al., 2006; Subarya et al., 2006). Following the 2010 MW8

.

8 Mauleearthquake (Melnick etal., 2012a) measureduplift bycomparing apre-earthquake lidarwithpost-earthquake differ-ential GPS tracks, which appears to be the ﬁrst record of using lidar todetect coseismiccoastal change.However, theavailability ofextensivelidarcoveragefrombothbeforeand aftertheKaik ¯oura earthquakeisunparalleled foramajorcoastalrupture.

3. Methods

3.1. Fieldmeasurements

Coseismiccoastal uplift was measured in theﬁeld at39sites using the post-earthquake elevation of algae that, prior to the

earthquake, livedup to andaround meanlow water(MLW) ele-vation (Fig. 2). Theupper limitof bullkelp(Durvillaeaantarctica)

andtheassociatedbandofcorallinealgaewasusedatshorelines experiencing highwave energy, whereas the upper limit of Car-pophyllummaschalocarpum andassociatedcorallinealgaewasused inareasoflowerwave energy.Surveyingwasundertaken usinga level, tripodandstaff to obtainthe difference inheight between theMLWmarkersandthetidelevelatthetimeofsurveying.Tide levelatthetimeofsurveyingwasthencorrectedrelativetomean sealevel.Thesurveypointswerenotputintoaglobalpositioning projection orverticaldatumasalllocalgeodetic benchmarkshad movedintheearthquake.Thenumberofsurveypointscollectedat each siteranged from1–21(mean

=

11).Surveyingcommenced ﬁve daysafter the earthquake andall points (except Lake Grass-mere)werecollected within2.5weeksoftheearthquake.Further detailsonthemethodologyofusingbiologicalmarkerstomeasure coastal deformation can be found in Supplementary Material A. The uncertainty on mean uplift measurements from each site is reported as a 95% conﬁdence interval estimated from propaga-tion ofthe systematicand statisticaluncertainties by summation inquadrature(SupplementaryMaterialA,TextA1).

3.2. Lidardifferencing

TheKaik ¯ouracoastlinewassurveyedwithairbornelidarinJuly 2012 and on 19th–21st November 2016 (details of lidar acqui-sition are in Supplementary Material A, Text A2), resulting in a

∼

0.5–4 km-wide,

∼

90km-longcoastal stripofrepeat, sub-meter resolutiontopographythat extendsfrom

∼

4.5yearsbeforeto

∼

1 weekaftertheKaik ¯ouraearthquake(Fig. 3a).Therewereno signif-icantearthquakesintheareaofjointcoveragebeforetheKaik ¯oura mainshock and the largest regional events, the 2013 Cook Strait sequence(Mw5

.

7,5.8and6.6),produced

<

5mmofvertical defor-mation intheKaik ¯ouraregion (Hamling etal., 2014).Over wave-lengthsofhundredsofmeters,differencesbetweenthetwo“bare earth” (classiﬁed groundreturn) datasets thereforereﬂect coseis-mic and earliest post-seismic deformation of the 2016 Kaik ¯oura earthquake. At shorterwavelengths, the paired lidar surveysalso captureerosionalanddepositional processes,includinglandslides androckfallstriggeredbytheKaik ¯ouraearthquake.

Toinvestigateverticaldeformationacrosstheareaofjointlidar coverage,wesubtracta1m-pixel2012lidardigitalterrainmodel (DTM)fromtheequivalent2016DTM.Inneglectinglateralmotions ofuptoseveralmetersintheearthquake,thissubtractiondoesnot yieldthetrueverticaldisplacement;rawelevationchangesare ad-ditionally inﬂuenced by local horizontal displacement magnitude andazimuth, slopeaspect andangle,and surface roughness(e.g.

Oskinetal.,2012).However,restrictingtheanalysistoﬂat,smooth partsofthelandscapelimitstheseeffects,suchthatthesimple el-evationchangecloselyresemblestheactualverticaldisplacement. Thereforeweeliminatedareaswithslopesof

>

5◦frombothDTMs anddifferencedonlythoseoverlappinglow-slope regionsthat ex-ceedacertaincut-offinsurfacearea.Thisreducedlocalizedscatter inelevationchangestotypically

∼

1 mforsmallareasof

>

10 m2, and

∼

0.5 m for larger areas of

>

2000 m2 ₍_{Fig. 4}_). _In _order _to limit the effects of erosionalanddepositional processes, we also removedriverbeds,beachesandlandslidesfromtheanalysis.

To verify that the remaining elevation change values are un-biasedby horizontalmotions,we independentlycomputed three-dimensional(3-D)surfacedisplacementsatafewdiscretelocations by applying the Iterative Closest Point (ICP) algorithm to win-dowed subsetsof the2012 and2016point clouds(Nissenetal., 2012) (further details on the ICP methods are inSupplementary Material A, TextA3). East–westandnorth–southICPcomponents accountforbroad-scalefeaturesofthehorizontaldeformationﬁeld including lateralslipacross themain surface ruptures,and

(5)

verti-Fig. 3. Mapsofcoastalverticaldisplacementoccurringinthe2016Kaik ¯ouraearthquake.(a)CoastlineimpactedbyKaik ¯ouraearthquake,dashedblacklineshowstheextent ofoverlappingpre- andpost-earthquakelidarcoverage,withinwhichthecolor shadedareasrepresentthe0–5degreeslopesoverwhichwemeasured2016–2012vertical displacement.Thedarkbluedashedlinesdelineatethehingepointsbetweenupliftandsubsidence.(b)–(f)Detailedmapsofareasofinterest,particularlyaroundfaultsurface ruptures.(Forinterpretationofthereferencestocolorinthisﬁgurelegend,thereaderisreferredtothewebversionofthisarticle.)

calICPcomponentsare incloseagreement withnearby elevation changeandfieldmeasurements(Fig. 4).Thisconfirmsthatfiltering ofthelidar elevationchangesforlowslopeangles (

<

5◦)is effec-tualineliminatingpotentialbiasingfromhorizontaldeformation.

3.3.Offshoredataandelasticdislocationmodeling

RuptureofoffshorefaultsincludingtheHundalee,Papatea, Nee-dles andnewly identiﬁed Point Kean fault (Fig. 1b) was mapped ontwopost-earthquakemarine voyagesusingKongsbergEM2040 andEM302multibeam systems(the detailedmethods ofoffshore datacollectionareinSupplementaryMaterialA,TextA5).Allthese faults show clear surface traceson the seaﬂoor and can be pin-pointedasbeingco-seismic withthe 2016earthquake from pre-andpost-earthquakesurveysorthepresenceofscarpsinthe shal-low water mobile sediment zone. In the case of the Point Kean fault which occurs on the outer shelf in

∼

50 m water depth, no knowledge of the Point Kean fault existed before the earth-quake and therefore the co-seismic movement is uncertain. The offshoreHopefaultwasalsore-surveyedinseverallocationsusing a TOPAS subbottom proﬁler butno change was found compared

tothepre-earthquakebathymetry.Thesedataindicatethatno off-shore ruptureoccurred onthe Hope fault.Verticaloffsets onthe 2016rupturescarpoftheNeedlesfaultweremeasuredat approx-imately 1 km intervals inArcGIS using fault normal bathymetric proﬁles, the offsets are assumed to equal the coseismic vertical slip(SupplementaryMaterialA,TableA5).

The slip model for the earthquake was developed using the methods described in Hamling et al. (2017). Their initial fault model incorporates InSAR displacements in the satellite line-of-sight,campaignandcontinuousscatteredGPSoffsetsandalimited numberofﬁeld measurements ofcoastal uplift.Ourreﬁned fault geometryisbased ontheir modelbutincorporatesthenewlidar measurements of coastal deformation and offshore fault vertical slipdata.

3.4. Comparisonbetweencoastaldeformationmeasurements

TheKaikouraearthquakecoastaldeformationdatasetis primar-ilyderived fromestimates ofuplift produced by two techniques, lidar differencing and ﬁeld measurements. It is important to as-sess the agreement between the two techniques because they

(6)

Fig. 4. (a)Field,lidar,andgeodeticmeasurementsofcoseismicverticaldeformationprojectedontoastraightlineofazimuth35◦thatroughlyparallelstheKaik ¯ouracoastline, centered(x=0)atthePapateafault(theendpointsoftheprojectionplaneareshownonFig. 3a).Verticalbarsindicate95%conﬁdenceboundsontheﬁeld,ICP,andGPS measurements(thelatterarebarelyvisibleatthisscale);verticaldashedlinesindicatelocationsofmajorfaultsatthecoastline.ThecolorscaleisthesameasFig. 3.Colored lidarpointsrepresentdifferentialverticalmovementcalculatedfromoverlappingsurfaceareas>2000 m2_;_the_grey_points_represent_surface_areas_>_{10 m}2_._{(b) Detail}_across

theHundaleefault.ThehorizontalbarsaroundtheICPdatapointsindicatethemeasurementapertureof∼1kmforthismethod(seeSupplementaryMaterialText A3). (c)Coastline-perpendicularupliftproﬁleshowingtiltingacrosstheKaik ¯ouraPeninsula.Here,theproﬁleazimuthis125◦andtheprojectioncenter(x=0)isatKaik ¯oura township.(d)Detailedcoastline-parallelupliftmeasurementsfromacrosstheHopeandPapateafaults.

cover different areas of coastal deformation along the rupture extent and systematic bias between the approaches could com-promiseinterpretations. We have statisticallyassessed the agree-ment between the main sources of coastal uplift data and find almost no bias between the field measurements and lidar dif-ferencing results at fourteen coincident sites, this indicates suf-ficient agreement to treat them as equivalent for the purposes of interpretation (Supplementary Material A, Text A6, Fig. A6). OnlythreecampaignGPSsiteslie withinthe areaoflidardouble coverage, but hereagain the agreement is within the 2

σ

uncer-tainties (Supplementary Material A, Fig. A7). The campaign GPS sites were measured within 1 week ofthe Kaik ¯oura earthquake, and prior to the earthquake they showed no signiﬁcant vertical movement (

<

−

2 mm/yr). Therefore, the campaign GPS and li-darshouldbemeasuringequivalentamountsofcoseismicvertical

movement.TheagreementbetweencampaignGPSandlidar, com-bined withtheobservationthatﬁeld andlidarmeasurementsare equivalent, demonstrates the robustness of the coastal deforma-tion datasetalongtheentire110kmofcoastlineresolved inthis study.

The correspondence between the ﬁeld, lidar and GPS uplift measurementswithsatelliteradar-basedestimatesismorevariable (Fig. 5).Inaddition totheir elasticdislocation modeling,Hamling et al. (2017) combine smooth line-of-sight displacements (from SAR interferometry)withnoisierhorizontaloffsets(fromSAR am-plitude pixelcorrelation)tomaptheregionaldeformation ﬁeldin three dimensions. Resulting vertical displacements reproduce the large-scale features ofthe other coastal uplift datasets, including broadswellsofupliftcenterednorthoftheHundaleefault,atthe Papateafault,andnorthoftheKekerengu fault(Fig. 5a). However,

(7)

Fig. 5. (a)Comparisonbetweenradar-derived(InSARplusSARpixeloffset)estimatesofcoastalverticaldeformation(smallblackdots)withtheﬁeld,lidar,andgeodetic measurementsshowninFig. 4a(hereshadedingrey).AswithFig. 4a,alldataareprojectedontoastraightlineofazimuth35◦thatparallelstheKaik ¯ouraDistrictcoastline (Fig. 3a).Theradar-derivedupliftpointsarefroma∼3 km-wideswathalongthecoastline.(b)DetailfromacrosstheHundaleefault.(c)Coastline- perpendicularproﬁle acrosstheKaik ¯ouraPeninsula(azimuth125◦,projectioncenteratKaik ¯ouratownship,asinFig. 4c).(d)DetailfromacrosstheHopeandPapateafaults.

ﬁne details such as slip across discrete faults – readily apparent inthelidarelevationchanges –areobscured inthenoisierradar upliftestimates(Fig. 5b,d).Thisindicatesthatalthoughtheradar measurementsspanafarbroaderregion(

>

100km)thanthe nar-rowcoastallidarswath,andnominallycapturedeformationin3-D, theycannotresolveupliftpatternsatthesameﬁnespatialscaleor withthesameprecisionasthelidar.

4. Coastaldeformation:south-westtonorth-east

Maps of coastal vertical displacement (Fig. 3) and projection ofcoastal deformation measurements onto a plane runningSW– NEalong the coastline (Fig. 4a)reveal a high level ofvariability over a range of length scales. Here we describe the coastal de-formationfromsouth-westto north-east,in thedirectionof rup-turepropagation,andrelateupliftmeasurementstofault distribu-tion.

4.1. Hundaleefault

The Hundalee fault marks the south-western limit of signiﬁ-cant coastal uplift; south-west ofthe fault only minor(

<

0.5 m) coseismic subsidence and uplift occurred (Figs. 3b, 4b). There is 1.2 m of vertical offsetat the main trace of the Hundaleefault; two minorfault traceswith

<

0.4 m of vertical displacement oc-cursouthoftheHundaleefault.Twopeaksincoastalupliftoccur north–eastof theHundalee fault:a peak of2.2 mat0.1 kmNE, and1.6m at1.8 km NE;thisshort-wavelength variabilityattests to distributed fault zone deformation on- and offshore (Fig. 4b). North–eastofGooseBayupliftgraduallydecreasesbutnotablyno distinct verticaloffsetis seenacross thecoastal projectionofthe N–SstrikingWhitesLineament,afaultthathasbeenidentiﬁedby InSAR andestimatedtohaveslip of1–2 mat depth(Hamling et al., 2017), and seen to have intermittentsurface rupture

>

3 km inland(Litchﬁeldetal.,2017) (Fig. 1b).

(8)

4.2. Kaik ¯ouraPeninsula

Upliftof0.8–1misrecordedfrom4kmsouthto16kmnorth oftheKaik ¯ouraPeninsula(Figs.3a,4a). Therelativelywide range (

∼

1 m) in differentiallidar upliftmeasurements acrossthis area reflects thewider inland extent ofjointlidar coverage combined withasmalldegree ofcoseismic landward tilting(Fig. 4c)which producesscatterwhenthepointsareprojectedtoaplane(Fig. 4a). Thereisarelativelyuniformupliftprofileof0.8–0.95 macrossthe Kaik ¯oura Peninsula (Fig. 4c), although slightlylower field and li-dar measurements on the southern coast show a degree of tilt, downtothesouthwest,acrossthePeninsula.TheKaik ¯oura Penin-sula fault has been inferred offshore of the peninsula by Barrell (2015)basedonthedistributionofHolocenemarineterraces,and thePointKeanfault(Fig. 1b) hasbeenidentified northeastofthe peninsula basedon seafloor scarps anda region of folded strata (erodedstrike ridges).The location ofthePoint Keanfault in re-spect to the broad, low amplitude coastal uplift seen along the

∼

25 kmstretchofcoastline surroundingKaik ¯ouraPeninsula, and thetentativelyidentiﬁedfreshrupturescarp, suggestsit couldbe acandidateforexplainingtheonshoredeformation,thisisfurther exploredinthediscussion.

4.3. HopefaulttoPapateafault

The 13 km stretch of steep and rocky coastline from 3 km south–west of the Hope fault to the Papatea fault displays large uplift values(mostly

>

2 m) andhigh variabilityrelated tothree surface fault ruptures, along with possible contributions from nearshorefaultsthatmayrunsub-parallelwiththecoast(Figs. 3e, 3f,4d).

Thereisadistinct 1mincrease incoastalupliftfrom2kmto 0.5kmsouth–westoftheHope faultandthenadecreasetoward thesurfacerupturetrace(Fig. 4d).Fieldmeasurements ofthe sur-face trace of the Hope fault, atHalf Moon Bay, showed

∼

0.4 m verticaloffset(uptotheNW)andnegligibledextralslip,although no other surface rupture was seen along the onshore Hope fault and resurveys of known traces of the offshore Hope fault show no new rupture. Minor offset (

<

0.5 m) in the lidar differencing measurements across the surface rupture is consistent with the ﬁeld observations. Slip modeling by Hamling et al. (2017) sug-gestedlocalized slip (predominantly strike-slip) of almost 3–4m intheupper3kmoftheseawardpartoftheHopefaultbutnoted thatlimiteddataandlocalinelasticeffectsmakethispoorly con-strained.ThustheroleoftheHopefaultintermsofcoseismicslip intheKaik ¯ouraearthquakeispoorlyconstrained,butofparticular interest,becauseifitdidnotslipintheearthquakethenorthward propagating surface rupture would havelargely jumped over the highestslipratefaultintheMarlboroughfaultsystem.Coastal up-lift south ofthe Hope fault suggests slip on a blind fault in the areabetweentheHopefaultandtheMangamaunufaults (Barrell, 2015) (Supplementary Material A,Fig. A8), andis therefore con-sistent withthegeneral locationof modeled slip(Hamling etal., 2017), butnotnecessarilywithit beingonthemain traceofthe Hopefault.

North–east of the Hope fault the coastal uplift is consistent at2–2.5 m for

∼

3km along the coast before increasing sharply within1kmofPaparoaPoint(Fig. 3f).AcrossPaparoaPointthree minorfaultscarpsof thePaparoaPointfault strikesubparallelto thecoast andhavevertical offsets of0.3–1 m,each upthrownto theNW. Therelationship betweenthe PaparoaPointandPapatea faults is not clear but the complex rupture pattern of the off-shore Papatea fault includes NNE-striking splays consistent with thestrikeofthePaparoaPointfaulttraces(Fig. 3e).

Between the Paparoa Point and Papatea faults, coastal uplift generally decreases (Fig. 4d). Minor variations on the order of

<

1 m of uplift suggest either slip on nearshore faults or vary-ingslipdistributionacrosstheoffshore,southwest-dippingPapatea fault. The Papatea fault splays into two strands at the coastline and thepopup block betweenthe strands recordsthe maximum coastalupliftfortheKaik ¯ouraearthquakeof6

.

6

±

0

.

5 m(Figs.3e,

4d). Verticaloffsetacrossthewestern faultstrandis 3.5mupto the east,andacross theeastern strandofthe Papateafault there is 4 m of vertical offset, down to the east. The relatively wide range (1–1.5 m) oflidar andﬁeld measurements across the Pap-ateafaultisduetotiltoftheupliftedfaultblocktowardtheproﬁle projection line(Fig. 4d).Halfa kilometer north–east ofthe west-ern strandofthePapateafault,anotherminorfaulttraceshowsa

∼

0.5 mverticalstep(downtoNE);furthernorth–eaststill,coastal upliftsteadilydecreases.

4.4. Kekerengu–Needlesfault

This earthquake has shown that the Kekerengu and Needles faultsarelikelytobethesamecontinuousfault.Offshore,thefresh rupture scarp of the Needles fault has been mapped to within 600 m of where the Kekerengu fault rupture meets the coast (Fig. 3a). Localized subsidence of up to 2.5 m over a width of 200–400 moccursonbothsidesoftheKekerengu fault(Figs.3d,

4a).Thismaybeduetointerplaybetweenobliquenormalbehavior of theKekerengu fault atthiseasterly-striking bendonthe fault, contributions fromminorfaults southofthemain trace(Fig. 3d), downthrow of the footwall of the Tinline Downs fault, and soft sedimentcompactioninthestreamvalley.

Theamountoflocalizedsubsidencemeasuredaroundthe Kek-erengu faultby lidardifferencing issurprisinglylarge,butitisof an order ofmagnitudegreaterthan thevertical uncertainty typi-caloflidar.Fieldobservationsalsoprovideevidenceofsubsidence: immediatelyaftertheearthquake,asignificantareaofponded wa-terappearedonbothsidesoftheKekerengufaultwheretherehad formerly beenonly a shallowstream (SupplementaryMaterial A, Fig. A9). Ponded waterstill persists at7 months afterthe earth-quake indicating the stream base level has permanently shifted. Satellite radar and optical imagery-based models of ground de-formation(Hamlingetal.,2017; Hollingsworth etal.,2017),have not observedthelocalized subsidencedespiteits significant mag-nitude; this demonstrates the advantage of lidar differencing in termsofcapturingfine-scalefaultzonedeformationandsecondary effectsoffaultrupture.

NorthoftheKekerengu fault,andinboardoftheNeedlesfault, there isachangetocoastal upliftontheorderof2–3m.From a peakof2.9mneartheWaimaRiver,thereisanorth–eastward de-creasein coastaluplift to0.4m atCapeCampbell (witha 0.5m stepacross theLighthousefault,2.5kmsouthofCapeCampbell). Thisnorth–eastwarddecreaselargelymimicsthescarpheightson the fresh rupture trace of the submarine Needles fault and is a function of both diminishing slip on the Needles fault (Supple-mentaryMaterialA,TableA5)andincreasingdistancebetweenthe coastlineandtheNeedlesfault(Fig. 3a).

5. Discussion

The coastal deformation record of the 2016 Kaik ¯oura earth-quake is primarily characterized by high spatial variability; no record of comparable complexity has been documented in re-cent times. The majority of recent coseismic coastal deforma-tion observationsare relatedto subduction earthquakesin which the deformation is characterized by arc-parallel zones of uplift and subsidencewithlong wavelengthvariability (

>

10’s kms) re-lated to heterogeneous slip on a single principal fault, and, less commonly, short wavelength variability due to splay fault rup-ture(e.g.Briggs etal.,2006; Melnick etal.,2012b; Plafker, 1969;

(9)

Fig. 6. (a)BestfittingslipmodelfortheKaik ¯ouraearthquakebasedoninversionofgeodeticandcoastaldeformationdata(seesection3.3fordetails).Momenttensors showtheUSGSCMTsolutionandtheequivalentfortheHamlingetal. (2017)modelandthemodelproducedforthisstudy.Heavyblacklinesdenotethetopedgeofthe faultsurface.NotethePapateafaultisnotincludedinthismodel,seeHamlingetal. (2017)fordiscussion.(b)Fitbetweenthemodeled verticaldisplacementandvertical displacementalongthe2016rupturescarpoftheNeedlesfault.(c)Fitbetweenthemodeled verticaldisplacementandcompiledverticaldisplacementmeasurements(see keyinFig. 4)fromtheConwayRivertonorthofKaik ¯ouraPeninsula.Thisplotdemonstratesimprovedfitusingtherefineddislocationmodelofthisstudycomparedtothe modelofHamlingetal. (2017)whichomittedacrustalfaultoffshoreofKaik ¯ouraPeninsula.

PlafkerandSavage,1970; Subarya etal.,2006).Coastaluplift ob-servations followingcrustal earthquakes hasusually been associ-ated with nearshore reverse faulting, and the observations have alsotypicallybeenconsistentwithsinglefaultruptures(e.g.Awata etal., 2008; Meghraouietal.,2004).Variablecoastalupliftinthe 2007MW7

.

0 Haiti earthquake reﬂected the complexity of multi-faultrupture,albeitonalesserscalethanKaik ¯ourawithmaximum upliftof0.65 m(Hayesetal.,2010).Thecomplexcoastal deforma-tionrecord presented inthis studyhasimplications forboth the fault source model of the Kaik ¯oura earthquake and paleoseismic reconstructionsofpastearthquakes.

5.1.Reﬁnedslipmodelandimplicationsforinvolvementofthe subductioninterface

Thehigh-resolutionrecordofcoastaldeformationpresentedin thisstudyenablesa morecompletepicture ofoffshorefault rup-ture, slip distributions, and tsunamigenesis associated with the earthquake.TheslipmodelfortheKaik ¯ouraearthquakedeveloped by Hamling et al. (2017) exploited some ﬁeld measurements of coastaluplift,butthegreatlyincreasedresolutionandcontinuityof thecoastal deformationrecord presented herewarrants reassess-ment of some offshore components of the slip model. Here we focusontwoareas,theNeedlesfaultandoffshoreKaik ¯oura Penin-sula, where new data contribute to reﬁning the earthquake slip model(Fig. 6a).

Alongthe Needlesfault, the precise post-earthquakemapping ofthe2016rupturescarpinshoreofthepre-earthquakefaulttrace resultinthemodeled faultplaneshiftingfurtherinshore,andthe amountof reverseslip isreduced by

∼

1–2 m. This isa rare ex-ample (cf. Escartín et al., 2016; Fujiwara et al., 2011) of where measurementofcoseismicverticalslipatthescarpofasubmarine rupturecanbe usedtoconstraina slipmodelofanoffshorefault (Fig. 6b).

The low-amplitude (

∼

1 m), broad wavelength uplift of the coastsurroundingtheKaik ¯ouraPeninsulawasnotpreviously well-resolved by radar (Fig. 5) andso the geometryof uplift deﬁned by lidar allows development of a more reﬁned fault model for offshoreKaik ¯ouraPeninsula.PreviouslyHamlingetal. (2017) mod-eled uplift of Kaik ¯oura Peninsula along a linear extension of the

Hundaleefaultthatstopped

∼

3kmNEofthePeninsula.However, marine surveying carried out following the Kaikoura Earthquake indicates that surfacerupture oftheHundaleefault probably ter-minatesneartherimoftheKaik ¯ouraCanyon(Fig. 1).Astructure identiﬁedontheoutershelf(thePointKeanfault)hasbeentested to see ifit could be responsible for the uplift extending 20 km north of Kaik ¯oura Peninsula. A NE-striking reverse fault, dipping 35◦ with

∼

3 mofpredominantlyreverseslipextendingtodepths of 20 km is added to the slip model (Fig. 6a). This fault plane continues for

∼

20 km north–eastof Kaik ¯oura Peninsula and the rupturewouldhavedisplaced abroad areaofthe shelf.Our pre-liminarytsunamimodelusingtherevisedoffshorefaultgeometries indicatethereversefaultprovides asigniﬁcantlybetterﬁtforthe 24 min,

∼

2 m drawdown of sea-levelobserved on the Kaik ¯oura tide gauge(Supplementary MaterialA,Fig. A10) comparedto the modelprovidedinHamlingetal. (2017)whichcouldnotreplicate theinitialprolongeddrawdownatKaik ¯oura.

It currentlyremains debated asto whetherthere was slip on thesouthernendoftheHikurangisubductioninterfaceduringthe 2016Kaik ¯ouraearthquake(FurlongandHerman,2017);generally, models using teleseismic data converge on a signiﬁcant compo-nent ofslip onthe subduction interface, while geodetic observa-tionspointtodominantlycrustalfaultingwithlimitedsubduction interfaceinvolvement.Hollingsworthetal. (2017)useoptical satel-lite imagery and inversion of teleseismic waveforms to develop a two-fault modelof the Kekerengu fault and a deeper shallow-dipping fault (possibly the subduction interface) underlying the Kaik ¯ouracoastline.Similarly,DuputelandRivera (2017)developed a four-point source inversion using teleseismic waves and iden-tiﬁed a shallowdipping thrust fault andsuggestedthis could be thesubductioninterface,oraforearcthrustfault.Baietal. (2017)

useiterativeforwardmodeling oftidegaugerecordsand teleseis-mic waves to model two regions of seaﬂoor deformation in the Kaik ¯ouraearthquake.Oneregioncorresponds totheNeedlesFault andtheother,northofKaik ¯ouraPeninsula,corresponds toa shal-lowdippingfaultwhichtheyinferrelatestoanMW7

.

6 ruptureof thesubductioninterface. Incontrast,usinganinversionof geode-tic data (InSAR and GPS) and coastal uplift ﬁeld measurements,

Hamlingetal. (2017)modeled slipof

∼

4 montheplateinterface inlandofKaik ¯oura,although theynotethecontributionofthe

(10)

in-terfacesourceto theseismicmoment isminor(15%iflimitedto anMW7

.

9)comparedwiththecrustalfaults.Furthermore,amodel withonlycrustalfaultsourcesprovidesasimilarlevelofﬁttothe geodeticdata.

Thecoastaldeformation recordpresentedinthisstudyis con-sistent with an offshore reverse fault northeast of the Kaik ¯oura Peninsula, and although the geometry at depth is poorly con-strained, the pattern of coastal uplift is more consistent with a shallowsource.MaximumupliftoftheKaikouraPeninsulareaches

∼

1 mbutdropstoclosetozeroovera4kmlengthscale(Fig. 4c). Atthislocation,thedepthtothe interfaceis

∼

19km–toodeep to explain such shortwavelengthvariations. In thereﬁned dislo-cationmodelpresentedhere(modiﬁedfromHamlingetal.,2017;

Fig. 6a), we still requireslip on thedeeper portion of the inter-face source to explain the observed subsidence inland. However, theaddition of thenew crustalfault source offshore ofKaik ¯oura Peninsula, which is predictedto have slipof

∼

3 m, reducesthe amountofsliprequiredontheinterface to1–2m,whilestill ﬁt-tingtheobservedupliftpattern(Fig. 6c)andglobalmomenttensor (Fig. 6a). Thedislocation modelofHamlingetal. (2017)that had upto4mofslipontheplateinterface,withoutanoffshorefault, doesnot ﬁtthe distinct decrease incoastal uplift south–west of Kaik ¯ouraPeninsula(Fig. 6c).

Currently, thestrongest evidence ofplateinterface slip inthe Kaik ¯oura earthquake comes from Bai et al. (2017) and although their fault modelﬁts theteleseismic andtide gauge datawell, it does not entirely reconcile with the coastal deformation record compiledinthisstudy.Compatibilitywiththecoastaldeformation record isachieved inshoreof theKekerengu–Needles fault where up to3 mof upliftis modeled andatKaik ¯oura Peninsula where 1 m ofupliftismodeled. However, southwestofKaik ¯oura Penin-sula a low-amplitude subsidence signal (largely driven by subsi-dence at the downdip end of the subduction interface rupture patch) contrasts to the coastal deformation record which shows upliftalong mostofthecoastlinefromKaik ¯ouraPeninsula tothe Hundaleefault(Figs.3a,4a).Baietal. (2017)modelabroadarea ofupliftnortheastofKaik ¯oura Peninsula,whichisofsimilar am-plitudetothecoastaldeformationrecordforapproximately15km northeastof thePeninsula. However, the modeled maximum up-lift of 0.9 m at 20 km northeast of Kaik ¯oura Peninsula occurs wherewemeasure2–2.5mofuplift.Furthermore,theirmodel ex-tendsupliftto40kmnortheastofKaik ¯ouraPeninsula,contrasting withmeasuredcoastalsubsidencenortheastofthePapateafault,at 30 kmnortheastofKaik ¯ouraPeninsula.Someofthesedifferences areexplainedbysliponthePapateafault,whichBaietal. (2017)

acknowledgeisnot includedasfaultsource(nor isitincludedin the dislocation model we present in Fig. 6a). To summarize, the 3-fault model of Bai et al. (2017), which includes up to 6 m of subduction interface slip, ﬁtsthe tide gauge andteleseismicdata wellbutareasofmismatchtothecoastaldeformationrecord indi-categreatercomplexityinoffshorefaultingthantheyaccountfor; wesuggestthatuntilthesedatasetsarereconciledtheamountand locationofsliponthesubductioninterfaceremainsuncertain.

5.2. Recognitionofmulti-faultrupturesinthecoastalpaleoseismic record

The complexity of coastal deformation inthe Kaik ¯oura earth-quake raises potential pitfalls in mapping the extent of past co-seismic uplift by along-coast continuity. Raised Holocene coastal geomorphology(e.g. marine terraces,beach ridges, tidalnotches) is often used to record past earthquakes (e.g. Merritts, 1996; Rockwell et al., 2016). Along-coast continuity of uplifted coastal geomorphologyand/or agecorrelationaretypicallyusedtoassign synchronicity ofuplift to features and thereby attribute them to asingle earthquake(e.g.Hsieh andRau,2009; Shawetal., 2008).

Thelengthandamountofcoastalupliftcanbeusedtoinfer pale-oearthquakemagnitudes(Berrymanetal.,2011; Ramosand Tsut-sumi,2010).

IftheKaik ¯ouraearthquakeweretobereconstructedfrom geo-logical evidence, we would likelyinfer 3–4 separate earthquakes based on the highpoints of coastal uplift. Based on comparison with global historical observations of coastal uplift in compara-tively simple fault ruptures, one marine terrace would likely be correlated with one planar offshore fault rupture, and therefore thepaleoearthquakemagnitudewouldalsobeunderestimated.Age correlation wouldnotgreatlyhelp inresolving synchronousuplift on spatially separated terracesbecause theresolution of conven-tionaldatingmethods(e.g.radiocarbon)couldnotdistinguish be-tween two spatially distant marine terracesuplifted in the same earthquake,asopposedtotwoormoreearthquakescloselyspaced intime.TherecordofhighlyvariablecoastalupliftintheKaik ¯oura earthquake should motivate re-thinking of coastal paleoseismic records togive greater consideration tomulti-fault ruptures, par-ticularly atcomplex plate boundaries. Such re-assessment of the ability ofvarious paleoseismic methods to determinepast occur-renceofcomplexmulti-faultearthquakesisakeylessonfromthe 2016MW7

.

8 Kaik ¯ouraearthquake.

6. Conclusions

The 2016 Kaik ¯oura earthquake is one of the most complex earthquakes ever recorded (Hamling et al., 2017), and the high variability in vertical deformation along 110 km of coastline re-flectstherupturecomplexity.Sharpchangesincoastaldeformation arerecordedatthesurfacerupturesoftheHundalee,Papateaand Kekerengu faults, whileloweramplitude variabilityincoastal up-lift reflects distributed deformation near the major fault surface ruptures,minorfaultsurfacerupturesandsliponnearshorefaults subparallel to the coastline. A high resolution dataset of coastal deformationcontributestobetterconstraintonoffshorefault rup-tures. In this case, broad uplift of the Kaikoura Peninsula area, precisely defined by lidar differencing but not well resolved by satellite geodesy, reveals the involvement of an offshore reverse faultintheforearccrust.Thehighlyvariablenatureofthecoastal deformation associatedwiththeKaik ¯ouraearthquakeshould lead to the re-examination of how uplifted coastal geomorphology is usedinpaleoseismicstudiestointerpretpastearthquakesand in-formseismichazard. Multi-faultrupturescenarios maybe under-represented in paleoseismology due to the prevalence of rela-tively simple fault sources in historic examples of coastal defor-mation.

Acknowledgements

Funding for the post-earthquake coastal uplift ﬁeld survey and subsequent analysiswas provided by GeoNet, withthe sup-port ofits sponsors New Zealand Earthquake Commission (EQC), GNSScience,andLand InformationNewZealand,andMinistryof Business, Innovation and Employment (MBIE) response funding, provided through the Natural Hazards Research Platform (grant 2017-GNS-01-NHRP), andGNS Science MBIEstrategic science in-vestment funding(GNS-SSIF-TSZ). Land InformationNew Zealand, New Zealand Transport Authority,Environment Canterbury, Marl-borough District Council, Aerial Surveys Ltd and AAM NZ Ltd are thanked for providinghigh quality lidar data.The 2012lidar data (http://dx.doi.org/10.5069/G98C9T67) and derived products arehostedanddisseminatedbytheOpenTopographyFacility(http: //www.opentopography.org/) with support from the US National ScienceFoundation underNSFAwards1226353and1225810. Bil-janaLukovicandDaveHeronarethankedforreadilyprovidingGIS

(11)

Supplementarymaterialrelatedtothisarticlecanbefound on-lineathttp://dx.doi.org/10.1016/j.epsl.2017.06.048.

References

Awata,Y.,Toda,S.,Kaneda,H.,Azuma,T.,Horikawa,H.,Shishikura,M.,Echigo,T., 2008.Coastaldeformationassociatedwith the2007 NotoHantoearthquake, centralJapan,estimatedfromupliftedandsubsidedintertidalorganisms.Earth PlanetsSpace 60,1059–1062.

Bai,Y.,Lay,T.,Cheung,K.F.,Ye,L.,2017.Tworegionsofseaﬂoordeformation gener-atedthetsunamiforthe13November2016,Kaikoura,NewZealandearthquake. Geophys.Res.Lett.http://dx.doi.org/10.1002/2017GL073717.

Barrell,D.J.A.,2015. GeneralDistributionandCharacteristics ofActiveFaultsand FoldsintheKaikouraDistrict,NorthCanterbury.GNSScienceConsultancy Re-port2014/210,EnvironmentCanterburyReportNo.R15/23.59pp.

Berryman,K.,Ota,Y.,Miyauchi,T.,Hull,A.,Clark,K.,Ishibashi,K.,Iso,N., Litch-ﬁeld,N.,2011.Holocenepaleoseismichistoryofupper-platefaultsinthe south-ernHikurangi subductionmargin,NewZealand,deducedfromMarineTerrace Records.Bull.Seismol.Soc.Am. 101,2064–2087.

Bodin,P., Klinger, T., 1986. Coastaluplift and mortality ofintertidal organisms causedbytheSeptember1985Mexicoearthquakes.Science 233,1071–1073. Briggs,R.W.,Sieh,K.,Meltzner,A.,Natawidjaja,D.,Galetzka,J.,Suwargadi,B.,Hsu,Y.,

Simons,M.,Hananto,N.,Suprihanto,I.,Prayudi,D.,Avouac,J.-P.,Prawirodirdjo, L.,Bock,Y.,2006.DeformationandslipalongtheSundaMegathrustinthegreat 2005Nias–Simeulueearthquake.Science 311,1897.

Duputel,Z.,Rivera,L.,2017.Long-periodanalysisofthe2016Kaik ¯ouraearthquake. Phys.EarthPlanet.Inter. 265,62–66.

Escartín,J.,Leclerc,F.,Olive,J.-A.,Mevel,C.,Cannat,M.,Petersen,S.,Augustin,N., Feuillet,N.,Deplus,C.,Bezos,A.,Bonnemains,D.,Chavagnac,V.,Choi,Y.,Godard, M.,Haaga,K.A.,Hamelin,C.,Ildefonse,B.,Jamieson,J.W.,John,B.E.,Leleu,T., MacLeod,C.J.,Massot-Campos,M.,Nomikou,P.,Paquet,M.,Rommevaux-Jestin, C.,Rothenbeck,M.,Steinführer,A.,Tominaga,M.,Triebe,L.,Campos,R.,Gracias, N.,Garcia,R.,Andreani,M.,Vilaseca,G.,2016.Firstdirectobservationof coseis-micslipandseaﬂoorrupturealongasubmarinenormalfaultandimplications forfaultsliphistory.EarthPlanet.Sci.Lett. 450,96–107.

FitzRoy,R.,1839.ProceedingsoftheSecondExpedition,1831–1836,Underthe Com-mandofCaptain RobertFitzRoy.VolumeIIofthe narrativeofthesurveying voyagesofHisMajesty’sshipsAdventureandBeaglebetweenYears1826and 1836,describingtheirexaminationofthesouthernshores ofSouth America, andtheBeagle’scircumnavigationoftheglobe.London,UK.

Fujiwara,T.,Kodaira,S.,No,T.,Kaiho,Y.,Takahashi,N.,Kaneda,Y.,2011.The2011 Tohoku–Okiearthquake:displacementreachingthe trenchaxis.Science 334, 1240.

Furlong,K.,Herman,M.,2017.Reconcilingthedeformationaldichotomyofthe2016 Mw7.8KaikouraNZearthquake.Geophys.Res.Lett.http://dx.doi.org/10.1002/ 2017GL074365.

Hamling,I., Hreinsdóttir, S.,Clark,K.,Elliott,J.,Liang,C., Fielding,E.J.,Litchﬁeld, N.,Villamor,P.,Wallace,L.,Wright,T.J.,D’Anastasio,E.,Bannister,S.,Burbridge, D.,Denys,P.,Gentle,P.,Howarth,J.,Mueller,C.,Palmer,N.,Pearson,C.,Power, W.,Barnes, P., Barrell, D., Van Dissen, R., Langridge, R., Little,T., Nicol, A., Pettinga,J.,Rowland,J.,Stirling,M.,2017.Complexmulti-faultruptureduring the2016Mw7.8Kaik ¯ouraearthquake,NewZealand.Science.http://dx.doi.org/ 10.1126/science.aam7194.

Hamling, I.J., D’Anastasio, E., Wallace, L.M., Ellis, S., Motagh, M., Samsonov, S., Palmer,N.,Hreinsdóttir,S.,2014.Crustaldeformationandstresstransfer dur-ingapropagatingearthquakesequence:the2013CookStraitsequence,central NewZealand.J.Geophys.Res.,SolidEarth 119,6080–6092.

Hayes,G.P.,Briggs, R.W.,Sladen,A.,Fielding,E.J.,Prentice, C.,Hudnut,K.,Mann, P.,Taylor,F.W.,Crone,A.J.,Gold,R.,Ito,T.,Simons,M.,2010.Complexrupture duringthe12January2010Haitiearthquake.Nat.Geosci. 3,800–805.

8.6Niasâ–Simeulueearthquakefromthejointinversionofseismicandgeodetic data.Bull.Seismol.Soc.Am. 97,S307–S322.

Langridge,R.M.,Ries,W.F.,Litchﬁeld,N.J.,Villamor,P.,VanDissen,R.J.,Barrell,D.J.A., Rattenbury,M.S.,Heron,D.W.,Haubrock, S., Townsend,D.B.,Lee,J.M., Berry-man,K.R.,Nicol,A.,Cox,S.C.,Stirling,M.W.,2016.TheNewZealandactivefaults database.N.Z.J.Geol.Geophys. 59,86–96.

Litchﬁeld,N.J.,VanDissen,R.J.,Sutherland,R.,Barnes,P.M.,Cox,S.C.,Norris,R., Bea-van,R.J.,Langridge,R.,Villamor,P.,Berryman,K.,Stirling,M.,Nicol,A.,Nodder, S.,Lamarche,G.,Barrell,D.J.A.,Pettinger,J.R.,Little,T.A.,Pondard,N.,Mountjoy, J.J.,Clark,K.,2014.AmodelofactivefaultinginNewZealand.N.Z.J.Geol. Geo-phys. 57,32–56.

Litchﬁeld, N.J., Benson, A., Bischoff, A., Hatem, A., Barrier, A., Nicol, A., Wan-dres, A., Lukovic,B., Hall, B., Gasston, C., Asher, C., Grimshaw, C., Madugo, C., Fenton, C., Hale, D., Barrell, D.J.A., Heron, D.W., Strong, D.T., Townsend, D.B.,Noble,D., Howarth,J.D.,Pettinga,J.,Kearse, J.,Williams,J.,Manousakis, J., Mountjoy,J., Rowland, J., Clark,K.J., Pedley,K.,Sauer, K., Berryman,K.R., Hemphill-Haley, M., Stirling, M.W., Villeneuve, M., Cockroft, M., Khajavi, N., Barnes,P.,Villamor, P.,Carne,R.,Langridge, R.M.,Zinke, R.,Van Dissen,R.J., McColl,S.,Cox,S.C.,Lawson,S.,Little,T.,Stahl,T.,Cochran,U.A.,Toy,V.,Ries, W.F., Juniper,Z.,2017. 4thNovember2016M7.8 Kaik ¯ouraearthquake. Sum-marysurfacefaulttracesandpreliminarydisplacementmeasurements.http:// dx.doi.org/10.21420/G21422RC21427C.

Meghraoui,M.,Maouche,S.,Chemaa,B.,Cakir,Z.,Aoudia,A.,Harbi,A.,Alasset,P.J., Ayadi,A.,Bouhadad,Y.,Benhamouda,F.,2004.Coastalupliftandthrustfaulting associatedwiththeMw=6.8Zemmouri(Algeria)earthquakeof21May,2003. Geophys.Res.Lett. 31,L19605.http://dx.doi.org/10.1029/2004GL020466. Melnick,D.,Cisternas,M.,Moreno,M.,Norambuena,R.,2012a.Estimatingcoseismic

coastalupliftwithanintertidalmussel:calibrationforthe 2010Maule Chile earthquake(Mw=8.8).Quat.Sci.Rev. 42,29–42.

Melnick, D., Moreno,M., Motagh, M., Cisternas, M., Wesson, R.L., 2012b. Splay fault slip during the Mw 8.8 2010 Maule Chile earthquake. Geology 40, 251–254.

Meltzner,A.J.,Sieh,K.,Abrams,M.,Agnew,C.D.,Hudnut,K.W.,Avouac,J.-P., Nataw-idjaja, D.H., 2006. Uplift and subsidence associated with the great Aceh– Andamanearthquake of2004. J.Geophys.Res. 111.http://dx.doi.org/10.1029/ 2005Jb003891.

Merritts,D.J.,1996.TheMendecinotriplejunction:activefaults,episodiccoastal emergence,andrapiduplift.J.Geophys.Res. 101,6051–6070.

Nissen,E.,Krishnan,A.K.,Arrowsmith,J.R.,Saripalli,S.,2012.Three-dimensional sur-facedisplacementsand rotationsfromdifferencing pre- andpost-earthquake LiDARpointclouds.Geophys.Res.Lett. 39,L16301.

Oskin,M.E.,Arrowsmith,J.R.,Corona,A.H.,Elliott,A.J.,Fletcher,J.M.,Fielding,E.J., Gold,P.O.,Garcia,J.J.G.,Hudnut,K.W.,Liu-Zeng,J.,Teran,O.J.,2012.Near-ﬁeld deformationfrom theEl Mayor–Cucapahearthquake revealed bydifferential LIDAR.Science 335,702–705.

Ota, Y., Pillans, B., Berryman, K., Beu, A., Fujimori, T., Miyauchi, T., Berger, G., Beu,A.,Climo,F.,1996.PleistocenecoastalterracesofKaik ¯ouraPeninsulaand theMarlboroughCoast,SouthIsland,NewZealand.N.Z.J.Geol.Geophys. 39, 51–73.

Plafker,G.,1969. TectonicsoftheMarch27,1964,AlaskaEarthquake.Geological SurveyProfessionalPaper543-1.174 pp.

Plafker,G.,Savage,J.C.,1970.MechanismoftheChileanEarthquakesofMay21and 22,1960.Geol.Soc.Am.Bull. 81,1001–1030.

Power,W.,Clark,K.,King,D.N.,Borrero,J.,Howarth,J.,Lane,E.M.,Goring,D.,Goff, J., Chagué-Goff,C., Williams,J., Reid,C., Whittaker, C.,Mueller, C.,Williams, S.,Hughes,M.W.,Hoyle,J.,Bind,J.,Strong,D.,Litchﬁeld,N.,Benson,A.,2017. Tsunamirunupandtide-gaugeobservationsfromthe14November2016M7.8 Kaik ¯oura earthquake, New Zealand. Pure Appl. Geophys. 174, 2457. http:// dx.doi.org/10.1007/s00024-017-1566-2.

Ramos,N.T.,Tsutsumi,H.,2010.Evidenceoflargeprehistoricoffshoreearthquakes deducedfromupliftedHolocenemarineterracesinPangasinanProvince,Luzon Island,Philippines.Tectonophysics 495,145–158.

(12)

Rockwell,T.K.,Clark,K.,Gamble,L.,Oskin,M.E.,Haaker,E.C.,Kennedy,G.L.,2016. LargetransverserangeearthquakesCauseCoastalUpheavalnearVentura, South-ernCalifornia.Bull.Seismol.Soc.Am. 106,2706–2720.

Shaw,B.,Ambraseys,N.N.,England,P.C.,Floyd,M.A.,Gorman,G.J.,Higham,T.F.G., Jackson, J.A., Nocquet, J.M., Pain, C.C., Piggott, M.D., 2008. Eastern Mediter-raneantectonicsandtsunamihazardinferredfromtheAD365earthquake.Nat. Geosci. 1,268–276.

Subarya,C.,Chlieh,M.,Prawirodirdjo,L.,Avouac,J.-P.,Bock,Y.,Sieh,K.,Meltzner, A.,Natawidjaja,D.,McCaffrey,R.,2006.Plate-boundarydeformationassociated withthegreatSumatra–Andamanearthquake.Nature 440,46–51.

Taylor,F.W., Briggs, R.W., Frohlich,C., Brown,A.,Hornbach, M.J.,Papabatu, A.K., Meltzner,A.,Billy,D.,2008.Ruptureacrossarcsegmentandplateboundaries inthe1 April 2007Solomonsearthquake.Nat.Geosci. 1,253–257.

Wallace,L.M.,Barnes,P.,Beavan,R.J.,VanDissen,R.J.,Litchﬁeld,N.J.,Mountjoy,J., Langridge,R.M.,Lamarche,G.,Pondard,N.,2012.Thekinematicsofatransition fromsubductiontostrike-slip:anexamplefromthecentralNewZealandplate boundary.J.Geophys.Res. 117,B02405.http://dx.doi.org/10.1029/2011JB008640.