JOURNALOFGEOPHYSICALRESEARCH:SOLIDEARTH,VOL.118,1277–1303,doi:10.1002/jgrb.50121,2013 Permanent upper plate deformation in western Myanmar during the great 1762 earthquake: Implications for neotectonic behavior of the northern Sunda megathrust Yu Wang,1,2 J. Bruce H. Shyu,3 Kerry Sieh,2 Hong-Wei Chiang,2,3 Chung-Che Wang,3 Thura Aung,4 Yu-nung Nina Lin,1 Chuan-Chou Shen,3 Soe Min,4 Oo Than,5 Kyaw Kyaw Lin,5 and Soe Thura Tun4 Received29November2012;revised4February2013;accepted5February2013;published28March2013. [1] The 1762 Arakan earthquake resulted from rupture of the northern Sunda megathrust and is one of those rare preinstrumental earthquakes for which early historical accounts document ground deformations. In order to obtain more comprehensive and detailed measurements of coseismic uplift, we conducted comprehensive field investigations and geochronological analyses of marine terraces on the two largest islands in western Myanmar.Weconfirm3–4mofcoseismiccoastalemergencealongsouthwesternCheduba Island,diminishingnortheastwardtolessthan1m.Farthernortheast,upliftassociatedwith the earthquake ranges from slightly more than 1m to 5–6m along the western coast of Ramree Island but is insignificant along the island’s eastern coast. This double-hump pattern of uplift coincides with the long-term anticlinal growth of these two islands. Thus, weproposethatthe1762earthquakeresultedfromsliponsplayfaultsundertheislands,in addition to rupture of the megathrust. Elastic modeling implies that fault slip during the 1762 earthquake ranges from about 9 to 16m beneath the islands and corresponds to a magnitudeofM 8.5iftherupturelengthofthemegathrustis~500km.Theisland’suplift w histories suggest recurrence intervals of such events of about 500–700years. Additional detailedpaleoseismologicalstudieswouldaddsignificantadditionaldetailtothehistoryof large earthquakes in this region. Citation: Wang,Y.,etal.(2013),PermanentupperplatedeformationinwesternMyanmarduringthegreat1762 earthquake:ImplicationsforneotectonicbehaviorofthenorthernSundamegathrust,J.Geophys.Res.SolidEarth,118, 1277–1303,doi:10.1002/jgrb.50121. 1. Introduction involved, as in the cases of the great 1964 Alaskan and 1946 Nankaido earthquakes [e.g., Plafker, 1965; Fukao, [2] Coseismic deformationabovesubductionmegathrusts 1979; Kato, 1983; Park et al., 2000]. Although they are is a key to understanding great earthquake ruptures along smallerthantheirassociated megathrusts, upperplatestruc- convergentplatemargins.Usually,deformationpatternsim- tures mayplaysignificantroles inthegeneration ofseismic ply rupture solely on the underlying megathrust, as in the shakingortsunami,asappearstohavebeenthecasewiththe 2005 Nias and 2007 Solomon Islands earthquakes [e.g., great 2004 Sumatran earthquake and tsunami [DeDontney Briggs et al., 2006; Konca et al., 2007; Taylor et al., and Rice, 2012]. Structures in the fore-arc region may be 2008]. Less commonly, upper plate structures are also also related to major asperities of large megathrust earthquakes[e.g.,Sugiyama,1994;Wellsetal.,2003].Nine- teenthcenturyreportsofcoastalupliftduringthegreat1762 1DivisionofGeologicalandPlanetarySciences,CaliforniaInstituteof Arakan earthquake in western Myanmar are intriguing in Technology,Pasadena,California,USA. this regard, because they imply that upper plate structures 2Earth Observatory of Singapore, Nanyang Technological University, played a role in the earthquake. Singapore. 3Department of Geosciences, National Taiwan University, Taipei, [3] At about the same time that Darwin [1845] was Taiwan. documenting and publishing his famous observations of 4Myanmar Earthquake Committee, Myanmar Engineering Society, deformation associated with the great 1835 Chilean earth- Yangon,Myanmar. quake,Britishnavalofficersdocumentedcoastalemergence 5DepartmentofMeteorologyandHydrology,Yangon,Myanmar. thatmayhaveoccurredduringthe1762Arakanearthquake. Correspondingauthor:J.BruceH.Shyu,DepartmentofGeosciences, Their observations suggested up to 7 m of coseismic uplift NationalTaiwanUniversity,Taipei10617,Taiwan.([email protected]) on Cheduba (Man-Aung) and neighboring Ramree Islands [Halsted, 1841; Mallet, 1878] (Figure 1). They also ©2013.AmericanGeophysicalUnion.AllRightsReserved. 2169-9313/13/10.1002/jgrb.50121 described associated flights of marine terraces. These 1277 WANGETAL:DEFORMATIONOF1762ARAKANEARTHQUAKE Figure 1. Cheduba (Man-Aung) and Ramree Islands are the expressions of two active antiforms above the Sunda megathrust offshore the western coast of Myanmar. (a) The last seismic ruptures of the northern Sunda megathrust, between the Indian and the Burma plates. Orange color depicts the inferred 1762 Arakan rupture from historical reports. This ~500km long seismic patch is the only megathrust-related rupture north of the 2004 patch (shown in purple, after Chlieh et al. [2007]) from the 18th century to present. Red lines are major active faults in Southeast Asia (after Le Dain et al. [1984]), where most of the major faults are strike-slip faults on the Burma and the Sunda plates. Blue box shows the area of Figure 1b. SAF: Sagaing fault system; WAF: West Andaman fault. (b) The accretion-related topography above the Sunda megathrust and our survey locations in Cheduba and Ramree Islands. This section of the megathrust receives ~23mm/yr of oblique plate convergence from the northeastward motion of the Indian plate [Socquet et al., 2006]. This plate convergence creates a series of megathrust-parallel underwater ridges within the accretionary prism. Cheduba and Ramree Islands are the two highest portions of these tectonic ridges. Black solid contours are modified from the U.S. Army topography maps [U.S. Army Map Service, 1955a, 1955b]. Grey dashed contours are from ETOPO-1 [Amante et al., 2009]. The high-resolution bathymetry along the trench front is digitized from Nielsen et al. [2004]. Yellow squares indicate the observation points in the 19th century [Halsted, 1841; Mallet, 1878]. White dots represent the survey locations of this study, between 2010 and 2011. 1278 WANGETAL:DEFORMATIONOF1762ARAKANEARTHQUAKE observationsledthemtospeculatethatthesecoastlineswere Ramree Island, 70–100km away from the trench, is the being permanently uplifted during similar successive earth- manifestation of the other (Figure 1b). Both antiforms are quakes [Halsted, 1841]. The permanence of uplift implied doubly plunging and are asymmetric, as evidenced by their bytheflightsofterracesdoesindeedsuggestrepeatedinelas- southwestern flanks being clearly steeper than their north- tic deformation within the accretionary prism. eastern flanks. In each case, cumulative uplift appears to [4] Though intriguing, the 19th century observations are havebeengreaterneartheirsouthwesternflanks,sincetheir too sparse to enable one to conclude much aboutthe nature highest topography is closer to their southwestern flanks. of the faulting that caused the deformations or about the Several studies have discussed the nature of these upper magnitude of the earthquake. One limitation is that most of plate structures. For example, Nielsen et al. [2004] docu- the observations were made decades after the earthquake, mented the active folds and faults within the accretionary so assignment of the observed deformations solely to the prism near the deformation front. Maurin and Rangin 1762 event is dubious. Another limitation of the historical [2009] suggested that a northeast dipping blind thrust fault observations is their small geographic spread. Most of the 20km west of Cheduba Island initiated after the late Plio- reliableobservationsare alongthe western side ofCheduba cene.Althoughtherearenoconstraintsontheratesofdefor- (Man-Aung) Island (Figure 1b), with just a few other ac- mation,theexistenceoftheantiformsstronglyimpliesthata counts from the west coast of Myanmar and Bangladesh. significantamountofIndian-Burmaplateconvergenceisoc- [5] This irregular and sparse distribution of observations curringwithintheaccretionarywedge.Thus,theupperplate and the uncertainty of the timing of uplift are inadequate structures are potential seismic sources in this area. for construction of a useful deformation pattern for the [9] In this context, it is not surprising that abundant 1762 earthquake. Thus, we decided to reevaluate the 19th evidence for geologically recent uplift exists on and in the centuryobservationsandtoimprovethequantityandquality vicinity of Cheduba and Ramree Islands. Flights of marine of observations via a field investigation that included new terraces have long been known along western Myanmar geomorphic measurements and precise geochronological coast.Brunnschweiler[1966]reportedpost-Pliocenemarine analyses of uplifted coastal features. terraces about 45–60m above sea level along western [6] Inthepagesthatfollow,wedescribeourobservations Cheduba Island and 30m high terraces along western of the vertical deformation along the coasts of Ramree and Ramree Island. Than Tin Aung et al.[2008] described a se- ChedubaIslandsassociatedwiththe1762eventviameasur- ries of marine terraces north of Ramree Island, the oldest ing several different sea level markers. U-Th dating ofwhichisabout~3000yearsoldand6–16mabovecurrent techniques[Shenetal.,2003,2012]onamulticollectorinduc- mean sea level (MSL). tively coupled plasma mass spectrometer (MC-ICP-MS), [10] Severalearlierobserverssuggestedthattheupliftoc- Thermo Fisher Neptune, at the High-Precision Mass Spec- curred during seismic events: Halsted [1841] observed that trometryandEnvironmentChangeLaboratory(HISPEC),Na- the elevation difference between each marine terrace on tionalTaiwanUniversity,wereusedtodeterminethetimeof westernChedubaIslandisidenticaltotheamountofthelat- uplift of these features. Ages of several carbonate samples est uplift there. Mallet’s [1878] observations suggested to werealsodeterminedbyradiocarbondatingtechnique.More- himthatnochangesoccurredbetweenCaptainHalsted’sob- over, we describe our mapping of regional geomorphic servations and his own visit in the late 19th century. More features, which provides the neotectonic context for under- recently, Shishikura et al. [2009] supported this view; they standing the dated uplifted features. We then discuss the suggestedthattheelevationofthelowestterraceonwestern possiblesources and seismic parametersof the 1762 Arakan Cheduba Island is similar to the elevation recorded by earthquake,includingitsearthquakemagnitudeandtherecur- CaptainHalsted.Thisimpliesthatnoappreciablenetvertical renceinterval. movement has occurred since the mid-19th century. Taken together,theseobservationsimplythatthemajorityofuplift occurs during or right after earthquakes and that recovery 2. Active Tectonic Context during the interseismicperiod is minimal. Thisdeformation [7] ThenorthernSundamegathrustisthenominalbound- behaviorthusprovidesusanexcellentopportunityforstudy- ary between the Indian and the Burma plates. In reality the ing the plausible coseismic coastal uplift that occurred boundary is not so simple, because thick sediments of the 250years ago. Bengal Fan sit atop the downgoing Indian Ocean litho- sphere,andmuchofthissedimentarysectionisbeingfolded 3. Sea Level Indicators rather than subducted [Curray, 1991; Curray et al., 2003]. These sediments sit at the boundary of two plates that are [11] To constrain land-level changes along the coasts of converging obliquely at about 23mm/yr [Socquet et al., the islands precisely, one must measure the elevations of 2006](Figure1b).Mostofthisdextral-obliqueconvergence upliftedsealevelindicatorsrelativetotheirmodernequiva- appears to be taken up by the megathrust and structures lents. These indicators may be either marine organisms above itin the accretionary prism. preservedintheirlivingpositionorerosionalanddepositional [8] In the vicinity of Cheduba and Ramree Islands, the features. Sea level indicators form at a range of locations BengalFansedimentsare8–12kmthickandexhibitawide between high- and low-water spring tides; therefore, on a zoneoffoldingandshorteningabovethedowngoingIndian mesotidal coast as in western Myanmar, where mean tidal Ocean lithosphere. Two active trench-parallel antiforms are range is>2m,theseindicators formovera vertical range of readilyapparentinthebathymetryandtopography.Cheduba about3m,asshowninFigure2. Island, 40–60km northeast of the deformation front, is the [12] The major types of sea level indicators that we used subaerial expression of the western of these two; and comprise coastal erosional features (shoreline angles, sea 1279 WANGETAL:DEFORMATIONOF1762ARAKANEARTHQUAKE Figure2. NaturalsealevelindicatorsandtheirrelationshipswiththetidallevelsintheareaofCheduba andRamreeIslands.Thisschematicdiagramshowsthemodernpositionsofvarioussealevelindicatorsin mesotidalenvironments(withtidalrangeof2–4m).Theuppergrowthlimitofoystersandthecoastalero- sionalfeatures(shorelineanglesandwave-cutnotches)aremostlyrelatedtothewaterlevelfrommeansea level(MSL)tohightide.Thetopofcoralmicroatolls,however,representsthewaterlevelthatis~1–2m lowerthantheotherfeatures.TheelevationofmicroatollsisinferredfromKayanneetal.[2007],andthe otherindicator’selevationsarefromthisstudy.MHWS:meanhigh-watersprings;MHHW:meanhigher high water; MLLW: mean lower low water; MLWS: mean low-water springs. notches and wave-cut platforms) and the living position of Islands,oursurveysdemonstratethattheuppergrowthlimit marine organisms (coral microatolls and oysters). Each of of living oysters (e.g., Saccostrea spp.) occurs between thesehasbeenextensivelyusedelsewherearoundtheworld mean higher high water (MHHW, ~1m above MSL) and to measure sea level histories in a range of tidal environ- meanhigh-waterspring(MHWS,~1.3maboveMSL).This ments [e.g., Chappell et al., 1983; Hull, 1987; ten Brink upper growth limit is slightly higher than documented et al., 2006;Meltzneret al.,2010]. elsewhere [e.g., Kelletat, 1988; Beaman et al., 1994; Lewis [13] To estimate the relationship between the sea level et al., 2008]. indicatorsandtheirassociatedwaterlevels,wefirstmeasure [17] Inourarea,livingoysterscommonlyadheretosand- the modern indicators’ elevations with respect to the water stone cliffs and isolated sandstone columns on wave-cut level at the time of survey. We then relate this measured platforms.Theverticalrangeofoystergrowthoverlapswith elevation to present MSL, using tidal predictions from the the zone of barnacle growth, but the highest barnacles are software package NLOADF (SPOTL v.3.2.4) [Agnew, generally higher than the highest oysters, extending above 1997] and the regional harmonic tidal solutions for the Bay MHWS.Becausemodernoystersinthelittoralzoneareeasily of Bengal from the Oregon State University [http://volkov. collectedbylocalfishermen,theyrarelyformprominentoys- oce.orst.edu/tides/BBay.html]. This method is reliable for terencrustations,instarkcontrasttoolder,fossilpopulations. estimation of water level in western Sumatra [e.g., Briggs Instead,theyusuallygrowasindividualsonrocksurfaces.In et al., 2006;Meltzneret al.,2006,2010]. localesnotfrequentedbyfishermen,weobservedlivingoys- [14] Our survey results and other studies imply that most tersformingverydensebeltsbeneathMHWS. of the indicators we used reliably constrain paleo–water levelswithprecisionsrangingfromabout(cid:1)0.25mtoabout 3.1.2. Coral Microatolls and Coral Heads (cid:1)1m [e.g., Chappell et al., 1983; ten Brink et al., 2006; [18] Ingeneral,coralmicroatollsprovideuswiththemost precise water-level indicators. Their upper growth limit de- Lewis et al., 2008]. This is a considerable improvement in velops between mean lower low water (MLLW) and mean precision from just correlating the average terrace elevation low-water spring (MLWS) in the mesotidal environment to current MSL, which may have more than 2m of uncer- (Figure2).Thisisconsistentwiththeirbeingabletosurvive tainty in thismesotidal environment. [15] Below,wedescribethefivemajorsealevelindicators short periods of exposure above the sea during the lowest monthly tides. This relationship of the highest level of sur- that we use and their relationship to the tidal datum. vival (HLS) of coral microatolls tolow-tide levels has been used recently to document sea level history [e.g., Chappell 3.1. Biological Indicators et al., 1983; Zachariasen et al., 1999; Natawidjaja et al., 3.1.1. Oysters 2007; Kench et al.,2009; Meltzner et al.,2010]. [16] Emerged oysters have been widely used to constrain [19] Recent studies have shown that the relationship of land-level changes [e.g., Davis et al., 2000; Awata et al., HLS to low tides varies with tidal environments and coral 2008; Lewis et al., 2008; Hsieh et al., 2009]. Their species.InmicrotidalenvironmentssuchaswesternSumatra upper growth limit is usually restricted below high-water (maximumtidalrange0.8–1m),HLSformassivespeciesof level [e.g., Kelletat, 1988; Beaman et al., 1994; Lewis the genus Porites is about 20cm above extreme low water et al., 2008; Hsieh et al., 2009]. On Ramree and Cheduba (ELW) [Meltzner et al., 2010]. For Goniastrea retiformis, 1280 WANGETAL:DEFORMATIONOF1762ARAKANEARTHQUAKE HLSisabout10cmhigherthere.Inmesotidalenvironments 3.2.2. Sea Notches (with2–6mtidalranges)suchastheGreatBarrierReef,the [25] We found two types of wave-cut notches: tidal uppermostlevelofliving coralsapproximatesMLWS[e.g., notches and surf notches. Each of these is distinguished by Chappell et al., 1983; Hopley, 1986], which is approxi- itsparticularshape.TidalnotchesareU-orV-shapedinden- mately70cm higher thanELW. tations that develop on cliffs or steep slopes in hard rock. [20] Along coasts with tidal ranges similar to that of the They result from wave action as the tides bring the sea western Myanmar coast (tidal range ~2m), the HLS of surface through the intertidal range. The deepest part of the microatolls is between MLLW and MLWS [Kayanne indentation occurs at the level of mean sea level (MSL) et al., 2007; Kench et al., 2009]. This elevation is similar [Pirazzoli, 1986]. In our area, tidal notches are most to HLS onthe GreatBarrier Reef. In addition, we observed commonly cut into sandstone cliffs. They commonly have that the HLS of living coral in a semiconfined tidal pool agentleUshape,withtheopening1–2mwidefromthebase is not higher than MLLW on northern Ramree Island, of the U. Oysters and other marine organisms commonly whereastheHLSofmicroatollsinopenwaterenvironments grow within the notches. mustbelowerthanMLLW.Thus,itisreasonabletosuggest [26] Surf notches exhibit far less erosional height than that the HLS of the microatolls of western Myanmar is tidalnotchesinourstudyarea.Wecommonlyfoundmodern at an elevation that is similar to the microatoll HLS in surfnotchesatactiveshorelineanglesandonsandstoneplat- other mesotidal environments and is not higher than the forms near MHWS, well above the tidal notches. These level of MLLW. notchesoftenformabovethehightidewherethecliffisreg- [21] Althoughtheupliftedmicroatollsareaprecisewater- ularly washed by waves [Pirazzoli, 1986]. Thus, unlike the level indicator, well-preserved microatolls are rarely found tidal notches, their heights are related to the energy of the inourfieldarea.Inplaceswherewedidnotfindmicroatolls, surf rather than to the tidal range. we compare the elevation of the highest coral colony to the [27] The accuracy with which marine notches reflect sea current MLLW. This yields a minimum water-level change level varies depending on the tidal range, the geomorphol- since the growth of corals. ogy of the site, and the slope of the bedrock [Pirazzoli, 1986]. For example, along one short stretch of coast, we found that the elevation of a modern tidal notch on a sand- 3.2. Erosional Coastal Features stone ridge facing the open ocean is nearly 1m lower than 3.2.1. Shoreline Angles that on another part of the same ridge, but at the top of a [22] The term “shoreline angle” refers to the locus of sandy beach. This elevation difference is very likely the re- points that form the join between a wave-cut platform and sult of differing wave runups in these two different settings a sea cliff. Uplifted shoreline angles are one of the most duringtidalsurges.Therefore,wesuggestthattheelevations common coastal features in our field area and have been of marine notches are uncertain by(cid:1)1m in our studyarea. widelyusedincoastalgeomorphicstudiestoreconstructhis- 3.2.3. Wave-Cut Platforms toriesof sea level change [e.g., Hull,1987; ten Brink et al., [28] Althoughmodernandupliftedwave-cutplatformsare 2006;Saillardetal.,2009].Inmacrotidalandmesotidalen- themostcommonfeaturesalongthecoastsofChedubaand vironments, field observations suggest that they usually de- Ramree Islands, they are not a precise sea level indicator in velopbetweenMHWSandmeanhigh-waterneap(MHWN) ourstudyarea.Wave-cutplatformsgenerallydevelopwithin [Hull,1987].Inplaceswherethetidalrangeissimilartoour the intertidal zone and commonly extend below it, where study area, modern shoreline angles develop in a more bedrock can be eroded by wave action [Trenhaile and restricted position within this range, near the elevation of Layzell, 1981]. Along mesotidal coasts, the elevation of the MHHW [ten Brink et al., 2006]. Our field surveys confirm platformmayramp3–4mfrombelowlowtidetohightide. that shoreline angles usually form in our area near MHWS, Thus, a direct comparison of the elevation difference be- about 20cm above MHHW. In rare cases, though, we tween a point on an uplifted platform and a point on the found that the shoreline angle has developed a bit higher, modern platform is not very useful in constraining uplift or above MHWS, perhaps due to erosion by waves during subsidence. Since wave-cut platforms commonly develop storm surges. between MHWS and MLWS, we suggest that their eleva- [23] Alluvialortalusdepositsatthebaseofaseacliffof- tions generally indicate MSL(cid:1)half of the tidal range. This ten obscuretheshoreline angle.Insuchcases,theelevation great uncertainty makes wave-cut platforms the worst sea ofashorelineanglewouldbeoverestimatedunlessitisdug levelindicatorsinourstudyarea.However,inplaceswhere outorexposedbyerosion.Duetotheverylimitedsurveying noother indicator isavailable, andwe wereable toconfirm timeinthefield,wedidnottrytodigtheshorelineangleout the sediments are thin on the uplifted platform (i.e., <1m), while surveying the profiles. Instead, we extrapolated the we estimated a minimal land-level change by measuring terraceprofileandtheseacliffslopetoestimatetheelevation theelevationdifferencebetweenthemodernshorelineangle of the shoreline angle to avoid the influence of later and anuplifted wave-cut platform. deposition. [24] The uncertainties in our measurements of shoreline angle elevations are likely greater than our measurements 4. Coastal Emergence ofbiologicalsealevelindicators,duetoboththeobscuration 4.1. Ramree Island bysedimentsonthewave-cutplatformandthevariabilityof the strength of storm surges. To account for these uncer- [29] Ramree Island lies ~70km east of the deformation taintiesandvariability,weassumedthatourshorelineangle frontandiselongateparalleltothestrikeofthemegathrust. measurements represent MHWS+1m in our studyarea. This 80 km long, 20 km wide island is connected to the 1281 WANGETAL:DEFORMATIONOF1762ARAKANEARTHQUAKE mainlandofMyanmarbyamarshthatisslightlyhigherthan oyster reef grew between 1417 and 1618 Common Era the intertidal zone (Figure 1b). (C.E.) (Figure 4 and Table 1). This date is suspect because [30] Our field observations on the island were limited by the local marine reservoir correction (Delta-R) is unknown. the availability of access roads. A semipaved road along Nonetheless, this date is similar to other radiocarbon ages thenorthernhalfoftheisland’swesterncoastprovidesgood of uplifted corals and oysters north of Ramree Island[Than access to its northwestern coastline, but the marshy north- Tin Aung et al., 2008]. Therefore, we believe that the eastern part of the island isdifficult toreach bycar. Farther uplifted oyster layer (KPU-15), together with the lowest south, overland access is even more limited by the lack of uplifted tidal notch, was elevated during a regional tectonic roads.Therefore,wereliedoncharteredboatstosailtosome event.Thefactthattheradiocarbonageisacenturyortwoear- largertownsonsoutheasternRamree.Accesstosmallervil- lier than the great 1762 Arakan earthquake encourages the lages along the southwestern coast was by foot. In places speculationthatitwasduringthisearthquakethatthislowest where chartered boats were unable to get close to shore, notchanditsassociatedoystersroseoutoftheintertidalzone. ourobservationswerelimitedtoviewsfromoffshore.These [33] To the east, the magnitude of late Holocene emer- logistical difficulties significantly limited our ability to per- gence is much smaller. Fossil mid-Holocene coral form detailed, high-precision surveys along the southern microatolls south of Kyauk-Pyu rest upon the T1 surface, coasts of Ramree Island. whichis only slightly above the current MHWS (Figure 3). 4.1.1. Northern Ramree Island (Kyauk-PyuArea) These microatolls are present from the terrace surface to [31] Ancient sea level indicators reveal that changes in the modern tidal flat, and the elevations of their upper sur- landleveldiffergreatlybetweennorthwesternandnortheast- faces are 1–1.5m above MSL or 2–2.5m above the current ern Ramree Island. Evidence for progressive uplift is abun- MLLW. U-Th analyses show that the ages of these corals dantly clear in the former and absent in the latter. About range from 5000 to 7100yearsB.P. (Table 2, KPU-102 to 3kmwestofKyauk-Pyu(Figure3),thelargestcityoftheis- KPU-110). Both the ages and the elevations of these corals land,wefoundaseriesofupliftedtidalnotchesandbandsof are consistent with the timing and water level of the mid- upliftedoystersonasandstoneridgebelowthesurfaceofthe HolocenehighstandoftheeasternIndianOcean[Woodroffe lowest marine terrace, T1 (KPU-15 in Figures 3 and 4). and Horton, 2005; Briggs et al., 2008]. Thus, these corals These stacked ancient coastal features indicate successive suggest negligible net uplift of northeastern Ramree Island upliftevents during the Holocene period. since the middle of the Holocene epoch. [32] The lowest uplifted tidal notch is ~1.5m above the [34] The broad morphology of Ramree Island reflects a modern notch. A layer of dead oysters encrusts the sand- northeastwardtiltthatiswhollyconsistentwith thecontrast stone cliff slightly above the uplifted tidal notch and is inupliftbetweenthesetwosites.Asingle,large,lowterrace ~1mabovethetopofthebandofmodernoysters(Figure4). (T1) and a plexus of estuaries and tidal channels dominate A radiocarbon date from the dead oysters (assuming the the surface of the northeastern part of the island (Figure 3). globalaveragemarinereservoircorrection)suggeststhatthis In great contrast to this, flights of marine terraces dominate Figure 3. The patterns of marine terraces, current drainages, and tidal flats show the eastward tectonic tilting in northern Ramree Island over the past several thousand years. The geomorphic characteristics in the northwestern part of the island are very different from those in the northeastern part of the island. Intheeast,mid-Holocenefossilcoralmicroatolls(KPU-102,KPU-106,andKPU-109)arepresentslightly above the modern high tide, reflecting a very small long-term uplift. To the west, however, a flight of wave-cutnotchesshowsclearsignsoflong-termsuccessiveuplift,andthelastuplifteventoccurredafter the16thcentury(KPU-15,Figure4).BluenumbersshowtheU-ThageofthecoralmicroatollsinyearsB. P. (yBP). 1282 WANGETAL:DEFORMATIONOF1762ARAKANEARTHQUAKE Figure4. (a)Photographand(b)linesketchofsiteKPU-15.Hereabeltofupliftedoysterfossilsanda wave-cut notch beneath T1 suggest about 1–1.5m of land-level change since the 16th century. Several levels of higher wave-cut notches on a sandstone ridge at the same site suggest successive uplift events in the past several thousand years. The radiocarbon age of uplifted oyster fossils (KPU-15) above the lowest uplifted sea notch suggests that the last land-level change event occurred after the 16th century. The uplifted oyster fossils (shown in orange) are ~1m above the modern oyster growth zone (shown in yellow). This elevation difference is similar to that between the modern sea notch (light blue arrows) and the uplifted sea notch (dark blue arrows). The elevation distributions of the oyster fossils and the wave-cutnotchareshownintheinsetofFigure4b.Thecolorcodeisthesameasthatinthelinesketch. Table1. RadiocarbonAgesObtainedinThisStudy MeasuredAge d13C ConventionalAge CalendarYeara,b(2s) Laboratory Sample ID Sample Type (B.P.) (%) (B.P.) From To NorthernRamreeIsland Beta-285817 KPU-15 Oyster 420(cid:1)50 (cid:3)0.10 830(cid:1)50 1417 to 1618 EasternChedubaIsland Beta-301002 KK-145 Oyster 280(cid:1)40 (cid:3)0.30 690(cid:1)40 1520 to 1686 Beta-301003 KK-146 Oyster 330(cid:1)70 (cid:3)0.10 740(cid:1)70 1454 to 1685 Beta-301004 KK-148 Oyster 350(cid:1)40 (cid:3)0.50 760(cid:1)40 1472 to 1647 NorthwesternChedubaIsland Beta-301005 TY-140 Oyster 380(cid:1)40 0.70 800(cid:1)40 1447 to 1623 aSamplesarecalibratedusingthemodeledoceanaverageMarine09calibrationcurve[Reimeretal,2009]. bWeassumethatΔR=0duetothelackofproperinformationalongtheeasternsideofBayofBengal. the geomorphology of the southwestern coast of the island. that are about 5m above their modern growth position This contrast implies northeastward tilt of the island, with (~MHHW; Figure 6a). These two sea level indicators dem- significant net uplift of the southwestern coast tapering onstratetheland-levelchangesincejustbeforethelastevent northeastward to zero. is about 5m alongthe central western coast. 4.1.2. CentralRamree Island [36] Samples from within the coral heads yielded very [35] Uplifted corals andothersealevel indicators demon- preciseU-Thages.Allthreeheadswerelivinginthemiddle strate that the central western coast of Ramree Island rose decades of the 18th century (Table 2). Among these U-Th several meters during the last emergence event, much more dates, the age of ZC-16 provides us the best timing than that on the northwestern tip of the island. In situ fossil constraint of the uplift event. There are four annual bands coralheads(ZC-16,ZC-118,andZC-119)restonaT1sur- between the dated annual band and the outermost band, face that is ~3.5m above MSL (Figures 5 and 6a). The fact which represents the date of death of the coral. Thus, the that the highest upper surface of these corals is about 5m coral appears to have died in 1762(cid:1)11 C.E., a perfect above the current MLLW implies that they are now about match for the 1762 earthquake. Although the growth bands 5mabovethemodernhighestlevelofcoralgrowth.Farther are not as clear as the ZC-16 sample, the U-Th ages from inlandarefossiloystersingrowthpositiononbedrockofT1 theothertwosamples(ZC-118andZC-119)collectedfrom 1283 WANGETAL:DEFORMATIONOF1762ARAKANEARTHQUAKE 59453129 93760 074413 75 25736 80 5059 g 82317825 65658 592687 74 35177 68 6932 n 01543001 77776 020542 44 89512 36 7667 e Year (cid:3)5(cid:3)5(cid:3)4(cid:3)4(cid:3)4(cid:3)5(cid:3)3(cid:3)4 11111 (cid:3)5(cid:3)5(cid:3)511(cid:3)5 1 1 (cid:3)(cid:3)1 1 [Ch endar (C.E.) totototototototo tototototo totototototo toto tototototo toto totototo 234orU al 36396706 67984 531536 98 43733 51 4696 f C 1822566652720519 7473727366 093507494532 4440 7291832026 3565 71616168 (cid:3)1 (cid:3)5(cid:3)5(cid:3)4(cid:3)4(cid:3)4(cid:3)5(cid:3)3(cid:3)4 11111 (cid:3)5(cid:3)5(cid:3)511(cid:3)5 1 1 (cid:3)(cid:3)1 1 ear y (cid:3)6 fP.) (cid:1)49(cid:1)48(cid:1)15(cid:1)127(cid:1)77(cid:1)323(cid:1)14(cid:1)18 (cid:1)11(cid:1)8(cid:1)19(cid:1)9(cid:1)8 (cid:1)22(cid:1)28(cid:1)24(cid:1)35(cid:1)14(cid:1)26 (cid:1)14(cid:1)18 (cid:1)54(cid:1)21(cid:1)160(cid:1)15(cid:1)6 (cid:1)6(cid:1)15 (cid:1)26(cid:1)37(cid:1)8(cid:1)22 (cid:4)10 Age s(B. 263 Year 70847127649864926399735449866128 192205202203278 7023727569974204837249 1487524 172101626272138681 5881285 1210129713231242 h,2.8 T 7 3 0 230 Age c,eCorrected (cid:1)14549(cid:1)18848(cid:1)55915(cid:1)55312(cid:1)46077(cid:1)41532(cid:1)04714(cid:1)18918 (cid:1)11252(cid:1)2668(cid:1)26319(cid:1)2649(cid:1)3388 (cid:1)08422(cid:1)33628(cid:1)05824(cid:1)48135(cid:1)54414(cid:1)26310 (cid:1)54814(cid:1)58518 (cid:1)23354(cid:1)07721(cid:1)16688(cid:1)19915(cid:1)7426 (cid:1)6496(cid:1)34615 (cid:1)27126(cid:1)35837(cid:1)3848(cid:1)30322 (cid:3)(cid:3)61yearfor 77666756 777 7 1 122 1 1111 0 1 (cid:4) Age Uncorrected (cid:1)718724(cid:1)722926(cid:1)656115(cid:1)666265(cid:1)653229(cid:1)773163(cid:1)505313(cid:1)619915 (cid:1)2643(cid:1)2734(cid:1)2815(cid:1)2803(cid:1)3463 (cid:1)709420(cid:1)735223(cid:1)706822(cid:1)5156(cid:1)5575(cid:1)731925 (cid:1)15598(cid:1)6035 (cid:1)2876(cid:1)10977(cid:1)284623(cid:1)221010(cid:1)7455 (cid:1)6525(cid:1)13596 (cid:1)12968(cid:1)13949(cid:1)13896(cid:1)13238 ntsare9.1577 a st 230Table2.U-ThCompositionsandThAgesforFossilCoralSamplesofMyanmarbyMC-ICP-MS d238232234230238230232UThU[Th/U][Th/Th]initial a,bcdSampleIDTypeactivityppmppbpptcorrected NorthernRamreeIsland(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)KPU-102aCoralremains24982283012148.91.50.07310.000210655(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)KPU-103Coralremains2516227649149.91.70.07360.000211065(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)KPU-104bCoral(Round)24902937148.91.70.06690.0001295682107(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)KPU-106Coralmicroatoll25052727338150.11.60.06800.00063874(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)1.70.06670.00035783KPU-107Coralmicroatoll26722509013149.7(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)KPU-108Coral,overturned267322255974148.41.70.07840.00061531(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)KPU-109aCoralmicroatoll273024467147.91.80.05190.0001524184(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)KPU-110aCoralmicroatoll262517677150.81.10.06340.0001358533CentralWesternRamreeIsland(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)ZC-16aCoral302338994146.91.60.02760.000031542(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)ZC-118aCoralmicroatoll?292935667149.01.70.02870.000042455(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)ZC-119aCoralmicroatoll?2740413548151.42.20.02960.00005992(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)ZC-119-1Coralmicroatoll?3332423167151.92.00.02950.00003701(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)ZC-04Coralblock304536045.4149.91.50.03630.000033034SouthernRamreeIsland(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)KYM-125aCoral262837167150.11.80.07220.0002437843(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)KYM-125bCoral2708311677150.02.20.07480.0002286619(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)KYM-127Coral271537107150.22.00.07200.0002454647(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)TK-130Coralblock2957527078153.12.50.05430.00006981(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)TK-131Coralblock2933410428153.12.20.05870.000052733(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)TK-132Coralblock286146427154.62.40.07470.0002549761Ka-Ma,ChedubaIsland(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)KM-143aCoralremains292649057147.22.20.01630.000078688(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)KM-144aCoralmicroatoll2733412897146.92.30.00630.000052212Ka-I,ChedubaIsland(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)KI-152aCoralmicroatoll21673310610149.22.30.00300.00006351(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)KI-154aCoralmicroatoll2390412727146.92.40.01150.000073563(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)KI-155aCoralmicroatoll21403903824150.42.70.02960.00021161(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)KI-156aCoral189635347149.62.50.02300.00009135018(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)KI-157aCoralmicroatoll233942066148.32.50.00780.00005146044Sachet,ChedubaIsland(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)SC-150aCoralremains254342067148.22.20.00680.00005139546(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)SC-151aCoral3056510837149.62.20.01420.000066625Man-Aung,ChedubaIsland(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)2MA-135Coral3142420569147.62.10.01350.00008341(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)MA-136aCoral,overturned2688325549145.21.90.01450.000092522(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)MA-136bCoral,overturned263333547148.91.90.01450.00006178234(cid:1)(cid:1)(cid:1)(cid:1)(cid:1)MA-138bCoral2393512847147.62.90.01380.000074253 sAnalyticalerrorsare2ofthemean.d(cid:3)(cid:4)a234234238U=([U/U]1)1000.activitylddd(cid:4)b234230234234234*TUcorrectedwascalculatedbasedonThage(T),i.e.,U=Ue,andTiscorrectedage.initialinitialmeasured(cid:3)l(cid:3)l(cid:3)l(cid:3)dlll(cid:3)c230238230T234(230234)T[Th/U]=1e+(U/1000)[/(-)](1e),whereTistheage.Decayconactivitymeasured230230234(cid:3)(cid:3)(cid:4)101238yearforU[Jaffeyetal.,1971].etal.,2000],and1.5512510d230230232ThedegreeofdetritalThcontaminationisindicatedbythe[Th/Th]atomicratioinsteadoftheactivityratio.(cid:1)e230232AgecorrectionswerecalculatedusinganestimatedatomicTh/Thratioof6.56.5ppm[Zachariasenetal.,1999].fYearsbeforepresent(B.P.)isreferencedto1950C.E. 1284 WANGETAL:DEFORMATIONOF1762ARAKANEARTHQUAKE Figure5. ThepatternsofmoderndrainagesandmarineterracesofthecentralwesterncoastofRamree Islandalsoshowaneastwardtilt.Thefluvialplainandterracesnortheastofthefoothillsshowcleareast- wardtiltingintheanalysisofaerialphotosanddrainagepatterns.Westofthefoothills,theelevationofthe lowestterracebetweenthevillagesofKyauk-Ka-Le(Kyauk-Galé)andKon-Baung-Gyi(Kon-Baung)was describedbyMallet [1878] to be ~6m above the water level at the timeof his visit. Figure6. OurfieldsurveysitesatthecentralwesternRamreecoast.(a)TheU-Thagesofupliftedcoral microatollsonthe lowest terrace (T1)show that ~5m of land-level change occurred in the18thcentury, mostlikelyduringthe1762earthquake.TheprofilelocationisindicatedinFigure5.Allsealevelindica- tors on T1 show identical amount of land-level change relative to their equivalent tide-water level. (b) Photograph and (c) line sketch of site ZC-04. Here a dated coral block within the terrace deposits of T1 also indicates that an uplift event occurred after the 17th century. The terrace surface elevation at ZC- 04ishigherthanthatattheprevioussite(ZC-16),at5.5mabovethemodernshorelineangle.Thisyields theminimalamountofland-levelchangeatZC-04.However,boththeamountsatZC-16andZC-04are lower thanthe 19th century account of Mallet [1878]. 1285 WANGETAL:DEFORMATIONOF1762ARAKANEARTHQUAKE thebandfurtherinsidethecoloniesalsoyieldadateofdeath increase in uplift from the northern to the central western verycloseto1762C.E.(Table2).Thisalsoimpliesthatthe coastindicatethetiltingresultsfromthegrowthofthedou- T1 wave-cut platform on the central western coast formed bly plunging anticline that has raised the island (Figure 1). long before 1762, then supported coral growth through the 4.1.3. Southern Ramree Island decades before the uplift in 1762. [39] Geomorphic evidence for young land-level changes [37] Thisextraordinaryamountofland-levelchangealong at the southern tip of the southwestern Ramree coast isalso central western section of Ramree Island attracted attention veryclear,althoughthetimingofthelastuplifteventisnot as early as the mid-19th century. This is where Mallet as well constrained. A flight of marineterraces between the [1878] observed a “raised beach about 20 feet above the currentshorelineandthewesternfoothillsindicatesprogres- sea” during his survey in 1877. Following the description siveupliftnearthesmallvillageofTet-Kaw(Figure7).The and the map in his report, we were able to survey the same amount of the last emergence is well constrained by the section of the coast between the villages of Kyauk-Ka-Le elevationofT1’sshorelineangle,whichisabout1.4–2mabove (Kyauk-Galé) and Kon-Baung-Gyi (Kon-Baung; Figure 5). its modern analogue at MHWS. The elevation of a group of WefoundthesurfaceofT-1thereis5.5mabovethecurrent small surf notches on a sandstone ridge is similar to that of shoreline angle, with a very thin sedimentary cover theshorelineangle(Figures8band8c).Thesefeaturessuggest (Figure6b).Althoughthisisourhighestmeasurementalong asmalleruplift,only1.4–2mfromthecurrentMHWS. thissectionofthecoast,itisslightlylowerthanMallet’sob- [40] AlackofdatableinsitumaterialsassociatedwiththeT1 servationin1877.The17thcenturyageofacoralfragment surface precluded determination of a date for the most recent (ZC-04) within the thin sediments is consistent with the upliftinthisarea.Agriculturalactivitiesappeartohaveremoved terrace being an active wave-cut platform a century or so mostofthefossilcoralsandoystersfromtheterrace.Nonethe- before the uplift in 1762 (Figure 6band Table 2). less,severalloosecoralblockswithinthethinsedimentcapof [38] Geomorphological evidence of a progressive north- T1provide some constraint.These coralblocksare10–30cm eastward tilt is even clearer for central Ramree Island than in diameter, significantly bigger than regular beach gravels itisforthenorthernsectoroftheisland.Alongmostofthis (<5cm)withinthemodernstormdeposits.Thus,itisunlikely section of coast, two to three major terrace treads along the thattheyweretransportedbynormalstormsorcyclonesuponto southwestern coast contrast with only one major terrace in theT1surfaceafteritsemergence.Moreover,becausewedid the northeast (Figure 5). Moreover, the highest terraces of notfindanyevidenceoftsunamidepositsalongthecoast,we the southwestern foothills show clear northeastward tilting believe that these coral blocks are not tsunami deposits, but of their surface in stereoscopic aerial photos. This eastward weredepositedwhentheT1surfacewasstilltheactivewave- tilt is also consistent with the predominance of northeast- cut platform. Thus, the youngest age of these coral blocks ward-flowing drainage networks over much smaller creeks mayrepresentthemaximumageoftheformationofT1. flowingtothesouthwesterncoast,similartowhathavebeen [41] The U-Th ages of these coral blocks range from observed on Makira (San Cristobal) Island, the Solomon early mid-Holocene to the 16th century (Table 2, TK-130 Islands [Chen et al., 2011]. The northeastward tilt of both to TK-132). The youngest U-Th age (1495–1564 C.E.) the northern and central sectors of Ramree Island and the provides a maximum limiting age for wave action on T1. Figure 7. The different geomorphic characteristics of the southwestern and southeastern Ramree coast indicate the long-term uplift and eastward tilt of southern Ramree Island. Similar to the northeastern Ramree coast, the U-Th age of fossil corals on T1 (KYM-125) suggests that the lowest terrace formed during the mid-Holocene period. However, the geomorphic characteristics west of the foothills are very different. A flight of marine terraces along the western coast suggests successive uplift during the past thousandsofyears.ColoredlinesaretopographicprofilesacrossthesemarineterracesshowninFigure8a. Red dots show the locations of dated corals. 1286