Marine Geology 257 (2009) 54–64

Contents lists available at ScienceDirect

Marine Geology

j ourna l homepage: www.e lsev ie r.com/ locate /margeo

Particle-size distribution of inferred tsunami deposits inSur Lagoon, Sultanate of Oman

S.V. Donato a, E.G. Reinhardt a,⁎, J.I. Boyce a, J.E. Pilarczyk a, B.P. Jupp b

a McMaster University, School of Geography and Earth Sciences, Hamilton, ON, Canada L8S 4K1b Marine Pollution and Coastal Zone Management Section, MRMEWR, P.O. Box 175 Mina Al Fahal P.C, 116, Muscat, Sultanate of Oman

⁎ Corresponding author.E-mail address: ereinhar@mcmaster.ca (E.G. Reinhar

0025-3227/$ – see front matter © 2008 Elsevier B.V. Adoi:10.1016/j.margeo.2008.10.012

a b s t r a c t

a r t i c l e i n f o

Article history:

The sedimentary characteris Received 8 February 2008Received in revised form 15 October 2008Accepted 23 October 2008

Keywords:Sur LagoonOmantsunami depositsparticle sizecluster analysis

tics of shell beds within an interpreted tsunami deposit from Sur Lagoon, Omanwere examined using shell taphonomy and high-resolution particle-size analysis. The tsunami bed wasdeposited by the 28 November, 1945 tsunami generated by the Makran subduction zone. Q-mode clusteranalysis of particle-size data was evaluated as a means of discriminating individual tsunamite shell layersfrom lagoonal intertidal deposits. Results showed that the tsunami shell bed was more poorly sorted, andheterogeneous (in both the digested and undigested samples) than the background lagoonal sediments. Thetsunami bed thickness correlated generally with the thickness of the shell-bed, however, cluster analysisextended the tsunami unit several centimeters above or below the shell bed in some cores. The particle-sizeanalysis also showed subtle textural trends in the tsunami unit, suggesting that the tsunami bed wasdeposited in several distinctive phases during tsunami incursion into Sur Lagoon. The findings indicate thatcluster analysis of particle-size data can be used to identify tsunami beds in intertidal environments andholds potential for identifying paleotsunami deposits in sediments from embayed intertidal–subtidalsiliciclastic systems where obvious sedimentary structures may be absent.

© 2008 Elsevier B.V. All rights reserved.

1. Introduction

The northeastern coast of Oman is locatedwithin ~120–170 km of theseismically-activeMakran subduction zone (Fig.1). On 28 November,1945a large thrust earthquake (Mw 8.1) epicentered on the eastern portion ofthe subduction zone off the modern coast of Pakistan, produced a largetsunami that impacted coastlines along India, Pakistan, Iran, and Oman(Byrne et al.,1992). Historical records detailing the tsunami and its impacton theOmani coast are limited, but the eventwas recorded in Indian newsreports (Ambreseys and Melville, 1982). Recently, Donato et al. (2008)identified and described thick, laterally extensive bivalve-rich shell bedsdeposited by the 1945 tsunami in Sur Lagoon ~500 km from the epicentreon the eastern margin of Oman (Fig. 1). In this paper, we further test thetsunami origin of the deposit by using Q-mode cluster analysis of high-resolution particle-size data. Our results demonstrate that cluster analysisis an effective method for recognizing subtle textural changes within thetsunami bed and for discriminating tsunami from lagoonal inter-tidaldeposits.

dt).

ll rights reserved.

1.1. The 1945 tsunami event

Historical records documenting tsunami impacts along the Omancoastline are lacking due to a sparsely-populated coastline and limitedcommunication with larger centres (i.e. Muscat). Records have provenelusive in Oman for the tsunami generated on 28 November 1945. Theepicentre was located on the seismically-active eastern portion of theMakran subduction zone (Fig. 1; Byrne et al., 1992). Details of thisevent were well documented by Indian and Pakistani media andmeteorological services, as the earthquake and tsunami caused over4000 deaths and extensive damage along the coastlines of Pakistan,Iran, western India, and Oman, with Pakistan recording a run-upheight of 13 m (Pararas-Carayannis, 2006). Tsunami damage andcasualties were reported in Muscat ~200 km north of Sur (Pararas-Carayannis, 2006, p. 362), although no supporting references wereprovided to substantiate the claim. However, a recent interview (May21st, 2008) by Dr. Abdulrahman A-Harthi (Sultan Qaboos University-Dept of Geology) and Dr. Emile Okal (Northwestern University,Department of Earth and Planetary Sciences) provide some corrobor-ating information. Mr. Ahmed M.J. Al-Alawi (approx. 85 yrs old) whowas living at Sur during 1945, witnessed the event and recalled theflooding and a tsumani water level rise of ca. 3 m above mean sea-level (msl). More interviews need to be conducted to corroborate thiswater level estimate, but it does indicate that the tsunami had someeffect at Sur.

Fig. 1. Location of Sur Lagoon, Oman. Location of Makran subduction zone and 1945 earthquake epicentre indicated (after Byrne et al., 1992 and Okal and Synolakis, 2008).

55S.V. Donato et al. / Marine Geology 257 (2009) 54–64

Models for the 1945 event show the tsunami predominantly directedto the south (Dominey-Howes et al., 2007) but also show it impacting theeastern corner of the Arabian Peninsula (Heidarzadeh et al., 2008b). Thesimulations, using the fault parameters of Bryne et al. (1992) showed thatthe observed tsunami was too large to be explained by the proposedsource mechanism. Other possible mechanisms include a submarinelandslide (Dominey-Howes, Cummins and Burbidge, 2007; Ambraseys &Melville,1982) and large displacements on splay faults (Heidarzadeh et al.,2008a). Heidarzadeh et al. (2008a) concluded that theMakranTrench hasthe potential for generating tsunamiswith considerable run-up heights infar-field areas. Additional simulations have been made for two hypothe-ticalMakran earthquakes (1765,1851) larger than theone in 1945but dataregarding the size and location of the events are scarce (Quittmeyer andJacob, 1979; Okal and Synolakis, 2008; Ambraseys & Melville 1982).

1.2. Tsunami shell taphonomy

Donato et al. (2008) identified the shell bed in Sur lagoon as atsunami deposit due to its distinctive shell taphonomy, shellprovenance, thickness and wide lateral extent in the lagoon. Thetaxonomic composition (and habitat) of the bivalves in the shell layersvaried due to proximity to source areas (e.g. lagoon vs offshore) butthe abundance of whole, articulated specimens and angular fragmentsin the layer was very distinctive. These consistent taphonomiccharacters were found from core to core (n=8) over a wide area(N1 km2) and at a similar depth (20–40 cm below modern surface)indicating a singular, large-scale event. The shell taphonomic traitsindicate live transport of the articulated shells, breakage of the shellsthrough turbulent flow and burial without subsequent reworking. Asimilar shell deposit in the Mediterranean (Caesarea, Israel) with nearidentical taphonomic characters was ascribed to an older tsunami (115A.D.; Reinhardt et al., 2006). The Sur deposit was attributed to the1945 tsunami based on its shallow depth below the sediment surfaceas no direct dating evidence was found (210Pb dating provedineffective; Donato et al., 2008).

Alternate interpretations on the origin of the shell unit at Sur relateto tidal channel migration and storm deposition, which can berejected (or minimized) based on the documented taphonomiccharacters. The shell bed was not deposited as a channel lag as shellconcentrations in such contexts do not contain articulated specimensand the shells are often heavily bioencrusted, bioeroded and rounded(e.g. Meldahl and Cutler, 1992). Articulated bivalves left exposed intidal channels disarticulate through tidal reworking and have littlechance of burial with both valves together.

Similarly, the distinctive taphonomic characters and the extensivesheet-like distribution of the bed in the lagoon were used to argue

against a storm origin. The presence of many articulated offshore andlagoonal bivalve species out of life position (along with angularfragments) is more consistent with rapid (minutes) deposition andburial by a tsunami event rather than more prolonged stormdeposition over several hours. The exhumation of offshore andlagoonal bivalves accompanied with suspended load transport anddeposition of shells in a massive, sheet-like geometry does not appearto be typical of storm deposition (e.g. see taphonomic charactersdescribed by Boyajian and Thayer, 1995; Meldahl and Cutler, 1992).Storm shell concentrations tend to form in more localized units orlenses (often shore parallel; Meldahl and Cutler, 1992) rather thanextensive sheet-like beds and often have a biofabric (shells are nested,imbricated and size-sorted; eg. compare with Meldahl and Cutler,1992 and see Brett, 2003 for a summary).

1.3. Sedimentological indicators

Tsunami deposits are difficult to identify in arid siliciclasticsandflat coastal settings due to numerous erosive post-depositionalprocesses acting on the sediment. The action of daily tides, wind-driven waves, storm waves, and even localized processes such asbioturbation can potentially erase traces of these events from thesediment record in short periods (i.e. 10 years or less, Dominey-Howeset al., 2006). In arid terrestrial settings, the deposits are furtherthreatened by wind erosion, deflation and fluvial sheet-wash erosionduring high-discharge flood events.

Perhaps due to the perceived lack of preservation potential, therehave been few attempts to identify tsunami deposits in aridenvironments outside the Mediterranean (e.g., Scheffers and Kelletat,2003; Scheffers and Kelletat, 2005; Scheffers and Scheffers, 2007).Consequently, most of the diagnostic characteristics used to identifytsunami deposits globally have been derived from temperate climateswith terrestrial deposits, coast proximal lakes, marshes and estuariesbeing the favored environments (Atwater et al., 2005). Most of thisprevious work has relied on the presence of intrusive marinesediment layers (sand sheets, gravel, etc.) as tsunami indicators,however, storms can also produce these deposits and recent researchhas tried to develop a criteria for separating the two (eg. Goff et al.,2004; Kortekaas and Dawson, 2007; Morton et al., 2007). Tsunamideposits are typically produced through suspended load transport andoften display one or more of the following characteristics: laterallyextensive and thickly-bedded sand sheets (often structureless)showing landward thinning, normal or inverse grading, and presenceof marine microfossils and macrofauna (Dawson and Smith, 2000;Goff et al., 2001; Tuttle et al., 2004; Morton et al., 2007). In contrast,storm deposits typically are produced through bedload transport and

Fig. 2. A. Satellite image of Sur Lagoon showing tidal channels and small birds foot delta. B. Digital elevation model (DEM) of Sur Lagoon. Core locations (1–8) shown by circles. Nodalpoint used to measure core distance from lagoon entrance depicted in Fig. 5.

56 S.V. Donato et al. / Marine Geology 257 (2009) 54–64

display extensive planar laminae, forsets, troughs, and climbingripples with a maximum bed thickness close to shore thinningabruptly landward (Morton et al., 2007).

Recently, research has been devoted to identifying tsunami units inshallow coastal and shelf settings where separation from stormdeposits becomes more problematic (i.e. Fujiwara et al., 2000; Prattand Bordonaro, 2007; Noda et al., 2007; Reinhardt et al., 2006, Donatoet al., 2008). Diagnostic sedimentary features of shallow watertsunami deposits include sharp erosive bases (Nanayama et al.,2007), graded beds (e.g. Noda et al., 2007), hummocky cross-stratification (Fujiwara and Kamataki, 2007) and mud drapes(Fujiwara and Kamataki, 2007).

Particle-size analysis of tsunami deposits has been employedpreviously (e.g. Switzer et al., 2005; Nanayama and Shigeno, 2006;Hawkes et al., 2007 Kortekaas and Dawson, 2007; Morton et al., 2007;Paris et al., 2007) but these studies have relied on conventionalgranulometric statistics (mean, median, mode, and standard devia-tion) to identify distinct tsunami units. Despite being widely acceptedfor their ease of use, these statistics lack the ability to characterizepolymodal distributions and to discriminate other significant texturalcharacteristics of poorly-sorted tsunami units. An alternate approach,as demonstrated in this paper, is to utilize high-resolution particle-size analysis (i.e. laser diffraction methods) and plotting of theparticle-size distributions (PSDs) as contoured surface plots (Beierle etal., 2002; van Hengstum et al., 2007). PSD plots display the particle-size distribution as an interpolated 2-dimensional surface with therelative percentage abundance of each particle-size class representedusing contours or colour shading (e.g. Fig. 4). The advantage overconventional plotting methods (i.e. cumulative particle-size curves)is that the complete PSD is displayed, allowing subtle textural trendsand multiple modes to be recognized. With the advent of laserparticle-size analysis (e.g. Coulter LS 230) high-resolution PSD datacan be obtained rapidly and cost-effectively. The wider adoption ofthe method, however, has been hindered by a reliance on conven-tional statistics and inadequate PSD plotting and presentationmethods (e.g. see Fig. 8 in Kortekaas and Dawson (2007). In addition,

statistical discrimination of the PSDs through cluster analysisprovides an objective means for classifying textural trends and foridentifying tsunami units in the sedimentary record (Hamilton,2007). Here we use it to detect particle-size contrasts betweentsunami layers and ‘background’ sediments in the lagoon and test theboundaries for the tsunami units interpreted by Donato et al. (2008)based on the shell taphonomy.

2. Regional setting

Sur Lagoon is a horseshoe-shaped lagoon covering approximately12 km2 on the eastern-most promontory of the Sultanate of Oman(Figs. 1, 2) and connects with the Gulf of Oman via a single entrancechannel. The lagoon lies approximately 500 km WSW of the Makrantrench, where the Arabian plate is being subducted below theEurasian plate along the ~1000 km long convergent boundary. Thenarrow shelf drops below 200 m depth within 5 km offshore of Sur(Szuman et al., 2006).

There is a paucity of Holocene sea-level data for theOmani coast in theenvirons of Sur. Lambeck (1996) modelled sea-level change for the Gulfwhich included a section between Musadan and Muscat and found thatsea level reached near present levels at approx. 6 kyrs with minimalevidence for higher than present sea-levels (b1 m) for the last 5000 yrs(see references therein).

The Gulf of Oman coastline is dominated by khors or lagoons, whichhave formed in themouth of wadis with the evolution of barrier bars andspits during the late Holocene. A full discussion of arid coastalenvironments is provided in Alsharhan and Kendall (2003) and Warren(2006)which both focus on systems in the ArabianGulf. TheGulf of Omancontains similar environments although they aremore clastic due towadisediment inputs from the Oman Mountains.

Sur lagoon ismicrotidal (mean tidal amplitude ~1.2m) andmost of thelagoon surface is exposed during low tide (Fig. 3). The lagoon surfaceelevationvaries between ≈±2mmean sea level (m.s.l.; Fig. 2). Several sub-tidal channelsmeander through the lagoon andmangroves are present onthe south, east, and northwestern lagoon margins. In the wadi mouths,

Fig. 3. A. Photo of tsunami unit (arrowed line) in Core 1 showing light brown coloration of unit. Diametre of core is 7.5 cm. B. Photograph of lagoon margin looking NW. Photo showsmangrove and muddy sabkha-like environment in wadi channel located in SW corner of Fig. 2A. C. Photo looking N near Core 1 (Fig. 2A) showing the sandflat environment of thelagoon basin and the barrier with buildings on top. (Photo credits: R Rothaus). (For interpretation of the references to colour in this figure legend, the reader is referred to the webversion of this article.)

57S.V. Donato et al. / Marine Geology 257 (2009) 54–64

behind the mangroves, muddy sabkha environments are present whichgrade to gravel moving up-wadi (Fig. 3B). The lagoon surface sediment(tidal sand flat) is dominantly fine to very-fine sand, while the elevatedlagoon margins in and behind the mangroves are mud-rich. On the sandflats there are few bivalve shells, but cerithid gastropods are abundantespecially in the small tidal pools (Fig. 3C). Crab burrows frequent the area.

The lagoon is enclosed by Paleocene–Eocene aged highlands (carbo-nates andwadi alluvium) that are incised by several largewadis that enterthe lagoon from the south (Fig. 2). The largest,Wadi Shamah, has created asmall delta in the southeastern arm, and sections are subaerially exposedat high tide and populated by thin mangrove thickets. The mouth of thelagoon sits at the eastern edge of a low-lying sand spit (2–3mabovem.s.l.)which has built out from exposed bedrock to the west (approximately1 km long by 800 m wide) through a combination of wave and windaction. The spit is characteristic of the spit-lagoon subsystemof Alsharhanand El-Sammak, (2004) or the barrier-lagoon coast of Feng et al. (2007).The spit may have formed around exposed bedrock that has been

subsequently covered by sand. The town of Sur (pop. ~77,000) is built onthe barrier spit and extends alongmuch of the lagoonmargin andmarineshoreline (Fig. 2).

Sur is not on the dominant cyclone pathway andGonu (June 1st to 7th2007)was the strongest cyclone on record to strike the Arabian Peninsula(windsof 150km/hrwhen it hit the coastline at Sur; IndianMeteorologicalDepartment, 2007). A visit to Sur in May 2008 showed that the cycloneproduced no shell accumulations in the lagoon, but the heavy rainfall (upto 610 mm) washed away the road and eroded much of the delta in thesoutheastern wadi (Wadi Shamah). However, most of the flow movedseaward bypassing the main lagoon basin causing little to no effect.

3. Methods

Eight sediment cores were collected in Sur Lagoon (March 2006;Fig. 2), and the coring locations were selected based on the presence ofthe allochthonous bivalve tsunami bed (TB) found close to the surface

Fig. 4. Lithologic logs for eight sediment cores collected in Sur Lagoon. Dashed lines represent the interpreted upper and lower boundaries of the tsunami bed identified in Donato et al.(2008).High-resolutionparticle-size distribution is shownasФ surfaceplots for bothdigested (D) andundigested (UD) sediment samples fromeach core.Differential volumebypercentagerepresented by the following colour scheme: blue=absent tominimal and red-pink=common–most common. Core top elevations (E) are indicated in cm abovem.s.l. Core scale shown incm below surface. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

58 S.V. Donato et al. / Marine Geology 257 (2009) 54–64

Fig. 5. Tsunami bed characters vs linear distance from nodal point in lagoon entrance. A) Elevation (msl) of tsunami bed, lagoon sediment surface and bivalve distributions in the shellunits (Donato et al., 2008). Dominant species grouped based on habitat (Bosch et al., 1995) include: Intertidal–Hiatula ruppelliana, Marcia mamorata, Marcia opima, Protapes sp.;Intertidal to Subtidal–Amiantis umbonella, Anodontia edentula, Diplodonta ct globosa, Dosina alta, Dosina contracta, Meropesta nicobarica, Serratina capsoides; Subtidal–Anadarauropigimelana, Diplodonta subrotunda, Dosinia ceylonica, Ervilia purpurea, Protapes cor, Protapes sinuosa, Tapes deshayesi, Tellina palatum. B) Tsunami bed thickness based on particle-size distributions and shell taphonomy (Donato et al., 2008) with the trend line representing the shell unit relationship with distance from node. C) Average mean and standarddeviation of the particle-size distribution (undigested) in the tsunami beds with the trend line representing the standard deviation relationship. D) Average skewness values(undigested) for the tsunami beds.

59S.V. Donato et al. / Marine Geology 257 (2009) 54–64

or exposed in tidal creek sections (Donato et al. (2008). The shell unitcould be traced along the tidal creeks (b100 m) and in addition manytest pits were dug checking the continuity of the unit (approx. 15). Ofthese test pits (40–60 cm deep), eight were selected to collect largesamples of bivalves and eight short percussion cores were collected todocument the detailed stratigraphy and continuity of the shell bed.

Surface-elevation data were collected within the lagoon using adifferential GPS (D-GPS) and interpolated to create a digital elevationmodel (DEM) of the surface topography (Fig. 2). The DEM surveyfollowed predetermined north–south transects at 100 m spacing andeast–west survey lines at 300mapart. The D-GPS surveywas conductedduring low tideandareas deeper than1.5mwere surveyedbyboatusinga 200 kHz echosounder. The D-GPS survey data were post-processedusing corrections from a fixed base station receiver to yield sub-centimetre survey positioning and elevation control. The sonar surveywas conducted in the lagoon tidal channels and in the nearshore outside

the lagoon to ~1 km from the coastline. The DEM was employed todetermine the core elevationsandprovided elevationdatumfordetailedstratigraphic correlation of the shell beds across the lagoon.

Cores were photographed immediately after opening and corelithology, colour, and general features were described and recordedprior to sub-sampling. Samples were analyzed for particle-sizedistribution using a Coulter LS 230 instrument and statistical datacomputed using the Fraunhofer optical model. Samples were analyzedin two states: 1) processed (sediment was subjected to HCl and H2O2

digestion to remove carbonates and organic components respec-tively), 2) unprocessed (sediment was sieved and run unaltered).Digested samples were prepared and run following the techniques ofvan Hengstum et al. (2007). Sub-samples of ~5–10 cm3 were sieved toremove the coarse fraction greater than 2000 µm (maximum particle-size that can be analyzed by the LS 230). Samples were centrifugedand the supernatant was decanted prior to treatment with HCl (10%)

Fig. 6.Q-mode clustered groupings (Ward'smethod) for undigested (e.g. Core 1u) and digested (e.g. Core 5d). Sample intervals shown are in cm below core top. Similarity shown alongtop axis and dashed line represents the selected cluster groupings. Lithologic key represents dominant sediment type at each sample interval.

60 S.V. Donato et al. / Marine Geology 257 (2009) 54–64

at room temperature to remove carbonate material. When thesamples were no longer reactive, they were treated with 40% H2O2

in a water bath (60 °C) to oxidize and remove any organic matter(Chambers and Upchurch, 1979; Wang and Ke, 1997). Samples werethen centrifuged, decanted, mixed and allowed to dry to a moist paste

to subsample for analysis. Sub-samples were mixed with a 1% sodiumhexametaphosphate solution and ultrasonically disaggregated beforeanalysis.

Conventional particle-size statistics (i.e. mean, median, mode, andstandard deviation) were all calculated in µm and phi (Ф) units.

Fig. 7. Conventional statistics for digested and undigested cores plotted against depth below surface (cm). Mean, mode, and standard deviation are plotted in µm.

61S.V. Donato et al. / Marine Geology 257 (2009) 54–64

Surface plots of the complete particle-size distributions (PSDs) foreach core were then generated following the methods describedBeierle et al. (2002). The PSDs were transformed to a phi log scale andthen gridded using a triangular irregular network algorithm (TIN;Sambridge et al., 1995). The surface plots allow visualization of subtlechanges in downcore particle-size distribution and the presence ofmultiple modes that are problematic for traditional statistics (mean,mode, standard deviation, etc.; i.e. Inman, 1952; Folk and Ward, 1957;Switzer et al., 2005). Multivariate analysis of the particle-size datawasconducted using Ward's algorithm (which emphasizes minimalwithin-group variance; Hammer and Harper, 2006) to produce a Q-

mode cluster dendrogram in the statistical program PAST™. Thisclustering technique creates hierarchical groupings based uponsimilarity determined from equally weighted Euclidean distancemeasurements of the entire grain size (0.0375–2000 µm) distributionfor each sample analyzed. A full discussion of cluster analysis and itsapplication to PSDs is provided in Hamilton (2007).

In order to investigate spatial changes in the tsunami bedcharacteristics, the derived particle-size parameters, tsunami bedthicknesses, and elevations were plotted along a E–W transect acrossthe lagoon (Fig. 5). A linear regression was then performed to identifytrends within the various parameters with distance from a nodal point

62 S.V. Donato et al. / Marine Geology 257 (2009) 54–64

selected close to the lagoon entrance channel. Other spatial relation-ships were examined (i.e. across barrier) but the channel nodal pointshowed the best relationship.

4. Results

4.1. Core lithology and taphonomy

Lithologic logs and particle-size surface plots for the eight corescollected in Sur Lagoon are shown in Fig. 4. The generalized corelithostratigraphy comprises lagoonal silty sand units interbedded withclast-supported shell bedswith shells ranging from5.6mm to N70mm insize (Donato et al., 2008). The shell beds ranged in thickness from1–25 cm(thinning primarily with distance from the lagoon entrance; Fig. 5), butcommonly did not exceed 10 cm in thickness.With the exception of Cores2, 5, and 8, all cores contained multiple shell beds (Fig. 4). Aside from theshell beds, the cores hadnodiscernible bedding structures. The siliciclasticcomponent (sand and silt) was commonly light to medium grey colour,while sediment in the shell beds was commonly reddish-brown in colour(Fig. 3A). The overall particle-size in cores ranged from coarse silt tomedium sand (undigested samples reached very-coarse sand), similar tothe modern lagoon surface sediment (Donato, 2008).

The shell units were interpreted as TBs based on the bivalvetaphonomy as described by Donato et al. (2008) and documented indetail in the data repository items of that publication. The TBs variedfrom 5–25 cm in thickness and thinned with distance from the lagoonmouth (Fig. 5). The average values of the diagnostic taphonomiccharacters for material in the TB were articulation (7%), whole valve(36%), fragmentation (58%), angular fragmentation (54%), edge round-ing (9%), encrustation (5%), and dissolution (2%). The allochthonousbivalve assemblage included intertidal –mudflat (23%), intertidal–lower shore (48%), and lower shore–offshore (29%) species (Fig. 5).Coarse shells and fragments greater than 2000 µm were largecomponents of the TB, representing between 16–65% of all identifiablefragments for each core. The high number of angular fragments withthe articulated bivalves allowed easy identification of the TB in thefield and in cores.

4.2. Particle-size distribution of digested and undigested data

Laser particle-size analysis conducted on both undigested anddigested sediment samples for all cores (Figs. 3, 6) showed markeddifferences between the PSDs. In general, the particle-size statistics foreach undigested sample reflected visual observations in the cores (i.e.fine sands with rare shell vs. shell- rich sands vs. shell beds), while thedigested samples (shell and organics removed) were generallyhomogenous throughout each core regardless of stratigraphic horizon(i.e. shell bed vs. clean sands). Undigested samples exhibit largerstandard deviations and typically contain a higher percentage of fine-grained particles within the TB compared to their digested counter-parts. The digested sediment from the TB samples was typically very-fine to fine-grained sand, which is consistent with the dominantparticle-size of the modern lagoon surface sediment (Donato, 2008).

The PSD plots (Fig. 4) show the differential volume (%) for bothsample runs and, as expected, particle-size of the undigested sampleswere generally coarser than those of the digested samples (oftenhaving mean values twice that of the digested samples) due to thepresence of the b2000 µm carbonate shell fraction. Conventionalparticle-size statistics showed that most undigested cores had meanvalues exceeding 200 µmwhile only digested Core 6 had any sampleswith a siliciclastic mean particle-size greater than 200 µm (Fig. 7).Modal values showed similar trends between undigested and digestedsamples but the variability in modal values and number of modes wasgreater for the undigested samples. In Core 8 for example, the PSD plotshows a distinctive bi-modal distributionwith modal peaks at 4 and 1phi for the TB at 17–27 cm. The corresponding summary statistics for

Core 8 (Fig. 7), in contrast, show only a large increase in the averagemodal and mean values within the tsunami bed. The conventionalsummary statistics assume a unimodal distribution and clearly do notdiscriminate the presence of the multiple modes.

The PSD plots showed distinct differences in differential volume ofa given particle-size between digested and undigested samples for agiven core (Fig. 4). Although both PSD plots tended to share the similarskewness and kurtosis values, the size distributions varied signifi-cantly in most cores between the TB and non-TB units. Most TB unitswere characterized by less peaked PSDs compared to the lagoonalsediments. This trend also existed between digested vs. undigestedsamples, where the digested samples had more peaked curves due tothe absence of organic and shell material. Overall though, the digestedsamples showed only minor variation in the PSD down-core while theundigested samples showed pronounced variation. PSD plots forseveral undigested cores (i.e. Cores 5, 6) showed an increase in thevolume ofmud and very coarse size particles, indicating that sedimentbecame more poorly sorted in the TB intervals than above and below.This was typically not mirrored in the digested PSD plots. The PSDplots of several undigested cores also show obvious to subtle gradingtrends. The PSD of Core 8 for example became finer immediatelybelow the TB, but became inversely graded and bi-modal in the TB,and was normally graded for ~10 cm above it.

4.3. Cluster analysis of PSD data

Cluster analysis of the PSD data produced groupings of samplesthat isolated the shell TB as a distinct unit in all cores except Core 7(Fig. 6). This strong correspondence was consistent between undi-gested and digested clusters indicating that the PSD for siliciclastic(digested) samples and undigested samples were distinct to the TB.Sample clusters were determined by similarity levels, which wereapproximately the same among both digested and undigestedclusters. Most dendrograms had between 2 to 4 clusters whichcontained two or more samples, while single non-clustered sampleswere considered outliers (e.g. sample 1.5 6d and 20.5 4d). A two to fourcluster level was chosen depending on the core as the TB formed adistinctive cluster from the normally deposited (background) sedi-ments which comprised the remaining cluster(s).

In general, most dendrograms grouped the TB separately fromoverlying and underlying sediment in each core, and did not include anyof the other shell units not formed by tsunamigenic processes (Fig. 6).The clusters were typically consistent between digested and undigesteddendrograms, but the undigested dendrograms clustered in greateragreementwith visual observations than did the digested core samples.For example, the dendrogram for Core 2u (undigested) had two clustersat a−5 similarity. One cluster was composed entirely of the shell-richsamples from theTBdetermined byDonato et al. (2008),while the otherwas composed of two sub-clusters, which were split into a shell-richsand, and a shell poor sand that loosely corresponded to sample depthbelow surface. The digested sample (also analyzed at the 2 cluster level)isolated the TB but the remaining sub-clusters showed less agreementwith visual core observations and depth below core top. Themajority ofcores behaved in this manner, with slight differences between thedigested and undigested dendrograms but ultimately clustering the TBseparately from the samples above and below (Fig. 6). The dendrogramfor Core 7 was anomalous because it clustered samples by depth, butgrouped the uppermost shell bed (5–10 years old; Donato et al., 2008)with the TB in the undigested sample. In the digested samples, theclusters separated the lowest shell bed (of the TB) from the upper twoshell beds showing a better relationship with the TB as determinedthrough the shell taphonomy.

In some instances, cluster analysis extended theupper boundaryof theTB beyond the original TB determination with the shells (Donato et al.,2008). In Core 8 it was extended by ~10 cm but cluster analysis of the PSDin Core 4 reduced the TBby ~25 cmand only included the upper shell bed.

63S.V. Donato et al. / Marine Geology 257 (2009) 54–64

In each of the remaining cores (with the exception of 7), clusteringmatched the original upper TB boundaries. In Cores 2, 3, and 6, clusteringlowered the basal boundary by a maximum of 5 cm, but the rest of thecores corresponded with the original TB. The clusters (both undigestedand digested) matched the stratigraphic range of the shell unit in mostinstances, indicating that the actual tsunami unit does not extend greatlyabove or below the TB identified by Donato et al. (2008). Clustering wasbest able to group tsunami horizons at core locations furthest from theprimary subtidal channels (i.e. Cores 1, 2, and 3) because these locationsexperienced lower flow velocities and less reworking than locationsnearer to subtidal channels, which would have acted as primary conduitsduring run-up and backwash (Umitsu et al., 2007).

5. Discussion

5.1. Tsunami sediment sources

Depositional processes in Sur Lagoonwould be primarily governedby tidal currents, wind driven waves, episodic wadi input, andoccasional storms (eg. Warren, 2006). These processes have tendedto homogenize the surface sediment distribution in the lagoonwinnowing and exporting mud-sized particles from the lagoon,leaving very fine to fine grained sands as the dominant sediment(Donato 2008). Sediment (shell and siliciclastic) within the TB wasderived from three principle sources, 1) marine, 2) lagoon, and 3)wadi. Marine sediment was mostly composed of shell material thatwas transported into the lagoon during the tsunami event mainlyfrom shallow coastal and the lagoon entrance areas. The presence ofmolluscan shells N70 mm in diameter indicates that the tsunami wasenergetic enough to exhume and transport subtidal shell materialdeep into the lagoon. The marine component also contained sandsized sediment, however, the quantity is difficult to assess as thelagoon contains similar particle-sizes (Donato, 2008). The modernlagoon is dominantly very fine to medium sand (Donato, 2008) andthis sediment would have been eroded by the tsunami as it enteredthe lagoon and incorporated into the deposit.

The mud-sized fraction present in several cores is derived from thelower wadi and deposited during tsunami backwash. Much of this fine-grained sediment would have been exported from the lagoon insuspension after the tsunami, but in many locations some of it wasretained in the TB. Generally absent in the TB were gravel-sized clasticparticles, as were rip-up clasts and other indicators of high-energytsunamiflow (Goff et al., 2004;Morton et al., 2007). This is largely due tothe distribution of surface sediment in Sur Lagoon with larger clasticparticles found only in the upper reaches of the wadis (Donato, 2008).Most of the easily eroded sediment in the lower reaches of the wadi ismuddy sediment which drapes over coarse gravel to boulder sizedsediment. The wadi mouths in the lagoon are filled with mangrovewhich likely baffled the tsunami surge by dissipating its energy, thuspreventing gravel entrainment with the backflow (Fig. 3C).

5.2. Stratigraphic development of tsunami bed

The cross-plots in Fig. 4 show that there are distinct spatial trends inthe TB thickness and particle-size parameterswithin the lagoon. Overallthe TBs thinned and became finer grained, better sorted and skewedwith distance inland from the lagoon entrance channel (Fig. 5). The r2

values are low, but nonetheless indicate that the TB geometry andsedimentary characteristics vary in proportion to distance from thelagoon entrance. These trends can be explained by a decrease in flowvelocities as the tsunami run-up entered the lagoon via the entrancechannel and underwent rapid flow expansion. The TB thickness andparticle-size would also be dependent to some extent on basincharacteristics such as water depth and bed topography, and it shouldbe expected that flow distance explains only part of the variations inthese parameters.

The heterogeneity of the PSD trends in the TB from core to core alsoindicates a complex relationship between erosion and deposition in thelagoonduring the run-up,maximumflooding, andbackwashphase. PSDtrends within the TB showed some commonality (eg. modes) betweencores although most exhibited different trends. No obvious size trendswere found with the shell distributions yet the PSD data showed cleartrends (eg. grading) in some cores. Cluster analysis of the PSDs extendedtheTBbelow the shell unit inCores2, 3, and6byamaximumof5 cmdueto an increase in mud-sized particles in these samples but this was notpresent in all cores. The broad features followed predictable trends(Fig. 5) but subtleties in the PSD trends were likely due to the complexinteraction of the tsunami with the bathymetry of the lagoon.

During the run-up phase marine and lagoonal (and to a lesserextent wadi) sediment was deposited onto the scoured substratethrough traction and suspension. Deposition continued during therun-up and maximum flooding stage as successive waves andrefracting waves continually reworked the initial coarse- graineddeposit. Suspended sediment settled out during the maximumflooding stage, and as the lagoon began to drain, wadi sediment waseroded and entrained through backwash transporting mud-sizedparticles. Erosion of sediment overlying the shell bed likely occurredduring backwash but the shell-rich TB resisted further erosion due tothe armoring effect of the densely packed shells. Any sandy sedimentcovering the shell bed immediately after the tsunamiwould have beenvulnerable towinnowing and reworking by normal processes inside ofthe lagoon.

Post-tsunami sediment accumulation in the tsunami scour wouldhave occurred at varying rates depending on site position inside of thelagoon, since some sites appear to accumulate sediment faster thanothers (e.g. Core 7 vs. Core 3). Sand-sized particles have covered the TBsthroughout the lagoon, andmost of the TBs are founddeeper than15 cmbelow the modern surface (sediment accumulation of ~0.25 cm/year).Channel and creek migration has exposed and eroded the TB in severallocations, while reworking of the shell bed has occurred at several sites(Cores 3 and 6) due to delayed burial or frequent winnowing related totheir proximity to major tidal channels.

In the absence of detailed eye witness accounts, it is impossible toknow howmany tsunamiwaves affected Sur lagoon, but it may be thatthere was more than one wave, and that the second or third wave wasactually the largest and caused the greatest disturbance of sediment(i.e. Nanayama and Shigeno, 2006; Fujiwara and Kamataki, 2007; Pariset al., 2007) which would have resulted in multiple scouring episodesprior to final deposition. Preliminary estimates of water height of +3mfrom the eyewitness account indicates that most of the water musthave surged through the entrance channel rather than overtop of thebarrier (2–3 m above m.s.l.). Lagoonal bathymetry would then furthercomplicate the run-up pattern, but the surge seemed to disperse withdistance into the lagoon, a scenario that seems to be supported by thebed thickness and textural data which varies with distance from theentrance (Fig. 5).

6. Conclusions

Morton et al. (2007) provided criteria for physically distinguishingtsunami and storm deposits. They posited, based on their modernexamples (and previous work), that tsunami deposits could bedistinguished by their thickness (generally b25 cm), homogeneity(often a single bed and normally graded), extent (100's of metresinland), presence of mud intraclasts or laminae, and coverage ofantecedent topography. In contrast, the storm deposits were typicallymore massive, contained abundant and variable sedimentary struc-tures and don't typically penetrate far inland (b300 m).

Physically, the Sur tsunami unit displays most of these tsunamigeniccharacters although the covering of antecedent topography is impossibleto assess in the lagoon and the mud likely percolated into the pore spaceamongst the shells so it is not a distinct entity. By adding bivalve

64 S.V. Donato et al. / Marine Geology 257 (2009) 54–64

taphonomy to this criteria—the massive (lacking biofabric) nature of theshell unit, the mixture of articulated bivalves and angular fragments fromoffshore provenance provides further confidence regarding a tsunami-genic origin. Further testing is required, but the results appear promising.

Inclusionof PSDplots of particle-sizedata andclusteranalysis providesadditional data on tsunami deposits over traditional PSD statisticalparameters. The PSD plots allow subtle changes in particle-size to beobserved in stratigraphic sequence with the cluster analysis permittingcomparison amongst large data sets. More comparative data is required,but this approach will be particularly useful for defining older tsunami-genic units in coreswhere shell taphonomywill be of limited use becauseof the small sample sizes.

Acknowledgements

The authors wish to thank Tom Vosmer who provided necessaryassistancewhile in Oman, and Brian Atwater and an anonymous reviewerwho provided insightful comments on the manuscript. Funding for thisresearch was provided by NSERC to E.G. Reinhardt.

References

Alsharhan, A.S., Kendall, C.G.S.C., 2003. Holocene coastal carbonates and evaporites of thesouthern Arabian Gulf and their ancient analogues. Earth Sci. Rev. 61, 191–243.

Alsharhan, A.S., El-Sammak, A.A., 2004. Grain-size analysis and characterization ofsedimentary environments of the United Arab Emirates coastal area. J. Coast. Res.20, 464–477.

Ambraseys, N.N., Melville, C.P., 1982. A History of Persian Earthquakes. CambridgeUniversity Press, Cambridge. 219 pp.

Atwater, B.F., Musumi-Rokkaku, S., Satake, K., Tsuji, Y., Ueda, K., Yamaguchi, D.K., 2005.The orphan tsunami of 1700; Japanese clues to a parent earthquake in NorthAmerica: U.S. Geol. Surv. Prof. Pap. 1707 133 pp. (http://pubs.usgs.gov/pp/pp1707/).

Beierle, B.D., Lamoureux, S.F., Cockburn, J.M.H., Spooner, I., 2002. A new method forvisualizing sediment particle size distribution. J. Paleolimnol. 27, 279–283.

Boyajian, G.E., Thayer, C.W., 1995. Clam calamity: a recent supratidal storm-deposit asan analog for fossil shell beds. Palaios 10, 484–489.

Bosch, D.T., Dance, S.P., Moolenbeek, R.G., Oliver, P.G., 1995. Seashells of Eastern Arabia.Emirates Printing Press, Dubai. 296 pp.

Brett, C.E., 2003. Taphonomy: sedimentological implications of fossil preservation. In:Middleton, G.V. (Ed.), Encyclopedia of Sediments and Sedimentary Rocks. Springer,Dordrecht, pp. 723–729.

Byrne, D.E., Sykes, L.R., Davis, D.M., 1992. Great thrust earthquakes and a seismic slipalong the plate boundary of the Makran Subduction Zone. J. Geophys. Res. 97,449–478.

Chambers, R.L., Upchurch, S.B., 1979. Multivariate analysis of sedimentary environmentsusing grain-size frequency distributions. Math. Geol. 11, 27–43.

Dawson, S., Smith, D.E., 2000. The sedimentology of Middle Holocene tsunami facies innorthern Sutherland, Scotland, UK. Mar. Geol. 170, 69–79.

Dominey-Howes, D.T.M., Humphreys, G.S., Hesse, P.P., 2006. Tsunami and paleotsunamidepositional signatures and their potential value in understanding the late-Holocene tsunami record. Holocene 16, 1095–1107.

Dominey-Howes, D., Cummins, P., Burbidge, D., 2007. Historic records of teletsunami inthe Indian Ocean and insights from numerical modeling. Nat. Hazards 42, 1–17.

Donato, S.V., 2008. Sedimentologic and taphonomic Investigation of a 1945 Tsunami Depositin Sur Lagoon, Sultanate of Oman. PhD Dissertation. McMaster University, Hamilton,Ontario, Canada. 91pp.

Donato, S.V., Reinhardt, E.G., Boyce, J.I., Rothaus, R., Vosmer, T., 2008. Identifying tsunamideposits using bivalve shell taphonomy. Geology 36, 199–202.

Feng, C., Hui-Mei, C., Xian-Ze, S., Dong-Xing, X., 2007. Analysis on morphodynamics ofsandy beaches in south China. J. coast. res. 23, 236–246.

Folk, R.L., Ward, W., 1957. Brazos river bar (Texas); a study in the significance of grainsize parameters. J. Sediment. Petrol. 27, 3–26.

Fujiwara, O., Kamataki, T., 2007. Identification of tsunami deposits considering the tsunamiwaveform: an example of subaqueous tsunami deposits in Holocene shallow bay onsouthern Boso Peninsula, Central Japan. Sediment. Geol. 200, 295–313.

Fujiwara, O., Masuda, F., Sakai, T., Irizuki, T., Fuse, K., 2000. Tsunami deposits in Holocene baymud in southern Kanto region, Pacific coast of central Japan. Sed. Geol. 135, 219–230.

Goff, J., Chagué-Goff, C., Nichol, S., 2001. Paleotsunami deposits: a New Zealandperspective. Sediment. Geol. 143, 1–6.

Goff, J., McFadgen, B.G., Chagué-Goff, C., 2004. Sedimentary differences between the2002 Easter storm and the 15th-century Okoropunga tsunami, southeastern NorthIsland, New Zealand. Mar. Geol. 204, 235–250.

Hawkes, A.D., Bird, M., Cowie, S., Grundy-Warr, C., Horton, B.P., Shau Hwai, A.T., Law, L.,Macgregor, C., Nott, J., Ong, J.E., Rigg, J., Robinson, R., Tan-Mullins, M., Sa, T.T., Yasin,Z., Aik, L.W., 2007. Sediments deposited by the 2004 Indian Ocean tsunami alongthe Malaysia–Thailand Peninsula. Mar. Geol. 242, 169–190.

Hamilton, L.J., 2007. Clustering of cumulative grain size distribution curves for shallow-marine samples with software program CLARA. Aust. J. Earth Sci. 54, 503–519.

Heidarzadeh, M., Pirooz, M.D., Zaker, N.H., Yalciner, A.C., 2008a. Preliminary estimationof the tsunami hazards associated with the Makran subduction zone at thenorthwestern Indian Ocean. Nat. Hazards. doi:10.1007/s11069-008-9259-x.

Heidarzadeh, M., Pirooz, M.D., Zaker, N.H., Yalciner, A.C., Mokhtari, M., Esmaeily, A.,2008b. Historical tsunami in theMakran Subduction Zone off the southern coasts ofIran and Pakistan and results of numerical modeling. Ocean Eng. 35, 774–786.

Hammer, O., Harper, D.A.T., 2006. Paleontological data analysis. Blackwell Publishing,Malden. 351pp.

Indian Meterological Department, 2007. Tropical Cyclone Advisory. (http://www.webcitation.org/5PNs2wqpW).

Inman, D.L., 1952. Measures for describing the size distributions of sediments.J. Sediment. Petrol. 22, 125–145.

Kortekaas, S., Dawson, A.G., 2007. Distinguishing tsunami and storm deposits: anexample from Martinhal, SW Portugal. Sediment. Geol. 200, 208–221.

Lambeck, K., 1996. Shoreline reconstructions for the Persian Gulf since the last glacialmaximum. Earth Planet. Sci. Lett. 142, 43–57.

Meldahl, K.H., Cutler, A.H., 1992. Neotectonics and taphonomy; Pleistocene molluscanshell accumulations in the northern Gulf of California. Palaios 7, 187–197.

Morton, R.A., Gelfenbaum, G., Jaffe, B.E., 2007. Physical criteria for distinguishing sandytsunami and storm deposits using modern examples. Sediment. Geol. 200,184–207.

Nanayama, F., Shigeno, K., 2006. Inflow and outflow facies from the 1993 tsunami insouthwest Hokkaido. Sediment. Geol. 187, 139–158.

Nanayama, F., Furukawa, R., Shigeno, K., Makino, A., Soeda, Y., Igarashi, Y., 2007. Nineunusually large tsunami deposits from the past 4000 years at Kiritappu marshalong the southern Kuril Trench. Sediment. Geol. 200, 275–294.

Noda, A., Katayama, H., Sagayama, T., Suga, K., Uchida, Y., Satake, K., Abe, K., Okamura, Y.,2007. Evaluation of tsunami impacts on shallow marine sediments: an examplefrom the tsunami caused by the 2003 Tokachi-oki earthquake, northern Japan.Sediment. Geol. 200, 314–327.

Okal, E.A., Synolakis, C.E., 2008. Far-field tsunami hazard frommega-thrust earthquakesin the Indian Ocean. Geophys. J. Int. 172, 995–1015.

Pararas-Carayannis, G., 2006. The potential of tsunami generation along the MakranSubduction Zone in the northern Arabian Sea. Case study: the earthquake andtsunami of November 28, 1945. Sci. Tsunami Hazards 24, 358–384.

Paris, R., Lavigne, F., Wassmer, P., Sartohadi, J., 2007. Coastal sedimentation associatedwith the December 26, 2004 tsunami in Lhok Nga, west Banda Aceh (Sumatra,Indonesia). Mar. Geol. 238, 93–106.

Pratt, B.R., Bordonaro, O.L., 2007. Tsunamis in a stormy sea: Middle Cambrian inner-shelf limestones of western Argentina. J. Sediment. Res. 77, 256–262.

Quittmeyer, R.C., Jacob, K.H., 1979. Historical and modern seismicity of Pakistan,Afghanistan, India, and southeastern Iran. Bull. Seismol. Soc. Am. 69, 773–823.

Reinhardt, E.G., Goodman, B.N., Boyce, J.I., Lopez, G., van Hengstum, P., Rink, W.J., Mart,Y., Raban, A., 2006. The tsunami of December 13, 115 A.D. and the destruction ofHerod the Great's harbour at Caesarea Maritima, Israel. Geology 34, 1061–1064.

Sambridge, M., Braun, J., McQueen, H., 1995. Geophysical parametrization andinterpolation of irregular data using natural neighbors. Geophys. J. Int. 122,837–857.

Scheffers, A., Kelletat, D., 2003. Sedimentologic and geomorphic tsunami imprintsworldwide—a review. Earth Sci. Rev. 63, 83–92.

Scheffers, A., Kelletat, D., 2005. Tsunami relics on the coastal landscape west of Lisbon,Portugal. Sci. Tsunami Hazards 23, 3–16.

Scheffers, A., Scheffers, S., 2007. Tsunami deposits on the coastline of west Crete(Greece). Earth Planet. Sci. Lett. 259, 613–624.

Switzer, A.D., Pucillo, K., Haredy, R.A., Jones, B.G., Bryant, E.A., 2005. Sea level, storm, ortsunami: enigmatic sand sheet deposits in a sheltered coastal embayment fromsoutheastern New South Wales, Australia. J. coast. res. 12, 655–663.

Szuman, M., Berndt, C., Jacobs, C., Best, A., 2006. Seabed characterization through arange of high-resolution acoustic systems—a case study offshore Oman. Mar.Geophys. Res. 27, 167–180.

Tuttle, M.P., Ruffman, A., Anderson, T., Jeter, H., 2004. Distinguishing tsunami from stormdeposits in eastern North America: the 1929 Grand Banks tsunami versus the 1991Halloween storm. Seismol. Res. Lett. 75, 117–131.

Umitsu, M., Tanavud, C., Patanakanog, B., 2007. Effects of landforms on tsunami flow in theplains of Banda Aceh, Indonesia, and Nam Khem, Thailand. Mar. Geol. 242, 141–153.

van Hengstum, P.J., Reinhardt, E.G., Boyce, J.I., Clark, C., 2007. Changing sedimentationpatterns due to historical land-use change in Frenchman's Bay, Pickering, Canada:evidence from high- resolution textural analysis. J. Paleolimnol. 37, 603–618.

Wang, X., Ke, X., 1997. Grain-size characteristics of the extant tidal flat sediments alongthe Jiangsu coast, China. Sediment. Geol. 112, 105–122.

Warren, J.K., 2006. Evaporites: Sediments, resources, and hydrocarbons. Springer, Berlin, pp.141–220.