RESEARCH ARTICLE

Assessing landslide movements in volcanic islandsusing near-shore marine geophysical data: south Picoisland, Azores

Neil C. Mitchell & Rui Quartau & José Madeira

Received: 7 September 2010 /Accepted: 18 August 2011 /Published online: 27 September 2011# Springer-Verlag 2011

Abstract Marine geophysical data from around the sub-marine flanks of volcanic islands can potentially help toresolve whether large-scale instability of an edifice has beengeologically recently active. We use geophysical data toinvestigate part of the coast of Pico Island of the Azoreswhere, above sea-level, a major slump of Topo volcano hasbeen interpreted previously from arcuate escarpments and arugged irregular topography seaward of them. Multibeamecho-sounder data collected offshore of this feature showremarkably little evidence for slump fault movements in theisland's submarine slope. Mid-slope benches, like thoseassociated with the Hilina Slump of Kilauea, are absent.The high-resolution data extends onto the island's shelf,allowing us to evaluate evidence for continuing activitythere. In particular, as the shelf's rock platform will havebeen last modified by surf erosion during the postglacialperiod of sea-level transgression, it provides a referencesurface of intermediate age (7–19 ka) that can potentiallyreveal whether any movements occurred in pre-historic

times. Where the arcuate escarpments are continuedoffshore, the modern seabed shows no bathymetric evi-dence for active faults where the shelf rock platform cropsout in the multibeam data. Elsewhere, mobile shelf sedi-ments could be disguising evidence for active faulting sowe examined boomer profiles able to image the rockplatform beneath them. The data reveal a platform that issteep (6.6°) compared with the dips of platforms that wehave studied previously around the coast of adjacent FaialIsland and steeper than the platform outside the proposedslump. This suggests that it was created by coastal erosionover a shorter period and hence is consistent with a youngerage of the coastline. As with the multibeam data, whereescarpments are continued offshore onto the shelf, the rocksurface imaged with these boomer data also shows no clearevidence of major slump-related fault displacements. Thisstudy therefore illustrates how high-resolution boomerseismic and multibeam data could usefully contribute tohazard assessment of volcanic islands, by helping toevaluate areas with no historical movements. Explanationsto reconcile the onshore and offshore data here are also putforward.

Keywords Volcano collapse . Slump . Hazards . Coastalerosion

Introduction

Identifying and dating collapse structures is important forhazard assessment of volcanoes as well as for understand-ing this significant geological process. Before sonar data-sets began to be collected, the incidence of lateral collapses(landslides) of volcanic ocean islands was considereduncertain by some researchers and for some locations

Editorial responsibility: B. van Wyk de Vries

N. C. Mitchell (*)School of Earth, Atmospheric and Planetary Sciences,University of Manchester,Manchester M13 9PL, UKe-mail: neil.mitchell@manchester.ac.uk

R. QuartauUnidade de Geologia Marinha—Laboratório Nacional de Energiae Geologia I.P., Estrada da Portela, Zambujal,Alfragide,2721-866 Amadora, Portugal

J. MadeiraFaculdade de Ciências, Departamento de Geologia, andLaboratório de Tectonofísica e Tectónica Experimental/InstitutoDom Luiz (Laboratório Associado), Universidade de Lisboa,Lisboa, Portugal

Bull Volcanol (2012) 74:483–496DOI 10.1007/s00445-011-0541-5

because tall sea cliffs can also be produced by the extremecoastal erosion often found around oceanic islands, andsemi-circular depressions (Masson 1996) could potentiallybe produced by vertical rather than lateral collapse. Fromthe 1980s, widespread surveying around island flanksbecame possible thanks to the routine use of sidescan andmultibeam sonars. Extensive debris fields characterised by“speckly” high backscatter were revealed in long-rangesonar images (Holcomb and Searle 1991; Moore et al.1989). Dating such collapse features is important for hazardassessment and for identifying possible causes (Carracedoet al. 1999; Keating and McGuire 2000; Paris et al. 2005;Quidelleur et al. 2008) but is difficult on land, for example,because erosion may have removed evidence or because ofdifficulty in assessing the time relationship betweencollapse and datable volcanic features. Researchers haveattempted to date their deposits underwater by identifyingcharacteristic volcanic turbidites that can be linked to theoriginal edifice that failed (Clague and Moore 2002;Lipman et al. 1988; Masson 1996), but this requiresidentifiable lithological and/or geochemical characteristics.

The instability of volcanic islands is a concern asmovements on slump faults can generate major earth-quakes, such as the M7.1 1975 Kalapana earthquake ofHawaii (Crossen and Endo 1982), as well as associatedrisks from tsunamis, more mobile landslides, and eruptions.While seismicity and geodetic (GPS) data can identify areasof present-day deformation, they may be unable to identifyfaults with long intervals between movements. In contrast,some other measurements, such as from dating raisedterraces or marine deposits or submerged corals (Mooreand Fornari 1984), provide evidence of vertical uplift orsubsidence that is often too long-term for meaningfulhazard assessment, or the evidence may be absent as inthe case of Pico. Methods are needed to evaluate move-ments over intermediate (ky) time-scales.

In Quartau et al. (2010), we adapted a coastal erosionmodel (Trenhaile 2000) to investigate how the rockplatforms underlying sediment on the shelves of volcanicocean islands can develop primarily from wave-force actingat sea level, which varies through multiple sea-level cyclesdepending on the age of the coastline. The study was basedon Faial Island. South Faial experiences a similar waveclimate to the nearby south Pico (Carvalho 2002, 2003) andis similarly unsheltered from waves, so comparisons can bemade with platforms around Topo volcano (Fig. 1a). Themodelling and observations suggest that coastal erosionproduces a relatively simple dipping platform surface thatcan potentially be used as a reference surface (Burbank andAnderson 2001) on which can be revealed abrupt disrup-tions such as small fault displacements produced by slumpmovements. It may potentially also contain depressions ifwave erosion has exploited weak fault rocks. Furthermore,

as rock platforms progressively widen with age (Menard1983; Ablay and Hurlimann 2000; Mitchell et al. 2003), theage of the coastline is roughly indicated by its platformwidth and dip. The shelf of an island thus provides furtherinformation to help assess island instability, in particular,revealing whether or not movements have occurred sincepostglacial transgression.

Chronology of Pico island lavas

The general chronology of the island's lavas was summa-rized by Madeira and Brum da Silveira (2003), which is ingeneral agreement with previous work (Zbyszewski et al.1963; Walker 1973; Woodhall 1974; Madeira 1998; Nunes1999; Scarth and Tanguy 2001; França 2002; França et al.2006). Geological boundaries, based on detailed fieldworkand analysis of aerial photos and landscape morphology, areoverlain on topography in Figs. 1b and 2. The 1,020-m-high summit of Topo mountain, on the south-central coastof Pico, is the remainder of a shield volcano formed by theoldest units of the island, comprising the Lajes VolcanicComplex (VC) (marked “Lj” in Fig. 1b). Madeira andBrum da Silveira (2003) mentioned a single K-Ar date of250±40 ka for this unit from Chovelon (1982). The westand east flanks of the Topo shield volcano are overlain bythe Calheta do Nesquim VC, which contains basalticHawaiian/Strombolian scoria cones and lava flows origi-nating from the Calheta do Nesquim Volcanic Complex(areas marked “Cn” in Fig. 1b encompassed by dashedlines). K-Ar ages obtained from the Calheta do NesquimVC vary from 270±150 to <25 ka (Chovelon 1982; Feraudet al. 1980). North of the Topo edifice and farther east andwest of there, younger lavas (Holocene to Late PleistoceneMadalena VC) were erupted from the chain of cones atopthe easterly ridge of the island. Topo volcano is tectonicallydisplaced by a major fault zone, the Topo fault, by about1 km dextrally, associated with the diffuse transtensionalplate boundary passing through the central and eastern

Fig. 1 a Topography and bathymetry of Pico Island of the Azoresarchipelago is shown along with adjacent Faial and São Jorge islands(contours are every 100 m, with coastlines in bold). Bathymetry isfrom Mitchell et al. (2008). Faial and Pico islands, along withsubmarine ridges of cones extending beyond them, form a ridge morethan 90 km long, with the old edifice Topo lying south of that trend. bRelief of southeast Pico Island, with elevations updated from InstitutoGeográfico do Exército (2001). Lj and CN mark areas (outlined incontinuous white and dashed white) where the Lajes and Calheta doNesquim volcanic series, respectively, have been located (Fig. 2 andMadeira and Brum da Silveira 2003). E1–E3 and associated solid andbold long-dashed lines are escarpments, the largest of which (E1) isthe Arrife escarpment. The three numbers (400, etc.) locate three coneelevations (m). Solid white circles with arrows orient the photosshown in Fig. 4b–d. Arrow in the lower right of the map is theviewpoint used for Fig. 4a. Contours are every 100 m (negative belowsea level). Maps are in Universal Transverse Mercator projection(zone 26, WGS84). Mid-gray used for areas without data

�

484 Bull Volcanol (2012) 74:483–496

4250000

4260000

4270000

4280000

340000 350000 360000 370000 380000 390000 400000 410000 420000

Faial Is.

Pico

Topo

Sao Jorge

Slump?Fig. 1b

4245000

4250000

4255000

4260000390000 395000 400000

Lj

Cn

No data

Topo

Cn

A

B 1000 m

C

713

400

989

Figure 3a

E1

E2E3

4b

4c

4d

Ribeiras lava delta

a

b

Bull Volcanol (2012) 74:483–496 485

Azores islands. If an age of 250 ka is assumed for the Topovolcano, a time-averaged slip rate of the order of 4 mm/yearis obtained.

In Fig. 1b, the southerly headland of the island has a2-km-wide rock platform with well-developed terraces,consistent with an old age of either Calheta do Nesquimor earlier Lajes units. The proposed slump surface affectsthe older Lajes VC lavas, and the depression is partiallyfilled by Calheta do Nesquim lavas erupted from cones713 and 400 (Figs. 1b and 2). The last unit emplaced,

covering most of the island and partly overlying theCalheta do Nesquim and Lajes VCs, is the Madalena VC,with Holocene radiocarbon dates reported (Madeira1998; Nunes 1999; Madeira and Brum da Silveira2003). Delicate submarine lava structures imaged withsonar on many other parts of Pico Island's shelf wereargued previously to be unlikely to have been preservedif subjected to erosion by surf (Mitchell et al. 2008).These flow units were therefore probably emplacedduring the Holocene high stand of sea level by subaerial lavas

Fig. 2 Geological map basedon aerial photos, fieldwork, andterrain morphology. E1, E2, andE3 are landslide escarpmentsand numbers 400 and 713 arealtitudes of cones mentioned inthe text. Landslide escarpmentson the east side are buried byyounger flows and thereforemore speculative than those onthe west side

486 Bull Volcanol (2012) 74:483–496

that penetrated the sea, suggesting extensive Holocenevolcanic activity.

Previous work on Topo volcano

The Topo edifice shown in Fig. 1b has a triangular-shapedsummit. Topographic maps and the Space Shuttle ImagingRadar band-C (SIR-C) image in Fig. 3 show a prominentarcuate escarpment (Arrife) on its southeast side. Figure 4ashows a three-dimensional rendering of the embayment,revealing three escarpments above sea level (E1–E3corresponding to those marked on Figs. 1b and 2). Thesethree escarpments form steps typical of slump headwallfaults, with any easterly escarpments of the embaymentnow obscured by lavas as described below. Three photosare shown in Fig. 4b–d, taken from the locations marked bytwo solid circles in Fig. 1b. The photo in Fig. 4b, takenfrom the extreme south of escarpment E1, shows a convexcoastline enclosing areas of subdued relief, which corre-spond to young lava deltas (Madeira 1998; Nunes 1999).Figure 4c shows a view from nearer the Topo summitlooking towards the coast with the Arrife escarpment to theright. Figure 4d also from that location shows the easterlyside of the embayment.

The depressed SE area in Fig. 3 was originallyinterpreted to be a vertical collapse structure (Woodhall1974; Zbyszewski et al. 1962), but more recently, it hasbeen reinterpreted as a result of a lateral collapse or moredeep-seated slump (Madeira 1998; Nunes 1999; Cruz andSilva 2001; Nunes 2002; Madeira and Brum da Silveira2003). Seismicity is at a high level, including M>7 eventsin the Azores region and probably M>6 for the Topo strike-

slip fault system immediately NE of Topo's summit(Madeira and Brum da Silveira 2003), so major landslideswould not be unexpected in the steep slopes of thesevolcanoes. In such an interpretation, the SW escarpment E1bounding this embayment represents the main side andheadwall fault of the slump, while the bounding faults tothe east have become obscured by lava flows emplacedsubsequently. E1 has a relief of 60–70 m in the coastal cliffsand 250–350 m in the landslide head region. The amount ofvertical displacement is larger, but it is not well constrainedbecause of subsequent infilling by younger volcanism.Nunes (2002) suggested that the age of the slump could beless than 37 ka based on units apparently offset in theescarpment and dated by Feraud et al. (1980) using wholerock K-Ar methods but older than the Holocene because thelava delta covering the east side of the slump embaymenthas developed at modern sea level. According to Nunes(1999), a Holocene age for the delta is supported by oneradiocarbon date of 3,520 years B.P. obtained on a paleosolcovered by the lava that forms the delta. According to him,the lava originated from a few hundred meters above sea-level up-slope from the coast.

Detailed mapping by one of us (Madeira) suggests thatsubsequent to the lava units within the slide area dated byFeraud et al. (1980), lavas were emplaced from cone 713and then from cone 400, partially covering the lavas datedby Feraud et al. (1980). In contrast to Nunes (1999), basedon aerial photo interpretation and geomorphological anal-ysis, we suggest that the lavas forming the coastal deltaoriginated from cone 989 on the axial chain of cones shownin Fig. 1b (illustrating the difficulty of making fieldobservations in terrain affected by a wet temperate oceanic

Fig. 3 GLORIA long-rangesidescan sonar data, adaptedfrom Mitchell (2003) with sonardata originally from Searle(1980) and SIR-C data abovesea level courtesy of the JetPropulsion Laboratory (Pase-dena, USA). Lighter tones in theimage represent high amplituderadar or sonar backscatter

Bull Volcanol (2012) 74:483–496 487

climate such as here). The high-resolution elevation dataalso reveal two further escarpments, though with moresubdued relief, close to the coast (E2 and E3 (Figs. 1b, 2,and 4a)).

The observations therefore suggest the following se-quence. Landslides or lateral collapses of the SE flank ofTopo volcano formed successively the E1, E2, and E3escarpments. This was followed by eruptions that createdcone 713 and then cone 400, along with associated lavaspartially filling the collapse depression. Eruptions from thevolcanic ridge to the NE of Topo produced lavas thatflowed down the slope reaching the sea in the Ribeiras area,the latest of which (the “3520 BP” lava) created theRibeiras lava delta. No anomalous seismicity has beenobserved here, and there are no historical records ofmovement (Madeira and Brum da Silveira 2003). Theproposed slump system therefore appears to be inactive atpresent, but whether the lack of seismicity instead is due toa long recurrence interval is unclear.

Sonar observations of south Pico

Island submarine slope

An image of combined SIR-C and GLORIA long-rangeside-scan sonar imagery from south Pico Island isshown in Fig. 3. Note that the sonar data were wronglypositioned to the southeast by 1–2 km because of a tow-fish navigation error (producing the dark gap betweenhigh backscattering and the coast) and is inaccuratelypositioned also because of a flat-bed assumption used inthe sonar travel-time to ground-range transformation(Mitchell 1991). Nevertheless, seaward of the coastal datagap, a high backscattering band running parallel to thecoast can be observed, produced by backscattering fromthe steep slope of the island. Farther seaward of the slope,the speckle pattern in the image is typical of debris fieldsof rocky avalanches caused by the acoustic contrastbetween blocks and surrounding sediments (Holcomb

400

720

989

E1 E2 E3

a c

b d

Fig. 4 a Three-dimensional view of the elevation data as marked inFig. 1b from N155°E, 15° elevation (2:1 vertical exaggeration).Annotation as in Fig. 1b. b Photo taken from the vantage pointmarked by a solid circle in Fig. 1b towards the east-northeast over theembayment towards. Low-relief coastal area with convex coastline is a

lava delta. c Photo taken looking roughly south-southwest from theheadwall of the proposed slump showing the Arrife escarpment. dPhoto taken from above the Arrife escarpment looking over the eastside of the embayment. This and photo c are also located by a solidcircle in Fig. 1b

488 Bull Volcanol (2012) 74:483–496

and Searle 1991). This was originally interpreted as such adebris field by Mitchell (2003).

In 2003, we collected high-resolution multibeam echo-sounder data around the coasts (acquisition and processingdescribed in Mitchell et al. (2008)). In the island's apronbelow 1,000 m (Fig. 1b), the debris field is not evident inthese data. Isolated hills such as “A” in the figure could belandslide blocks, but without in situ observations, it isdifficult to be certain. Cores collected in 1970 by R/VTrident in deep water among the central Azores islands eastof here contained primarily calcareous ooze, and deep-towsidescan records also suggest a muddy (weakly backscat-tering) bed away from the submarine extension of Picoeasterly ridge (Stretch et al. 2006). Perhaps explaining theapparent different appearance in the high-frequency multi-beam data, then, some high-backscattering areas reportedpreviously from the GLORIA data may have been imagedthrough a surface veneer of sediment (associated with its

low 6.5 kHz acoustic frequency, the sonar is able topenetrate mud and image surfaces buried by several meters(Mitchell 1993)). Indeed, small mounds near “C” appearsmooth as though more angular blocks have been partiallycovered in sediment.

Also suggesting little recent slump movement, the slopeof the island here does not show obvious landslideembayments or headwalls of secondary slope failures. Incontrast to the mid-slope benches found in the slopes ofvolcanoes such as Kilauea (Moore and Chadwick 1995),the slope has no prominent tabular bench or terrace withinit; one terrace immediately SW of “B” is of limited extentand does not correspond well to the subaerial embayment.Depth contours show that the slope steepens upwards toaround 29° immediately below the shelf edge, as we haveobserved in the constructional sectors of other volcanicislands and in contrast to the lower gradients of sectors withdebris avalanche deposits (Mitchell et al. 2002) (referring to

Lavadelta

Rock platform

Line 20Line

21

Line

22

Line 28

Line 29

Line

23a

Rocky outcrops

Arrife escarpment

Line

23b

Chirp

E1

E2

E3

Fig. 5 High-resolution image of the multibeam data on the shelf.Contours are every 20 m and coordinates as Fig. 1. White continuouslines show paths of boomer seismic profiles (those in bold are shownin Fig. 7). Dotted lines mark where Chirp sediment profiler recordswere collected (white highlighted line annotated “Chirp” is shown in

Fig. 6). Four white-filled circles overlying plus symbols locate grabsamples on the shelf. Circled plus symbol on line 29 lies at thecontinuation of the fault seen in the subaerial topography data andcorresponds with the arrow marked “F?” on line 29 in Fig. 7

Bull Volcanol (2012) 74:483–496 489

the landslide types defined by Moore et al. (1989)). Thissteepness is also typical of rock talus at the angle of repose(Mitchell et al. 2000) and unlike those where particles haverun out (Lee et al. 1994). Finally, the smooth slope NE of“B” lies down-slope of the lava deltas (Nunes 1999;Mitchell et al. 2008). The slope's subdued relief here(smooth at the scale of the sonar data) is typical of volcanicslopes mantled by debris created from lava–water inter-actions or surf erosion (Moore and Chadwick 1995).

Island shelf

Figure 5 shows a higher-resolution image of the shelf regionwith the artificial sun placed at N045°E to enhance any shelf-crossing features (this illumination also unfortunately enhan-ces artefacts of the data binning, motion sensor errors, andslight movements of the sonar transducer mounting duringvessel motion, mainly affecting the data on the island slope).A rugged fine texture in this imagery nearest the coastlineindicates bedrock exposures where marked “Rock platform”,none of which shows evidence of faulting, in particularadjacent to escarpments E1 and E2. Small isolated low-reliefmounds occur at the shelf edge, indicating rock exposuresnear the surface (“Rock outcrops”), also showing no obviousdisruptions associated with a fault offset. The change ingradient marking the shelf edge where rocks outcrop here liesin the region of 100–120 m, only marginally shallower thanthe Last Glacial Maximum eustatic level of ∼130 m(Yokoyama et al. 2000) and suggesting that little significantsubsidence has occurred since then. Visually continuingescarpment E1 offshore to the shelf edge (at about the “R”of “Rock outcrops”), there is no obvious depression of theshelf edge to the northeast of that location.

The shelf edge seaward of escarpment E3 in Fig. 5 (nextto “line 20” annotation around UTM 395000 mE,4249500 mN) is shallower than 100 m and appears to lielandward of the trend made by the shelf edge farthersouthwest of there. Furthermore, a feature with a curvededge and rugged surface texture in the multibeam data liessouth of the lava delta and east-southeast of “E3”. This typeof feature was interpreted previously as the eroded surfaceof a lava delta (Mitchell et al. 2008). It implies that around200 m of coastal retreat has occurred and that the deltaoriginally extended nearer to E3. The surface of this featurealso contains no evidence of faulting.

Chirp sediment profiler records were taken along thedotted lines marked in Fig. 5, which were useful forexamining areas of outcropping bedrock not reached withthe multibeam sonar. Those profiles unfortunately were notplaced optimally for studying the faults but one collectedclose to the coast and shown in Fig. 6 (from the white-highlighted dotted line in Fig. 5) just crosses the end ofescarpment E2. It reveals a rugged seabed that undulates(an artifact of the irregular track of the vessel on theinclined rock platform), but no abrupt steps that might beexpected if there were recent fault movements.

Elsewhere the shelf has a smooth texture typical ofsediments. Grab samples from the four sites marked onFig. 5 comprise a mixture of volcanic and biogenic gravellysand (average 70% sand and 30% gravel, of which 30% ofparticles by mass are carbonate). In settings such as this,which experience strong effects of surface waves anddown-welling currents driven by wind shear during storms,particles produced by coastal erosion and biological activityare vigorously mobilized into clinoform-shaped bodies(Field and Roy 1984; Hernández-Molina et al. 2000).There is therefore a need for high-resolution seismicreflection data able to image bedrock surfaces beneath thesediment. Figure 7 shows images of boomer seismic datacollected along the lines marked in bold solid white inFig. 5 as part of an aggregate assessment (Quartau et al.2003). In these images, depth is approximately 7.5 m per10 ms of seismic two-way time in the water column.Seismic velocities are typically 1,700–1,800 m/s in sandcompared with 1,500 m/s in water (Hamilton and Bachman1982), so seismic “pull-up” is minor (the followingmeasurements were derived assuming 1,800 m/s in sand).

Figure 8 shows depth-converted interpreted basementand seabed reflectors in four boomer seismic lines that run∼30° from perpendicular to the coast (Fig. 5). Line 21 (the

0

20

SW NE

Two-

way

tim

e (m

s)

100 mFig. 6 Example Chirp profilerrecord collected close to thecoastline (location highlighted bywhite dotted line in Fig. 5). In thevertical scale (milliseconds oftwo-way time), 10 ms correspondsto 7.5 m in the water column

Fig. 7 Boomer seismic profiles (Quartau et al. 2003) collected alongthe shelf as shown by white bold lines in Fig. 5 (locations are alsogiven against the coastline of Pico in the map to left of line 28).Numbers across the top of each profile are the shot point numbers andthe vertical axes are the seismic two-way travel times in milliseconds(10 ms is equivalent to 7.5 m in water). All lines have the samevertical scale but horizontal scales differ as marked. White dashedlines above each profile locate where seismic lines cross. Theannotation represents M seismic multiple, R bedrock reflector undersediment, S sub-basement reflectors, F? location of the fault ifextended from the coastline

�

490 Bull Volcanol (2012) 74:483–496

1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000 2100 2200 2300 2400 2500

57m

SW NE75

100

125

150

100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900

57m

SW NE 50

75

100

125

2400 2200 2000 1800 1600 1400

46m

SW NE75

100

125

800 700 600 500 400 300 200

49m

N S 25

50

75

100

125

150

2200 2100 2000 1900 1800 1700 1600 1500 1400 1300 1200 1100 1000 900 800 700 600 500 400 300 200

51m

SW

NE

50

75

100

125

150

Line 20

Line 28

Line 29

Line 22

M

M

MR

R

R

R

R

S

Sei

smic

two-

way

tim

e (m

s)Turn Turn

F?

F?

Mass-movement deposit?

Diffractionhyperbolae?

Bedrock slope

Bedrock slopeHorizontalbedrock

Line 21

Line 28 Line 29

Line 20

Line 21Line 22

Bull Volcanol (2012) 74:483–496 491

farthest northeast of the lines) shows a steep, irregular rocksurface dipping typically ∼10% beneath surface sediments.The basement in line 22 dips 9.6%. Smaller dips weremeasured off lines 23a and 23b (6.3% and 4.8% for thelower segments of those lines). Allowing for the 30°obliquity of line 21, the true dip of basement is ∼11.5%(i.e., 6.6°).

The rugged feature within the center of line 29 (Fig. 7)has a relief of 20 ms or around 18 m. It lies seaward ofwhere the multibeam sonar image shows rugged bedrockoutcropping nearest to the coastline (Fig. 5). We initiallysuspected this feature to be a lava flow but some curioussub-basement reflectors (“S”) suggest instead that it couldbe a deposit produced by a small mass movement. Itappears to overlie the abrasion platform, dating it as post-transgression. Rates of modern coastal erosion can be higharound volcanic islands depending on the timescale of

measurement and the materials (Mitchell et al. 2008), e.g.,an extreme >100 m of a lava apron overlying pyroclasts ofCapelinhos on nearby Faial island over only 7 months(Machado et al. 1962) and a more modest 0.2 m/yearmedian retreat rate over 33 years from aerial photo analysisof coasts of Sao Miguel island (Borges et al. 1997). Wesuspect that this imaged rugged feature would not havesurvived surf erosion if it had existed at sea-level, and thatit was therefore emplaced below sea level during transgres-sion, for example, by failure of retreating coastal cliffs. Inthat interpretation, the blocks exposed at the shelf edgecould represent more distal deposits emplaced by suchfailure.

Where the Arrife escarpment (E1) is continued offshore,no obvious major step is observed at its intersection withline 29 in the sub-sediment reflector (where marked “F?” inFig. 7). Where the fault crosses line 22, the basement

0 200 400 600 8000 200 400 600 800

60

40

9.6%

120

80

-40

100

60

40

20

39.7%

8.1%

15.5%

5.3%

10.1%

24.2%

3.8%

80

0

40

0

6

0

15.6%

6.3%

120

80

100

60

16.3%

7.2%

4.8%

Distance scale (m)

Line 21

Line 22

Line 23a

Line 23b

Dep

th (

m)

Seabed reflector

Acousticbasement

Fig. 8 Depth-converted inter-pretations of the shore-normalboomer seismic profiles show-ing the seabed and acousticbasement reflectors (bold graylines). Values above basementreflectors are their gradients inpercent. Profiles are located inFig. 5

492 Bull Volcanol (2012) 74:483–496

reflector is overlain by diffraction hyperbolae (a possibleindication of a fault), and there is perhaps a small change ofgradient of the basement reflector either side of this feature.However, the small angle of the track relative to the faultmakes interpretation difficult and the gradient of thesoutherly part of the profile is complicated by the vessel'schange of course. The diffractions could be associated withdistal blocks of the interpreted mass movement deposit online 29. Nevertheless, the presence of a basement escarp-ment of significant relief (>10 m) can be ruled out by thesedata.

Discussion

Character of the landslide

The subaerial data (SIR-C and topography) give a markedlydifferent view of the proposed Topo slump than do the newmultibeam sonar data. Whereas the former suggests a majorslump, the latter reveal no evidence of one. Indeed, thesteep slope of the island and lack of a mid-slope bench donot support the idea of seaward-directed movement withinthe island flank associated with deep-seated slumping. Themorphology contrasts strongly with that of the HilinaSlump of Kilauea—there, the subaerial Hilina fault escarp-ments are subdued from over-emplaced lavas, whereasthrusts of the slump toe have produced a prominent mid-slope bench offshore (Smith et al. 1999).

There seems to be several potential explanations. First,the embayment on land may not have been produced bydeep-seated slumping at all, but rather is a vertical collapsefeature (caldera) as was originally suggested. A largeBouguer anomaly beneath the east side of the Topo massif(Nunes et al. 2006) suggests a dense intrusive complex thatcould represent a solidified magma chamber. However,although calderas of this size are not uncommon in theAzores, they are related to acidic (trachyte) explosive(plinian type) eruptions, not basaltic effusive (hawaiian/strombolian) eruptions as on Pico Island. In contrast,vertical collapse structures in basaltic volcanoes tend to besmall relative to the overall size of the volcano. This isshown by the calderas of the Hawaiian (Walker 1988) andGalapagos (Simkin 1972) volcanoes. In a compilation ofmid-ocean ridge (largely basaltic) seamounts shown inMitchell (2001), the caldera diameter is on average 1.5times the height of the edifice. If such a relation were tohold here, the caldera would be ∼3 km across, somewhatsmaller than the embayment (Fig. 1). Furthermore, a largeseaward-directed embayment such as the one here would bein an unusual location, away from the summit of thevolcano, and no evidence for a southerly wall of such acaldera has been found offshore.

Second, emplacement of the Calheta do Nesquim andlater lavas and associated volcaniclastic deposits havesupplied sufficient material to the slope to obscure anydeeper terrace or other evidence on the slope. Wide-spread cover of talus in this latter explanation seemslikely, given the lava delta and associated smooth slopeimaged with our equipment. In the Canary Islands,many of the embayments produced by debris avalancheshave become partially infilled by subsequently emplacedlava (Masson et al. 2002). These have not been suffi-ciently voluminous to obscure the submarine morpholog-ical evidence of landsliding, but a prolonged phase oferuption could conceivably do so. In such a scenario, theTopo landslide (either a slump or a debris avalanche) hasleft deposits on the ocean floor that have become obscuredby later sediment deposits as well as debris from lavaemplaced in the embayment obscuring morphologicevidence on the slope. On the other hand, in somevolcanoes undergoing gravity slumping, the broadeninggeometry of the movement away from the summit leads tosmaller benches associated with thrusting than would beexpected from the size of headwall scarps, a situationobserved at Casita volcano, Nicaragua (van Wyk de Vrieset al. 2000). Whether or not the embayment here wascreated by a sector collapse is unclear as the easterly sideis now obscured, although the possibility cannot be ruledout. Clearly, further data such as deeply penetratingseismic reflection data like those collected around Kilauea(Morgan et al. 2000) and samples from drilling theonshore deposit would be needed to resolve this issueless equivocally.

Assessing the age of movements

Stable coasts can develop simple rock platforms fromerosion by surf at the coastline. Here, the pattern ofplatform dips (Fig. 8) suggests a possible sequence. Thesteeper lines (21 and 22) both lie northeast of thecontinuation of the Arrife escarpment and have arounddouble the gradient of the lower segments of lines 23a and23b. Modeling suggests that platform dip should declinewith age, as erosion during successive high-stands of sea-level widens the rock platform (Trenhaile 2000). Attenua-tion of waves as they propagate across a wide shelf canreduce the rate of platform widening, but is unlikely to beimportant here given the narrow width of the shelf (Quartauet al. 2010). Our visual observations made from sea alsosuggest similar lava sequences outcrop from the cliffs oneither side of the escarpment so differences of coastlineerodibility are probably not an issue. A more resistantlithology in the landslide area would be needed to explainthe difference. We therefore suggest that the differentplatform dips on either side of the Arrife escarpment are

Bull Volcanol (2012) 74:483–496 493

due to different ages of these parts of the coast, thecoastline to the northeast of the escarpment beingyounger. The shallower shelf edge and its deviation eastof E3 may suggest that the escarpment was producedlater, after the LGM.

To get a rough idea of “abrasion” age, we initiallycompared the bedrock dips with those of nearby Faial,where different sectors of the island have been radio-metrically dated (Quartau et al. 2010). Away from theHolocene Capelo Peninsula of westerly Faial, whererecently emplaced lavas are likely to have complicatedthe shelf geometry, the older coastal platforms typicallydip at 3.5° up to a maximum of 9° near the coastlines. Thesteeper dips near the coastlines there might also corre-spond with recent lavas, but, where we have sonar andboomer seismic observations of bedrock in the outerplatform, it everywhere dips less than 5°. Interpreting thesteeper dips of lines 21 and 22 then implies that the Toposhelf has an age younger than 30 ka, the youngest sector(Almoxarife Formation) of Faial away from Capelinhos(Quartau et al. 2010).

The lack of any clear offset of the platform surface in theboomer seismic and multibeam and chirp sonar datasuggests that the surface was created or modified by waveerosion during sea-level transgression, and there has beenno significant movement since then. This is generally inagreement with the suggestion of Nunes (2002) that theslump is older than the Holocene and has been inactivesince then.

Conclusions

This study illustrates how high-resolution marine geophys-ical data can be used to assess slump or other volcano-tectonic movements of the coasts of volcanic islands. Theseassessed movements would be intermediate in age (kys)because the rock platforms developed around their coastswere last modified by surf erosion during the post-glacialsea-level transgression. Using multibeam data, we havebeen able to assess areas of outcropping rock platform forsigns of recent faulting, while boomer seismic data wereused in areas where the rock platform is obscured bysediments. The depth of the rocky edge of the platformprovides a further test of vertical movements as it shouldcoincide with the Pleistocene glacial low stands of sealevel. The dip and width of the platform are also clues to itsrelative age. In this study, the platform is narrow and steep(6.6°) within the proposed slump area, indicating a youngcoastline. The depth of platform edge and lack of any clearvertical steps in the acoustic basement imaged with boomerand multibeam here suggest that the Topo slump and

overlying deposits have largely not moved after transgres-sion. The submarine slope is smooth, in contrast with thescarped subaerial surface, suggesting either only a partiallydeveloped slump or, more likely in our view, because ofcovering by debris associated with emplacement of the lavadelta.

Acknowledgements The marine surveys were difficult efforts toundertake in small research vessels working in sometimes tempera-mental seas. They required the help of a large number of people. Wewould like especially to thank Eduardo Isidro, Frederico Cardigos,Thierry Schmitt, João Figueiredo, Francisco Teixeira, Serguei Bouriak,Pedro Brito, Célia Pires, Joana Gafeira, Henrique Duarte, João CarlosNunes, and the crews of R/Vs Águas Vivas and Arquipélago for alltheir help in preparing and carrying out the geophysical and sedimentsampling surveys. Funding was provided by the Royal Society, theBritish Council, the Higher Education Funding Council for Wales(JREI), the Secretarias Regionais da Ciência e Tecnologia e doAmbiente dos Açores and Portuguese projects MARINOVA, MAROV(PDCTM/P/MAR/15249/1999) and GEMAS. R.Q. was funded by apostdoctoral fellowship (SFRH/BPD/27135/ 2006) of the Fundaçãopara a Ciência e Tecnologia. The authors and the Laboratório Nacionalde Energia e Geologia also acknowledge the support of LandmarkGraphics Corporation via the Landmark University Grant Program.The Direcção Regional do Ordenamento do Território e dos RecursosHídricos of the Azores is acknowledged for sharing elevation data.Figures were created with the “GMT” software system (Wessel andSmith 1991). We finally also thank reviewers Andrea Argnani andRaphael Paris, and editor Ben van Wyk de Vries, for helpfulcomments, which significantly improved this paper.

References

Ablay G, Hurlimann M (2000) Evolution of the north flank of Tenerifeby recurrent giant landslides. J Volcanol Geotherm Res 103:135–159

Borges PA, Cruz JV, Andrade CF (1997) O recuo da arriba litoral nolocal da Rocha dos Campos (Água d'Alto, São Miguel)—umestudo de caso. Açoreana 8:391–401

Burbank DW, Anderson RS (2001) Tectonic geomorphology. Black-well, Malden MA, p 288

Carracedo JC, Day SJ, Guillou H, Perez Torrado FJ (1999) GiantQuaternary landslides in the evolution of La Palma and El Hierro,Canary Islands. J Volcanol Geotherm Res 94:169–190

Carvalho F (2002) Apuramentos climatológicos mensais em Açorescentral no período 1989–2002. Instituto de Meteorologia, Lisboa

Carvalho F (2003) Elementos do clima de agitação marítima no grupocentral dos Açores. Instituto de Meteorologia, Lisboa

Chovelon P (1982) Évolution volcanotectonique des iles de Faial et dePico, Archipel des Açores—Atlantique Nord. PhD Thesis.Université de Paris-Sud, Centre D'Orsay, France, Paris, p 193

Clague DA, Moore JG (2002) The proximal part of the giantsubmarine Wailau landslides, Molokai, Hawaii. J Volcanol Geo-therm Res 113:259–287

Crossen RS, Endo ET (1982) Focal mechanisms and locations ofearthquakes in the vicinity of the 1975 Kalapana earthquakeaftershock zone 1970–1979: implications for tectonics of thesouth flank of Kilauea volcano, island of Hawaii. Tectonics1:495–542

Cruz JV, Silva MO (2001) Hydrogeologic framework of Pico Island,Azores, Portugal. Hydrogeol J 9:177–189

494 Bull Volcanol (2012) 74:483–496

Instituto Geográfico do Exército (2001) Pico-Açores - Folha 12. In:Carta Militar de Portugal na escala de 1:25 000—Série M8892nd edn. Instituto Geográfico do Exército, Lisboa

Feraud G, Kaneoka I, Allegre CJ (1980) K/Ar ages and stress patternin the Azores: dynamic implications. Earth Planet Sci Lett46:275–286

Field ME, Roy PS (1984) Offshore transport and sand-body formation—evidence from a steep, high-energy shoreface, southeasternAustralia. J Sed Petrol 54:1292–1302

França Z (2002) Origem e evolução petrológica e geoquímica dovulcanismo da Ilha do Pico—Açores. PhD Thesis. University ofthe Azores, Ponta Delgada, Azores. Portugal, p 391

França Z et al (2006) Petrology, geochemistry and Sr–Nd–Pb isotopesof the volcanic rocks from Pico Island—Azores (Portugal). JVolcanol Geotherm Res 156:71–89

Hamilton EL, Bachman RT (1982) Sound velocity and relatedproperties of marine sediments. J Acoust Soc Am 72:1891–1904

Hernández-Molina FJ et al (2000) The infralittoral prograding wedge:a new large-scale progradational sedimentary body in shallowmarine environments. Geo-Mar Lett 20:109–117

Holcomb RT, Searle RC (1991) Large landslides from oceanicvolcanoes. Mar Geotech 10:19–32

Keating BH, McGuire WJ (2000) Island edifice failure and associatedtsunami hazards. Pure Appl Geophys 157:899–955

Lee HJ, Torresan ME, McArthur W (1994) Stability of submergedslopes on the flanks of the Hawaiian Islands, a simplifiedapproach. Open-File Report 94–638:1–54

Lipman PW, Normark WR, Moore JG, Wilson JB, Gutmacher CE(1988) The giant submarine Alika debris slide, Mauna Loa,Hawaii. J Geophys Res 93:4279–4299

Machado F, Parsons WH, Richard AF, Mulford JW (1962) Capelinhoseruption of Fayal volcano, Azores 1957–1958. J Geophys Res67:3519–3529

Madeira J (1998) Estudos de Neotectónica nas ilhas do Faial, Pico e S.Jorge: uma contribuição para o conhecimento geodinâmico dajunção tripla dos Açores. PhD thesis, Lisbon Univ, pp 481

Madeira J, Brum da Silveira A (2003) Active tectonics and firstpaleoseismological results in Faial, Pico and S. Jorge islands(Azores, Portugal). Annals Geophys 46:733–761

Masson DG (1996) Catastrophic collapse of the volcanic island ofHierro 15 ka ago and the history of landslides in the CanaryIslands. Geology 24:231–234

Masson DG, Watts AB, Gee MJR, Urgeles R, Mitchell NC, Le Bas TP(2002) Slope failures on the flanks of the western Canary Islands.Earth-Sci Rev 57:1–35

Menard HW (1983) Insular erosion, isostasy, and subsidence. Science220:913–918

Mitchell NC (1991) Improving GLORIA images using Sea Beamdata. J Geophys Res 96:337–351

Mitchell NC (1993) A model for attenuation of backscatter due tosediment accumulations and its application to determine sedimentthickness with GLORIA sidescan sonar. J Geophys Res98:22477–22493

Mitchell NC (2001) The transition from circular to stellate forms ofsubmarine volcanoes. J Geophys Res 106:1987–2003

Mitchell NC (2003) Susceptibility of mid-ocean ridge volcanic islandsand seamounts to large-scale landsliding. J Geophys Res 108.doi:10.1029/2002JB001997

Mitchell NC, Tivey MA, Gente P (2000) Slopes of mid-ocean ridgefault scarps from submersible observations. Earth Planet Sci Lett183:543–555

Mitchell NC, Masson DG, Watts AB, Gee MJR, Urgeles R (2002) Themorphology of the flanks of volcanic ocean islands: a compar-ative study of the Canary and Hawaiian hotspot islands. JVolcanol Geotherm Res 115:83–107

Mitchell NC, Dade WB, Masson DG (2003) Erosion of the submarineflanks of the Canary Islands. J Geophys Res 108. doi:10.1029/2002JF000003

Mitchell NC, Beier C, Rosin P, Quartau R, Tempera F (2008)Submarine lava flows around the coasts of Pico Island, Azores.Geochem Geophys Geosyst. doi:10.1029/2007GC001725, 9Paper Q03024

Moore JG, Chadwick WW (1995) Offshore geology of Mauna Loaand adjacent areas, Hawaii. In: Rhodes JM, Lockwood JP (eds),Mauna Loa Revealed: Structure, Composition, History andHazards. American Geophysical Union, pp 21–44

Moore JG, Fornari DJ (1984) Drowned reefs as indicators of the rateof subsidence of the Island of Hawaii. J Geol 92:752–759

Moore JG et al (1989) Prodigious submarine landslides on theHawaiian Ridge. J Geophys Res 94:17465–17484

Morgan JK, Moore GF, Hills DJ, Leslie S (2000) Overthrusting andsediment accretion along Kilauea's mobile south flank, Hawaii:evidence for volcanic spreading from marine seismic reflectiondata. Geology 28:667–670

Nunes JC (1999) A actividade vulcânica na Ilha do Pico doPlistocénico Superior ao Holocénico: mecanismo eruptivo ehazard vulcânico. PhD Thesis. Universidade dos Acores, PontaDelgada, p 355

Nunes JC (2002) Lateral collapse structures in Pico Island (Azores):mechanism, constraints and age, 3ª Assembleia Luso Espanholade Geodesia e Geofísica (3rd Portuguese-Spanish Assembly ofGeodesy and Geophysics), Valencia

Nunes JC et al (2006) Gravity anomalies and crustal signature ofvolcano-tectonic structures of Pico Island (Azores). J VolcanolGeotherm Res 156:55–70

Paris R, Pérez Torrado FJ, Carracedo JC (2005) Massive flank failuresand tsunamis in the Canary Islands: past, present, future.Zeitschrift für Geomorphologie Suppl 140:37–54

Quartau R, Curado F, Bouriak S, Monteiro JH, Pinheiro L (2003)Projecto Gemas—localização e distribuição de areias em redor dailha do Pico, Relatório Técnico INGMARDEP 16/2003. Dep deGeologia Marinha, IGM, Lisboa, Portugal

Quartau R, Trenhaile AS, Mitchell NC, Temera F (2010) Developmentof volcanic insular shelves: insights from observations andmodelling of Faial Island in the Azores Archipelago. Mar Geol275:66–83

Quidelleur X, Hildenbrand A, Samper A (2008) Causal link betweenQuaternary paleoclimatic changes and volcanic islands evolution.Geophys Res Lett. doi:10.1029/2007GL031849, 35 PaperL02303

Scarth A, Tanguy J-C (2001) Volcanoes of Europe. Terra, Harpenden,England, p 243

Searle RC (1980) Tectonic pattern of the Azores spreading centre andtriple junction. Earth Planet Sci Letts 51:415–434

Simkin T (1972) Origin of some flat-topped volcanoes and guyots.Geol Soc Am Mem 132:183–193

Smith JR, Malahoff A, Shor AN (1999) Submarine geology of theHilina slump and morpho-structural evolution of Kilaueavolcano, Hawaii. J Volcanol Geotherm Res 94:59–88

Stretch R, Mitchell NC, Portaro RA (2006) A morphometric analysisof the submarine volcanic ridge of Pico Island. J VolcanolGeotherm Res 156:35–54

Trenhaile AS (2000) Modeling the development of wave-cut shoreplatforms. Mar Geol 166:163–178

van Wyk de Vries B, Kerle N, Petley D (2000) Sector collapse formingat Casita volcano, Nicaragua. Geology 28:167–170

Walker GPL (1973) Lengths of lava flows. Phil Trans Roy Soc LondA274:107–118

Walker GPL (1988) Three Hawaiian calderas: an origin throughloading by shallow intrusions? J Geophys Res 93:14773–14784

Bull Volcanol (2012) 74:483–496 495

Wessel P, Smith WHF (1991) Free software helps map and displaydata. Eos Trans Am Geophys Union 72:441

Woodhall D (1974) Geology and volcanic history of Pico IslandVolcano, Azores. Nature 248:663–665

Yokoyama Y, Lambeck K, De Deckker P, Johnston P, Fifield LK (2000)Timing of the Last Glacial Maximum from observed sea-levelminima. Nature 406:713–716

Zbyszewski G, Ferreira CR, Ferreira OV (1962) Étude géologique del'ile de Pico (Açores). Comunicações dos Serviços Geológicos dePortugal 46:5–66

Zbyszewski G, Ferreira CR, Ferreira OV, Assunção ECT (1963). CartaGeológica de Portugal na escala de 1:50 000. Notícia explicativadas Folhas A e B, Ilha do Pico (Açores). Serviços Geológicos dePortugal, Lisboa

496 Bull Volcanol (2012) 74:483–496