Insights into the emplacement dynamics of volcanic landslides from high-resolution 3D seismic data acquired offshore Montserrat, Lesser Antilles

Marine Geology 335 (2013) 1–15

Contents lists available at SciVerse ScienceDirect

Marine Geology journal homepage: www.elsevier.com/locate/margeo

Insights into the emplacement dynamics of volcanic landslides from high-resolution 3D seismic data acquired offshore Montserrat, Lesser Antilles G.J. Crutchley a,⁎, J. Karstens a, C. Berndt a, P.J. Talling b, S.F.L. Watt b, M.E. Vardy b, V. Hühnerbach b, M. Urlaub b, S. Sarkar b, D. Klaeschen a, M. Paulatto c, A. Le Friant d, E. Lebas d, F. Maeno e, f a

Helmholtz Centre for Ocean Research Kiel, GEOMAR, Wischhofstr. 1-3, 24148 Kiel, Germany National Oceanography Centre, Southampton SO14 3ZH, UK Department of Earth Sciences, University of Oxford, South Parks Road, Oxford OX1 3AN, UK d Institut de Physique du Globe de Paris, UMR 7154, CNRS, Paris, France e Department of Earth Sciences, Wills Memorial Building, University of Bristol, Bristol BS8 1RJ, UK f Earthquake Research Institute, University of Tokyo, 1-1-1 Yayoi, Bunkyo-ku, Tokyo, Japan b c

a r t i c l e

i n f o

Article history: Received 9 September 2011 Received in revised form 7 October 2012 Accepted 11 October 2012 Available online 23 October 2012 Communicated by D.J.W. Piper Keywords: submarine landslide debris avalanche volcanic flank collapse 3D seismic imaging Montserrat

a b s t r a c t We present results from the first three-dimensional (3D) marine seismic dataset ever collected over volcanic landslide deposits, acquired offshore of the Soufrière Hills volcano on the island of Montserrat in the Lesser Antilles. The 3D data enable detailed analysis of various features in and around these mass wasting deposits, such as surface deformation fabrics, the distribution and size of transported blocks, change of emplacement direction and erosion into seafloor strata. Deformational features preserved on the surface of the most recent debris avalanche deposit (Deposit 1) reveal evidence for spatially-variant deceleration as the mass failure came to rest on the seafloor. Block distributions suggest that the failure spread out very rapidly, with no tendency to develop longitudinal ridges. An older volcanic flank collapse deposit (Deposit 2) appears to be intrinsically related to large-scale secondary failure of seafloor sediments. We observe pronounced erosion directly down-slope of a prominent headwall, where translational sliding of well-stratified sediments was initiated. Deep-reaching faults controlled the form and location of the headwall, and stratigraphic relationships suggest that sliding was concurrent with volcanic flank collapse emplacement. We also identified a very different mass wasting unit between Deposit 1 and Deposit 2 that was likely emplaced as a series of particle-laden mass flows derived from pyroclastic flows, much like the recent (since 1995) phase of deposition offshore Montserrat but at a much larger scale. This study highlights the power of 3D seismic data in understanding landslide emplacement processes offshore of volcanic islands. © 2012 Elsevier B.V. All rights reserved.

1. Introduction Some of the largest landslides on Earth occur around volcanic islands, with some events involving up to several thousand cubic kilometers of material (Moore et al., 1989; Masson et al., 2006). Here, we describe submarine mass wasting processes offshore of the currently active Soufrière Hills volcano on the island of Montserrat in the Lesser Antilles, using the first ever three-dimensional (3D) marine seismic survey of volcanic landslide deposits. Landslides have played an important part in the evolution of this volcano (Druitt and Kokelaar, 2002; Wadge et al., 2010; Watt et al., 2012a). Understanding how volcanic island landslides are triggered and emplaced is important because of the hazard they pose directly, and because they can generate potentially very destructive tsunami (Masson et al., 2006; Løvholt et al., 2008). Recent mapping has shown that submarine landslide deposits are common around volcanic islands worldwide ⁎ Corresponding author. Tel.: +49 431 6002269; fax: +49 431 6002922. E-mail address: [email protected] (G.J. Crutchley). 0025-3227/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.margeo.2012.10.004

(e.g. Moore et al., 1989; Watts and Masson, 1995; Urgeles et al., 1997; Krastel et al., 2001; Oehler et al., 2004; Boudon et al., 2007). Tsunami hazard depends strongly on the way in which the landslides are emplaced, especially their volume, initial acceleration, and single or multistage nature (Løvholt et al., 2005; Harbitz et al., 2006; Masson et al., 2006). Better constraining how landslides are emplaced is challenging because we are yet to monitor directly a large-volume volcanic landslide that enters the sea, meaning that much of our understanding must be based on the rock record of previous landslide deposits and advanced through experiments and modeling strategies. In recent decades, our understanding of submarine landslide deposits has been improved by advances in various marine geophysical techniques, including swath bathymetry, side-scan sonar, 2D multichannel seismic acquisition, and high-frequency echo-sounding (e.g. Masson, 1996; Krastel et al., 2001; Le Friant et al., 2004; Boudon et al., 2007). All of these methods are, however, two-dimensional approaches, with inherent limitations on the spatial resolution at which landslide morphology and internal structures can be imaged. The high spatial resolution provided by 3D seismic data has the capacity to advance our

G.J. Crutchley et al. / Marine Geology 335 (2013) 1–15

understanding of landslide emplacement processes, imaging surfaces and structures within deposits at a level of detail that is necessary to test models of how material moves and evolves during the emplacement process. Montserrat (Fig. 1A, B) offers an ideal location to acquire such 3D seismic data. Due in part to its ongoing eruption since 1995, Montserrat is one of the best-studied island arc volcanoes on Earth (e.g. Druitt and Kokelaar, 2002, and references therein; Wadge et al., 2010). There is a wealth of knowledge regarding dome evolution and collapse, pyroclastic flow emplacement, and the distribution of large landslide deposits (Fig. 1B) (Deplus et al., 2001; Harford et al., 2002; Le Friant et al., 2004; Trofimovs et al., 2006; Boudon et al., 2007; Le Friant et al., 2009, 2010; Lebas et al., 2011; Watt et al., 2012a, 2012b). In 2010, during research cruise JC45/46 aboard RSS James Cook offshore Montserrat's eastern coast, a seismic volume was acquired in an area where deposits derived from pyroclastic flows generated by partial dome collapses during the ongoing eruption of Soufrière Hills

63

62 [°W] 61

A

5

[x 10 m East]

5.85

60

5.95

6.05

B

1.86 0

Silver Hills

Anguilla St. Martin 18 St. Bathelemy Barbuda Saba St. Kitts Antigua Nevis Montserrat

have been emplaced offshore (Trofimovs et al., 2006; Le Friant et al., 2009, 2010). These deposits overlie much more extensive and blockier volcanic flank collapse deposits, including the near-surface Deposit 1 and the deeper Deposit 2 (Deplus et al., 2001; Le Friant et al., 2004). The western extent of the 3D survey lies approximately 7.5 km east of Soufrière Hills, overlapping with the tip of the recent pyroclastic flow-derived deposits, and covering much of Deposit 1 and the northern reaches of Deposit 2 (Fig. 1B, C). The aim of this paper is to provide insight into the emplacement processes of volcanic flank collapse deposits offshore Montserrat beyond that which could be constrained previously from 2D seismic data (e.g. Le Friant et al., 2004; Lebas et al., 2011; Watt et al., 2012a, 2012b). We target three different deposits: Deposit 1, Deposit 2, and a more coherent stack of reflections that separates Deposit 1 from Deposit 2 — hereafter referred to as the “Intermediate Unit”. The first objective is to determine what the 3D data can tell us about the character of Deposit 1 and how it was emplaced. Are

[mbsl]

1200 pyroclastic deposits emplaced from W-E since 1995 (white boundary) Dep. 2, initially emplaced from W-E, runs out further to the south.

6

embayment east of Soufrière Hills

Rough seafloor (protruding blocks)

Centre Hills

Guadeloupe 16

1.85

P-Cable 3D survey

Dominica [°N]

Soufrière Hills Martinique

14

[x 10 m North]

2

Dep. 1 (emplaced from W-E)

St. Lucia St. Vincent Grenadines

Dep. 2b (upper part)

Dep. 2a (lower part)

South Soufrière Hills

5 km

Grenada

C

800

1045

mbsl

1 km

headwall scarp

1 km 1850000

p ll scarp

morphological slab - displaced from headwall?

headwa

rough seafloor in the west of the survey from protruding blocks

1851000

UTM m North

1849000 pockmarks

1848000

598000

600000

602000

UTM m East

604000

N

Fig. 1. A) The Lesser Antilles Volcanic Arc. Enlarged field of view in (B) is outlined by the box. B) The island of Montserrat (gray shades) with volcanic provinces annotated. Bathymetry around the island is color-shaded with a semi-transparent slope map superposed. The 3D seismic survey is given by the black box. Spatial extents of mass transport deposits in this area identified from previous studies (Deplus et al., 2001; Le Friant et al., 2004; Boudon et al., 2007; Le Friant et al., 2009; Lebas et al., 2011; Watt et al., 2012a, 2012b) are shown by the various colored lines: white — recent pyroclastic deposits, yellow — Deposit 1, orange and red — Deposit 2 (two phases; 2a in red, 2b in orange). Unlike Deposit 1, Deposit 2 bends around and extends much further to the south. Rough seafloor east of Montserrat is the manifestation of protruding blocks. C) Enlarged field of view from the black box in (B). Seafloor bathymetry derived from the 3D seismic data. Annotated are; deposit extents as in (B), prominent headwall scarp (broken white line), morphological slab apparently displaced from the headwall (Watt et al., 2012a, 2012b) (dotted white line), series of pockmarks down-slope of the slab, and the rough seafloor in the west of the survey.

G.J. Crutchley et al. / Marine Geology 335 (2013) 1–15

there internal reflections that might suggest multiple lobes, or stages of emplacement? Are all of the blocks protruding above the seafloor derived from the same flank collapse event that formed Deposit 1? This is important because the presence of such blocks provides insight into the composition and origin of the failed material, while the size and spatial distribution of blocks can help to constrain emplacement processes. The 3D seismic data provide more detailed constraints on seafloor morphology than previous swath bathymetry data have, so we examine the surface of Deposit 1 for any evidence of emplacementrelated deformation. The second objective is to understand the nature of the Intermediate Unit and determine its most likely composition and origin. It has been proposed that the Intermediate Unit represents the submarine continuation of deposits derived from pyroclastic flows (Watt et al., 2012a), similar to those generated by the 1995–recent eruptions (Trofimovs et al., 2006; Le Friant et al., 2010). This proposal is critically evaluated using the more detailed 3D seismic data. Are the seismic characteristics, external shape, and run out distance of the Intermediate Unit consistent with such an origin? Understanding its origin is important for constraining the evolution and hazards associated with this volcano, and the types of events that precede or post date major landslides. The third objective is to understand better the evolution and emplacement dynamics of Deposit 2. We focus on the origin of a major headwall scarp (Fig. 1C) that marks the northeastern extent of Deposit 2. How does this feature record remobilization and incorporation of seafloor sediment into Deposit 2, which seems to have been initially emplaced as a volcanic flank collapse-style failure (Le Friant et al., 2004; Lebas et al., 2011; Watt et al., 2012a, 2012b)? We examine this area in detail in order to shed light on how this area of the seafloor failed. Understanding how volcanic island landslides can destabilize adjacent seafloor sediment is important for understanding landslide evolution and tsunami generating potential. 2. Geological setting Montserrat is an island within the Lesser Antilles Volcanic Arc, formed by subduction of the North American Plate beneath the Caribbean Plate (Macdonald et al., 2000). The island is dominated by andesitic rocks existing in a wide range of deposits, including remnants of lava domes, pyroclastic flow deposits and lahar and debris avalanche deposits (Harford et al., 2002; Le Friant et al., 2004, 2008). Three main volcanic centers make up the island; the Silver Hills massif (~ 2.6 to ~ 1.6 Ma), the Centre Hills massif (at least ~ 950 ka to ~ 550 ka), and the Soufrière Hills–South Soufrière Hills massif (at least ~ 170 ka to present) (Fig. 1B) (Harford et al., 2002). After a dormant period of ~ 350 years, Soufrière Hills began erupting in 1995 (Boudon et al., 2007) and has continued to do so until the present day. The pyroclastic deposits emplaced offshore Montserrat since 1995 are derived from pyroclastic flows that were converted into waterladen mass flows as they entered the ocean (Trofimovs et al., 2006; Le Friant et al., 2009). Most of these deposits were transported to the east of Soufrière Hills, and form a lobe offshore of the Tar River Valley, extending approximately 8 km from Montserrat's eastern shoreline (Fig. 1B, C). Deposit 1 is a sub-circular, blocky volcanic flank-collapse deposit that covers ~50 km2 on the seafloor and has an estimated volume of ~1.7 km3 (Lebas et al., 2011) (Fig. 1B). It was emplaced in a west to east direction, spreading laterally as it entered the ocean (Le Friant et al., 2004). In 2D seismic data it is characterized by incoherent and low-amplitude reflectivity, and the shape of the deposit suggests that it was emplaced as a debris avalanche, characterized by a freely-spreading cohesionless flow of heterogeneous, coarse grained material (Lebas et al., 2011). The Intermediate Unit lies directly beneath Deposit 1 and is characterized by more coherent and laterally continuous, albeit irregular reflections (Fig. 2). The unit, which thins to the southeast (away from the island), may represent a stack of deposits derived from

3

pyroclastic flows, formed in a similar way to the 1995–present pyroclastic deposits on top of Deposit 1 (Watt et al., 2012a). Deposit 2 lies beneath the Intermediate Unit (Fig. 2). It covers a significantly larger area (~ 190 km 2) than Deposit 1 and has a total volume of ~ 8.4 km 3, which can be subdivided into two internal subunits (Deposit 2a, ~ 2.8 km 3 and Deposit 2b, ~ 5.6 km 3) (Lebas et al., 2011; Watt et al., 2012a) (Fig. 1B). It is has also been interpreted that Deposit 2 was initially emplaced in a west to east direction (Le Friant et al., 2004), but the deposit extends further to the east where it then bends around and runs out to the south. The two sub-units in the deposit, most simply interpreted as representing two distinct mass wasting phases, each comprise a substantial proportion of comparatively well-bedded seafloor sediment in addition to regions of less-coherent, low-amplitude reflectivity similar to that of Deposit 1. The existence of well-bedded, presumably fine-grained seafloor sediment within Deposit 2 is primarily what makes it different from Deposit 1. Watt et al. (2012a) interpret that large-scale secondary failure of seafloor sediments accompanied the emplacement of volcanic material from a flank collapse, resulting in the much more extensive run-out of Deposit 2 to the south. They interpreted a seafloor scarp in the eastern part of the survey area (Fig. 1C) as the north-eastern boundary of the deposit and a headwall marking the up-dip limit of the sedimentary component of the slope failure. A morphological slab, in the form of a smooth-surfaced rectangular patch of seafloor, directly down-slope of the headwall and upslope of a group of pockmarks (Fig. 1C), was interpreted as a displaced slab of seafloor sediment that slid away from the headwall (Watt et al., 2012a). 3. Methods 3.1. Seismic acquisition The 3D seismic data were acquired with the P-Cable system belonging to the National Oceanography Centre of Southampton. The system consisted of 12 streamers attached to a cross cable that was towed perpendicular to the ship's steaming direction. The volume, covering a surface area of 38.5 km 2, was acquired over a period of ~ 7 days. The source consisted of two 150 in 3 GI guns towed at a depth of 3 m directly behind the vessel, which were fired at 7 s intervals resulting in an average shot spacing of 10–15 m (average ship speed of 3–4 knots). Data were sampled at 1 ms. 3.2. Seismic processing The key seismic processing steps involved: navigation quality control, geometry calculations, trace editing, band-pass filtering (with corner frequencies of 24, 50, 300 and 400 Hz), normal move-out correction with a constant 1500 m s−1 velocity field, stacking, trace interpolation, and 3D time migration. Initial stacks were generated from a 15 m by 15 m stacking grid — the smallest bin size achievable given the distribution of traces in the survey. Such coarse grid spacing proved undesirable for the frequency content in the data, resulting in some spatial aliasing of energy. In order to decrease spatial aliasing we conducted 2D post-stack interpolations, first in the cross-line direction, then in the in-line direction. The interpolated stack, with a bin spacing of 7.5 m × 7.5 m, was then time-migrated with a 3D post-stack Kirchhoff routine assuming a constant velocity field (1500 m s −1). More details pertaining to all processing steps are given in Supplementary Material A. 3.3. Seismic attributes In order to extract more information from the volume, we used the OpenDtect software to calculate 3D seismic attributes (e.g. Chopra and Marfurt, 2007). We found the most useful attribute for this study to be seismic similarity (see details in Supplementary Material B), which

4

G.J. Crutchley et al. / Marine Geology 335 (2013) 1–15

1.1

West

?

B

Recent pyroclastic deposits

?

1.2

C

TWT (s)

750 m

East

A

N

Dep. 1

D

A

1 km lens

Dep. 2 base

1.5

50 m @ 2000 m/s

Intermediate Unit

1.4

1.6

inset

Intermediate Unit (between Dep. 1 and Dep. 2)

Dep. 2

not imaged

8x vert. exag. at seafloor South

North 750 m

B

Dep. 1 1.3 well-defined reflections

Intermediate Unit

1.4

Dep. 2

?

50 m @ 2000 m/s

TWT (s)

Dep. 1

?

6.5x vert. exag. at seafloor

C South

Recent pyroclastic deposits (above Dep. 1)

North

1.1

D West

1.2

East

Recent pyroclastic deposits: (more reflective unit above Dep. 1)

N

TWT (s)

Dep. 1

s ple lenxam e

1.2

in map view

1 km

Dep. 1 lens

1.4

Dep. 2 750 m

TWT (s)

50 m @ 2000 m/s

Intermediate Unit

500 m lens

1.4

6x vert. exag. at seafloor

Intermediate Unit

Dep. 2 6x vert. exag. at seafloor

Fig. 2. Four example profiles (A–D) extracted from the volume to highlight the deposits discussed in this paper. Intersections between sections are marked by vertical red lines. The inset sub-figure (upper right of (A)) shows the bathymetry and deposits extents as plotted in Fig. 1C, as well as the locations of the profiles (A–D) displayed in this figure. A) In-line showing Deposit 1 beneath the seafloor, the Intermediate Unit beneath Deposit 1, and Deposit 2 beneath the Intermediate Unit. Interpreted deposit boundaries are marked by broken white lines. A lens of low reflectivity within the Intermediate Unit is marked by the broken yellow line. B) Cross-line showing the same deposits. Relatively well-defined concave-down reflections within Deposit 1 are highlighted by the finely dotted line. Well-defined, sub-horizontal reflections in the Intermediate Unit are annotated. C) Cross-line showing Deposit 1, overlying pyroclastic deposits, the Intermediate Unit, and Deposit 2 at depth. Broken yellow lines within the Intermediate Unit and the recent pyroclastic deposits highlight some of the well-defined reflections in these units that often separate lens-shaped sub-units of low-reflectivity. D) In-line section showing the same units as (C). Broken yellow lines as in (C). A low-amplitude lens shape between well-defined reflections in the Intermediate Unit is labeled (the same lens as that labeled in both (A) and (C)). This lens can be traced laterally in the dataset. It thins to the east, and in map view it delineates a lobe-shaped expression (black outline in inset map in this sub-figure). The Intermediate Unit extends further east than the recent pyroclastic deposits above Deposit 1.

yields information about the lateral and temporal continuity of seismic facies. This assisted in the spatial mapping of seismically incoherent mass wasting material in comparison to well-bedded, in-situ sedimentary strata. 4. Results 4.1. Data resolution and quality Post-stack trace interpolation resulted in a significant improvement in both horizontal and vertical resolution after migration (Fig. 3 —

comparing A and B). The dominant frequency in the dataset is ~80 Hz and significant energy exists up to 150 Hz (Fig. 3C). We predict the maximum vertical resolution in the data to be approximately 2.5 m, i.e. quarter of a wavelength assuming 1500 m s −1 sound speed in shallow seafloor sediments. For higher sound speeds this resolution decreases, for example to ~4 m at 2400 m s −1. The well-stratified sediments in the east of the survey area reveal the high signal to noise ratio (SNR) in the data. Steeply-dipping normal faults can be easily identified at offset reflections in both cross-lines (Fig. 3D) and in-lines (Fig. 3E), with no noticeable difference in the SNR between the two orthogonal directions. Incoherent

G.J. Crutchley et al. / Marine Geology 335 (2013) 1–15

East

1.6 West

AA

D

TWT (s)

TWT (s)

100 m @ 2000 m/s

1.3

750 m

1.4

500 m

15 m bin size East

TWT (s)

1.6 West

BB

1.5

1.7

500 m

7.5 m bin size

West

50 m @ 2000 m/s

100 m @ 2000 m/s

1.6

1.8

1.9

North

Cross-line direction coherent reflections normal faults

1.8

1.9

South

5

12x vert exag. at seafloor

low-amplitude, incoherent deposits

1.2

EE

East

In-line direction 750 m coherent reflections, sub-horizontal

coherent reflections, folded and faulted (extensional faults)

TWT (s)

1.3

1.4

1.5

CC

E

A,B

Amplitude

1 km

50 m @ 2000 m/s

1.0

D

D

0.8

0.4 0.2 0

N

inset

12x vert exag. at seafloor

Frequency Spectrum

0.6

80

160

240

280

Frequency [Hz]

Fig. 3. A) An example section of data from the migrated volume with 15 m trace spacing (see inset map, lower left, for location). B) The same section location as in (A) after having applied post-stack trace interpolation to reduce the trace spacing to 7.5 m. The pronounced improvement in both spatial and temporal resolution is apparent. C) Frequency spectrum of the data, showing a dominant frequency of ~80 Hz and significant energy up to ~150 Hz. D) An example cross-line in the volume (see inset subfigure for location). Vertical red line shows intersection with profile in (D). E) An example in-line from the volume (see inset subfigure for location). Vertical red line shows intersection with profile in (D).

deposits in the west of the 3D survey are characterised by lowamplitude, chaotic reflectivity; a stark contrast to the well-stratified sediments in the east (Fig. 3E). 4.2. Transported blocks The high resolution of the seismic data enables a detailed analysis of the distribution and individual sizes of blocks on the seafloor. By picking the seafloor reflection, a bathymetric grid with 7.5 spatial resolution

was generated (Figs. 1C, 4). The grid reveals significant seafloor detail, with a subtle E–W grain that is an artifact of acquisition (ship track artifact, Fig. 4A). In addition to mapping the seafloor reflection, some blocks can also be identified and mapped to some extent beneath the seafloor. This proved most feasible for larger blocks (Fig. 5). Protruding blocks occur directly east (down slope) of an embayment east of Soufrière Hills (Figs. 1B, C and 4). Most of the blocks have surface areas of b~3.5 × 104 m 2 (Fig. 4B), and they are randomly distributed without preferential accumulation of blocks within the central or

6

G.J. Crutchley et al. / Marine Geology 335 (2013) 1–15

20 Number of blocks

A 598686 599686

(m E

ast) 601686

apparently random block distribution

Block 1

B

Histogram showing surface area distribution of blocks in Deposit 1

10

0 0

large marginal blocks

ship

track

head

wall

4 2 3 1 surface area (m2 x 10 4 )

artef

acts

scar

p

1 km 2

morp slab hological

(m

No

rth

)

ent s ext k ern 1 bloc t s ea ep. D f o

5

Block 2 100 m high and ~100 m buried

1848413

1847413 7.5x vert. exag.

Fig. 4. A) Bathymetry within the survey area derived from picking the seafloor reflection in the seismic data. Blocks to the west of the broken white line (including “Block 1”) are buried relatively shallow; shallower than the base of Deposit 1. They have an apparently random distribution. Blocks to the east of this line (including “Block 2”) appear to be rooted in the deeper Deposit 2. Also labeled are: the pronounced headwall scarp, the morphological slab and acquisition artifacts (stripes on the seafloor parallel to the data acquisition direction). B) Histogram of the surface area of blocks rooted in Deposit 1 — i.e. those to the west of the broken white line in (A).

1.2

A West

(C)

East

(B)

1000 m

disturbed and discontinuous internal stratification

Dep. 1 Block 2

1.3

TWT (s)

Intermediate Unit

? irregular upper surface of Dep. 2

?

?

50 m @ 2000 m/s

Dep. 2

1.5 5.7x vert. exag. at seafloor

B

B

inset

North

(A)

Block 2

Headwall

500 m reflections abutted against block

reflections abutted against block

1 km A

TWT (s)

1.3

South

B

C C

1.5

basal reflections?

TWT (s)

50 m @ 2000 m/s

1.2

C

N

Southwest 500 m

Northeast

(A) Block 1

Dep. 1

1.3 Intermediate Unit

1.7 2.6x vert. exag. at seafloor

7x vert. exag. at seafloor

Fig. 5. A) E–W oriented profile (see inset in sub-figure (B) for location) cutting through Block 2, whose interpreted extent is shown by broken white lines. The interpreted upper surface of Deposit 2 and the lower surface of Deposit 1 are marked by solid white lines. B) N–S oriented profile cutting through Block 2 showing the interpreted extent of the block, as in (A). Interpreted basal reflections are circled yellow. C) SW–NE profile showing a smaller block (Block 1) lodged in Deposit 1 — interpreted extent annotated as in (A). Red vertical lines in each sub-figure mark the lines of intersection between profiles.

G.J. Crutchley et al. / Marine Geology 335 (2013) 1–15

marginal parts of the block-field (Fig. 4A). However, a smaller number of unusually large blocks occur at the margins of the blocky area. Some of these large blocks occur to the south of the 3-D cube, but the 3-D cube includes the largest block that is hereafter termed Block 2 (Fig. 4A). Block 2, oriented in an east–west direction, extends ~900 m east– west and ~700 m north–south, occupying a surface area on the seafloor of 40 ha (400,000 m 2) and standing 100 m high at its highest point (Fig. 4A). This block is by far the largest of all blocks mapped in the 3D data. Reflections within Block 2 are relatively disturbed and discontinuous, but there is evidence for internal stratification (Fig. 5A, B). The extent of the block beneath the seafloor can be estimated by mapping the truncations of shallow, sub-horizontal strata that abut against the block, and by picking, in places, what appears to be the block's basal reflection (Fig. 5B). This exercise required close examination of all in-lines and cross-lines transecting the block. The buried part of the block appears to be at least as voluminous (possibly twice as voluminous) as the part protruding out of the seafloor, indicating that its total volume lies somewhere between 5 × 107 and ~8 × 107 m 3. Although it is difficult to determine precisely where the base of the block is, it certainly appears to be deeper than the base of Deposit 1 (Fig. 5A, B). Most of the seafloor blocks within the 3D volume lie to the west of Block 2 and are buried significantly shallower; shallower than the base of Deposit 1 (reflections of the underlying Intermediate Unit can be clearly imaged beneath them). We counted a total 97 of these shallow blocks, having surface areas ranging from 1200 m 2 to 50,000 m 2 (Fig. 4B). The largest block in this distribution, i.e. excluding Block 2, has a volume above the seafloor of just over 1 × 106 m 3, and the sub-seafloor imaging suggests that approximately the same volume again is buried (Block 1, Fig. 5C). Therefore, Block 1 is more than twenty times smaller than Block 2. 4.3. Deposit 1 The seismic unit that defines Deposit 1 is characterized by chaotic reflectivity of relatively low amplitude (Fig. 2). There are, however, continuous reflections within the deposit, such as the pronounced concave-down reflections highlighted in Fig. 2B. These concave down-reflections are the most laterally-continuous internal reflections identifiable within Deposit 1, existing undisrupted over several hundreds of meters. Although locally disrupted, the reflections can be traced over lateral extents of more than 1 km, but less than 2 km (Fig. 2B). Other internal reflections within Deposit 1 are much less spatially-extensive, typically traceable over extents of a few hundred meters up to 500 m. Although individual blocks on the seafloor dominate the surface topography of Deposit 1 (Fig. 4), there are also deformation patterns that can be identified on a larger scale. By calculating the deviation of local slope angles from the mean slope angle in the western half of the survey area, north–south trending wave patterns are illuminated (Fig. 6A, B). The waves define stepped ridge topography with dominant wavelengths of ~200–400 m and peak amplitudes up to ~5 m. Although the wave patterns are quite well-developed, in some places they are not laterally continuous, but rather disrupted and offset (Fig. 6B). In the northern reaches of the volume, the wave-patterns bend around slightly in the up-slope direction. Because the volume does not extend far enough south to capture the southern extent of Deposit 1, we examined multi-beam bathymetry data outside the survey area for evidence of similar patterns further south. The slope gradient map reveals these same wave patterns (Fig. 6C) bending around in the up-slope direction (i.e. defining a convex-downslope regional wave pattern). 4.4. The Intermediate Unit The Intermediate Unit between Deposit 1 and Deposit 2 comprises spatially-continuous, high-amplitude reflections, which bear similarities

7

to packages of reflections in the recent pyroclastic fan deposits above Deposit 1 (Fig. 2C, D). There is no evidence for blocks within the Intermediate Unit like those mapped on the seafloor as part of Deposit 1, and there is a marked difference in the seismic appearance of the Intermediate Unit when compared to both Deposit 1 and Deposit 2. The unit has a thickness of approximately 50 m in the proximal extents imaged by our data, and thins with distance from Montserrat. The thickness is generally slightly thicker than that of Deposit 1 (Fig. 2A, C, D). The Intermediate Unit extends ~4 km farther east than the distal extent of the recent pyroclastic fan deposits, approximately as far eastward as the distal extent of Deposit 1. The base of the unit infills the irregular upper surface of incoherent reflectivity that marks the top of Deposit 2 (Fig. 5A). Internal reflections can be traced locally over distances of several hundred meters throughout the Intermediate Unit, in places defining upper and lower bounds of lens-shaped, lower-amplitude seismic facies (Fig. 2C, D). An example of a lens-shaped sub-unit that was traced throughout the dataset is shown in Fig. 2D (labeled “lens”). This lens thins and narrows in map view with increasing distance from Montserrat. In other places, well-defined stacks of reflections extend vertically over two-way time intervals of more than 500 ms (e.g. Fig. 2B).

4.5. Deposit 2 and the headwall scarp The nature of the north-eastern extent of Deposit 2 is particularly interesting, as this is the region where Watt et al. (2012a) suggest that large-scale sedimentary failure was initiated at a prominent headwall after loading of the seafloor by material from initial failure of the volcanic edifice. The headwall is a well-defined escarpment with ~30 m of relief on the otherwise very smooth seafloor in the east of the survey area (Figs. 1C, 4A). The general trend of the headwall is from NW to SE, but it is characterised by sharp bends, in particular by a 90° bend at its north-eastern-most point (Fig. 1C). Directly down-slope of this 90° bend is the morphological slab interpreted as a section of stratified sediments displaced down-dip of the headwall (Watt et al., 2012a) (Fig. 1C). Fig. 7 shows three seismic sections extracted from the volume that highlight the stratigraphic relationships between the low-amplitude, incoherent body of Deposit 2 and the well-stratified sediments that lie beneath it and directly east of it. The sections show truncation of prominent well-bedded reflections up against the incoherent body of Deposit 2, over a vertical extent of as much as ~ 60 m (Fig. 7A–C). In the central parts of the survey area, directly east of Block 2, the reflection truncation extends over the entire vertical extent of the incoherent deposit (Fig. 7A), whereas further south, truncation is limited to the lower section of the deposit (Fig. 7B). Deposit 2 itself in this region is separated into two parts (Deposit 2a (lower part) and 2b (upper part)) by coherent, sub-horizontal reflections (Fig. 7B). A relatively thin lens of incoherent reflectivity at the eastern extent of Deposit 2 can be traced through parts of the dataset, overlying well-stratified strata (Fig. 7B, C). In order to map out spatially the nature of the boundary between the incoherent body of Deposit 2 and the sediments to the east of it, we extracted similarity along a time-slice at a depth that lies within Deposit 2 in this region (Fig. 8A, B). The similarity distribution on this horizontal slice (Fig. 8B) clearly reveals a curved boundary marking the margin between the seismically-incoherent portion of Deposit 2 and the well-stratified sediments to the east. Profiles extracted across the headwall reveal deep-seated faults that extend from much deeper than the base of Deposit 2 up to the seafloor expression of the headwall (Fig. 9A, B). A time slice of the similarity attribute extracted at 1.76 s two-way travel time (TWT) (approximately 500 m below the seafloor) (Fig. 9C) reveals the low-similarity streaks of these deep reaching faults in map view. The E–W trending section of the headwall is aligned with an E–W trending, S-dipping fault, and the N–S trending section is aligned with a N–S trending, W-dipping fault.

8

G.J. Crutchley et al. / Marine Geology 335 (2013) 1–15

5.96

5.98

5

(x10 m E)

6.00

5.96

5.98

6.00

A

B 1 km

1 km

(x106m N)

1.850

1.848

N

N

gentlest east dip 5.96

steeper east dip 5.98

5

(x10 m E)

6.00

6.02

6.04

C 1.852

extent of view in (A) and (B)

N

1 km

(x10 6 m N)

1.850

1.848

3D survey extent

1.846

Fig. 6. A) Deviation of local seafloor gradients from the mean gradient (calculated along the in-line direction, from west to east) in the western half of the volume. Areas of the slope that are equal to or less than the mean gradient are shaded white. Other colors (yellow through to black) represent east-dipping seafloor in excess of the mean gradient, with yellow highlighting the gentlest — and black the steepest east-directed dips. The white stripes amid other colors highlight a ridge-like morphology. B) Ridge morphology highlighted by annotated black lines that track contrasts between white regions and colored regions. C) Multi-beam bathymetry map plotted as in Fig. 1B, with slopes enhanced by superposition of a semi-transparent slope map. Wavy ridge patterns on the seafloor are indicated by arrows, which are drawn perpendicular to the ridges.

Southeast of the morphological slab, the headwall scarp can be traced beneath the seafloor, merging into a well-defined reflection (termed “Reflection A”, Fig. 10A, A′) that marks a boundary between more steeply-dipping reflections below and more gently dipping reflections above. The thin lens of incoherent reflectivity, previously shown in Fig. 7B and C, lies directly on top of Reflection A (Fig. 10A, A′). Well-defined, gently-dipping strata on top of the lens lap onto the headwall in the up-slope direction. Reflection A (Fig. 10A′) can be traced laterally in the data on the down-slope side of the headwall. Further to the northwest, in a profile through the morphological slab

(Fig. 10B, B′), Reflection A is again observed to merge in the up-dip direction with the sub-seafloor continuation of the headwall. We note two clear differences in profiles extracted through the displaced morphological slab compared to profiles southeast of the slab. Firstly, there is no thin lens of incoherent reflectivity beneath the morphological slab (Fig. 10B, B′). Secondly, a package of reflections (shaded green, Fig. 10B′) above Reflection A exhibits the same dip as strata to the northeast (up-slope) of the headwall. There are, however, also strata above Reflection A in Fig. 10B′ that dip back towards the headwall, directly upslope of the green-shaded polygon.

G.J. Crutchley et al. / Marine Geology 335 (2013) 1–15

A

BB West

East

West

9

East

750 m

750 m

Block 2

C B

1.4

1.6

50 m @ 2000 m/s

Dep. 2b

TWT (s)

50 m @ 2000 m/s

TWT (s)

Dep. 2

N

A

over-topping lens 1.4

inset

1 km

Dep. 2a

1.6

C C E

N

ng lens

1 km

over-toppi

1.4

TWT

~60 m Dep. 2

TWT (s)

base of thin over-topping lens

surface of trunc ated refl (projec ections te from se d forward in 3 D ismic lin e)

ata

d de

ed

ll-b

we

str

1.6

Fig. 7. Examples of reflection truncation on the eastern margin of the incoherent body of Deposit 2 (see inset in sub-figure (B) for the locations of profiles). Broken white lines highlight the outline of the incoherent Deposit 2 body; solid white lines highlight well-defined, truncated reflections. A) Example of reflection truncation directly east of Block 2, where truncation extends from the base to the top of the deposit. B) Example of truncation in the southern part of the volume, where truncation occurs adjacent to the lower section of the deposit. The upper section forms a comparatively thin lens that overlies coherent reflectivity. Deposit 2 in this profile is clearly separated into two sub-units (Deposit 2a and 2b), separated by the coherent reflections (dotted yellow line). C) 3D view showing the 3D manifestation of the surface where reflections are truncated, and the base of the thin over-topping lens. Vertical exaggeration is approximately a factor of 8 in this sub-figure.

5. Discussion 5.1. Mass movement emplacement mechanisms Submarine landslides, often also referred to as mass movements, can be characterized by a range of differing emplacement styles and mass transport deposits (Nardin et al., 1979; Locat and Lee, 2002; Tripsanas et al., 2008). Here, we give a brief overview of three classes of submarine mass movements that are relevant to this study, namely debris avalanches, particle laden flows, and translational slides. Debris avalanches typically originate from relatively deep-seated failures of cohesive, lithified material on steep slopes. In the case of Montserrat they are sourced from the steep slopes of the volcano. Initial slope failure develops into a debris avalanche through a process of shearing along relatively weak surfaces (like bedding planes or faults), frictional fragmentation and dilation (Pollet and Schneider, 2004; Tripsanas et al., 2008). The collision of individual clasts produces a granular material that may, in some cases, drive progressive evolution from sliding to flowing behavior (Pollet and Schneider, 2004). Particle laden flows offshore Montserrat form when pyroclastic flows enter the sea (McLeod et al., 1999; Trofimovs et al., 2006; Le Friant et al., 2009). Coarse particles are retained in the basal part of the flow, which behave as a gravity

current that runs out over the seafloor, ultimately resulting in the deposition of pyroclastic lobes. Translational slides refer to the movement of coherent masses of sediment above a basal shear plane that is subparallel to the surface slope (Coleman and Prior, 1988; Mulder and Cochonat, 1996). Headwalls delimit the up-slope extent of translational slides. 5.2. The emplacement of deposit 1 Deposit 1 was emplaced as a debris avalanche, where material that collapsed from the flanks of Montserrat was channeled offshore and dispersed onto the seafloor as a blocky, seismically-incoherent, fan-shaped deposit (Le Friant et al., 2004; Lebas et al., 2011; Watt et al., 2012a). Close inspection of the deposit revealed internal reflections, some of which can be traced laterally over distances of more than a kilometer (e.g. Fig. 2B). Well-defined, spatially-continuous internal reflections separating Deposit 2 into two distinct units have been interpreted as representing a boundary between two different phases of volcanic flank collapse (Watt et al., 2012a). The internal reflections in Deposit 1 are much more poorly developed and are not as spatially-extensive. The origin of these internal reflections is ambiguous. They may represent divisions between separate phases (or pulses)

10

G.J. Crutchley et al. / Marine Geology 335 (2013) 1–15

West

A 1 km N A

50 m @ 2000 m/s

TWT (s)

inset

East

1 km

Block 2

1.3

1.4

incoherent seismic facies of Dep. 2

B

Well-stratified sediments

timeslice: 1.465 s in (B) 1.5

B

low similarity disaggregated volcanic material

1 km high similarity sediments

1.0

16°43'30''

seismic section in (A)

similarity

(N)

N 16°42'30''

0.1 62°4'30''

(W)

62°3'30''

62°2'30''

Fig. 8. The location of the seismic section in (A) is marked by the black line labeled “A” in the inset sub-figure (upper left). The areal extent of the time-slice in (B) is shown by the red box. A) E–W oriented section highlighting Block 2, the incoherent seismic facies of Deposit 2 (within the broken white line), and well-stratified sediments to the east of Deposit 2 (solid white lines). Reflection truncation up against Deposit 2 is marked by the broken yellow line. The horizontal red line shows the depth of extraction of the time-slice given in (B). The blue dot marks the point on this line between incoherent reflectivity to the left and well-stratified sediments to the right. B) Seismic similarity extracted from the time slice highlighted in (A) over the region highlighted in the inset sub-figure. The seismically-incoherent facies of Deposit 2, west of the annotated blue line, manifests itself as much lower similarity than the well-stratified sediments to the east (color scale in left of figure). Annotated white arrows show interpreted emplacement directions of disaggregated volcanic material in Deposit 2.

of one major event (e.g. Lebas et al., 2011), or they may represent a period of normal sedimentation between completely separate volcanic flank collapse events. We prefer the former interpretation of Lebas et

A

al. (2011), as we do not see evidence for multiple parallel reflections like those seen in Deposit 2, which would indicate a time gap where a sequence of stratified sediments was deposited.

SW

NE

1.2

B

50 m @ 2000 m/s

B

W

500 m

E

1000 m

C N

1.76 s similarity time-slice

Dep. 2

TWT (s)

C

16° 44'

1.4

1.6

1.6

100 m @ 2000 m/s

1 km

AA

TWT (s)

1.4

inset inset

1.8

16° 43.5'

-62° 2'

-62° 1.5'

Fig. 9. The nature of the headwall scarp at depth. Black lines in the inset sub-figure show the locations of seismic sections in (A) and (B); the white box shows the spatial extent of similarity time slice shown in (C). A) SW–NE oriented section through the headwall; broken yellow line highlights a fault extending to great depths from the headwall at the seafloor. The extent of the seismically incoherent body of Deposit 2 is annotated by the broken white lines. B) E–W oriented section through the headwall; broken yellow line highlights a deep-reaching fault. C) Time-slice (at 1.76 s — i.e. much deeper than Deposit 2) showing seismic similarity from the region of the white box in the inset sub-figure. Similarity is plotted in grayscale; pale shades represent high similarity, darker shades represent low similarity. The solid blue line shows the projection of the headwall location to this time slice. Dotted yellow lines highlight low similarity streaks in plan view of the prominent faults shown in the sections given in (A) and (B). They clearly mimic the shape of the headwall.

G.J. Crutchley et al. / Marine Geology 335 (2013) 1–15

B

N A

Block 2

NE

A 750 m

1 km

NE

SW 500 m

50 m @ 2000 m/s

1.3 TWT (s)

A

B B

1.3

TWT (s)

SW

1.4

1.4

1.5

NE

SW 750 m

50 m @ 2000 m/s

NE-dipping reflections

500 m headwall

gently-dipping reflections

B’ SW

1.3

headwall

NE

strata with same dip as those upslope (NE) of headwall

TWT (s)

A’

1.3

1.4 lens nA pping eflectio o t r ve R

nA

TWT (s)

inset

11

ectio

Refl

o

Dep. 2

1.4

1.5

Fig. 10. SW–NE trending sections extracted from the volume, highlighting relationships between the headwall and the strata down-slope of this region. The inset map in (A) shows the locations of the sections given in (A) and (B). The morphological slab is delineated by the dotted white outline southwest of the headwall. A and B) NE-trending seismic sections through the headwall. A′ and B′) Interpreted line drawings of (A) and (B), respectively. Reflection annotated green is “Reflection A”. Broken black lines are strata beneath Reflection A and up-dip of the headwall. Solid black lines are strata above Reflection A. Yellow lines are faults. The red body (in A′) is the upper part of disaggregated volcanic material in Deposit 2. The blue line is the headwall. Above Reflection A, a package of reflections with the same dip as reflections upslope of the headwall is shaded green (in B′). Other reflections elsewhere above Reflection A that dip differently from those in the shaded green polygon are also annotated (A′ and B′).

The distribution of the blocks in a fan covering the foot of the Montserrat slope below a chute-like indentation of the volcanic edifice suggests that all the blocks were sourced from the volcanic edifice and were emplaced during collapse events that led to debris avalanche style mass transport (Le Friant et al., 2004; Lebas et al., 2011). The significantly smaller blocks in Deposit 1 compared to the large block (Block 2) in Deposit 2 suggest that the degree of fracturing of the failed volcanic material was probably more extensive, resulting in more extensive fragmentation during the edifice collapse that led to Deposit 1. More energetic or more frequent inter-clast collision during emplacement may have also been important in the disintegration of blocks. The near-random distribution of blocks in Deposit 1 and lack of longitudinal ridges (parallel to the emplacement direction) point towards a freely spreading avalanche where flow velocities in the emplacement direction (mainly east-directed) were not significantly higher than flow perpendicular “spreading” velocities (Dufresne and Davies, 2009). The ridges identified in the high-resolution 3D seismic data-derived bathymetry on the surface of Deposit 1 (Fig. 6) are perpendicular to the emplacement direction and record compression due to deceleration. Such convex down-slope compressional ridges have been reproduced in analog modeling of mass failure emplacement in the regions where

the failure decelerates on a more gently dipping surface (Major, 1997; Moriwaki et al., 2004), and have also been observed in a range of different mass transport deposits in nature (Hampton et al., 1996; Siebert, 2002; Dufresne and Davies, 2009). We suggest that lateral discontinuity observed along the ridges is the manifestation of spatially-variable emplacement velocities across the failing mass. Recent analogue modeling of debris avalanches has shown that strike-slip faults develop within the failing mass in order to accommodate strain generated by spatiallyvariable emplacement velocity (Crosta et al., 2012; Longchamp et al., 2012). This phenomenon has also been observed in rock avalanches on land, for example in the Sherman glacier rock avalanche in Alaska (Shreve, 1966) and the Flims sturzstrom in the Swiss Alps (Pollet and Schneider, 2004). It is likely that the separation between individual ridge segments records such lateral variations in deceleration. 5.3. The origin of the Intermediate Unit The similarities in the seismic appearances of the Intermediate Unit and the recent pyroclastic fan deposits (Fig. 2C, D) suggest that the Intermediate Unit was also likely deposited from particle-laden mass flows derived from pyroclastic flows (Watt et al., 2012a). The

12

G.J. Crutchley et al. / Marine Geology 335 (2013) 1–15

absence of large blocks suggests that this unit was not emplaced as a debris avalanche. The individual lens body mapped out in Fig. 2D has an elongated lobe shape in map view. Such elongated lobe shapes are typical of individual pyroclastic deposits (derived from dome collapse events) that have been mapped on the seafloor in this area with repeated bathymetric surveys during the current eruption (Trofimovs et al., 2006; Le Friant et al., 2009). This gives further support to the interpretation that the parts of the Intermediate Unit were deposited in a similar way to the recent (since 1995) phase of deposition. However, we do not expect that particle-laden mass flows constitute the entire sedimentary succession of the Intermediate Unit. Packages of relatively undisturbed reflections (e.g. Fig. 2B) indicate that processes other than mass transport, such as turbidite and hemipelagic sedimentation, have contributed significantly to the infill. Assuming that the Intermediate Unit has origins similar to the recent pyroclastic deposits, we expect that it is composed of multiple beds, some being more voluminous and probably containing pyroclastic blocks in a coarse-grained matrix, whereas others are likely finer grained turbidite deposits, ash layers, and hemipelagic sediments (e.g. Trofimovs et al., 2006). Short cores taken from the offshore continuation of recent pyroclastic flows indicate that a dense, mass flow fraction is limited to the proximal extents of offshore deposition, while the finer grain sizes are transported much further by turbidity currents on the more gently-dipping (b2°) slope gradients (Trofimovs et al., 2006, 2012). A notable difference between the Intermediate Unit and the recent pyroclastic fan deposits is that the Intermediate Unit extends approximately 4 km farther eastwards (Fig. 2D). If the Intermediate Unit was emplaced by the submarine continuation of pyroclastic flows, these flows were most likely more powerful than those generated in the 1995–recent dome collapses. This is somewhat surprising given that the 2003 dome collapse is one of the largest volume dome collapses recorded in historical times, with a volume in excess of 0.2 km 3 (Trofimovs et al., 2008). This longer run-out suggests that higher energy pyroclastic flows have probably been generated in the past. Additionally, the stratigraphic thickness of the Intermediate Unit is significantly greater than that of the modern pyroclastic fan (Fig. 2D), indicating that the duration of this older deposition phase was probably more prolonged than the modern phase of pyroclastic flow activity. 5.4. Destabilization of seafloor slopes by volcanic flank collapse (Deposit 2) It is important to understand how an initial volcanic flank collapse might trigger secondary sediment failure, which could increase greatly the landslide volume, and therefore have important implications for tsunami hazard prediction. Previous observations from 2D seismic data, including the existence of coherent reflections and smooth marginal upper surfaces, led Watt et al. (2012a) to the interpretation that a large proportion of Deposit 2 consists of seafloor sediments that were incorporated into the failing mass. The prominent scarp on the seafloor in the eastern part of the survey area (Figs. 1B, 4A) was interpreted to mark the northeastern headwall of Deposit 2, and the morphological slab directly down-slope of the headwall was thought to be a slab of well-bedded pre-existing seafloor sediment that detached from the headwall and moved a short distance down slope (Figs. 1B, 4A). The 3D imaging in the region of the seafloor headwall scarp allows us to develop an improved model of the evolution of Deposit 2 with respect to seafloor sediment incorporation. 5.4.1. The headwall and translational sliding The east-striking and north-striking sections of the headwall are clearly aligned with prominent normal faults in well-bedded sediments (Fig. 9). These faults likely acted as planes of weakness that have controlled the shape and location of the headwall, explaining the notably sharp 90° bends in the headwall. Reflection A, which merges with the sub-seafloor continuation of the headwall, is interpreted as a slide

plane on which translational sliding occurred. The strata above the slide plane that exhibit the same dip as in-situ strata up-dip of the headwall (Fig. 10B′) are interpreted as a displaced slab, which did not slide far (~500 m) from the headwall. The slab is draped by postfailure sediments that lap onto the headwall (Fig. 10B′, NE-dipping reflections). Southeast of the slab, all strata above the slide plane dip more gently and also lap onto the sub-seafloor projection of the headwall (Fig 10A′). These strata are also interpreted as post-sliding drape. There is no evidence in this area for any other displaced slabs of sediments that were once conformable with the strata up-slope of the headwall, indicating that additional seafloor sediment that failed from this area was more extensively disaggregated and mobilized down-slope, so that it is no longer preserved on top of the slide plane. 5.4.2. Preconditioning of the translational slide to failure Factors that pre-conditioned the slope to translational sliding are not clear, but spatial relationships between seismically-incoherent, lowamplitude seismic facies and the well-bedded sediments (e.g. Figs. 7 and 8) may give insight. We interpret the incoherent, low-amplitude material in Deposit 2 as extensively disaggregated volcanic material, but avoid the term debris avalanche when referring to this material as we do not have evidence for a style of failure similar to that of Deposit 1. It is likely that much of this material west of the headwall was ultimately derived from a volcanic flank collapse event (Le Friant et al., 2004; Lebas et al., 2011), but there is also evidence in Deposit 2, especially further south of the headwall, for large-scale failure and incorporation of seafloor sediment into the failing mass (Watt et al., 2012a). The curved eastern boundary of the disaggregated volcanic material (Fig. 8B) indicates a pronounced change in emplacement direction, from east-directed to south-directed. We interpret that a volcanic flank collapse event, initially emplaced from west to east (Le Friant et al., 2004), collided with locally-elevated topography and as a result was deflected to the south over lower-lying seafloor topography. Extensive erosion into seafloor sediments in the region where the disaggregated volcanic material bends around to the south is indicated by truncated reflections (Figs. 7, 8A). We propose three possible hypotheses for how the erosion surface formed. First, as the volcanic flank collapse event curved around to the south it scoured into the seafloor sediments, resulting in significant marginal erosion and entrainment of seafloor sediments into the failing mass. That is, the erosion was contemporaneous with flank collapse emplacement. Second, the erosion may have occurred prior to the flank collapse as a result of focused flow of ocean bottom currents. Third, also preceding the flank collapse, the erosion surface may represent the upslope termination of an older, buried sedimentary slope failure. We suggest it is very unlikely that such pronounced (up to 60 m) erosion could have been caused by ocean-bottom currents, and there is also no evidence in regional 2D seismic lines down-slope of this locality for an older sedimentary failure event (Watt et al., 2012a). Therefore, we prefer the first hypothesis; that erosion was caused by scouring into marginal sediments during emplacement of a volcanic flank collapse event. In any case, the erosion of sedimentary strata at the toe of the modern day headwall likely left well-stratified sediments up-slope more prone to sliding (e.g. Mienert et al., 2002). Therefore, there is likely to be a link between the buried erosion surface and translational sliding away from the scarp on the modern seafloor. A series of pockmarks directly south of the morphological slab (Fig. 1C) indicates focused fluid flow in this area, a process that may also have led to destabilization of the slide plane through the generation of high pore fluid pressure. A 3D construction of stratigraphic relationships in the region around the headwall is given in Fig. 11A. Our interpretation of these observations is that the disaggregated volcanic flank collapse material (in red) was emplaced from west to east, encountered elevated topography, and bent around to the south. This process likely caused marginal erosion into seafloor sediments (labeled grey region), leaving strata up-slope prone to sliding. A translational slide occurred directly

G.J. Crutchley et al. / Marine Geology 335 (2013) 1–15

599000

1.0

600000

m East

601000

13

602000

603000

A

1851000

1.2 TWT (s)

present day seafloor

1850000

1.4

N

m North

headwall lab +

ced s

displa

Block 2

drape

1849000

slide plane

he

ad

nts

Dep. 2 (upper surface) stra

Dep. 2 tion 1

erosion

edime tified s

extent of over-topping lens

wa

ll

on slide plane

Sec

b a

over-topping lens Dep. 2 stratified sediments

Section 2

volcanic flank collapse

B B

bending during emplacement

Deposit 2: Schematic emplacement model (view looking NW)

headwall

pronounced erosion

W

translational sliding

N

Fig. 11. A) 3D construction of stratigraphic relationships around the headwall and where the disaggregated volcanic material in Deposit 2 bends around to the south. Two interpreted vertical sections (Section 1, striking NE and Section 2, striking E) show slices through the key horizons. In these sections shaded red regions delimit the extents of disaggregated volcanic material in Deposit 2 and horizontal black lines are well-stratified sediments. The over-topping lens at the top of Deposit 2 in Section 2 lies directly on top of the slide plane (the green surface). The spatial extent of the over-topping lens on the slide plane is shown by the broken red line. Other surfaces are: dark blue — present day seafloor; pale blue — headwall scarp; red — upper surface of disaggregated material in Deposit 2; gray (immediately left of the slide plane) — base and eastern margin of disaggregated material in Deposit 2. Cylindrical arrows show the interpreted emplacement direction of the volcanic flank collapse (i.e. originally ~ east-directed, bending around to the south). Erosion at the eastern margin of the disaggregated volcanic material is annotated, as are black arrows that show the interpreted direction of translational sliding on the slide plane. B) Schematic representation of our interpreted model for the relationship between volcanic flank collapse emplacement and seafloor destabilization. The volcanic flank collapse material bends around to the south upon impact with elevated topography. Unstable seafloor sediment, directly up-slope of where the volcanic landslide bent around, fails by translational sliding and is incorporated into the failing mass. The approximate extent of the field of view in (A) is shown by the broken red polygon.

up-slope, probably concurrent with emplacement of the volcanic flank collapse. A remnant of the translational slide that was displaced just a short distance from the headwall is preserved as an intact slab, but the rest of the slide plane within the survey area is draped by post-sliding sediment with no evidence for intact slabs. In line with interpretations of Watt et al. (2012a), we suggest that (with the exception of the remnant slab) the failed sediments from around the headwall became disaggregated and were mobilized down-slope, to become incorporated into the volcanic flank collapse (represented schematically in Fig. 11B). These sediments formed a significant part of the upper part of Deposit 2, which is imaged in the SE part of our 3D volume as an over-topping lens covering part of the exposed slide plane. The model we propose for volcanic flank collapse (the primary failure) destabilizing gently-dipping sedimentary slopes to translational sliding (the secondary failure) depends on pronounced erosion of the primary failure into the substrate onto which it is emplaced. Le Friant et al. (2004) first noted the bending during emplacement in Deposit 2 from the shape of the deposit, and Watt et al. (2012a) then showed and discussed the headwall scarp and the secondary failure of seafloor sediment into the flank collapse body of Deposit 2 from this region. Here, by imaging this region in 3D, we could reconstruct

the relationships between the primary and secondary failures and, in particular, the large-scale erosion that has occurred at the margin of the flank collapse. We conclude that the rheology of the flank collapse body in this region of emplacement was relatively dense, enabling it to hug the topography and erode efficiently into marginal sediments. Deflection of mass transport deposits during emplacement has been documented in various places around the world in both the terrestrial (e.g. Kelfoun et al., 2008; Dufresne and Davies, 2009) and the marine environment (e.g. Gee et al., 2001, 2007), but to our knowledge we present the first observations of such bending being associated with large-scale marginal erosion directly down dip of a secondary seafloor slope failure. We postulate that certain conditions are required if this interaction is to result in secondary failure. First, a primary mass failure (a flank collapse in our example) must be topographically confined such that it interacts with slopes that dip toward the failing mass. Second, the substrate must be sufficiently unconsolidated that it can be eroded by the impinging primary failure, which itself must be sufficiently dense and energetic to erode effectively. Third, potential secondary failure planes, likely to be aligned with lithological boundaries in shallow seafloor strata, must dip in the down-slope direction towards the primary failure. These processes may be applicable to mass

14

G.J. Crutchley et al. / Marine Geology 335 (2013) 1–15

transport deposits on other island chain volcanoes, but as yet there are no 3D seismic data to corroborate this.

6. Conclusions The first 3D seismic dataset acquired over volcanic island landslides gives new insight into emplacement processes that cannot be constrained from 2D seismic data. 3D imaging enabled us to quantitatively analyze block sizes on the seafloor and recognize subtle ridge patterns hidden amid blocks. Erosion patterns, deposit over-topping and deposit bending were reconstructed in 3D, yielding a much improved understanding of interactions between volcanic flank collapse deposits and the seafloor topography onto which they were emplaced. The dataset, located offshore Montserrat, images two volcanic landslides (Deposit 1 and Deposit 2) as well as an Intermediate Unit between these two deposits. Deposit 1 contains relatively poorly-developed internal reflections, in comparison to internal reflections in Deposit 2, which suggest Deposit 1 was emplaced as a single event but with more than one pulse of debris avalanching. Spatially-disrupted wave fabrics on the surface of Deposit 1 likely represent variable compression rates at the distal extent of deposition. Our data confirm that most blocks on the seafloor are associated with Deposit 1. Their distribution, devoid of longitudinal block trains, indicates that the debris avalanche spread out rapidly, likely with emplacement-perpendicular “spreading” velocities being not significantly lower than emplacement-parallel velocities. It was previously proposed that the Intermediate Unit with strong, continuous reflections between Deposits 1 and 2 was deposited by the submarine continuation of pyroclastic flows. Our data confirm that this Intermediate Unit has a seismic character resembling that of the 1995–recent submarine pyroclastic deposits. The rapid distal thinning of the Intermediate Unit is consistent with deposition from submarine pyroclastic density currents. However, the run out of the Intermediate Unit is significantly greater than that of the recent submarine pyroclastic deposits that were generated from dome collapse. This indicates that the Intermediate Unit was most likely emplaced by relatively vigorous pyroclastic flows, perhaps generated by open vent explosions rather than dome collapse. The thickness of the Intermediate Unit is also greater than that of the recent pyroclastic deposits, suggesting a more prolonged period of pyroclastic flow generation. Deeper seismic imaging has enabled us to map out stratigraphic relationships in an area where a volcanic flank collapse (part of Deposit 2), upon collision with elevated seafloor topography, changed its emplacement direction from east-directed to more south-directed. Pronounced erosion of well-stratified sediments at the outer margin of the flank collapse deposit, and a translational slide directly up-slope, point to a likely correlation between the two events. The up-dip extent of sliding was controlled by deep seated faults that influenced the shape and location of the headwall. A slab of failed material that did not move far from the headwall is preserved above part of the slide plane, but elsewhere above the slide plane there is no evidence for the material that failed from the headwall. We predict that translational sliding was concurrent with emplacement of the volcanic flank collapse, and that much of the remobilized seafloor was incorporated into the failing body of the volcanic flank collapse material. Our model for secondary failure of the seafloor is that extensive erosion by the impingent volcanic landslide destabilized the slope, initiating the translational sliding. Our results advance the understanding of submarine landslide emplacement offshore Montserrat and show that the outlook for application of high-resolution 3D seismic acquisition for understanding volcanic landslide emplacement is very promising. The detail at which stratigraphic relationships can be imaged, and the spatial control delivered by the data to test theories of emplacement mechanisms, make the 3D method a powerful tool in this respect.

Acknowledgements We are very grateful for the constructive and helpful comments made by Sharon Allen, two anonymous reviewers and David Piper that greatly improved the manuscript. We thank the captain and crew of the RRS James Cook and the technicians for their assistance with data acquisition. Appendix A. Supplementary data Supplementary data to this article can be found online at http:// dx.doi.org/10.1016/j.margeo.2012.10.004. References Boudon, G., Friant, A.L., Komorowski, J.-C., Deplus, C., Semet, M.P., 2007. Volcano flank instability in the Lesser Antilles Arc: diversity of scale, processes, and temporal recurrence. Journal of Geophysical Research 112 (B08205). http://dx.doi.org/10.1029/ 2006JB004674. Chopra, S., Marfurt, K.J., 2007. Seismic Attributes for Prospect Identification and Reservoir Characterization. Society of Exploration Geophysicists. (464 pp.). Coleman, J.M., Prior, D.B., 1988. Mass wasting on continental margins. Annual Review of Earth and Planetary Science 16, 101–119. Crosta, G., Blasio, F.D., Caro, M.D., Volpi, G., Frattini, P., 2012. Granular flows on erodible layers: type and evolution of flow and deposit structures. Geophysical Research Abstracts, EGU General Assembly 2012, 14. Deplus, C., Le Friant, A., Boudon, G., Komorowski, J.-C., Villemant, B., Harford, C., Ségoufin, J., Cheminée, J.-L., 2001. Submarine evidence for large-scale debris avalanches in the Lesser Antilles Arc. Earth and Planetary Science Letters 192, 145–157. Druitt, T.H., Kokelaar, B.P., 2002. The eruption of Soufrière Hills volcano, Montserrat, from 1995 to 1999. Geological Society, London, Special Publications, Memoir, 21 (645 pp.). Dufresne, A., Davies, T.R., 2009. Longitudinal ridges in mass movement deposits. Geomorphology 105, 171–181. Gee, M.J.R., Masson, D.G., Watts, A.B., Mitchell, N.C., 2001. Passage of debris flows and turbidity currents through a topographic constriction: seafloor erosion and deflection of flow pathways. Sedimentology 48, 1389–1409. Gee, M.J.R., Uy, H.S., Warren, J., Morley, C.K., Lambiase, J.J., 2007. The Brunei slide: a giant submarine landslide on the North West Borneo Margin revealed by 3D seismic data. Marine Geology 246, 9–23. Hampton, M.A., Lee, H.J., Locat, J., 1996. Submarine landslides. Reviews of Geophysics 34 (1), 33–59. Harbitz, C.B., Løvholt, F., Pedersen, G., Masson, D.G., 2006. Mechanisms of tsunami generation by submarine landslides: a short review. Norwegian Journal of Geology 86, 255–264. Harford, C.L., Pringle, M.S., Sparks, R.S.J., Young, S.R., 2002. The volcanic evolution of Montserrat using 40Ar/39Ar geochronology. In: Druitt, T.H., Kokelaar, B.P. (Eds.), The Eruption of Soufrière Hills Volcano, Montserrat, From 1995 to 1999: Memoirs of the Geological Society of London, pp. 93–113. Kelfoun, K., Druitt, T., van Wyk de Vries, B., Guilbaud, M.-N., 2008. Topographic reflection of the Socompa debris avalanche, Chile. Bulletin of Volcanology 70, 1169–1187. Krastel, S., Schmincke, H.U., Jacobs, C.L., Rihm, R., Le Bas, T.M., Alibés, B., 2001. Submarine landslides around the Canary Islands. Journal of Geophysical Research 106, 3977–3997. Le Friant, A., Harford, C.L., Deplus, C., Boudon, G., Sparks, R.S.J., Herd, R.A., Komorowski, J.C., 2004. Geomorphological evolution of Montserrat (West Indies): importance of flank collapse and erosional processes. Journal of the Geological Society 161, 147–160. http://dx.doi.org/10.1144/0016-764903-017. Le Friant, A., Lock, E.J., Hart, M.B., Boudon, G., Sparks, R.S.J., Leng, M.J., Smart, C.W., Komorowski, J.C., Deplus, C., Fisher, J.K., 2008. Late Pleistocene tephrochronology of marine sediments adjacent to Montserrat, Lesser Antilles volcanic arc. Journal of the Geological Society of London 165, 279–289. Le Friant, A., Deplus, C., Boudon, G., Sparks, R.S.J., Trofimovs, J., Talling, P., 2009. Submarine deposition of volcaniclastic material from the 1995–2005 eruptions of Soufrière Hills volcano, Montserrat. Journal of the Geological Society 166, 171–182. http:// dx.doi.org/10.1144/0016-76492008-047. Le Friant, A., Deplus, C., Boudon, G., Feuillet, N., Trofimovs, J., Komorowski, J.-C., Sparks, R.S.J., Talling, P., Palmer, M., Ryan, G., 2010. Eruption of Soufrière Hills (1995–2009) from an offshore perspective: insights from repeated swath bathymetry surveys. Geophysical Research Letters 37 (L11307). http://dx.doi.org/10.1029/2010GL043580 (6 pp.). Lebas, E., Le Friant, A., Boudon, G., Watt, S., Talling, S., Feuillet, N., Deplus, C., Berndt, C., Vardy, M., 2011. Multiple widespread landslides during the long-term evolution of a volcanic island: insights from high-resolution seismic data, Montserrat, Lesser Antilles. Geochemistry, Geophysics, Geosystems 12 (5), Q05006. http://dx.doi.org/ 10.1029/2010GC003451. Locat, J., Lee, H.J., 2002. Submarine landslides: advances and challenges. Canadian Geotechnical Journal 39, 193–212. http://dx.doi.org/10.1139/T01-089. Longchamp, C., Charrière, M., Jaboyedoff, M., 2012. Analogue modelling of rock avalanches and structural analysis of the deposits. Geophysical Research Abstracts, EGU General Assembly 2012, 14. Løvholt, F., Harbitz, C.B., Haugen, K.B., 2005. A parametric study of tsunamis generated by submarine slides in the Ormen Lange/Storegga area off western Norway.

G.J. Crutchley et al. / Marine Geology 335 (2013) 1–15 Marine and Petroleum Geology 22, 219–231. http://dx.doi.org/10.1016/j.marpetgeo. 2004.10.017. Løvholt, F., Pedersen, G., Gisler, G., 2008. Oceanic propagation of a potential tsunami from the La Palma Island. Journal of Geophysical Research 113, C09026. http:// dx.doi.org/10.1029/2007JC004603. Macdonald, R., Hawkesworth, C.J., Heath, E., 2000. The LesserAntillesvolcanicchain: a study in arc magmatism. Earth-Science Reviews 49 (1–4), 1–76. Major, J.J., 1997. Depositional processes in large-scale debris-flow experiments. Journal of Geology 105, 345–366. Masson, D.G., 1996. Catastrophic collapse of the volcanic island of Hierro 15 ka ago and the history of landslides in the Canary Islands. Geology 24, 231–234. Masson, D.G., Harbitz, C.B., Wynn, R.B., Pedersen, G., Løvholt, F., 2006. Submarine landslides: processes, triggers and hazard prediction. Philosophical Transactions of the Royal Society A 364, 2009–2039. McLeod, P., Carey, S., Sparks, R.S.J., 1999. Behaviour of particle-laden flows into the ocean: experimental simulation and geological implications. Sedimentology 46, 523–536. Mienert, J., Berndt, C., Laberg, J.S., Vorren, T.O., 2002. Slope instability of continental margins. In: Wefer, G., et al. (Ed.), Ocean Margin Systems. Springer-Verlag, Berlin Heidelberg, pp. 179–193. Moore, J.G., Clague, D.A., Holcomp, R.T., Lipman, P.W., Normark, W.R., Torresan, M.E., 1989. Prodigious submarine landslides on the Hawaiian Ridge. Journal of Geophysical Research 94 (17465–17484). Moriwaki, H., Inokuchi, T., Hattanji, T., Sassa, K., Ochiai, H., Wang, G., 2004. Failure processes in a full-scale landslide experiment using a rainfall simulator. Landslides 1, 277–288. http://dx.doi.org/10.1007/s10346-004-0034-0. Mulder, T., Cochonat, P., 1996. Classification of offshore mass movements. Journal of Sedimentary Research 66, 43–57. Nardin, T.R., Hein, F.J., Gorsline, D.S., Edwards, B.D., 1979. A review of mass movement processes, sediment and acoustic characteristics, and contrasts in slope and baseof-slope systems versus canyon-fan-basin floor systems. SEPM Special Publications 27, 61–73. Oehler, J.F., Labazuy, P., Lenat, J.F., 2004. Recurrence of major flank landslides during the last 2 Ma-history of Reunion Island. Bulletin of Volcanology 66, 585–598. Pollet, N., Schneider, J.-L.M., 2004. Dynamic disintegration processes accompanying transport of Holocene Flims sturzstrom (Swiss Alps). Earth and Planetary Science Letters 221, 433–448. Shreve, R.L., 1966. Sherman landslide, Alaska. Science 154, 1639–1643.

15

Siebert, L., 2002. Landslides resulting from structural failure of volcanoes. Reviews in Engineering Geology 15, 209–235. Tripsanas, E.K., Piper, D.J.W., Jenner, K.A., Bryant, W.R., 2008. Submarine masstransport facies: new perspectives on flow processes from cores on the eastern North American margin. Sedimentology 55, 99–136. http://dx.doi.org/10.1111/ j.1365-3091.2007.00894.x. Trofimovs, J., Amy, L., Boudon, G., Deplus, C., Doyle, E., Fournier, N., Hart, M.B., Komorowski, J.C., Le Friant, A., Lock, E.J., Pudsey, C., Ryan, G., Sparks, R.S.J., Talling, P.J., 2006. Submarine pyroclastic deposits formed at the Soufrière Hills volcano, Montserrat (1995–2003): what happens when pyroclastic flows enter the ocean? Geology 34 (7), 549–552. http://dx.doi.org/10.1130/G22424.1. Trofimovs, J., Sparks, R.S.J., Talling, P.J., 2008. Anatomy of a submarine pyroclastic flow and associated turbidity current: July 2003 dome collapse event, Soufriere Hills volcano, Montserrat, West Indies. Sedimentology 55, 617–634. Trofimovs, J., Foster, C., Sparks, R.S.J., Loughlin, S., Friant, A.L., Deplus, C., Porritt, L., Christopher, T., Luckett, R., Talling, P.J., Palmer, M.R., Bas, T.L., 2012. Submarine pyroclastic deposits formed during the 20th May 2006 dome collapse of the Soufrière Hills Volcano, Montserrat. Bulletin of Volcanology 74, 391–405. Urgeles, R., Canals, M., Baraza, J., Alonso, B., Masson, D.G., 1997. The most recent megaslides on the Canary islands: the El Golfo Debris avalanche and the Canary Debris flow. Journal of Geophysical Research 102, 20 305–20 323. Wadge, G., Herd, R., Ryan, G., Calder, E.S., Komorowski, J.C., 2010. Lava production at Soufrière Hills Volcano, Montserrat: 1995–2009. Geophysical Research Letters 37 (L00E03). Watt, S.F.L., Talling, P.J., Vardy, M.E., Heller, V., Hühnerbach, V., Urlaub, M., Sarkar, S., Masson, D.G., Henstock, T.J., Minshull, T.A., Paulatto, M., Le Friant, A., Lebas, E., Berndt, C., Crutchley, G.J., Karstens, J., Stinton, A.J., Maeno, F., 2012a. Combinations of volcanic-flank and seafloor-sediment failure offshore Montserrat, and their implications for tsunami generation. Earth and Planetary Science Letters 319–320, 228–240. Watt, S.F.L., Talling, P.J., Vardy, M.E., Masson, D.G., Henstock, T.J., Hühnerbach, V., Minshull, T.A., Urlaub, M., Lebas, E., Le Friant, A., Berndt, C., Crutchley, G.J., Karstens, J., 2012b. Widespread and progressive seafloor-sediment failure following volcanic debris avalanche emplacement: landslide dynamics and timing offshore Montserrat, Lesser Antilles. Marine Geology 323–325, 69–94. Watts, A.B., Masson, D.G., 1995. A giant landslide on the north flank of Tenerife, Canary islands. Journal of Geophysical Research 100, 24487–24498.