Repeated slope failure linked to fluid migration: The Ana submarine landslide complex, Eivissa Channel, Western Mediterranean Sea

Earth and Planetary Science Letters 319–320 (2012) 65–74

Contents lists available at SciVerse ScienceDirect

Earth and Planetary Science Letters journal homepage: www.elsevier.com/locate/epsl

Repeated slope failure linked to fluid migration: The Ana submarine landslide complex, Eivissa Channel, Western Mediterranean Sea Christian Berndt a, b,⁎, Sergio Costa c, Miquel Canals c, Angelo Camerlenghi c, Ben de Mol d, Martin Saunders b a

Leibniz-Institute for Marine Sciences (IFM-GEOMAR), Kiel, Germany National Oceanography Centre, Southampton, UK GRC Geociències Marines, Departament d'Estratigrafia, Paleontologia i Geociències Marines, Universitat de Barcelona, Spain d Parc Científic de Barcelona, Barcelona, Catalonia, Spain b c

a r t i c l e

i n f o

Article history: Received 11 July 2011 Received in revised form 28 November 2011 Accepted 30 November 2011 Available online 21 January 2012 Editor: P. Shearer Keywords: slope stability fluid migration submarine landslide 3D seismic imaging Eivissa Channel Western Mediterranean Sea

a b s t r a c t Submarine slope failures are a well-known geohazard. They are able to destroy seafloor installations along their path and by generating tsunamis they may threaten coastal infrastructures. While the mechanisms involved in submarine landslide generation remain poorly known, there are observations that slope stability can be reduced in the presence of free gas. Here, we present new high-resolution 3D seismic data from the Eivissa Channel between the Iberian Penninsula and the Balearic Promontory in the Western Mediterranean Sea. The data reveal slope stability reduction in this area at least since mid-Quaternary times, and an intimate relationship between fluid migration and slope stability. We show that two landslides, i.e. pre-Ana Slide and Ana Slide, occurred at almost the same location above an erosional channel in the Messinian unconformity. There is seismic evidence that fluids including gas are migrating upwards through this erosional surface and that they charge sedimentary layers at the base of the Ana Slide possibly reducing its strength and predisposing it to failure. Our data show in unprecedented detail the ways in which the presence of gas influences slope stability. The findings illustrate the importance of including high-resolution 3D seismic data in slope stability and tsunami risk assessments to identify shallow gas distribution as one of the main controls on slope stability in gas prone areas. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Submarine landsliding is a widespread geological phenomenon that shapes submarine slopes on continental margins, ocean islands and seamounts. Some of these landslides involve several thousand cubic kilometres of material that move in a short period of time (Bull et al., 2009; Canals et al., 2004). Slope failures occurring on sediment covered continental margins have been the subject of significant research efforts over the past decades as hydrocarbon exploration moves into deeper waters where landslides threaten submarine installations, and because studies of salt marsh, river estuary, and lake deposits surrounding the northeast Atlantic show that submarine landslides can cause significant tsunamis that endanger coastal populations (Assier-Rzadkiewicz et al., 2000; Bondevik et al., 2005; Harbitz, 1992; Lopez-Venegas et al., 2008; ten Brink et al., 2009; Watts et al., 2003). The geological processes that lead to unstable slopes are still not fully understood. The most detailed investigation of any submarine landslide, i.e. the Storegga Slide off mid-Norway, concluded that the ⁎ Corresponding author at: Leibniz-Institute for Marine Sciences (IFM-GEOMAR), Kiel, Germany. E-mail address: [email protected] (C. Berndt). 0012-821X/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.epsl.2011.11.045

particular sequence of hemipelagic deposition and glacial debris flows led to overpressures that make this glaciated continental margin inherently unstable (Berg et al., 2005; Bryn et al., 2005; Kvalstad et al., 2005a, 2005b), and that the trigger of this particular landslide must have been an earthquake of M >7 (Løvholt et al., 2005). However, the spatial relationship of a major gas reservoir and the most deeply incised part of the landslide could also suggest that upwards gas migration played a role in the destabilisation of the slope (Bünz et al., 2005). Unfortunately, the findings from the Storegga Slide area cannot be generalised for the geological situation away from the influence of glacial processes, where large submarine landslides occur on continental margins in areas of gentle seafloor gradient, e.g. the margins of NW Africa (Henrich et al., 2008; von Rad et al., 1982; Wynn et al., 2000), of the east coast off the US (Dugan and Flemings, 2000, 2002) and in the Gulf of Mexico (Flemings et al., 2008). For these landslides other destabilising mechanisms have been proposed such as overpressuring of the toe of the slopes due to lateral fluid migration from major deposition centres to the distal parts of the continental margin (Dugan and Flemings, 2002). As a result of IODP Expedition 308 to the Gulf of Mexico, a flow-focusing model was produced showing how sedimentation, overpressure, fluid flow, and deformation can be coupled in a passive margin setting and how extremely rapid deposition of fine-

66

C. Berndt et al. / Earth and Planetary Science Letters 319–320 (2012) 65–74

grained sediment might lead to a rapid build-up of pore pressure in excess of hydrostatic (overpressure), under-consolidation, and finally sedimentary mass wasting (Flemings et al., 2008). The Mediterranean Sea is a largely land-locked basin where a variety of tectonic and sedimentary environments co-exist in a relatively small area. Here submarine landslides of different sizes and ages are widespread. However, there is no obvious correlation of landslide abundance with tectonic activity or with zones of high sediment accumulation (Camerlenghi et al., 2009). This reinforces the view that it is a combination of geologically driven pre-conditioning factors that control conditions of continental slope instability in the submarine environment (Canals et al., 2004; Lee, 2009; Sultan et al., 2004). The landslides in the Eivissa Channel at the southern termination of the Valencia Trough (Fig. 1a) lend themselves to the study of the general processes of submarine slope instability as they occur in a low seismicity, rifted continental margin setting characterised by low rates of carbonate-dominated sediment accumulation influenced by bottom currents (Alonso et al., 1988; Canals and Ballesteros, 1997; Lastras et al., 2004). The structure of the submarine landslides in the Eivissa Channel is relatively simple compared to the large multi-phase landslides off Norway or NW Africa, and the run-out length of less than 10 km facilitates complete surveying. A bathymetric survey conducted in 1995 showed that there are at least four landslides originating on the eastern slope (i.e. the Balearic slope) of the channel and running out to the west, named Jersi, Nuna, Joan and Ana slides from north to south after Lastras et al. (2004) (Fig. 1b). High-resolution sub-bottom profiling yielded volumes of less than 0.4 km 3 for each of these landslides (Lastras et al., 2004) and MAK1 side scan sonar data showed translational ridges in the central part of the landslides and compressional ridges in the accumulation zone. There are pockmarks at water depths between 400 and 700 m,

which are similar to the water depth at the headwalls of the landslides. A sediment core reaching below the glide plane within the Ana Slide headwall area sampled warm foraminifera fauna of MIS 5 that may be considered the oldest possible age of the Ana Slide. Based on dating of overlying sediments and sequence stratigraphic considerations Cattaneo et al. (2011) proposed a possible age of 60,000 years. In 2006 we collected a cube of high-resolution 3D seismic data to investigate the processes that lead to slope failure in this area focusing on Ana Slide, the southernmost of the landslides discovered on the eastern slope of the Eivissa Channel. The first objective of this work was to reconstruct the history of slope instability in the area of the Ana Slide. The second objective was assessing the role of different geological processes such as fluid migration, structural deformation and sedimentation history in predisposing the slope to fail and to constrain possible triggers. 2. Method 2.1. P-Cable seismic data The high-resolution 3D seismic data of this survey (Cruise CD178) were acquired using the RRS Charles Darwin of the National Environment Research Council (NERC) and the P-Cable 3D acquisition system of the National Oceanographic Centre, Southampton (Perez-Garcia et al., 2011). The P-Cable consists of a cross wire (P-wire) extending perpendicularly to the ship's steaming direction. This wire is held in place by paravanes attached to both ends of the wire and towed by the ship. In the normal configuration this system consists of twelve 30 m-long, single-channel Teledyne Instruments analogue streamers, which are connected to the P-wire and towed in line with the ship's movement.

Fig. 1. a: Geological compilation for the western Mediterranean showing the main tectonic structures (adapted from Camerlenghi et al., 2009). Main geological structures of the Balearic Promontory and surrounding areas (Acosta et al., 2002, 2004; Camerlenghi et al., 2009; Maillard and Mauffret, 1999; Mauffret et al., 2004; Roca and Desegaulx, 1992). b: Relative position of the surface-near submarine landslides in the Eivissa Channel (after Lastras et al., 2004). c: New bathymetry data collected during the CD178 cruise showing the detailed bathymetric expression of the Ana Slide. b + c: illumination with the dip map.

C. Berndt et al. / Earth and Planetary Science Letters 319–320 (2012) 65–74

During cruise CD178, only eleven streamers were used because one streamer was malfunctioning. The P-wire and the streamers were towed at a depth of 1 m. The source consisted of four 40 in 3 Bolt 600B airguns spaced 0.75 m apart and towed at a depth of 1.5 m about 20 m behind the stern of the vessel. The seismic cube in the Eivissa Channel covers an area of 10 km 2 over and around the Ana Slide (Fig. 1c). The data were frequency filtered from 45 to 220 Hz and binned at 10 m bin interval before a Stolt time migration with a migration velocity of 1500 m/s was carried out. The resolution of the data is approximately 5–6 m vertically and for the 10 m inline and crossline spacing the horizontal resolution is 10–15 m. Interpretation of the data was carried out in KingdomSuite. In order to map the gas cloud and the erosion channel we calculated the similarity attribute in Opendtect and re-imported the attribute data into KingdomSuite. 2.2. Multibeam bathymetry In addition to seismic data, we present high-resolution multibeam bathymetry data that were collected onboard the Spanish R/V Hespérides by the University of Barcelona during three surveys of the Eivissa Channel in 1995 and 2002 using the SIMRAD EM-12 (13 kHz, 81 beams) and in 2007 using the EM-120 (13 kHz, 191 beams) swath bathymetry systems. Further multibeam data were also acquired during the 3D seismic survey. All these bathymetry data were merged and processed with Caris HIPS/SIPS software in order to obtain high quality images of individual landslides. Positioning of the vessel was ensured by GPS (1995) and DGPS (2002, 2006 and 2007). The resulting data are displayed in Fig. 1c. 3. Results 3.1. Seismic constraints on the geological setting The 3D seismic data have a penetration of approximately one second two-way-travel time (TWTT) below the sea floor corresponding to some 1–1.5 km. In areas with most penetration, i.e. in the western part of the study area, km-scale rotated fault blocks are visible at approximately 1.6 s TWT (Fig. 2). There are no clear fault-related amplitude anomalies and for the three eastern faults the interpretation is entirely based on the rotation of the reflections and the fact that there are areas with completely chaotic reflections as opposed to areas with more consistent sub-parallel reflections. This seismic pattern is typical for seismic imaging of deeply buried half grabens and most likely this fault pattern represents the Mesozoic multiphase rifting of the basin (Fernandez et al., 1995). The tops of the fault blocks are truncated by an erosional unconformity (Unconformity 4). Unconformity 4 is the deepest and oldest unconformity that can be distinguished in the 3D seismic data. It separates a transparent unit above from an east-dipping seismic reflection facies below. The unconformity is quite smooth, but in many places poorly imaged because of its great burial depth. The transparency of the unit between Unconformity 4 and the overlying Unconformity 3 is not the effect of automatic gain control as it is also visible in seismic displays of the data for which this process was turned off. Except for the general change in seismic character, Unconformity 4 is clearly defined by the toplap of reflections in the western part of the cube. Paleogene compressional tectonic events produced regional uplift of the Balearic Promontory with prolonged subaerial erosion of a large portion of the Mesozoic limestones and non-deposition of sediments throughout the Paleogene (Fernandez et al., 1995; Fontbote et al., 1990). The Neogene–Mesozoic unconformity found in the Ibiza Amposta Well, on the northern side of the Ibiza Channel, separates eroded Jurassic massive shallow-water carbonates from overlying Mid-Miocene marls (Hispanoil-Eniepsa, 1978). The new seismic data show the typical transparent seismic character of the transgressive

67

Miocene marls as a reflection-free seismic unit overlying Unconformity 4. Thus we interpret Unconformity 4 as the Neogene–Mesozoic unconformity. An 150 to 200 ms TWTT-thick seismic unit overlies Unconformity 4. It is mostly reflection-free except at the top where particularly strong reflections separate two erosional unconformities (Unconformity 2 and 3), which are 10–80 ms TWTT apart. Unconformity 3 incises deeply into the underlying reflection-free Miocene unit. It partly consists of a NE-SW striking sinuous channel in the western part and a 2 km-wide depression east of the channel, tracing the tectonic graben underlying Unconformity 4. On-lapping reflections of upward-increasing amplitude show that the depression was in-filled with sediments. Further eastwards (up-slope), the surface of Unconformity 3 has a rough morphology with many short high-amplitude reflection segments in an irregular pattern. Unconformity 2 is also erosional but smoother than Unconformity 3. It shoals from 1500 ms TWTT in the western part of the study area to 1200 ms TWTT in the east of the 3D seismic block (Fig. 2e). It is characterised by hummocky topography and the incision of a NE-SW striking sinuous erosional channel of about 200 m width in the lower western part in the same place as the channel in Unconformity 3 (Fig. 2e). As both channels incise into the unconformities without a significant change in vertical separation between the unconformities, it is likely that the channel was actively incising throughout the entire Messinian salinity crisis (see below) as small-scale deviations of the course of this erosion suggests that it is not a late feature that incised all the way down to Unconformity 3 during formation of Unconformity 2. In many parts of the Mediterranean continental margin the emergence of parts of the continental slopes during the Messinian sea level drop led to the development of a dendritic river pattern which locally incised deeply into the soft substratum (Urgeles et al., 2011). Because of its stratigraphic position just above the reflection-free seismic facies and because of the well-developed channel system we interpret Unconformities 2 and 3 as the lowermost and uppermost subaerial erosion surfaces produced during the Messinain sea level drop of about 1500 m, which caused a short-lived (about 0.5 Ma) emersion of the Eivissa Channel in the Late Miocene (Escutia and Maldonado, 1992; Maillard et al., 2006). The multiple character of the Messinian unconformity is characteristic of poly-phase erosion caused by repeated Messinian sea-level oscillations (Maillard et al., 2006; Ryan, 2009). Apparently, the Messinian emergence of the Eivissa Channel only caused the erosion of part of the Neogene sediments (reflector-free unit) but did not exhume the Mesozoic carbonate basement. The discontinuous high-reflectivity reflections included between the two unconformities most likely represent reworked, probably continental or transitional transgressive coarse lag sediments (sand-gravel) alternating with fine-grained homogenous infill sediments in the depression of Unconformity 3. On top of Unconformity 2 lie 80 to 100 ms TWTT of seismically opaque sediments that are bounded by Unconformity 1 at the top. The sequence on top of Unconformity 1 is well stratified, but offset by a number of normal faults (Fig. 2). The faults strike NE to SW and are restricted to the southern half of the 3D seismic cube (Figs. 2, and 4). These faults are clearly visible as offsets in the sedimentary reflections and show local amplitude decreases at the fault cuts. The fault throw is as much as 20 ms TWTT for the longest and highest fault (Fig. 2). Altogether we map seven normal faults above Unconformity 1. Two continue beyond the southern limit of the seismic cube and all gradually terminate towards the North. All faults that are not antithetic can be mapped down to Unconformity 1. The data do not conclusively show whether these faults continue to greater depth (e.g. Fig. 2). None of the faults reach the seafloor but two have a topographic expression at the top of the pre-Ana Slide (Fig. 4). These faults may be the result of reactivation of faults from the latest Late Oligocene–Early Miocene rift phase that accompanied the opening of the southwestern part of the Gulf of Valencia (Fernandez et al., 1995).

68

C. Berndt et al. / Earth and Planetary Science Letters 319–320 (2012) 65–74

Fig. 2. a) Inline throughout the entire cube showing the mass wasting deposits Ana Slide and pre-Ana Slide. In the interpreted section (e) black arrows indicate high amplitude anomalies and grey arrows indicate sense of shear. Location in Figs. 3 and 4. Vertical axis, TWTT in seconds; horizontal axis, offset in metres. Colour bar (refers to a and e) shows relative amplitude variation as grey scale with peaks in yellow and troughs in red. The insets b,c, and d show details of projected sections throughout the cube onto a). The locations of these data examples are shown in Fig. 3. For enhancing the amplitude range a different colour scale was used.

The stratified succession above Unconformity 1 includes two mass transport deposits just below the sea floor (within the resolution of our seismic data) and about 100 m TWTT below, which are Ana Slide (Lastras et al., 2004, 2006) and a larger, older landslide that we call “pre-Ana Slide”. The nearby DSDP Sites 124, 134, and 371 (Ryan, 1976) suggest that the low amplitude seismic unit above Unconformity 2 represents sediments of Pliocene age and a deep water nannofossil or calcareous

ooze lithology. Above Unconformity 1 the frequency and amplitude of seismic reflections increases towards the present day seafloor and it is likely that the upper reflective unit consists of calcareous ooze sediments alternating with coarse sands to fine grained turbidites (Acosta et al., 2004; Nelson and Maldonado, 1990; Ryan, 1976). Unconformity 1 has a moderately erosional character that defines broad depressions and onlap reflector terminations indicating sedimentary filling by the overlying turbidites.

C. Berndt et al. / Earth and Planetary Science Letters 319–320 (2012) 65–74

69

Fig. 3. Time structure map of the base of the Ana Slide (colour coded) and time structure map of the top of the Ana Slide (contours), i.e. the sea floor within the resolution of our seismic data. Red line: inline seismic reflection profile shown in Fig. 2. Coordinates in UTM Zone 31 m. Thin green lines and numbers indicate inlines and crosslines.

3.2. Mass wasting deposits The 3D seismic reflection data cover the entire Ana Slide and provide further information on its internal structure and size. The Ana Slide and the pre-Ana Slide moved from east to west (Lastras et al., 2004). The Ana Slide has a maximum run-out distance from headwall to the toe accumulation zone of 4.05 km. The excavation area is 1.45 km across with a total along strike length of the headwall of 2.9 km. These measurements are accurate to within the resolution of the 3D seismic data (see Section 3). Contouring the top of the landslide shows that the excavation area comprises almost the entire eastern half of the landslide, i.e. the upslope part of landslide. This can be deduced (Micallef et al., 2009) from the upslope eastward-protruding contours (Figs. 3 and 4), and it opens within a few hundred metres into the depositional zone, as shown by down slope westward-protruding contours. The internal structure of the landslide was already described by Lastras et al. (2004) based on TOPAS PS 018 very high-resolution 2D seismic lines. It shows a transition from a chaotic blocky facies in the upslope part of the landslide through a translational zone that is characterised by slightly tilted intact blocks into a depositional zone with fault bounded blocks and chaotic material at the toe of the landslide. The 3D seismic data confirm this pattern throughout the landslide.

Assuming an interval velocity of 1700 m s − 1 the volume of the landslide is approximately 0.1 km 3 +/− 10% due to uncertainty in the interval velocity. An additional uncertainty of about 5% is due to an unclear lower boundary of the landslide particularly in the toe region where, locally, it is difficult to distinguish blocky landslide material from material that has not moved at all. The volume stated above is the volume between the landslide top and base surfaces tracked through the 3D cube and limited laterally by the landslide outline that can be easily picked up in a 3D representation of the landslide surface. The deeper pre-Ana Slide (Fig. 2) is barely visible in the data of Lastras et al. (2004), (cf. their Fig. 4b). This landslide also moved from east to west and its internal structure mimics that of Ana Slide with an upslope excavation area in the east, a translation area where the landslide deposit narrows in the north–south direction, and a broader down slope depositional area in the west. About three quarters of its way down slope, the pre-Ana Slide deposit covers two normal faults (Fig. 2). East of the faults, and up to the headwall, the landslide deposit thins continuously, while west of the faults the deposit thickness is quite uniform (Fig. 2). The pre-Ana Slide has a maximum run out distance of 4.9 km. The headwall of this landslide was not completely covered by the seismic survey, but the segment that can be measured is 2.0 km long and the overall shape of the

Fig. 4. Time structure map of the top of the pre-Ana Slide. Ana Slide outline (red hatched area). Bold black lines are fault cuts. Thin green lines and numbers indicate inlines and crosslines.

70

C. Berndt et al. / Earth and Planetary Science Letters 319–320 (2012) 65–74

landslide would suggest that the total length is approximately 3–4 km. Assuming the same interval velocities for the pre-Ana Slide as for the Ana Slide the volume of sediment involved in pre-Ana Slide was at least 0.234 km 3 and perhaps in excess of 0.3 km 3. For most of their lengths the headwalls of both landslides are less than 300 m apart and the overall shapes of the landslides are similar. Seismic cross sections show that the headwall of the overlying Ana Slide is in continuation of the headwall of the underlying pre-Ana Slide. In the following we call both landslides the Ana submarine landslide complex noting that both slide events are obviously separated in time as evidenced by the stratified hemi-pelagic sedimentation between both deposits (Fig. 2). The upper Quaternary sediments of the Balearic slope of the Eivissa Channel, where the Ana submarine landslide complex is located, mainly consist of carbonate-rich (~ 50% CaCO3 is common) hemipelagic silty clays with biogenic sands (~10%, with more than 20% at some levels) (S. Lafuerza, person. comm.). Since the Ana Slide and the other three landslides occupy the same stratigraphic position in the Eivissa Channel and share the same slip plane (Lastras et al., 2004) it is possible that they are all of the same age, which may in turn suggest a destabilisation event affecting several parts of the Balearic slope of the Eivissa Channel. No age control exists on the pre-Ana Slide. We place it tentatively in the Mid-Upper Quaternary because of its stratigraphic position and the stratified character of the seismic facies above and below, which corresponds to the Quaternary part of the Plio-Quaternary sequence in the area.

3.3. Evidence for gas and fluid migration The high-resolution 3D seismic data show three types of seismic amplitude anomalies that may indicate presence of fluids and in particular free gas. Underneath Unconformity 4 (Fig. 2) there is a zone of increased amplitudes in the western part of the survey area. These amplitude anomalies are confined to 30–50 ms underneath the unconformity. The seismic data do not allow the polarity of these seismic events to be measured, as the high amplitudes are not restricted to a single reflector. The reflections with increased amplitudes are dipping to the east as do the other reflections underneath, and they terminate in the west against the unconformity. The area with high amplitude reflections extends some 400–500 m in east–west direction and extends for at least 1 km throughout the northern half of the seismic cube.

The well-stratified seismic unit above Unconformity 1 and below the pre-Ana Slide also shows significantly increased seismic amplitudes in particular below the depositional area of the pre-Ana Slide (Figs. 2 and 5). These amplitude anomalies consist of local brightening of the sedimentary reflections. This means the reflections continue uninterrupted beyond the brightening to both sides. The amplitudes in the area of brightening are stronger than the sea floor reflection amplitude in the same region, therefore indicating a significant change in acoustic impedance. The amplitude anomalies are mostly confined to a 100 to 120 ms TWTT-thick interval that occurs in a band that strikes from NE to SW through the seismic cube and is approximately 1500 m wide (anomaly labelled as “gas cloud” in Fig. 6). It is not possible to determine the phase of these reflections (Fig. 5). Apparent polarity varies between normal and reverse and in many parts of the anomaly the phase seems to be shifted by 90 . Nevertheless the observation of the strong amplitude anomalies, the abrupt change of amplitude along continuous reflections, and the observed phase shifts of the seismic signal strongly suggest that these anomalies are caused by free gas in the sediments and in the following we refer to these reflections as the “gas cloud”. There are some less strong instances of reflector brightening about 50 ms TWTT higher up in the western part of the cube and 20 to 50 ms TWTT above the eastern termination of this anomaly (black arrows in Fig. 2). The third area with seismic evidence for the presence of free gas includes the top 50 ms below the seafloor just beneath the headwall of the Ana Slide (Fig. 2d). In this area a clear polarity reversal and modest brightening of reflectors signifies a decrease in acoustic impedance. These amplitude anomalies connect down slope with the brightening of the slip plane of the Ana Slide. Brightening of this reflector is restricted to the section up dip of the termination of the gas cloud (Fig. 2c). The strongest reflections observed in the seismic data are related to the erosion surfaces of Unconformity 1 and 2. These reflections are clearly not due to free gas as they are of normal polarity and their erosional character is documented by the abrupt termination of the underlying sedimentary reflections and the onlap relationship of overlying reflections. Perhaps the most unequivocal indication is the observation of the incised channel (Fig. 6). The fact that the reflections from these unconformities are stronger than the seafloor reflection or any of the other amplitude anomalies that we interpret as evidence for free gas cautions that the amplitude anomalies on their own cannot be used for postulating the presence of fluids. 4. Discussion 4.1. Mass wasting history

Fig. 5. Wiggle plot of inline 165 showing the polarity of the amplitude anomalies. In this section they are predominantly reversed as expected for a negative acoustic impedance contact caused by free gas. Throughout the seismic cube the polarity of this event is variable indicating that there is a superposition of several reflections. Vertical axis, TWTT.

The headwalls of the Ana Slide and the pre-Ana Slide are very close together (Fig. 4). Along in-lines 20 to 50 they are congruent. In the northern part of the landslide complex, and within the 3D seismic data coverage the Ana Slide extends up to 800 m further up the slope than pre-Ana Slide, while in the southern part the pre-Ana Slide extends up to 400 m further up the slope than Ana Slide, with an unknown additional distance beyond the edge of the surveyed area further south. The close proximity of the headwalls suggests the possibility that the same geological processes control slope stability in this area. Lastras et al. (2004) noted that three more landslides exist further north on the eastern, Balearic slope of the Eivissa Channel. While this observation demonstrates that most of the eastern slope of the Eivissa Channel is prone to failure on the weak layer at the base of the four slides, the clustering of two landslides in the same location after a significant time (~ 100 k years assuming approximately constant sedimentation) of quiescence with continuous sedimentation of fine grained sediments may indicate that they had the same failure mechanism. It is conceivable that the presence of the first landslide caused changes in later sedimentation that made the

C. Berndt et al. / Earth and Planetary Science Letters 319–320 (2012) 65–74

71

Fig. 6. Similarity plots (+/−18 ms window) of time slices 1.294 s through the high amplitude reflections (top) and 1.444 s through Unconformity 2 (bottom). Note the spatial correlation of the amplitude anomaly and the course of the erosion channel. Thin green lines and numbers indicate inlines and crosslines.

same area prone to fail again, e.g. more rapid infilling due to greater accommodation space or differential subsidence causing additional stress in the overburden. However, there are no seismic observations, e.g. downward deflection of seismic reflectors, to suggest this. A second possibility is that landslide location is controlled by additional factors such as long-term fluid seepage at fixed locations. There is ample evidence for a local fluid migration system that may destabilise the slope exactly in the area of Ana and pre-Ana slides. This evidence includes the gas cloud above Unconformity 1, the small-scale amplitude anomalies beneath the head wall area, and possibly the amplitude anomalies in the rotated fault blocks below Unconformity 4 (Fig. 2e). Additionally, the pockmarks mapped in the multi-beam bathymetry data (Fig. 1c) suggest that there are active fluid seepage systems in the area (Lastras et al., 2004, 2006). There is also isotopic evidence from the geochemical analysis of micro-fossils found in sediment cores from the headwall of the Ana Slide, that there is a protracted history of methane seepage in this area (Panieri et al., in press). An additional factor favouring landsliding could be the presence of weak

layers continuous along the eastern slope of the Eivissa Channel acting as slip planes. Lastras et al. (2004) show that the four landslides including Ana Slide share the same slip plane. Our observation of fluid migration indicators in the area opens the possibility that the weak layer is weak because of fluids, for example because of a permeability barrier that may cause subsurface overpressures. Both the Ana Slide and the pre-Ana Slide are fairly simple structures. Judging from the deflection of the seabed contours (Figs. 3 and 4) approximately the lower half of the landslide is the accumulation or deposition area, whereas the up-dip half of the landslide can be divided into a source and a translation area. Two steps in the base of the Ana Slide close to the headwall (Fig. 3) show that stepping of the slip plane occurred. This possibly indicates a retrogressive evolution of the landslide caused by footwall erosion. However, the upper slip plane only developed in an area 200 to 300 m west of the main easternmost headwall. This suggests that initial sliding took place on the main lower slip plane on a width of up to 3 km. This coincides with the area of increased seismic amplitude anomalies of the

72

C. Berndt et al. / Earth and Planetary Science Letters 319–320 (2012) 65–74

Fig. 7. Amplitude anomalies suggest fluid migration from underneath Unconformity 4 into the gas cloud and further up along through the overlying sediments and along the slide plane of the Ana Slide towards the head wall.

reflection that constitutes the slip plane (Fig. 2a and c). This increase of seismic amplitudes is not very strong and only slightly lower amplitudes are also observed down-dip (Fig. 2). Nevertheless, the coincidence of the high amplitudes and the area where gas rises from the main gas cloud suggests that gas accumulated at the base of the slip plane and may reduce the strength of this layer. We cannot rule out, however, that this variation in amplitudes and possibly fluid distribution is a late, i.e. post-slide effect, due to the permeability barrier imposed by the slide material. Perhaps readjustment of the seafloor close to the head wall took place in the late stage of slide event when the upper slip plane was active. The seismic data show that most of the landslide material consists of internally well-stratified sediments representing intact blocks that have moved only a small distance down slope. This is different for the evacuation area in the upper-most part of the landslide where a significant amount of material is missing which can be found as a 10–30 ms TWTT-thick chaotic unit further down slope. The seismic data also suggest that the extent of disintegration was greater for the pre-Ana Slide, with fewer intact blocks and more chaotic material. The internal structure (Fig. 2a) and thickness measurements of the pre-Ana Slide indicate that there was some ponding of landslide material against the faults in the lower part of the landslide, thus suggesting that faulting was already active at the time when the pre-Ana Slide took place. 4.2. Fluid migration pathways The origin of fluids migrating through the surface sediments in the Eivissa Channel is poorly constrained without direct sampling. The observation of amplitude anomalies in the top of rotated fault blocks underneath Unconformity 4 (Fig. 2) may indicate thermogenic gas production at depth as previously proposed for the area (Acosta et al., 2001). This would be consistent with the thermogenic gas systems further north in the Valencia Trough, where petroleum systems have developed from a source rock located in the Jurassic–Cretaceous shales associated with shallow water carbonates of the Thetyan margin (Gibbons and Moreno, 2002). Upward fluid migration has occurred through the fault paths opened during rifting with the reservoirs formed as a result of secondary porosity, of karstic origin, produced during Paleogene subaerial exposure of Lower Cretaceous

carbonates. The mid-Miocene marls overlying the major Paleogene erosional unconformity provide the trap. Geological conditions for organic matter maturation and hydrocarbon generation have been met by rapid subsidence below the Ebro sedimentary system. In the Eivissa Channel, the lack of burial and subsidence has prevented the development of hydrocarbon reservoirs. A second line of evidence for a deep source of the fluids is the spatial relationship between the extent of the gas cloud and the location of the erosional channel in Unconformity 1. The gas cloud extends like the channel in NE-SW direction. This may indicate that thermogenic gas formed in the Mesozoic limestones cannot accumulate below Unconformity 4 due to either the absence of the karstic secondary porosity, as suggested by porosity logs in Amposta Marino Well, or/and because the Messinian erosional channel has eroded into the underlying sediments to such an extent that the seal above the rotated fault blocks was severed and that fluids may leak through the seismically transparent unit between Unconformity 2 and Unconformity 3 in the area of the channel. Although not evident as high amplitude seismic reflections (Fig. 2b), the fluids may rise vertically through the sediments overlying Unconformity 2 up towards the gas cloud where they accumulate. The seismic character of the sediments above the incised channel is slightly more disrupted than away from the channel possibly indicating past fluid migration (Fig. 2b), but it has to be pointed out that this may be the result of seismic imaging artefacts close to the strong hummocky reflection of Unconformity 1. The fact that there are no seismic amplitude increases may mean that fluid migration is intermittent and has presently stopped. Another possibility is that gas is rising in solution through the sediments above Unconformity 2 and that it comes out of solution when the hydrostatic pressure drops below the threshold. The seismic response of the gas cloud consists of at least three strong reflectors that are characterised by abrupt lateral amplitude changes. The fact that there is no lower boundary, i.e. flat spot, suggests that the gas most likely is trapped stratigraphically in more porous lithological units that possibly have larger grain sizes. This seismic response is typical for shallow gas and often found in gas accumulations below gas hydrates (Berndt et al., 2004). Whether increased pore pressures characterise such systems is an ongoing debate (Hornbach et al., 2004). Overpressures seem more likely for

C. Berndt et al. / Earth and Planetary Science Letters 319–320 (2012) 65–74

deeply buried and compacted systems because compaction would lead to an overall reduction of permeability. Vertically wellconnected gas pockets in such systems would imply buoyancyrelated overpressure, which is supported by drilling results for the Blake Ridge (Flemings et al., 2003). However, assuming seismic Pwave velocities of 1600 m/s for the surface sediments in the Eivissa Channel, the observed gas cloud is only between 150 and 220 m below the sea floor, and the vertically discontinuous character of the seismic anomalies does not suggest good connection of the gas inclusions. This would imply that overpressures may be low, but drilling or long-offset seismic experiments would be necessary to test this. The only evidence for buoyancy driven flow is the observation of a 50–100 m wide and 90 ms TWTT high zone of moderately increased amplitudes of the sedimentary reflections at the up-dip limit of the gas cloud (Fig. 2c). We interpret this as gas seeping from the gas cloud. The fact that these amplitude anomalies only occur predominantly at the up-dip limits of the gas cloud may suggest that the sediments overlying the gas cloud are impermeable to the extent that they deflect fluid migration laterally (Fig. 2c). There is a trail of increased seismic amplitudes indicating that fluid migration continues laterally and up-dip along the slip plane of the Ana Slide towards the third type of potentially gas-related amplitude anomaly right beneath the head wall of the Ana Slide (Fig. 2d). However, both the seismic anomalies at the updip end of the gas cloud and at the base of the pre-Ana Slide are not significantly higher than in some other places above Unconformity 1 and this interpretation has to be taken with caution. 4.3. Effects of fluid flow on slope stability The spatial relationship between the landslides and the fluid flow indicators (Fig. 7) strongly suggests that fluid migration has played a role in the repeated destabilisation of the slope sediments in the Eivissa Channel. This was proposed also for the Storegga Slide off midNorway where the most deeply incised slope failures occur precisely above a leaking gas reservoir (Bünz et al., 2005). However, off Norway it was not possible to document the temporal evolution of slope stability because the Storegga Slide consists of a very complex succession of sliding events. Our data from the Eivissa Channel show that repeated failure can occur in the same area over long geological time spans. Understanding the role of gas in slope stability is not straightforward. Both laboratory measurements (Wheeler, 1988) and numerical modelling studies (Grozic et al., 2005) show that free gas in the pore fluids increases the undrained shear strength of the sediment as the gas will go into solution under normal loading and the sediments enter a state of partial drainage. Gas will only have a detrimental effect on slope stability when additional processes, such as sudden unloading due to a fast sea level drop or a landslide, act on the sediments. The reason for this is that the gas expands at a drop of hydrostatic pressure. This raises the pore pressure, which in turn means decreased effective stress. Other processes like rapid loading of the slope may also cause increased pore pressures (Dugan and Flemings, 2002; Förster et al., 2011). However, there is no evidence for rapid loading in the Eivissa Channel. As fluid flow itself is pressure driven it will not develop overpressure unless it is episodic and flow focusing occurs (Dugan and Stigall, 2010) but again, in the laterally continuous sediment deposition of the Eivissa Channel this seems unlikely. Thus, dewatering of deep sediments and related slow fluid migration is not likely to generate significant amounts of overpressure either. If there is a significant free gas fraction buoyancy forces may contribute to a pore pressure increase (Crutchley et al., 2010). This would require an interconnected gas column of approximately 20% pore space (Henry et al., 1999). However, the distribution of probably gas-related seismic amplitude anomalies (Fig. 2) does not suggest a

73

vertically well-connected gas column, e.g. there is no proper flat spot at the base of the amplitude anomalies. The new 3D seismic data suggest that presently there is little if any overpressure in the shallow sediments and the 50–100 m of hemipelagic sediments that have accumulated between the occurrence of the pre-Ana Slide and the Ana Slide would suggest that there are prolonged times of continuous sedimentation and stability in the Eivissa Channel. Thus, it is likely that both events were caused by dynamic changes of the pore pressure regime. Three processes are conceivable. First, pore pressures may rise during times of rapid sea level fall as gas comes out of solution reducing effective stress. Second, the fluid advection may change due to deep-seated tectonic processes leading to a situation in which pore pressure cannot dissipate quickly enough due to low permeability. Finally, earthquake loading may induce overpressure (Jackson et al., 2004). The whole of the Mediterranean must be considered seismically active due to the collision between Africa and Eurasia, although long-term observation for the past 100 years did not reveal significant seismicity (Vannucci et al., 2004). This obviously does not preclude major earthquakes during the past 100,000 years and it cannot be ruled out that tectonic activity and seismicity was higher during the Pliocene and Pleistocene due to residual activity of the Neogene rifting phase and volcanic activity. However, these processes alone would not explain the striking spatial relationship between fluid flow indicators and the location of the slope failures. 5. Conclusions The discovery of a second, previously unknown landslide below the Ana Slide shows that the eastern Balearic slope of the Eivissa Channel has been unstable for an extended period of time, probably since the mid-Quaternary. The similar size and the fact that its source area is to a large part congruent with the Ana Slide suggests that the same process was active at least since the pre-Ana Slide occurred. Previous studies proposed that fluid migration and the presence of gas in the sediments have destabilised the slopes of the Eivissa Channel (Lastras et al., 2004). The new high-resolution 3D seismic data show in unprecedented detail the distribution of free gas in the sediments imaging a significant portion of the fluid migration pathway. These data also provide some seismic constraints on the overpressure regime. The fact that direct indicators for overpressure such as a seismic flat spot at the base of the gas accumulations are absent and that there seems to be little vertical connection of high-amplitude reflections suggest that pore pressures are not significantly increased at least at present. Although processes such as earthquake loading, deposition of weak layers, or weakening of the sediments due to fluid percolation might have played a role, we propose based on the spatial relationship between landslide deposits and evidence for gas and fluid migration that the Ana Slide and the pre-Ana Slide were caused by changes of pore pressure. Buoyancy due to the exsolution of gas in times of lowering sea level is a possible driver for increased pore pressure as we do not observe signs of rapid sediment deposition in the study area. Buoyancy-related overpressure would have decreased the effective stress and made the slopes less stable. Drilling and dating of the timing of the landslides are required to corroborate this geophysical inference. Acknowledgements This work is part of the EC-funded HERMIONE project (ref. 226354-HERMIONE). Also the Spanish CONSOLIDER-INGENIO 2010 “GRACCIE” project (ref. CSD2007-00067) contributed to this research. SC, MC, AC and BDM are grateful to Generalitat de Catalunya for its excellence research group grant to GRC Geociències from Universitat de Barcelona. The PhD studentship of SC was funded through a

74

C. Berndt et al. / Earth and Planetary Science Letters 319–320 (2012) 65–74

Generalitat de Catalunya grant. We thank the master and the crew of RRS Charles Darwin who facilitated data acquisition during voyage 178, and Frode Eriksen of VBPR for technical support on the same cruise. Landmark Graphics supported this work by providing ProMAX to the University of Southampton through the Landmark Graphics university grant programme. Seismic Micro Technology provided academic licenses for KingdomSuite. References Acosta, J., et al., 2001. Pockmarks in the Ibiza Channel and western end of the Balearic Promontory (western Mediterranean) revealed by multi-beam mapping. Geo-Mar. Lett. 21 (3), 123–130. Acosta, J., et al., 2002. The Balearic Promontory geomorphology (western Mediterranean): morphostructure and active processes. Geomorphology 49, 177–204. Acosta, J., et al., 2004. Sea floor morphology and Plio-Quaternary sedimentary cover of the Mallorca Channel, Balearic Islands, western Mediterranean. Mar. Geol. 206 (1–4), 165–179. Alonso, B., et al., 1988. Los sedimentos de la plataforma continental balear. Acta Geol. Hisp. 23 (3), 185–196. Assier-Rzadkiewicz, S., Heinrich, P., Sabatier, P.C., Savoye, B., Bourillet, J.F., 2000. Numerical modelling of a landslide-generated tsunami: the 1979 Nice event. Pure Appl. Geophys. 157, 1707–1727. Berg, K., Solheim, A., Bryn, P., 2005. The Pleistocene to recent geological development of the Ormen Lange area. Mar. Pet. Geol. 22 (1–2), 45–56. Berndt, C., Bünz, S., Clayton, T., Mienert, J., Saunders, M., 2004. Seismic character of bottom simulating reflectors: examples from the mid-Norwegian margin. Mar. Pet. Geol. 21, 723–733. Bondevik, S., Mangerud, J., Dawson, S., Dawson, A.G., Lohne, Ø., 2005. Evidence for three North Sea tsunamis at the Shetland Islands between 8000 and 1500 years ago. Quat. Sci. Rev. 24, 1757–1775. Bryn, P., Berg, K., Forsberg, C.F., Solheim, A., Kvalstad, T.J., 2005. Explaining the Storegga Slide. Mar. Pet. Geol. 22 (1–2), 11–19. Bull, S., Cartwright, J., Huuse, M., 2009. A review of kinematic indicators from masstransport complexes using 3D seismic data. Mar. Pet. Geol. 26 (7), 1132–1151. Bünz, S., Mienert, J., Bryn, P., Berg, K., 2005. Fluid flow impact on slope failure from 3D seismic data: a case study in the Storegga Slide. Basin Res. doi:10.1111/j.13652117.2005.00256.x Camerlenghi, A., Urgeles, R., Fantoni, L., 2009. A database on submarine landslides of the Mediterranean Sea. In: Mosher, D.C., et al. (Ed.), Submarine Mass Movements and Their Consequences. Advances in Natural and Technological Hazards Research. Springer, pp. 491–501. Canals, M., Ballesteros, E., 1997. Production of carbonate particles by phytobenthic communities on the Mallorca–Menorca shelf, norhwestern Mediterranean Sea. Deep Sea Res. II 44 (3–4), 611–629. Canals, M., et al., 2004. Slope failure dynamics and impacts from seafloor and shallow sub-seafloor geophysical data: case studies from the COSTA project. Mar. Geol. 213 (1–4), 9–72. Cattaneo, A., et al., 2011. Age constraints and sediment properties of Ana Slide (Balearic Sea, Western Mediterranean) and implications on age dating of submarine landslides. EGU General Assembly, Vienna. Crutchley, G.J., Pecher, I.A., Gorman, A.R., Henrys, S.A., Greinert, J., 2010. Seismic imaging of gas conduits beneath seafloor seep sites in a shallow marine gas hydrate province, Hikurangi Margin, New Zealand. Mar. Geol. 272 (1–4), 114–126. Dugan, B., Flemings, P., 2000. Overpressure and fluid flow in the New Jersey continental slope: Implications for slope failure and cold seeps. Science 289, 288–291. Dugan, B., Flemings, P.B., 2002. Fluid flow and stability of the US continental slope offshore New Jersey from the Pleistocene to the present. Geofluids 2, 137–146. Dugan, B., Stigall, J., 2010. Origin of overpressure and slope failure in the Ursa Region, Northern Gulf of Mexico. Submarine Mass Movements and Their Consequences. Springer, pp. 167–178. Escutia, C., Maldonado, A., 1992. Palaeogeographic implications of the Messinian surface in the Valencia trough, northwestern Mediterranean Sea. Tectonophysics 203 (1–4), 263–284. Fernandez, M., Foucher, J.P., Jurado, M.J., 1995. Evidence for the multi-stage formation of the south-western Valencia Trough. Mar. Pet. Geol. 12 (1), 101–109. Flemings, P.B., Liu, X., Winters, W.J., 2003. Critical pressure and multiphase flow in Blake Ridge gas hydrates. Geology 31 (12), 1057–1060. Flemings, P.B., et al., 2008. Pore pressure penetrometers document high overpressure near the seafloor where multiple submarine landslides have occurred on the continental slope, offshore Louisiana, Gulf of Mexico. Earth Planet. Sci. Lett. 269 (3–4), 309–325. Fontbote, J.M., et al., 1990. The Cenozoic geodynamic evolution of the Valencia Trough (Western Mediterranean). Rev. Soc. Geol. Esp. 3, 249–259. Förster, A., Ellis, R.G., Henrich, R., Krastel, S., Kopf, A.J., 2011. Geotechnical characterization and strain analyses of sediment in the Mauritania Slide Complex, NW-Africa. Mar. Pet. Geol. 27 (6), 1175–1189. Gibbons, W., Moreno, T., 2002. The Geology of Spain. The Geological Society, London.

Grozic, J.L.H., Nadim, F., Kvalstad, T.J., 2005. On the undrained shear strength of gassy clays. Comput. Geotech. 32 (7), 483–490. Harbitz, C.B., 1992. Model simulations of tsunamis generated by the Storegga Slides. Mar. Geol. 105, 1–21. Henrich, R.D., Hanebuth, T.J.J., Krastel, S., Neubert, N., Wynn, R.B., 2008. Architecture and sediment dynamics of the Mauritania Slide Complex. Mar. Pet. Geol. 25 (1), 17–33. Henry, P., Thomas, M., Clennell, M.B., 1999. Formation of natural gas hydrates in marine sediments; 2. Thermodynamic calculations of stability conditions in porous sediments. J. Geophys. Res. 104 (B10), 23,005–23,022. Hispanoil-Eniepsa, 1978. Report on a Geochemical Evaluation of the Ibiza Marino AN-1 Well, Offshore Spain. Hornbach, M.J., Saffer, D.M., Holbrook, W.S., 2004. Critically pressured free-gas reservoirs below gas-hydrate provinces. Nature 427, 142–144. Jackson, P.D., Gunn, D.A., Long, D., 2004. Predicting variability in the stability of slope sediments due to earthquake ground motion in the AFEN area of the western UK continental shelf. Mar. Geol. 213, 363–378. Kvalstad, T.J., et al., 2005a. The Storegga slide: evaluation of triggering sources and slide mechanics. Mar. Pet. Geol. 22 (1–2), 245–256. Kvalstad, T.J., Nadim, F., Kaynia, A.M., Mokkelbost, K.H., Bryn, P., 2005b. Soil conditions and slope stability in the Ormen Lange area. Mar. Pet. Geol. 22 (1–2), 299–310. Lastras, G., Canals, M., Urgeles, R., Hughes-Clarke, J.E., Acosta, J., 2004. Shallow slides and pockmark swarms in the Eivissa Channel, western Mediterranean Sea. Sedimentology 51, 837–850. Lastras, G., et al., 2006. Eivissa slides, western Mediterranean Sea: morphology and processes. Geo-Mar. Lett. 26, 225–233. Lee, H.J., 2009. Timing of occurrence of large submarine landslides on the Atlantic Ocean margin. Mar. Geol. 284, 53–64. Lopez-Venegas, A.M., ten Brink, U.S., Geist, E.L., 2008. Submarine landslide as the source for the October 11, 1918 Mona Passage tsunami: observations and modeling. Mar. Geol. 254, 35–46. 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. Mar. Pet. Geol. 22 (1–2), 219–231. Maillard, A., Mauffret, A., 1999. Crustal structure and riftogenesis of the Valencia Trough (north-western Mediterranean Sea). Sediment. Basins 11, 357–379. Maillard, A., et al., 2006. Offshore evidence of polyphase erosion in the Valencia Basin (Northwestern Mediterranean): scenario for the Messinian Salinity Crisis. Sediment. Geol. 188–189, 69–91. Mauffret, A., Frizon d Lamotte, D., Lallemant, S., Gorini, C., Maillard, A., 2004. E–W opening of the Algerian Basin (Western Mediterranean). Terra Nova 16, 257–264. Micallef, A., Masson, D.G., Berndt, C., Stow, D.A.V., 2009. Development and mass movement processes of the north-eastern Storegga Slide. Quat. Sci. Rev. 28 (5–6), 433–448. Nelson, C.H., Maldonado, A., 1990. Factors controlling late Cenozoic continental margin growth from the Ebro Delta to the western Mediterranean deep sea. Mar. Geol. 95 (3–4), 419–440. Panieri, G. et al., in press. Tracing seafloor methane emissions with benthic foraminifera: results from the Ana Slide submarine slide (Eivissa Channel, Western Mediterranean Sea). Mar. Geol. Perez-Garcia, C., et al., 2011. Linked halokinesis and mud volcanism at the Mercator mud volcano, Gulf of Cadiz. J. Geophys. Res. 116 (B5), B05101. Roca, E., Desegaulx, P., 1992. Analysis of the geological evolution and vertical movements in the Valencia Trough area, western Mediterranean. Mar. Pet. Geol. 9 (2), 167–176. Ryan, W.B.F., 1976. Quantitative evaluation of the depth of the western Mediterranean before, during and after the Late Miocene salinity crisis. Sedimentology 23 (6), 791–813. Ryan, W.B.F., 2009. Decoding the Mediterranean salinity crisis. Sedimentology 56 (1), 95–136. Sultan, N., Cochonat, P., Foucher, J.-P., Mienert, J., 2004. Effect of gas hydrate melting on seafloor slope instability. Mar. Geol. 213 (1–4), 379–401. ten Brink, U.S., Lee, H.J., Geist, E.L., Twichell, D., 2009. Assessment of tsunami hazard to the U.S. East Coast using relationships between submarine landslides and earthquakes. Mar. Geol. 264 (1–2), 65–73. Urgeles, R., et al., 2011. New constraints on the Messinian sealevel drawdown from 3D seismic data of the Ebro Margin, western Mediterranean. Basin Res. 23 (2), 123–145. Vannucci, G., Pondrelli, S., Argnani, A., et al., 2004. An Atlas of Mediterranean Seismicity. Ann. Geophys. Suppl. 47, 247–306. von Rad, U., Hinz, K., Sarnthein, M., Seibold, E., 1982. Geology of the Northwest African Continental Margin. Springer Verlag, Heidelberg. Watts, P., Grill, S.T., Kirby, J.T., Fryer, G.J., Tappin, D.R., 2003. Landslide tsunami case studies using a Boussinesq model and a fully nonlinear tsunami generation model. Nat. Hazards Earth Syst. Sci. 3 (391–402). doi:10.5194/nhess-3-391-2003. Wheeler, S.J., 1988. The undrained shear strength of soils containing large gas bubbles. Geotechnique 38 (399–413). Wynn, R.B., Masson, D.G., Stow, D.A.V., Weaver, P.P.E., 2000. The northwest African slope apron: a modern analogue for deep-water systems with complex seafloor topography. Mar. Geol. 17, 253–265.