The Tagus River Delta Landslide, Off Lisbon, Portugal. Implications For

Marine Geology 416 (2019) 105983

Contents lists available at ScienceDirect

Marine Geology

journal homepage: www.elsevier.com/locate/margo

The Tagus River delta landslide, oﬀ Lisbon, Portugal. Implications for Marine geo-hazards T ⁎ Pedro Terrinhaa,b, Henrique Duartec, Pedro Britoa, João Noivad, Carlos Ribeirod, , Rachid Omiraa,b, Maria Ana Baptistae,b, Miguel Mirandaa,b, Vitor Magalhãesa,b, Cristina Roquef,b, Tagusdelta cruise team1 a Instituto Português do Mar e de Atmosfera – IPMA, I.P.; Rua C do Aeroporto, 1749-077 Lisbon, Portugal b Instituto D. Luiz – IDL; Faculdade de Ciências da Universidade de Lisboa, Campo Grande, Edifício C8, Piso 3, 1749-016 Lisbon, Portugal c GeoSurveys - Consultants in Geophysics, Lda.; Avenida Araujo e Silva n.59, 3810-049 Aveiro, Portugal d Departamento de Geociencias, – Universidade de Évora; Instituto de Ciências da Terra – ICT; Colégio Luís António Verney, Rua Romão Ramalho, 59, 7000-671 Évora, Portugal e Instituto Superior de Engenharia de Lisboa – ISEL; Rua Conselheiro Emídio Navarro, N. 1, 1959-007 Lisboa, Portugal f Estrutura de Missão de Extensão da Plataforma Continental – EMEPC; Rua Costa Pinto, n. 165, 2770-047 Paço de Arcos, Portugal

ARTICLE INFO ABSTRACT

Editor: Michele Rebesco The stratigraphy of the Tagus river ebb-tidal delta off Lisbon (Portugal) is investigated using high resolution fl fi Keywords: multichannel seismic re ection pro les with the purpose of searching for sedimentary or erosive features as- Tagus River – NE Atlantic sociated with landslides. The Tagus delta is sub-divided in two prograding seismic units of 17 ky to 13 ky and Ebb-tidal delta 13 ky to Present based on the calibration of seismic lines using gravity and box-cores in the Tagus pro-delta. We Seismic stratigraphy report the existence of a buried landslide with 11 km of length, 3.5 km of width and a maximum thickness of Submarine landslide 20 m that accounted for the collapse of half of the upper unit of the Tagus river delta front in Holocene times. Tsunami hazard The non-collapsed half of the delta front contains extensive shallow gas of still unknown origin and nature. An estimated age of ~8 ky BP for the Tagus delta landslide is proposed based on stratigraphic correlation. The trigger mechanisms of the newly identified Tagus landslide are discussed as well as of the several landslides also found in the lower delta unit. These findings present a first step towards a future assessment of the susceptibility of the nearby coastal areas and the off-shore infrastructures to hazards related to such large collapses.

1. Introduction aquaculture, harbors, etc.). Delta-front collapses are potential triggers of tsunami that could affect both port areas and coastal towns (Hughes Deltas are sedimentary bodies with high sedimentation rates that Clarke et al., 2012). A recent example is the 1979 Nice submarine are very sensitive to changes in environmental conditions, such as cli- landslide that occurred along the Var delta front in the SE coast of mate changes or anthropogenic alteration of hydraulic regimes of the France, followed by a tsunami that caused several casualties and in- rivers and their sedimentary load (e.g. construction of dams, irrigation frastructural damage of the Nice-Antibes airport (Pelinovsky et al., for agriculture). For these reasons the study of deltas, together with pro- 2002; Sultan et al., 2010; Sahal and Lemahieu, 2011). deltas and estuaries are key features to investigate the record and Slope failure of the delta front is the major mechanism for the evolution of environmental changes in Holocene and Late Pleistocene supply of coarse sediment into the pro-delta, contributing for its de- times. Deltas and nearby coastal areas are regions where some of the velopment and growth (Maillet et al., 2006; Kim and Chough, 2000). highest concentrations of population in the world take place in both Slides and slumps formed by failure of delta front can be up to tens of modern times and old cultures. Physical instability of deltas is an im- meters thickness and hundreds of meters in length (Nichols, 2009). portant societal concern as it can affect a variety of aspects, from de- Gravity flows, such as debris flows and turbidity currents, can feed and/ struction of civil engineering facilities (dwelling or industry), to agri- or erode the pro-delta and reach further distances into the continental culture, land, marine or submarine facilities (communication cables, slope and abyssal plains. The pro-delta sediments usually display

⁎ Corresponding author. E-mail address: [email protected] (C. Ribeiro). 1 Marcos Rosa, Paulo Alves, Francisco Teixeira. https://doi.org/10.1016/j.margeo.2019.105983 Received 24 February 2019; Received in revised form 12 June 2019; Accepted 6 July 2019 Available online 07 July 2019 0025-3227/ © 2019 Elsevier B.V. All rights reserved. P. Terrinha, et al. Marine Geology 416 (2019) 105983 pervasive small-scale sediment deformation and slope failure dipping to the north-west. The NW-SE striking segments of the coast (Correggiari et al., 2001; Lykousis et al., 2003). between 9°20′W and 9°30′W correspond to dextral strike-slip faults that Delta front slopes can vary from very shallow dips of 0.5–2° in mud- are imaged in the seismic profiles in this work (see Fig. 2)(Terrinha rich deltas (Lykousis et al., 2009) to steep slopes of ~30° in coarse et al., 2017). grained sediments (Nichols, 2009). Forcing mechanisms for delta front The Pliocene and Quaternary siliciclastic sandstones and conglom- failures can be: i) sediment load arising from mouth bar accretion, ii) erates outcrop at the top of the cliffs on the Tagus southern bank, lying changes in pore pressure due to wave conditions (Nemec, 1990; Mello unconformably on top of the Upper Miocene (Pais et al., 2012). and Pratson, 1999; Hill and Christian, 2003), iii) very low tides or river The Tagus delta depicts a crescent shape to the west of the estuary flood peaks (Hughes Clarke et al., 2012, 2014; Hill, 2012), iv) events inlet, it has an estimated volume of 4 km3 (Ferraz, 2014) to 5.5 km3 that cause motion of crescent bedforms, v) hyperpycnal river flood (Vis, 2009), with a maximum sediment thickness of about 50 m and discharge (Mulder et al., 2003), and vi) ground shaking in active con- mean surface dimensions of approximately 13 km long - across shore - tinental margins with important seismicity (Trincardi et al., 2004; by 15 km wide - along shore (Fig. 1). The delta top surface corresponds Lykousis et al., 2009; ten Brink et al., 2009). to a shallow flat platform that varies in height from 25 m below mean The ebb-tidal delta of the Tagus river estuary (from now on referred sea level (bsl) to a couple of meters above sea level (Fig. 1B). The pro- as the Tagus delta), is a shallow water, Pleistocene-Holocene sediment delta lies at the base of the delta front at depths around 90 m bsl. body located in the high energy, sediment starved, west Iberian con- The present-day position of the Cascais and Lisbon canyons (CC and tinental shelf (Lantzsch et al., 2009). The acquisition of the 2D multi- LC in Fig. 1B) controls the irregular shape of the shelf break and the channel seismic reflection profiles, under the project TAGUSDELTA, latter determines the southern limit of the Tagus delta front as bathy- was carried out with the main goal of searching for high energy deposits metric contours drop abruptly from ca. 50 m bsl to 200 m bsl. or erosive structures associated with past tsunamis, mapping the extension of the landslide and the gas bearing sediments (Fig. 1). 2.2. Core data in Tagus pro-delta The objectives of this work are, i) to report the existence of buried landslides in the Tagus delta, ii), to propose a stratigraphic model of the Since the identification of a deltaic system in the mouth of the Tagus study area and iii) to discuss the trigger mechanisms for the Tagus delta River by Vanney and Mougenot (1981) several works have been done landslides and their potential hazard. both in the delta and the pro-delta using sediments dredged, collected in box, gravity and piston cores (Fig. 1C). Determinations of 14C and 2. Previous works 210Pb led to the proposal of age models and estimates of the variations of the accumulation rate through time. Erosive hiatuses and detrital 2.1. Geologic and geomorphologic setting deposits arguably associated with the 1969 and 1755 earthquakes, with instrumental magnitude M = 7.9 and estimated magnitude M~8.5, The Tagus river flows westwards across Spain and Portugal for > respectively, have been reported (Abrantes et al., 2005; Abrantes et al., 1000 km with a catchment area of > 80,600 km2 across semi-arid lands 2008) based on the analysis of box cores and shallow gravity cores of the Iberian Meseta (Fig. 1A and B). It flows into an estuary with a collected in the condensed sequence of the Tagus pro-delta. total area of 320 km2 that includes an extensive area of tidal flats and The information reported from three cores was used in this work to marshes (Fig. 1B). These intertidal zones cover about 40% of the inner calibrate our seismic reflection lines and propose stratigraphic models. estuary area and are located essentially in the continuity of areas of The D13902 and D13882 long piston cores were acquired during the terrigenous sandstones and gravels of Pliocene and Early Quaternary Discovery 249 cruise and the GeoB8903-1 gravity corer was acquired age around the southern river bank. It is an upper mesotidal estuary, during Poseidon PO304 cruise (Table 1 and Fig. 1C). Detailed in- with a mean tidal prism of 650 × 106 m3 and tidal ranges varying from formation on age models and sedimentation rates for the pro-delta can 0.75 m near the mouth during neap tides, to 4.3 m in the upper estuary be found in Abrantes et al. (2005, 2008), Alt-Epping et al. (2009), during spring tides (Fortunato et al., 1999; Silva, 2013). Due to high Burdloff et al. (2008), Mil-Homens et al. (2009); Bartels-Jónsdóttir et al. sedimentation rates, regular dredging of the estuary is needed to (2006) and Rodrigues et al. (2009). maintain the navigation of large vessels. Inspection of the bathymetry of the Tagus estuary (Fig. 1B) shows 3. Data and methods two strikingly distinct areas: 1) a wide (over 13 km in some places) and shallow (maximum depths of 15 m) main estuarine basin and 2) a The bathymetry used in this paper is a compilation by Baptista et al. narrow bottleneck channel about 30 m deep, 2 km wide and 12 km long (2011) with a grid cell size of 200 m covering the study area. (Fortunato et al., 1997) that connects the main estuarine basin to the The seismic reflection dataset used in this work consists of Chirp ocean. (Lisboa98 dataset), Sparker single channel (PACEMAKER 2011 dataset) The striking morphologic difference between the main estuarine and multichannel ultra-high resolution seismic (UHRS) data (TAGUS- basin, the inlet channel and the northern and southern river banks are DELTA 2013 dataset). The UHRS lines interpreted in this work were due to different geologic terrain constitution, as well as tectonic obtained during the TAGUSDELTA 2013 survey done with the R/V structure. The geology and tectonics of the region around Lisbon are Noruega, in November/December 2013 (Fig. 1C). The UHRS was ac- summarized in Fig. 1B, where it is clear that the northern bank is quired using a 800 tips Geo-Sparker seismic source operated with an dominantly covered by hard carbonate sedimentary and volcanic rocks energy of 2000 J (2.5 J per tip), and a shooting period of 0.9 s (shooting of Cretaceous age (Dinis et al., 2008; Miranda et al., 2009). Patches of rate ~1.1 Hz). The used receiver array was a Geo-Sense 24 channels Lowermost Miocene carbonate formations occur west of Lisbon on the streamer with group length and spacing of 0.5 m and 3.125 m, respec- north coast dipping shallowly towards the Tagus delta. On the southern tively, and 3 elements per group. bank, the cliff consists of Miocene limestones, marls and poorly con- The TAGUSDELTA dataset consists of a regular grid of 2D UHRS in- solidated fine-grained sandstones. lines with about 4 km length and a spacing of 200 m, covering a total The Mesozoic sedimentary rocks of the Lisbon area are included in area of approximately 40 km2, as shown in Fig. 1C. Four cross-lines the Lusitanian Basin (Fig. 1B), a rift basin formed from Triassic through were also acquired crossing the sampling sites done in previous works latest Early Cretaceous times, which was affected by alkaline magma- (Abrantes et al., 2005, 2008). tism during Late Cretaceous. The Lisbon part of the Lusitanian Basin The UHRS seismic data were processed using the Promax seismic suffered tectonic shortening in Paleogene and Miocene times and the processing software. The main applied processing steps included south-east part of the Tagus estuary is a foreland basin, shallowly bandpass filtering, signature deconvolution, spherical divergence

2 P. Terrinha, et al. Marine Geology 416 (2019) 105983

Fig. 1. Location, physiography and geology of the study area. A- the Tagus river hydrographic basin in the Iberian Peninsula; B- Geology and bathymetry of the study area, ebb-tidal delta depicted by 10 m spaced contour lines estuary an inlet of the Tagus River. L- Lisbon, C- Cascais, LC- Lisbon Canyon, red rectangle- area in C; Q- unconformable erosive contacts topped by Quaternary coarse marine siliciclastics; TR- Tagus River; SR- Sado River; blue circular arrow: Currents induced by estuarine outflow and main predominant wave climate from the NW; C- location of the seismic profiles, gravity cores and landslide and gas areas. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

3 P. Terrinha, et al. Marine Geology 416 (2019) 105983

Fig. 2. Seismic profile (dip line TD8200) and seismic stratigraphy of the Tagus delta area (for location see Fig. 1B). A) Seismic line with inset of interpreted landslide with stacked slid units separated by detachment faults (in blue). B) Seismic units U1 to U4 and discontinuities D1 to D3 and D3A. Note that the distal part of the basal landslide surface detaches on top of the top foresets of the lower deltaic unit. Note the existence of lobes of chaotic facies in the lower deltaic unit that are interpreted as MTDs that cannot be correlated across seismic profiles, i.e. smaller scale landslides. The pro-delta is approximately 10× thinner than the delta front. Insets show details of interpretation and the sea level curve for the last 50 ky according to de Boer et al. (2010). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) correction, denoise, multiple attenuation, swell filter, pre-stack migra- 1/50.000 scale and 1/10.000 of the Lisbon area. tion, stack and tide correction. For depth conversion a velocity of 1650 m/s in the pro-delta was used based on velocity analysis of the multichannel seismics. 4. Results All the processed seismic datasets were imported to an OpenWorks (from Landmark Graphics Corporation) database, and interpreted with 4.1. Seismic stratigraphy other Landmark tools (SeisWorks, PowerView, AssetView) to better correlate the profiles with core logs and to map out horizons and in- The analysis of the 2D seismic reflection profiles of the study area terpolate the geological surfaces. Contour maps were produced using allowed identifying 4 seismic units (U1 to U4) separated by dis- MIRONE software (Luis, 2007). continuities D1, D2 and D3 (Figs. 2 and 3). The synthesis on the morphology and geology of the study area were The oldest seismic unit (U1) is topped by erosive discontinuities D1 based on field work and inspection of published geological maps on the or D2. Seismic unit U1 consists of a series of high and low amplitude reflections with good lateral continuity corresponding to folded and

Table 1 Core list. Geographic coordinates, datum WGS84a.

Core n° Type Lat (°N); Lon (°W) Water depth (m) Length Oldest age (ky calBP ± y)/mbsfb Cruise

D13902 Long piston 38.554; 9.336 90 6.00 8.002 ± 55/4.40d D249 D13882 Long piston 38.635; 9.454 88 13.61 13.011 ± 70/11.4c D249 GeoB8903-1 Gravity corer 38,625; 9,508 102 5.60 2.58 ± 144/4.98c PO287

a Data from www.pangaea.de. b mbsf- meters below seaﬂoor. c Abrantes et al., 2005, 2008. d Rodrigues et al., 2011.

4 P. Terrinha, et al. Marine Geology 416 (2019) 105983

Fig. 3. Seismic profile (strike line TD010) and seismic stratigraphy of the Tagus pro-delta area (for location see Fig. 1C). A - Seismic line TD010. B - Interpreted line showing seismic units U1 to U4 and discontinuities D1 to D3. Note the two lobes of the Tagus landslide within U4. U2 is well imaged confined to channels; internal unconformities and soft sediment deformation structures. Shaded area shows fluid migration structures. See text for detailed description. HAR - high amplitude reflectors. faulted strata. Faults are steep, and it is not possible to correlate stra- variation of the seismic facies, i.e. onlap terminations of the basal re- tigraphic markers on both flanks of faults, i.e. suggesting strike-slip flectors towards northeast on top of D2 and downlap terminations of faulting. the basal reflectors indicating progradation towards southwest. In be- Seismic unit U2, limited at the base by D1 and by D2 at the top, tween, U3 thins down to nil as it is interrupted by hard-rock reliefs of corresponds to sediments deposited in paleo-valleys sculpted in U1. The U1 (see Fig. 2A–B, SP~2200, SP~1400 and SP~800). The acoustic seismic image in Fig. 3 A–B (a strike-line across paleo-channels) shows character of U3 reflections is very heterogeneous. In the distal part strata confined to channels, containing various internal discontinuities reflectors continuity is poor with wrinkled, chaotic patterns, as well as separating packages of different acoustic facies, like: i) coherent hor- segments of transparent facies. Fairly continuous horizons separating izons with good lateral continuity, ii) transparent units, and iii) dis- packages of poorly organized reflections also occur (Fig. 2 A–B from SP continuous horizons with high amplitude contrast. Also, various types 3000 to SP 2200 and Fig. 3 A–B from SP 8400 to SP 10200). Incisions of of deformation of the sedimentary units are observed, such as: i) dis- small channels are also observable (Fig. 3 A–B, SPs ~9200, ~9600 and harmonic folding, ii) upward injection of material of the transparent ~10,000). Neither the continuous reflectors nor the channels can be units into overlying sediments, iii) disrupted layers, and iv) hetero- mapped at the present scale of observation and 2D seismic reflection geneous bulk deformation of the transparent units, as shown schema- survey, indicating that their lateral extension is smaller than the dis- tically in (Fig. 3B). Neither folding nor faulting affecting U1 propagated tance between seismic lines, ~200 m. into U2 sediments. On the other hand, deformation of U2 sediments is Seismic unit U4 lies on top of discontinuity D3, it is topped by the confined laterally and the deformation style is compatible with water seafloor and comprehends the Tagus delta and pro-delta sediments with saturated sediments within channels under the influence of downslope downlap terminations prograding towards the southwest. From a geo- movements. metric point of view, reflections in the delta can be grouped in three Seismic unit U3 lies between discontinuities D2 and D3 (Table 2, sets, namely progradational, aggradational, and chaotic (Figs. 2 and 3). Figs. 2 and 3). Inspection of along dip profiles (Fig. 2) shows a lateral The progradational reflections correspond to the foresets of the delta

5 P. Terrinha, et al. Marine Geology 416 (2019) 105983

Table 2 Synthesis of characteristics of seismic units and discontinuities.

Seismic unit (U)/ Seismic facies Age Observations discontinuity (D)

U lds Chaotic, hummocky, transparent reflections Late Pleistocene and Holocene Landslides within U3a and U3b. U3b Moderate and high amplitude reflections with good Undetermined, probably mid- Delta plain and delta front facies lateral continuity Holocene to Present D3a Continuous high amplitude reflection Undetermined, probably mid- Delta plain and delta front geometry; separates delta in Holocene two packages U3a Moderate amplitude reflections, delta plain sub- Undetermined, probably Late Delta plain and delta front facies that downlap D2 horizontal and frontal delta prograding foresets Pleistocene D2 High amplitude reflector across study area Late Pleistocene Covers conformably U2; base of the Tagus delta topped by prograding delta units; downlap surface U2 Internal highly discontinuous reflections with Pleistocene Mean thickness of 20 ms TWT, i.e. approximately, 16 m transparent areas thick; with onlap terminations over D1 D1 Good lateral continuity, high amplitude reflection Post -Cretaceous through Regional erosion unconformity that cuts through folded Pliocene and faulted strata of unit U1 U1b High and low amplitude reflectors; lateral Paleogene through Pliocene Channels within U1a filled in by Cenozoic sediments continuity interrupted by various internal unconformities U1a High and low amplitude reflectors with good Lower Cretaceous Folded and faulted strata, intrusions of Lower-Upper lateral continuity Cretaceous alkaline and volcanics deeply incised by channels

Fig. 4. Isochrone maps of discontinuities D2 and D3, a) and b) respectively (TWT in milliseconds, ms). The slope is much less than that of the present day delta front (cf. Fig. 1B). A) Note the ﬂat geometry of D2 surface up to ~150 ms (TWT) (shot point ~2400 in Fig. 2). From this point towards shallower depths D2 is wrinkled and displays jogs that limited sedimentation depocenters (shot points ~2200, ~1400 and ~600 in Fig. 2). The deepening of D2 towards the present day position of the Lisbon canyon indicates the existence of the canyon previous to D2. B) Note that the shape of D3 although less steep that the present day delta front displays an arcuate shape with a jog at ~110 ms TWT possibly associated with a proto-delta towards the continent (cf. shot point ~600 in Fig. 2). front dipping a maximum of 6° towards the oﬀshore. These are con- discontinuities separating packages of foresets are also found. These tinuous along strike and dip and may vary in amplitude indicating discontinuities are typical of deltaic systems and are associated with lateral variations of the deposited sediments. Minor internal variations of the depositional currents and sedimentation

6 P. Terrinha, et al. Marine Geology 416 (2019) 105983

Fig. 4. (continued) accommodation space (Fig. 2). The aggradation package, shown in inset unconformity D2 (Fig. 4a) depicts an oceanwards dipping surface in Fig. 2B, corresponds to delta platforms located upstream of the delta spread across the study area with a mean dip of < 1°, except in its front. The position of delta platform to delta front transition varies southeastern termination where it shows the existence of a buried upstream and downstream in the stratigraphic record (Fig. 2 A and B). paleo-valley (Fig. 4A). It is worthwhile noting the geometrical con- This indicates oscillations that are typical of a very dynamic sedi- tinuity between the TWT isochrones of D2 and the present-day bathy- mentation system in which sedimentation is a function of various en- metry in the Lisbon canyon headscarp that indicates that the canyon vironmental factors, such as littoral currents, hydrological regime of the headscarp predates and controlled the position of D2. Tagus River, sea level fluctuations, and variations of the sediment ac- Discontinuity D3 is a downlap surface of reflections of seismic unit commodation space. At present, the wash over platform of the Tagus U4, imaged as a high amplitude reflection with good lateral continuity delta is located to the south of the estuary inlet above −10 m bathy- and an average dip ~3° (Figs. 2 and 5). The isochrone map of D3 metric contour in Fig. 1A. Swash sand bars occasionally build up on top (Fig. 4B) shows concave and convex areas separated by isochrones of the wash over surface up to a couple of meters above sea level. The 100 ms and 120 ms with a higher frequency curvilinear pattern. accommodation space that controls the submarine delta platform can Discontinuity D3a (Fig. 2) separates two sub-units of seismic unit U4 vary from daily to decadal time scale. Lensoid bodies made of chaotic (U4a and U4b). The internal reflections of these two sub-units have reflections (up to 3 to 4 km long in the NE-SW seismic profiles) occur at similar geometries, as both consist of sigmoidal bodies with a pro- various stratigraphic positions and different locations within the delta. gradation front passing upstream to an aggradation surface. The stra- They display top irregular discontinuities and basal planar well-defined tigraphic relationship of D3a with the underlying sediments of U4a is discontinuities. These bodies are the main object of this study and they the same of seafloor with present day delta sediments of U4b. Thus, the are interpreted as mass transport deposits as described and discussed geometric relationship of discontinuity D3a with the underlying re- later in the text. flectors can be of erosive truncations, toplap or conformable reflections. Discontinuity D1 (Fig. 2 A and B) is a very irregular unconformity Discontinuity D3a truncates U4a foresets reflections upstream and be- cutting > 100 ms into underlying strata. It is restricted to the southern comes conformable with foresets downstream; D3a is conformable with part of the study area near the present-day area of the canyons and the aggradation horizons of U4a. The bulk shape of U4a and U4b is also gullies that are indenting the continental slope break (see Fig. 1C for similar. However, as the target of the seismic survey was the upper part location of seismic profiles). D1 is scarcely sampled by the present of the Tagus delta the lower part is not well covered as it lies upstream seismic dataset thus not allowing for producing maps. of the surveyed area. Discontinuity D2 is imaged in the seismic profiles as a very con- Two-way time isopach maps of seismic units U3 (Fig. 5) and U4 tinuous high amplitude reflection. D2 is an erosive surface that cuts (Fig. 6 A–C) were produced by subtracting two way travel time depths through folded and faulted strata of seismic unit U1 and channels and of discontinuities D3-D2 and seafloor-D3, respectively. Seismic unit U3 horizons of seismic unit U2 (Figs. 2 and 3). The isochrones map of has a maximum thickness of 20 ms TWT, i.e. approximately, 16 m thick.

7 P. Terrinha, et al. Marine Geology 416 (2019) 105983

Fig. 5. Isopach maps of seismic unit U3 (thickness in TWT milliseconds). Note that the two depocenters are separated by a W-E trending ridge-like structure, suggesting a drainage system towards the canyon to the southeast with an internal lagoon or ria-like feature in the north. It is possible that this ridge and the structural jogs (see Fig. 2) constituted environmental barriers associated to Quaternary low stand sea levels (see Discussion).

Thickness variation of unit U3 down to nil is essentially controlled by stratigraphy of the delta forming the main landslide detachment fault, topographic variations of the inherited paleo-topography of its base an extensional listric fault that becomes parallel to the autochthonous (D2), while the top of U3 is essentially parallel to the internal reflec- stratigraphy, i.e. the landslide basal detachment (see inset in Fig. 2A tions, meaning that U3 is not incised by D3. and B). Isopach maps of U4, U4a and U4b (Fig. 6A, B and C, respectively) The upper part of the landslide detachment fault cuts at a high angle are coherent with the geometry of the delta as depicted by the bathy- the delta front forming a typical headscarp at the base of which bodies metry (see Fig. 1B). They show a thickest part (~65 ms TWT, ap- of transparent acoustic facies are observed. The transparent acoustic proximately 52 m thick) at the delta front rapidly thinning towards the character is an indication of lack of stratification probably due to abrupt prodelta; the delta platform is also thinner than the delta front. The deposition of fallen heterogeneous material from the top of the head- frontal lobe area of U4b is longer, more arcuate and thicker than the scarp (Fig. 2). equivalent of U4a. The frontal lobe of U4b is located downstream with The chaotic aspect is given by vertically and laterally juxtaposed respect to the equivalent part of U4a, which is consistent with growth of discontinuous high and low amplitude reflections intermingled with the delta towards the offshore (cf. Figs. 1B and 2). transparent areas. Detachment faults are displayed as high amplitude continuous curvilinear discontinuities separating bodies of the chaotic facies. These are stacked slid units and/or highly extended or dis- 4.2. The Tagus delta landslides membered parts of the main landslide (Fig. 2). The top surface of the landslide is irregular as it wraps around dismembered and/or amalga- Groups of chaotic reflections and transparent areas comprehended mated mass transported bodies of sediment of the delta front. between well-defined bottom curved discontinuities and a top irregular The mapped Tagus landslide is > 9 km of length (parallel to the surface, both with good lateral continuity are observed within U4b. delta front) and > 3 km of width with a maximum thickness of ~20 m Identical seismic facies are observed within U4a; however, the present (> 24 ms TWT, Fig. 7) with a volume of approximately 0.27 km3. The seismic database does not allow for ensuring their lateral continuity. top of headscarp and toe of the landslide lie, respectively, at approxi- These types of seismic facies are typical of landslides (Figs. 2 and 3). mately 45 m and 105 m below present-day sea level. The chaotic-transparent facies units are limited at their base by a smooth erosive concave upward surface, i.e. the landslide detachment that flattens towards the prodelta and steepens towards the inner part 4.3. Gas-bearing sediments of the delta, defining the landslide headscarp (Fig. 2B blue line). This surface cuts progressively downslope through the autochthonous The observed acoustic blanking, acoustic turbidity, the presence of

8 P. Terrinha, et al. Marine Geology 416 (2019) 105983

Fig. 6. Isopach maps of the Tagus delta seismic units (thickness in TWT milliseconds), U4 (A) and its sub-units U4a (B) and U4b (C). Note that U4b is thicker, longer (along strike length) than U4a; U4a depocenter lies SW of U4a depocenter consistently with the prograding sedimentation facies (cf. Fig. 2). bright spots and shallow enhanced reflectors (Fig. 8) are indicators of established by comparison with the adjacent onshore geology directly the presence of shallow gas the in sediments (Davis, 1992; Judd and observed along the coast on both river banks, as shown in Fig. 1B and Hovland, 1992). The sediments containing shallow gas cover a total with the three available cores and respective 14C ages (Table 1). area of 33 km2 (Fig. 1C). Inspection of the onshore field geology, whose stratigraphy is shown Acoustic blanking is the most common evidence of shallow gas and in Fig. 1, allows inferring that seismic unit U1 consists of folded and it typically starts at the shallowest stratigraphic levels, within the top faulted Cretaceous, Paleogene and Miocene formations. These units 10 m of sediments (Fig. 8). The shallow gas seismic evidences are ob- were affected by the polyphase Alpine orogenic shortening that ac- served in a restricted area, at the northwestern part of the delta, located counted for thrusting and strike-slip faulting (Fig. 1B) (Terrinha et al., between 20 and 80 m water depth (Fig. 1), i.e. approximately the 2017). Discontinuities within U1 (Fig. 2) probably can thus correspond maximum water depths of the landslide. It is worthwhile noting that in to Cretaceous unconformities (such as lava flows on top of sediments), some areas the gas is trapped under impermeable layers and thus must or Paleogene and Miocene age unconformities depicted as geological be overpressured (Fig. 8). contacts in the geological map of Fig. 1B. The areas of occurrence of shallow gas and the Tagus delta landslide The occurrence of fluvial and shallow marine deposits of fine within unit U4b show that they are almost mutually exclusive. Both, grained to coarse siliciclastic of Pliocene and Quaternary formations shallow gas and landslide cover equivalent areas in size, with 3% of (Fig. 1B) south of the Tagus River is extensive and its base is an erosive overlap only (Fig. 1C). The possibility that the shallow gas acoustic unconformity. They cut down through the Miocene to Mesozoic sedi- blanking could be concealing the mapped landslide cannot be totally ments either as thin, patchy deposits on elevated erosive surfaces or as ruled out. However, considering that, i) where the gas does not have a thick deposits within valleys, the most developed of which are the full blanking effect landslides were not observed in the northwestern Tagus and Sado rivers estuaries and their tributaries (Fig. 1B). Ac- part of the delta, and ii) the tipping out of the landslide thickness to- cordingly, discontinuity D1 (Fig. 2) is interpreted as the equivalent of wards the northwest (Fig. 8) it is suggested that there are no large the base of the onshore Pliocene-Quaternary formations both by its landslides in the gas area. erosive nature and by the filling of the submarine valleys by seismic unit U2 (see Q contacts in Fig. 1B). Seismic unit U2 is heterogeneous both horizontally and vertically, 5. Discussion indicating varied lithological constitution and lateral facies change. The transparent units of U2 have significant lateral thickness variations and 5.1. Chrono-stratigraphic model injections into the overlying well-layered units, suggesting syn-sedimentary deformation of the transparent units and forced space The chronostratigraphic constraints of these seismic units are

9 P. Terrinha, et al. Marine Geology 416 (2019) 105983

Fig. 6. (continued) accommodation of the overlying ones (Fig. 3B). The disharmonic This age is compatible with the stratigraphic calibration (Fig. 9) using folding is also consistent with rheological variations of alternating information from core D13882 showing that U3 deposits and D3 dis- poorly sorted water saturated sediments with well bedded sediments. continuity lie well below the oldest cored sediments of 13 ky BP Seismic unit U2 is thus interpreted as the offshore correlative of the (Table 1). Hence, D2 is interpreted as a transgressive ravinement sur- onshore Pliocene-Quaternary formations. Internal unconformities face sculpted during various Quaternary eustatic cycles including the within U2 are compatible with various generations of erosion and infill LGM. possibly associated with several sea level low stands and/or environ- The Tagus delta unit (U4) that comprises the delta and pro-delta of mental changes. Folds and faults within U2 do not have any corre- the Tagus River, it lies on top of D4, a smooth slightly convex surface spondence with deformation structures within U1. This is again com- with a shallow dip (~0.5°) towards the sea. D4 can thus be interpreted patible with soft sediment deformation mechanisms associated with as a high-stand unconformity, possibly the maximum inundation un- gravity driven processes, such as mass movements within the channels, conformity formed during the post LGM sea level rise. fluidization during earthquakes or differential compaction. Comparison of geometric patterns between U4a and U4b isopachs Inspection of the onshore geology has not shown a unique strati- (Fig. 7) show an oceanwards and northwestwards development of the graphic unit equivalent to unit U3 so a direct age by correlation with delta with time. Growth towards the open ocean is coherent with onshore deposits is not unequivocal. However, because U3 lies un- progradation of the river sediment deposition in a high-stand sea level conformably on top of the channels infill of U2 that are supposed to be position. Development of the delta towards the northwest suggests in- of Pliocene through Quaternary age, a Late Pleistocene age is proposed. fluence of the littoral currents from southeast to northwest that ac- The seismic facies of unit U3 made of chaotic to short lateral con- counted for the growth of the sand spits and sand banks deposited in the tinuity reflections, occasionally interlayered with thin continuous re- last 2 to 3 ky on the south banks of the Tagus and Sado Rivers. These flections, suggests a poorly organized internal structure, i.e. possibly currents result from interference of the Roca and Espichel capes with made of coarse clastic materials, blocks and boulders interbedded with the main NW to SE currents that create anti-clockwise whirls at theses sands strands These type of deposits are found as on the nearby bea- rivers mouths (see Fig. 1B). ches, where these deposits cover directly a wave cut surface on Jurassic- Although core D13882 that yielded the oldest age in the Tagus Cretaceous and Miocene rocks. Internal planar erosive surfaces overlain prodelta area, 13 ky BP, (Table 1, Fig. 1) lies near the delta front, it is by aggradational deposits can correspond to beach-rock deposits or not possible to unequivocally bring this stratigraphic horizon into the remnants of old beaches, thus suggesting various depositional cycles Tagus delta survey area due to the existence of erosive truncations. (Figs. 2 and 3). The onlap terminations towards the coast on top of D2 Thus, a maximum age for the lowest sediments of unit U4 of the Tagus indicates deposition during sea level rise transgressive cycles. It is delta in the study area is proposed based on the position of the sea level possible that the uppermost parts of unit U3 were deposited during the at approximately 80 m bsl (~110 ms TWT) as shown in Fig. 2B (sea last transgressive cycle, i.e. after the Last Glacial Maximum (~18 ky). level curve inset), which is ~17 ky BP according to de Boer et al.

10 P. Terrinha, et al. Marine Geology 416 (2019) 105983

Fig. 6. (continued)

(2010). Following the same rationale, the top of U4a, i.e. the older sub- than the landslide. unit of the Tagus delta, lying at ~55 m bsl (~70 ms TWT), would be ~13 ky BP. 5.3. The Tagus Delta Landslide

5.2. Origin of gas in sediments 5.3.1. Age The stratigraphic calibration of the seismic profile TD3200 (Fig. 9) The shallowest evidences of gas occur just below the seafloor re- using the closest located core (D13902 in Table 1) shows that the 8 ky flection or bound by the top of the landslide or by a specific strati- BP horizon drapes the frontal part of the main landslide within U4b, graphic horizon at approximately 10 m below sea floor (see Fig. 8 in indicating an age just older than ~8 ky cal BP, a time when the sea level this work and Fig. 15 in Neres et al., 2014). This shallow setting of the was approximately 11 m below present day sea level (Fig. 2b), ac- gassy sediments can be interpreted as an indicator of a probable mi- cording to de Boer et al., (2010). This implies that the main Tagus delta crobial origin of the gas. However, this does not preclude a deeper landslide was a submarine landslide because the top of its headscarp source for the origin of gas. The reported (Gaspar and Monteiro, 1977) lies at around 45–60 m below present day sea level (for sea level curve organic carbon content of superficial sediments in the Tagus delta area see inset in Fig. 2B). range between 0.5 and 2.3% (dry weight %). The organic content of the The landslides within unit U4a would have occurred between 17 ky pelagic sediments in this high sedimentation rate area can be sufficient, and 13 ky according to the proposed ages for this seismic unit. according to Rice and Claypool (1981), to generate microbial methane in shallow sediments. The observed absence of shallow gas in most of 5.3.2. Geometry the area of the landslide can have several interpretations. If we assume The Tagus delta landslide dimensions, ~11 km in length, > 3 km a deep source for the gas (which can only be tested through organic and width and ~20 m maximum thickness, with a volume of approximately isotopic geochemistry), then the landslide and the cover sediments have 0.27 km3, and external geometry are shown in Fig. 6. Four features are to be sufficiently permeable in order to allow the escape of the as- worthwhile pointing out, firstly, the contrast of the outline of the 5 ms cending gas. If the gas is of shallow microbial origin than at the places contour at the headscarp and toe zones of the landslide, secondly, the where the gas is trapped at some meters of depth, the shallow sediments interruption of the 5 ms contour in the central part of the landslide, have to be impermeable to trap it. In any case, either the shallow se- thirdly, a thicker central part and fourthly the open contours towards diments in the landslide and gas areas have different permeability or the NW. the gas is only being produced in the northern part of the delta. The 5 ms contour in the headscarp zone is fairly linear if compared The clear evidence of gas trapped underneath the stratigraphic top with the equivalent toe zone of the landslide, indicating a fairly curvi- of the landslide together with gas hosted at very close to the seafloor planar steep headscarp fault surface and gradual shallowing and thin- sediments (Fig. 8) indicates that the accumulation of gas is more recent ning out of the toe of the landslide. The 5 ms contours in the landslide

11 P. Terrinha, et al. Marine Geology 416 (2019) 105983

Fig. 7. Two way travel time isopach map of the landslide (thickness in millisecond TWT). Note the landslide has two main thicker areas and it is wider towards the central part of the delta thinning out towards the Lisbon canyon. toe also depict a roughly radial pattern that can be interpreted as radial and various landslides within and tens of kilometers beyond the Lisbon- spreading of the slid material. This is compatible with the interruption Cascais canyons systems based on side scan imagery attesting for recent of the 5 ms contour in the central part of the landslide, suggesting lat- mass transport processes. Masson et al. (2011) identified two episodes eral spreading during the slide process and immediately afterwards. of widespread landsliding in the Setúbal and Cascais canyons (see The thickest area of the landslide lies closer to the headscarp than to Fig. 1b for location) simultaneously with the emplacement of turbi- the toe zone. Inspection of the seismic profiles shows that thickening is dites > 1.5 m thick in the Tagus Abyssal Plain at ~6600 and accommodated by stacking up of slid bodies where listric extensional ~8100 Cal yr BP. According to these authors, the synchronous nature of faults become thrusts at their toe (Fig. 2). Individual duplexes of slid such events and their great extent requires a regional trigger such as an bodies in the thickest part of the landslide are at a distance of ap- earthquake. proximately 600 m of the headscarp suggesting this was a maximum The relationship between submarine landslide triggering and travelled distance of these bodies. The latest slid bodies filled up the earthquakes is addressed by ten Brink et al. (2009) who present the landslide scar suggesting that most of the space left by the slide was relationship between the maximum distances from a rupturing fault to filled up by collapsed structureless sediments. sites where submarine failure is expected to occur. This distance in- The open 5 and 10 ms contours towards the NW are in contrast with creases with the magnitude of the parent earthquake and with seabed the equivalent ones to the SE, suggesting that the Tagus landslide ex- slope. The study shows that the maximum triggering distance changes tends further to the NW. with seabed slope for the same earthquake magnitude. For example, a magnitude 7.5 earthquake, this distance varies from 60 km to 100 km for seabed slopes of 2° and 6° respectively (ten Brink et al., 2009). 5.3.3. Trigger mechanism Considering these results together with the chronological calibra- According to Hampton et al. (1996) submarine landslides may occur tion of the seismic data, we can infer that earthquakes of magnitude up fl in areas with thick sedimentary deposits, sloping sea oor and high to 8.0 located either in the Gloria Fault area or in the Horseshoe Abyssal fi fi “ ” environmental stresses. Lee (2009) de nes ve landslide territories plain, located in the central part of the Northeast Atlantic, are a very one of them being active river deltas. This author proposes a number of unlikely cause for landslide at the Tagus delta. This is because those ff di erent trigger mechanisms such as gas hydrate dissociation, excess earthquake source areas are located at > 600 km and 300 km, respec- pore pressure in the sediments, and salt movement, but earthquakes tively. Nevertheless, the Estremadura Spur (morphologic high west of and climate change have also been invoked as the most common trigger Sintra Magmatic Complex, see Fig. 1 for location) is a low to moderate mechanism for submarine landslide generation in continental margins magnitude seismicity cluster, which comprehends the Tagus landslide (Masson et al., 2006). area, is a source for submarine ground shaking. Lastras et al. (2009) reported the existence of important scouring

12 P. Terrinha, et al. Marine Geology 416 (2019) 105983

Fig. 8. Gas in the Tagus river delta. A) MCS line with gas hosted on the shallowest sediments casting a blanking eﬀect below, AT- Acoustic turbidity, AB- acoustic blanking; B) MCS line with (A) gas virtually seeping at seaﬂoor and (B) gas trapped within the landslide sediments.

Important climate changes have been reported in Early Holocene significant mass of water leading to the generation of tsunami waves. times. As an example, the 8.2 ka event marked by the onset of cool Large landslides involving sediment volumes of hundreds to thousands summers and dry winters can be mentioned (Gràcia et al., 2010). These of km3 such as the 8150 BP Storegga and the 1929 Grand Banks and the instability conditions associated with the sea level rise have (Bondevik et al., 2005; Fine et al., 2005) can generate transoceanic certainly imposed unstable conditions on the evolution of the Tagus tsunamis. However, more often landslides occur with smaller volumes, delta. and produce only local tsunamis (Harbitz et al., 2014). The character- It could be argued that the speculative age of > 8000 yr assigned to istics of a tsunami generated by a submarine landslide are mainly de- the Tagus delta landslide is in agreement with the time window of termined by the volume mobilized, the initial acceleration, the max- occurrence of landslide and turbidite events recognized by Masson et al. imum velocity, and the water depth (Harbitz et al., 2006). The distance (2011). Considering this, it is possible to speculate about a linkage to the shoreline, on the other hand, controls the level of coastal impact. between this landslide and the ~8100 Cal yr BP episode referred by In the southwest Iberia Margin region, damage to the submarine Masson et al. (2011) suggesting an earthquake as a probable triggering communication cables between Brest (France) – Casablanca (Morocco) mechanism. However, at the present state of knowledge it is not known, and Brest (France) – Dakar (Senegal) was reported after the M8.3 1941 whether earthquakes, tsunami waves, floods, storms or simply the local Gloria earthquake (Debrach, 1946) and was attributed to a possible environmental conditions (balance between hydraulic regime and se- local underwater landslide (Baptista et al., 2016). Moreover, the in- dimentation rate, and nature of the sediment containing gas) were the vestigation of submarine mass failures (SMFs) in the deep-water of main cause of landsliding in the Tagus delta. southwest Iberia Margin region has showed the importance of large SMFs close to Gorringe Bank and Hirondelle Seamount as a significant contributor to tsunami hazard along the Iberian coast (Lo Iacono et al., 5.4. Landslide-related hazard 2012; Omira et al., 2016a). Earthquake activity is one of the most likely pre-conditioning and Submarine landslides account as a major source of marine geo-ha- triggering mechanisms of submarine landslides (Masson et al., 2006). zards. They have the potential to significantly change the morphology of Many large historical episodes, recognized by their consequences (i.e. the seafloor and, in some cases, to damage offshore infrastructures, breaking communication cables or causing the collapse of river deltas particularly communication cables (Fine et al., 2005). The large sub- or tsunamis), have followed earthquakes (Tappin et al., 2001; Fryer marine slope failure (200 km3), that followed the M = 7.2 Grand Banks et al., 2004; Fine et al., 2005; Masson et al., 2006). The proximity of the earthquake, was responsible of breaking 12 telegraph cables (Fine et al., Tagus Delta from highly seismic active zones (i.e. the Gulf of Cadiz) 2005). The sudden failure of a submarine slope can also perturb a

13 P. Terrinha, et al. Marine Geology 416 (2019) 105983

Fig. 9. Chrono-stratigraphic calibration of seismic profile TD3200 using core D13902 (Table 1 and Fig. 1C for location). Calibration using other cores could not be used due to erosive truncations. 8 ky is a minimum age as there are sediments deposited between the top of the landslide and the 8 ky horizon that drapes the crest of the landslide. favours the occurrence of future events in the region. The occurrence of 2016b). Moreover, further investigation of the tsunami hazard from the a submarine landslide with similar dimensions of the Tagus landslide Tagus Delta landslide must take into consideration the observed char- (10 km × 3.5 km × 20 m) in the present-day delta configuration may acteristics of the deposited sediments and therefore employ adequate have a destructive impact on communication cables present in the re- numerical models for tsunami generation and impact prediction. gion. Moreover, the dimension of such a submarine landslide and its source location (delta front head lying at 30–40 m below present-day 6. Conclusions mean sea level), favor the generation of a tsunami following a sudden failure. Therefore, we cannot rule-out the hypothesis of tsunamigenic The research carried out described and discussed above allow for landslides in the Tagus Delta front head and their consequent tsunami the following direct conclusions: hazard at Lisbon coasts that stand few kilometers from the source zone. Further work is required to investigate the tsunamigenic potential of i) The Tagus ebb-tidal delta started its formation not before 17 ky BP submarine mass failure occurring in the Tagus Delta front mainly in and it consists of two main stratigraphic units, the younger of which what concerns mapping the total extension of the Tagus landslide, de- (U4b) initiated its formation at approximately 13 ky BP when sea termining the origin of the gas in sediments and extent of the gas level was at ~55 m below present-day sea level. overpressured area. ii) The upper delta unit (U4b) revealed the existence of a large land- It is also worth to mention here that the mapped landslide deposit slide, of ~11 km × 3.5 km × 0.02 km dimensions, which accounted presents a reduced run-out. Submarine landslides of such a small run- for the collapse of approximately half of the frontal part of the delta. out, if tsunamigenic, often led to the generation of tsunamis with re- The age of the Tagus delta landslide is estimated at just over 8 ky latively short wavelengths and prominent dispersion (Løvholt et al., BP, based on core data from previous studies. 2015). In these cases, locally focused tsunami impact is expected since iii) The lower deltaic unit U4a contains several kilometer scale land- short wavelength tsunamis undergo rapid energy dissipation when slides that could not be mapped with the present dataset indicating propagating in the open ocean towards far-field coasts (Omira et al., continuous instable conditions of the delta.

14 P. Terrinha, et al. Marine Geology 416 (2019) 105983 iv) Shallow gas occurs in the northern half of the delta front. In most of de Boer, B., van de Wal, R.S.W., Bintanja, R., Lourens, L.J., Tuenter, E., 2010. Cenozoic the area the gas occurs at the seafloor but in some areas it is global ice-volume and temperature simulations with 1D ice-sheet models forced by benthic delta-18O records. Ann. Glaciol. 51, 23–33. probably overpressured as it is trapped under ~8 m of sediments, Debrach, J., 1946. Raz de marée d’origine sismique enregistré sur le littoral atlantique du which can ease landsliding by reducing the normal stress of sedi- Maroc. Service de Physique du Globe et de l’institut scientifique Chérifien, Annales, ment load. Maroc. Dinis, J.L., Rey, J., Cunha, P.P., Callapez, P., Dos Reis, R.P., 2008. Stratigraphy and al- v) The trigger mechanism for the Tagus landslide is still to ascertain. logenic controls of the western Portugal cretaceous: an updated synthesis. Cretac. The 8 ky BP age for the Tagus landslide is compatible with large Res. 29 (5), 772–780. turbidites at the mouth of the Lisbon and Cascais canyons in the Ferraz, M., 2014. Infl uence of Delta Morphodynamics on Coastal Response to Climate Tagus Abyssal Plains reported by Masson et al. (2011), who sug- Change. Ph.D. thesis. Faculty of Science, University of Sydney. Fine, I.V., Rabinovich, A.B., Bornhold, B.D., Thomson, R.E., Kulikov, E.A., 2005. The gested an earthquake cause for these. The parameters of distance Grand Banks landslide-generated tsunami of November 18, 1929: preliminary ana- and slope versus earthquake magnitude provided by ten Brink et al. lysis and numerical modeling. Mar. Geol. 45–57 (215 (1-2 SPEC ISS). (2009) suggest that the Africa-Iberia plate boundary tectonic Fortunato, A., Baptista, A.M., Luettich Jr., R.A., 1997. A three-dimensional model of tidal currents in the mouth of the Tagus estuary. Cont. Shelf Res. 17, 1689–1714. sources in the Atlantic could be excluded. However, the proximal Fortunato, A., Oliveira, A., Baptista, A.M., 1999. On the effect of tidal flats on the hy- seismicity off Lisbon, only tens of kilometers away from the Tagus drodynamics of the Tagus estuary. Oceanol. Acta 22, 31–44. Fryer, G.J., Watts, P., Pratson, L.F., 2004. Source of the great tsunami of 1 April, 1946: a delta cannot be excluded as a possible cause (Custódio et al., 2015). – – fi landslide in the upper Aleutian forearc. Mar. Geol. 203 (3 4), 201 218. Our ndings of sediment bearing gas insert a new parameter that is Gaspar, L.C., Monteiro, J.H., 1977. Matéria orgânica nos sedimentos da plataforma not straightforward to interpret. Further work is needed in order to continental entre os cabos Espichel e Raso. Commun. Serv. Geol. Port. 69–83. map the area containing overpressured gas and the origin of the gas. Gràcia, E., Vizcaino, A., Escutia, C., Asioli, A., Rodes, A., Pall, R., Garcia Orellana, J., Lebreiro, S., Goldfinger, C., 2010. Holocene earthquake record offshore Portugal (SW vi) The reported occurrence of one major landslide in the shallow Iberia): Testing turbidite paleoseismology in a slow convergence margin. Quat. Sci. water Tagus delta of ~8 ky and various kilometer scale landslides Rev. 29, 1156–1172. with ages estimated between 13 ky and 17 ky should be considered Hampton, M.A., Lee, H.J., Locat, J., 1996. Submarine landslides. Rev. Geophys. 34 (1), ff 33–59. as a potential source for marine geo-hazard because these can a ect Harbitz, C.B., Lovholt, F., Pedersen, G., Masson, D.G., 2006. Mechanisms of tsunami submarine cables and cause tsunami threating coastal facilities and generation by submarine landslides: a short review. Nor. J. Geol. 86, 255–264. population. Harbitz, C.B., Løvholt, F., Bungum, H., 2014. Submarine landslide tsunamis: how extreme and how likely? Nat. Hazards 72 (3), 1341–1374. Hill, P., 2012. Changes in submarine channel morphology and slope sedimentation pat- Acknowledgements terns from repeat multibeam surveys in the Fraser River delta, western Canada. International Association of Sedimentologists. Special Publication 44, 47–70. The following projects are acknowledged: TAGUSDELTA (Fundação Hill, P.R., Christian, H.A., 2003. Monitoring in situ pore pressures for prediction of slope failure on the prodelta slope of the Fraser River Delta, Canada. In: Geophysical para a Ciência e Tecnologia, Portugal; PTDC/MAR/113888/2009), Research Abstracts, EGS/AGU/EUG Joint Assembly, 5, 13783. France, Nice. PACEMAKER (Seventh Framework Programme, European Union - FP7/ Hughes Clarke, J.E., Brucker, S., Muggah, J., Church, I., Cartwright, D., Kuus, P., 2007-2013/ERC grant agreement 226600), ASTARTE (Seventh Hamilton, T., Pratomo, D., Eisan, B., 2012. The Squamish Prodelta, monitoring active landslides and turbidity currents. Canadian Hydrographic Conference 2012, 1–15. Framework Programme, European Union; FP7- Coasts at threat in Hughes Clarke, J.E., Videra Marques, C.R., Pratomo, D., 2014. Imaging active mass Europe: tsunamis and climate-related risks; ENV.2013.6.4-3; Grant wastingon a fjord delta, Squamish, British Columbia. In: Krastel, S., Behrmann, J.-H., agreement no: 603839), TAGUSGAS (Fundação para a Ciência e Völker, D., Stipp, M., Berndt, C., Urgeles, R., Chaytor, J., Huhn, K., Strasser, M., Harbitz, C.B. (Eds.), Submarine Mass Movements and Their Consequences, Advances Tecnologia, Portugal; PTDC/CTA-GEO/031885/2017) and in Natural and Technological Hazards Research Vol. 37. Springer International MAGICLAND (Fundação para a Ciência e Tecnologia, Portugal; PTDC/ Publishing, Switzerland, pp. 249–260. CTA-GEO/30381/2017). Judd, A.G., Hovland, M., 1992. The evidence of shallow gas in marine-sediments. Cont. Shelf Res. 12, 1081–1095. Kim, J.W., Chough, S.K., 2000. A gravel lobe deposit in the prodelta of the Doumsan fan References delta (Miocene), SE Korea. Sediment. Geol. 130, 183–203. Lantzsch, H., Hanebuth, T.J.J., Bender, V.B., Krastel, S., 2009. Sedimentary architecture of a low-accumulation shelf since the Late Pleistocene (NW Iberia). Mar. Geol. 259, Abrantes, F., Lebreiro, S., Rodrigues, T., Gil, I., Bartels-Jonsdottir, H., Oliveira, P., Kissel, 47–58. C., Grimalt, J.O., 2005. Shallow-marine sediment cores record climate variability and Lastras, G., Arzola, R.G., Masson, D.G., Wynn, R.B., Huvenne, V.A.I., Hühnerbach, V., earthquake activity off Lisbon (Portugal) for the last 2000 years. Quat. Sci. Rev. 24, Canals, M., 2009. Geomorphology and sedimentary features in the Central 2477–2494. Portuguese submarine canyons, Western Iberian margin. Geomorphology 103, Abrantes, F., Alt-Epping, U., Lebreiro, S., Voelker, A., Schneider, R., 2008. 310–329. Sedimentological record of tsunamis on shallow-shelf areas: the case of the 1969 AD Lee, H.J., 2009. Timing of occurrence of large submarine landslides on the Atlantic Ocean and 1755 AD tsunamis on the Portuguese Shelf off Lisbon. Mar. Geol. 249, 283–293. margin. Mar. Geol. 264 (1), 53–64. Alt-Epping, U., Stuut, J.-B.W., Hebbeln, D., Schneider, R., 2009. Variations in sediment Lo Iacono, C.L., Gràcia, E., Zaniboni, F., Pagnoni, G., Tinti, S., Bartolomé, R., Zitellini, N., provenance during the past 3000 years off the Tagus River, Portugal. Mar. Geol. 261, 2012. Large, deep-water slope failures: implications for landslide-generated tsunamis. 82–91. Geology 40 (10), 931934. Baptista, M.A., Miranda, J.M., Omira, R., Antunes, C., 2011. Potential inundation of Løvholt, F., Pedersen, G., Harbitz, C.B., Glimsdal, S., Kim, J., 2015. On the characteristics Lisbon downtown by a 1755-like tsunami. Nat. Hazards Earth Syst. Sci. 11 (12), of landslide tsunamis. Philos. Trans. R. Soc. A Math. Phys. Eng. Sci. 373 (2053), 3319–3326. 20140376. Baptista, M.A., Miranda, J.M., Batlló, J., Lisboa, F., Luis, J., Maciá, R., 2016. New study Luis, J.F., 2007. MIRONE: a multi purpose tool for exploring grid data. Comput. Geosci. on the 1941 Gloria Fault earthquake and tsunami. Nat. Hazards Earth Syst. Sci. 16, 33, 31–41. 1967–1977. https://doi.org/10.5194/nhess-16-1967-2016. Lykousis, V., Sakellariou, D., Roussakis, G., 2003. Prodelta slope stability and associated Bartels-Jónsdóttir, H.B., Knudsen, K.L., Abrantes, F., Lebreiro, S., Eiríksson, J., 2006. coastal hazards in tectonically active margins: Gulf of Corinth (NE Mediterranean). Climate variability during the last 2000 years in the Tagus Prodelta, western Iberian In: Locat, J., Mienert, J. (Eds.), Submarine Mass Movement and their Consequences. Margin: benthic foraminifera and stable isotopes. Mar. Micropaleontol. 59, 83–103. Kluwer Academic Publishers, pp. 433–440. Bondevik, S., Løvholt, F., Harbitz, C., Mangerud, J., Dawson, A., Inge Svendsen, J., 2005. Lykousis, V., Roussakis, G., Sakellariov, D., 2009. Slope failure and stability analysis of The Storegga Slide tsunami-comparing field observations with numerical simulations. shallow water prodeltas in the active margins of western Greece, northern Mar. Pet. Geol. 22 (1), 195–208. Mediterranean Sea. Int. J. Earth Sci. 98, 807–822. Burdloff, D., Araújo, M.F., Jouanneau, J.M., Mendes, I., Monge Soares, A.M., Dias, J.M.A., Maillet, G.M., Vella, C., Berné, S., Friend, P.L., Amos, C.L., Fleury, T.J., Normand, A., 2008. Sources of organic carbon in the Portuguese continental shelf sediments during 2006. Morphological changes and sedimentary processes induced by the December the Holocene period. Appl. Geochem. 23, 2857–2870. 2003 flood event at the present mouth of the Grand Rhône River (southern France). Correggiari, A., Trincardi, F., Langone, L., Roveri, M., 2001. Styles of failure in Late Mar. Geol. 234, 159–177. Holocene highstand prodelta wedges on the Adriatic shelf. J. Sediment. Res. 71 (2), Masson, D.G., Harbitz, C.B., Wynn, R.B., Pedersen, G., Løvholt, F., 2006. Submarine 218–236. landslides: processes, triggers and hazard prediction. Philos. Trans. R. Soc. A Math. Custódio, S., Dias, N.A., Carrilho, F., Góngora, E., Rio, I., Marreiros, C., Morais, I., Alves, Phys. Eng. Sci. 364 (1845), 2009–2039. P., Matias, L., 2015. Earthquakes in western Iberia: improving the understanding of Masson, D.G., Arzola, R.G., Wynn, R.B., Hunt, J.E., Weaver, P.P.E., 2011. Seismic trig- lithospheric deformation in a slowly deforming region. Geophys. J. Int. 203 (1), gering of landslides and turbidity currents offshore Portugal. Geochem. Geophys. 127–145. https://doi.org/10.1093/gji/ggv285. Geosyst. 12 (12). Davis, A.M., 1992. Shallow gas - an overview. Cont. Shelf Res. 12, 1077–1079. Mello, U.T., Pratson, L.F., 1999. Regional slope stability and slope failure mechanics from

15 P. Terrinha, et al. Marine Geology 416 (2019) 105983

the two-dimensional state of stress in an infinite slope. Mar. Geol. 154, 339–356. Q07U06. https://doi.org/10.1029/2008GC002367. Mil-Homens, M., Branco, V., Vale, C., Boer, W., Alt-Epping, U., Abrantes, F., Vicente, M., Rodrigues, T., Grimalt, J.O., Abrantes, F., Flores, J., Lebreiro, S., 2011. Decadal-to-cen- 2009. Sedimentary record of anthropogenic metal inputs in the Tagus prodelta tennial scale variations during the Holocene reconstructed from sediment core (Portugal). Cont. Shelf Res. 29, 381–892. D13902. PANGAEA. https://doi.org/10.1594/PANGAEA.762035. (Supplement to: Miranda, R., Valadares, V., Terrinha, P., Mata, J., Azevedo, M.R., Gaspar, M., Kullberg, Rodrigues, T et al. (2009): Holocene interdependences of changes in sea surface J.C., Ribeiro, C., 2009. Age constraints on the Late Cretaceous alkaline magmatism on temperature, productivity, and fluvial inputs in the Iberian continental shelf (Tagus the West Iberian margin. Cretac. Res. 30, 575–586. mud patch). Geochemistry, Geophysics, Geosystems, 10(7), Q07U06, doi:10.1029/ Mulder, T., Syvitski, J.P.M., Migeon, S., Faugères, J.-C., Savoye, B., 2003. Marine hy- 2008GC002367). perpycnal flows: initiation, behavior and related deposits. A review. Mar. Pet. Geol. Sahal, A., Lemahieu, A., 2011. The 1979 Nice airport tsunami: mapping of the flood in 20, 861–882. Antibes. Nat. Hazards 56 (3), 833–840. Nemec, W., 1990. Aspects of sediment movement on steep delta slopes. In: Colella, A., Silva, T.A.A., 2013. Sedimentologia e morfodinâmica de sapais do estuário do Tejo em Prior, D.B. (Eds.), Coarse-Grained Deltas. vol. 10. International Association of cenários de alteração climática. Master thesis. Faculdade de Ciências, Departamento Sedimentologists Special Publication, pp. 29–73. de Geologia, Universidade de Lisboa. Neres, M., Bouchez, J.L., Terrinha, P., Font, E., Moreira, M., Miranda, R., Launeau, P., Sultan, N., Savoye, B., Jouet, G., Leynaud, D., Cochonat, P., Henry, P., Stegmann, S., Kopf, Carvallo, C., 2014. Magnetic fabric in a cretaceous sill (Foz da Fonte, Portugal): flow A., 2010. Investigation of a possible submarine landslide at the Var delta front (Nice model and implications for regional magmatism. Geophys. J. Int. 199, 78–101. slope - SE France). Can. Geotech. J. 47 (4), 486–496. Nichols, G., 2009. Sedimentology and Stratigraphy. John Wiley & Sons. Tappin, D.R., Watts, P., McMurty, G.M., Lafoy, Y., Matsumoto, T., 2001. The Sissano, Omira, R., Ramalho, I., Terrinha, P., Baptista, M.A., Batista, L., Zitellini, N., 2016a. Deep- Papua New Guinea tsunami of July, 1998—offshore evidence on the source me- water seamounts, a potential source of tsunami generated by landslides? The chanism. Mar. Geol. 175 (1–4), 1–24. Hirondelle Seamount, NE Atlantic. Mar. Geol. 379, 267–280. ten Brink, U.S., Lee, H.J., Geist, E.L., Twichell, D., 2009. Assessment of tsunami hazard to Omira, R., Baptista, M.A., Lisboa, F., 2016b. Tsunami characteristics along the Peru–Chile the US East Coast using relationships between submarine landslides and earthquakes. trench: analysis of the 2015 Mw8. 3 Illapel, the 2014 Mw8. 2 Iquique and the 2010 Mar. Geol. 264 (1), 65–73. Mw8. 8 Maule tsunamis in the near-field. Pure Appl. Geophys. 173 (4), 1063–1077. Terrinha, P., Emilio, L., Pueyo, E., Aranguren, A., Kullberg, J.C., Kullberg, M.C., Casas- Pais, J., Cunha, P.P., Pereira, D., Legoinha, P., Dias, R., Moura, D., Brum da Silveira, A., Sainz, A., Azevedo, M.A., 2017. Gravimetric and magnetic fabric study of the Sintra Kullberg, J.J., Gonzálex-Delgado, J.A., 2012. The Paleogene and Neogene of Western Igneous complex: laccolith-plug emplacement in the Western Iberian passive margin. Iberia (Portaugal). Springer-Verlag. Int. J. Earth Sci. https://doi.org/10.1007/s00531-017-1573-7. Pelinovsky, E., Kharif, C., Riabov, I., Francius, M., 2002. Modelling of tsunami propa- Trincardi, F., Cattaneo, A., Correggiari, A., Rident, D., 2004. Evidence of soft sediment gation in the vicinity of the French coast of the Mediterranean. Nat. Hazards 25 (2), deformation, fluid escape, sediment failure and regional weak layers within the late 135–159. Quaternary mud deposits of the Adriatic Sea. Mar. Geol. 213, 91–119. Rice, D.D., Claypool, G.E., 1981. Generation, Accumulation, and Resource Potential of Vanney, J.R., Mougenot, D., 1981. La plate-forme continental du Portugal et les provinces Biogenic Gas. 65. American Association of Petroleum Geologists Bulletin, pp. 5–25. adjacents: analyse géomorphologique. Memórias dos Serviços Geológicos de Rodrigues, T., Grimalt, J.O., Abrantes, F., Flores, J., Lebreiro, S., 2009. Holocene inter- Portugal 28. dependences of changes in sea surface temperature, productivity, and fluvial inputs Vis, G.-J., 2009. Fluvial and Marine Sedimentation at a Passive Continental Margin. The in the Iberian continental shelf (Tagus mud patch). Geochem. Geophys. Geosyst. 10, Late Quaternary Tagus Depositional System. PhD thesis. VRIJE UNIVERSITEIT.