L. Moscardelli, M. Hornbach, and L. Wood
Abstract The study area is situated along the obliquely converging boundary of the Caribbean and the South American plates and proximal to the Orinoco delta.
Several Plio-Pleistocene-age seafloor destabilization events have been identified in the continental margin of eastern offshore Trinidad. These mass wasting processes are thought to have been of sufficient scale to produce tsunamigenic waves.
This work concentrates on the modeling of mass-failure-event-generated tsunami waves in eastern offshore Trinidad. Three different models were generated on the basis of geomorphological characteristics and causal mechanisms of mass transport complexes (MTCs): (1) slope-attached, (2) shelf-attached and (3) detached
MTCs. Present-day geologic conditions suggest that detached and slope-attached mass failure events (MFEs) are more likely to occur today. In addition, modeling results indicate that detached MFEs that can occur on the collapsing flanks of mudvolcano ridges represent a higher tsunamigenic risk. These modeling results are a first approach to try to establish a tsunamigenic risk assessment in the region.
Keywords Mass transport complex • mass failure event • submarine slide • tsunami • offshore Trinidad
The term MTC groups a variety of gravity-induced deposits that can co-occur in the same event or depositional unit (slides, slumps, and debris flows) (Dott 1963). It is well known that a sudden displacement of the seafloor through catastrophic sliding
L. Moscardelli ( ) and L. Wood
Bureau of Economic Geology, Jackson School of Geosciences,
The University of Texas at Austin, TX, USA e-mail: lorena.moscardelli@beg.utexas.edu
M. Hornbach
Institute for Geophysics, Jackson School of Geosciences, The University of Texas at Austin,
TX, USA
D.C. Mosher et al. (eds.), Submarine Mass Movements and Their Consequences,
Advances in Natural and Technological Hazards Research, Vol 28,
© Springer Science + Business Media B.V. 2010
733

734 L. Moscardelli et al.
Fig. 1 Map showing the area of study located in eastern off-shore Trinidad. The area covered by the 3D seismic data and that was used in previous studies is outlined or slumping has the potential to displace large volumes of water, generating tsunamis that can affect coastal areas and offshore infrastructure (Murty 1979; Jiang and
LeBlond 1992). The 1998 Papua-New Guinea slide-triggered tsunami (Immamura and Hashi 2003) exemplifies the devastation that MFE-generated tsunamis can cause.
The focus of this study is to evaluate the tsunamigenic risks associated with the potential occurrence of MFEs in the southeastern Caribbean (Fig. 1). Our main premise is that MFEs, similar to those documented in the Plio-Pleistocene stratigraphic succession of eastern offshore Trinidad (Moscardelli and Wood 2008), could pose a tsunami risk to the coastal areas of eastern Trinidad and northeastern
Venezuela if they occurred today.
Moscardelli and Wood (2008) documented that Plio-Pleistocene MTCs in the study area present remarkable differences in terms of their source regions, dimensions, geometries, and associated causal mechanisms. This variability raises questions in terms of the tsunamigenic potential that different types of MFEs might have.

Tsunamigenic Risks Associated with Mass Transport Complexes 735
In this paper, we show preliminary results associated with modeling of tsunamis that could be triggered by a variety of MFEs. The questions that we seek to address are: (1) what is the tsunamigenic hazard associated with the occurrence of MFEs in eastern offshore Trinidad? and (2) what types of MFEs present the greatest tsunamigenic risks in this region?
The area of this study is located on the southeast margin of the tectonically active
Caribbean Plate Boundary Zone in eastern offshore Trinidad (Fig. 1). The geomorphology of the Plio-Pleistocene MTCs that were used as proxies to estimate the spatial dimensions of potential future occurrences were obtained from previous work (Moscardelli and Wood 2008). Boundary conditions for each case scenario were defined on the basis of the causal mechanisms associated with each type of
MFE, as well as on differences regarding their area, thickness, length, and width.
MTCs in eastern offshore Trinidad were originally described using 10,708 km 2 of high-quality 3D seismic data (Fig. 1). Moscardelli and Wood (2008) described three different types of MTCs in the study region: (1) shelf-attached MTCs, (2) slope-attached MTCs, and (3) detached MTCs (their figure 13). On the basis of this classification scheme, three case scenarios were defined in this work, each modeled to estimate the associated relative tsunamigenic risk.
The following parameters were used as input in the modeling package: location, length, width, area, and thickness of individual MTCs, as well as estimated prefailure slope angles and water depths at which failures might occur today (Table 1).
Because MFEs tend to evacuate most sediment from their source area, values associated with prefailure slope angles and water depths at which failures might occur
(Table 1) were obtained using present-day bathymetry data in analogous areas that have not failed (shelf-break region and flanks of mud-volcano ridges).
In this work, kilometer-scale regional bathymetric data were used in tsunami modeling to obtain first-order estimates of influences that MFEs might have in the generation of tsunami waves. More accurate tsunami wave run-up estimates require
Table 1 Input parameters to model tsunami waves generated by the occurrence of different MFEs
Case scenario
1
2
3
Latitude
(degrees)
10.4204
10.4669
10.8518
Longitude
(degrees)
Length
(km)
−60.3645
−60.3343
−59.9571
5
1.8
5.6
Width
(km)
Area
(km 2 )
4
2.5
5.5
20
4.5
30.8
Thickness
(m)
72
50
60
Slope angles
0.5
0.5
6
Water depth
(m) MFE type
175.5
Slope attached
175.5
Shelf attached
952.5
Detached

736 L. Moscardelli et al.
high-resolution bathymetry data. Despite this limitation, results presented should reveal whether a significant MFE-related tsunami risk exists in this region.
To model the potential tsunami wave generated by MFEs, we first estimated slide motion, and from this, ocean-water displacement. As a first-order approach, changes in MTC shape due to dispersion during slide evolution were ignored. For bathymetry, we interpolated GEBCO 1-min, satellite-based bathymetry onto a
** 230 ** -m grid and used a Mercator projection. To approximate the shape of the initial slide motion, we assumed a smoothly shaped 3D Gaussian beam that roughly matched slide dimensions.
Displacement S of the slide’s center of mass as a function of time t is given by:
( g + C m
)
2 d S dt 2
=
( g −
) ( q − C n cos q
) d p
2
B
⎝
⎛
⎜ dS dt
⎞
⎠
⎟
2
(1) where B = length of MTC, t = time, g = MTC specific density, Cm is the mass density coefficient, Cd is the drag coefficient of the slide through the water, and Cn is the basal Coulomb coefficient of friction (Watts 1997; Grilli and Watts 2005)
In general, because higher slide accelerations have the potential to produce significantly larger tsunamis (Harbitz 1992), robust estimates of slide acceleration are necessary to accurately predict tsunami wave height. To determine velocity and acceleration, we adopted the method of Watts (1997) and Grilli and Watts (2005).
From Watts (1997), an analytical solution for slide velocity (dS/dt) in Eq. 1 is
( ) = u t tanh
⎡ a t ⎤
⎢ u t
⎥ (2)
Where u t
is the terminal velocity of the slide and a
0
denotes initial slide acceleration which equals a
0
= g
( g −
1
) ( g −
C m
) sin q
For MTCs, we estimated g = 2, q ≈
3°, and assume Cm = 1, Cd = 2 and Cn = 0.32, consistent with other slide tsunami models (e.g. Watts 1997; Grilli and Watts 2005)
(Table 1).
This method may over predict slide terminal velocity (Hornbach et al. 2007), as indirect measurements of slide velocities based on timing of slide-associated submarine cable breaks suggest velocities of 6 to < 30 ms-1 for slides consisting of low-density hemipelagic mud (Heezen and Ewing 1952; Bjerrum 1971); we therefore apply an upper limit of 30 ms-1 for terminal velocities of all modeled MFEs.
The initial water wave generated by the first few seconds of motion was modeled by coupling seafloor deformation to sea level using nonlinear shallow-water wave approximations (e.g., Jiang and LeBond 1992); however, the propagating wave and

Tsunamigenic Risks Associated with Mass Transport Complexes 737 run-up estimates were modeled using the standard fully nonlinear Boussinesq wave model, FUNWAVE (Wei and Kirby 1995). The model uses a semi-implicit leapfrog approach over 308,000 km area (600 × 600 grid), with a square cell size of
925 m per side.
Incidence of tsunamis is often related to occurrence of major earthquakes (Bryant
2001); however, the final triggering mechanism that is related to a specific tsunamigenic event can be associated with a variety of causal mechanisms, including submarine mass movements (MTC) (Tappin and Suppl 1999). The variety of processes that can trigger tsunamis suggests that not all tsunamigenic events have the same characteristics in terms of magnitude and propagation mechanisms. For example, tsunamis generated by submarine slides (e.g., PNG tsunami) have large run-up heights close to the failure area (Tappin and Suppl 1999) but have far-field effects that are more limited than earthquake tsunamis (e.g., Indian Ocean tsunami)
(Okal and Synolakis 2004).
The magnitude of the run-up generated by a MFE will depend primarily on (1) volume of material that moves, (2) water depth in which the failure occurs, (3) acceleration of submarine sediment mass, (4) rheology of the material that fails, and (5) distance from shore (Pelinovsky and Poplavsky 1996). In eastern offshore Trinidad, six MTCs have been identified in the shallow subsurface. Estimated sediment volumes of these deposits and water depths at which failure occurred vary from 1 to
242 km 3 and from 600 to 1,400 m, respectively. In addition, significant differences exist in terms of the rheology of sediments that fed these MTCs. The smallest MTCs were derived from collapsing flanks of mud-volcano ridges (detached MTCs).
Sedimentary sources of regional MTCs were associated with shelf-edge deltas (paleo-
Orinoco) and with upper-slope sediments (attached MTCs), where sandier intervals are more likely to occur (Sullivan et al. 2004; Moscardelli and Wood 2008).
According to Skvortsov and Bornhold (2007), 70% of total energy transfer from the submarine slide to surface waves occurs during initial phases of failure, and, hence, determining specific initial conditions is crucial. For this reason, when we evaluated the risks associated with MFE-generated tsunamis, we decided to define three different case scenarios that took into consideration not only dimensions of the MTCs (morphology), but also their associated geologic settings and causal mechanisms (Moscardelli and Wood 2008).
Slope-attached MTCs are regionally extensive and can occupy hundreds to thousands of square kilometers. Slope-attached MTCs have scarps that can be tens of kilometers long and hundreds of meters deep. These scarps define their updip boundaries in the shelf break and upper-slope region (Fig. 2). Slope-attached

Fig. 2 (a) 3D visualization of horizon defining base of slope-attached MTC_2 near the shelf break. (b) Root-mean-square (RMS) amplitude extraction on a 20-ms window above the horizon defining base of slope-attached MTC_2. Note scar zone near shelf break and downslope mass wasting. (c) 3D visualization of horizon defining base of shelf-attached MTC_1. (d) RMS amplitude extraction on 20-ms window above the base of shelf-attached MTC_1. Note low-amplitude seismic response on axis of paleocanyon. (e) Dip seismic line across shelf break and upper-slope region showing seismic response and strata architecture of shelf-attached (MTC_1) and slopeattached (MTC_2) MTCs

Tsunamigenic Risks Associated with Mass Transport Complexes 739
MTCs are associated with catastrophic and almost instantaneous collapse of the upper continental slope that may be preconditioned by a variety of processes, including earthquakes, long-shore currents, hydrate dissociation, and salt deformation (Mosher et al. 2004). Slope-attached MTCs in offshore Trinidad are thought to have been preconditioned catastrophically by earthquakes or gas hydrate dissociation; these deposits cover 60 to 626 km 2 in area, 20 to 40 km in length, and 3 to 10 km in width. These morphological values include total area affected by the mass wasting event, which involves initial sediment that failed upslope and sediments that were incorporated as the flow scoured the seafloor downslope. A reexamination of 3D seismic data allowed us to measure the morphological parameters that we thought corresponded to the shape of original upslope collapses, and these measurements were used as input for the model (Table 1)
(Fig. 2).
In the case of slope-attached MTCs, we used the upslope MTC_2 as described by Moscardelli and Wood (2008). According to this description, several collapse features can be seen in the upper-slope area near the paleoshelf break (Fig. 2). On the basis of an estimate of the missing vertical section and the area of the original failure, we calculated that the original material that failed was 72 m thick. Present location of the shelf break suggests that a similar failure could occur today at water depths of
∼
180 m, affecting the shelf-break region where slope angles are near 0.5°
(Table 1). During modeling, we assumed that an upper-slope collapse generated by a slope-attached MTC occurring today would have a slightly southwest-northeast trajectory following maximum inclination of the slope-to-basin-floor transition.
Shelf-attached MTCs are also regionally extensive but are usually fed by canyon systems that cut across the outer shelf and upper-slope region, funneling considerable amounts of sediment downdip (Fig. 2). Causal mechanisms of failure can include oversteepening in the upper parts of clinoform foresets that generate gravitational instabilities (Moscardelli and Wood 2008).
Shelf-attached MTCs may be generated by relatively small retrogressive failures that occur over time and that eventually accumulate to generate the bulk of the
MTC deposit downslope. For modeling purposes, we used MTC_1 as described by
Moscardelli and Wood (2008); however, we recalculated length, area, and thickness of the upslope parts of MTC_1 because previous values included the total area covered by the deposit. Our estimations suggest that the area affected by initial upslope failures could reach between 4 and 5 km 2 with a thickness of
∼
50 m. If a similar collapse occurred today, it would affect sediments sitting at the shelf break at slope angles of 0.5° and at water depths of
∼
180 m (Table 1). Initial upslope failures associated with a shelf-attached MTC would move southwest-northeast toward the basin.

740
L. Moscardelli et al.
Detached MTCs cover less than tens of square kilometers in area and a few kilometers in width and length. Detached MTCs can originate in any type of localized bathymetric high, and resulting sediments are derived locally within the minibasin of deposition. Triggering mechanisms of detached MTCs are controlled by local gravitational instabilities that can occur on the flanks of mud volcanoes and other bathymetric highs (Moscardelli and Wood 2008).
Several detached MTCs were documented in eastern offshore Trinidad
(MTC_2.2, MTC_2.3, and MTC_2.4) (Moscardelli and Wood 2008); these deposits covered
∼
30 km 2 in area. Detached MTCs could occur today at water depths of 950 m, in which bathymetric data clearly show modern expression of active mud-volcano ridges on the seafloor (Sullivan et al. 2004) (Fig. 3). Initial failures could involve slides that can be 60 m thick. The average slope angle of mud-volcano flanks is 6° (Table 1), but maximum reported slope angles reach 18°
(Sullivan et al. 2004). In this case, a northwest–southeast-moving collapse affecting the southern flank of the Darien Ridge was used to model the associated tsunamigenic event.
Fig. 3 Geobody between the structural surface that defines the base of slope-attached MTC_2 and the seafloor. Mud-volcano heights can be >165 m, and mud-volcano ridges have northeastsouthwest orientations

Tsunamigenic Risks Associated with Mass Transport Complexes
741
Our initial working hypothesis indicated that the worst-case scenario for an
MFE-generated tsunami would involve an event with the following characteristics:
(1) initial failure remobilizing large amounts of sediments in a relatively short period of time and (2) failures occurring at shallower depths (
∼
175 m). The suddenness of regional tectonic instabilities that result in the occurrence of slope-attached MFEs implies an almost instantaneous remobilization of huge volumes of sediment from the upper-slope region. The sudden tectonically triggered failure and release of energy that is unique to slope-attached MFEs prompted us to think that this type of event would pose the greatest tsunamigenic risk of any studied in the area. Our purely qualitative interpretation also suggested that detached MFEs would pose the lowest tsunamigenic risk for coastal areas because these events occur farther (>120 km) from the coast and in the deeper parts of the basin (
∼
950 m water depth).
Results of tsunami modeling are considerably different from our original qualitative expectations (Fig. 4). Modeling results revealed that detached MTCs affecting
Fig. 4 Wave propagation in MTC-generated tsunamis. T1 occurs shortly after slide generation and t3 one-half hour after slide generation, when the wave reaches the east coast of Trinidad

742 L. Moscardelli et al.
the flanks of mud-volcano ridges in the deeper parts of the basin are the events that pose the greatest tsunamigenic risk for the east coast of Trinidad. Controlling factors seem to be steepness of mud-volcano flanks, as well as slide trajectories because flank collapses on mud-volcano ridges can occur in many different orientations. It is when sliding occurs perpendicular to shore that the model indicates a significant wave (
∼
1 to 2 m high) could propagate towards the east eventually hitting the
Trinidadian coast (Fig. 4). Prefailure slope angles on the flanks of mud-volcano ridges also play a crucial role in increasing the magnitude of water-column perturbation. Slope angles near the shelf break are usually < 1, but average prefailure slope angles at the flanks of mud-volcano ridges are
∼
6° and may even reach 18° (Sullivan et al. 2004). Steeper slope angles have the capacity to increase initial acceleration of the slide. Modeling also suggests that the Orinoco delta region could be affected by a tsunamigenic wave generated by a detached MFE in offshore Trinidad.
The tsunami model for slope-attached MFEs shows that an event like this could generate
∼
1-m-high tsunami waves that could reach the east coast of Trinidad. The propagation pattern of the wave, however, seems to suggest that effects would be more noticeable on the southern coast of Tobago (Fig. 4). The model also shows that much of the energy would propagate towards the Central Atlantic, whereas a negative-positive tsunami wave would hit the eastern coast of Trinidad first. Similar behavior, in terms of wave propagation, is observed for the tsunami wave modeled for shelf-attached MFE. However, in this case, wave heights are reduced (<0.5 m high) because the initial volume of sediments that failed was considerably lower.
Our analysis also suggests that the Orinoco delta region could be affected by a tsunami if a slide occurred off Trinidad, although significant wave dispersion occurs in the delta.
The study area is within an active tectonic margin. More than 30 seismic events with varying magnitudes of Mw 1 to 5 have been reported in the study area from
2006 to 2008. Since 1983, at least three earthquakes of Mw 6 and 7 have occurred in the same geographic area (USGS online Earthquake Center). Location of some earthquake epicenters observed in this region coincides with the location of mudvolcano ridges, demonstrating that these ridges are currently active and potentially unstable. Our model of case scenario III suggests that 2-m-high tsunami waves could strike the east coast of Trinidad if a detached MFE were generated on the south flank of the Darien Ridge. Tsunami wave heights and propagation directions, however, would vary according to the volume of material that fails and the plane and orientation of failure. Historically events like this have not been reported in
Trinidad.
The minimum earthquake magnitude necessary to trigger a slope-attached MFE is unknown and would probably depend on a variety of factors. Some data suggest

Tsunamigenic Risks Associated with Mass Transport Complexes 743 that slope-attached MFEs associated with the occurrence of earthquakes are usually triggered by at least M > 7 earthquakes (e.g., Papua New Guinea and Grand Banks events) (Imamura and Hashi 2003). However, Mosher et al. (1994) concluded that more frequent, lower magnitude (M > 3) earthquakes have been the likely cause of large-scale slope failures in the Scotian slope of Canada. Seismicity records from eastern offshore Trinidad indicate that relatively low intensity earthquakes are common, whereas M > 7 events are rare in this region. We do not think that the occurrence of a catastrophic slope-attached MFE is imminent in the region today; however, the geologic record shows that at least one slope-attached MFE occurred during the Plio-Pleistocene (Fig. 2) (Moscardelli and Wood 2008).
From the three scenarios that were modeled in this work, case scenario II
(shelf-attached MFE) represents the least likely to occur under present-day conditions because shelf-attached MFEs have been associated with failure of shelf-edge deltas. Today the modern active Orinoco delta front is several hundred kilometers to the southwest of the modern shelf break, and current highstand sea-level conditions define a different scenario from the one prevalent at the time of MTC_1 deposition.
Acknowledgements This research was made possible through the generosity of the member companies of the Quantitative Clastics Laboratory Industrial Associate Program. We also thank the reviewers Drs. D. Mosher and F. Dias for critiquing and improving this manuscript. Publication was authorized by the Director of the Bureau of Economic Geology, The University of Texas at
Austin.
Bjerrum L (1971) Subaqueous slope failures in Norwegian fjords. Nor Geotech Inst Bull 88:
1–8.
Bryant E (2001) Tsunami; the underrated hazard. Cambridge University Press, Cambridge.
Dott RH (1963) Dynamics of subaqueous gravity depositional processes. Am Assoc Pet Geol Bull
47: 104–128.
Grilli ST and Watts P (2005) Tsunami generation by submarine mass failure. I: modeling, experimental validation, and sensitivity analysis. J Waterway Port Coast Ocean Eng 131: 283–297.
Harbitz CB (1992) Model simulations of tsunamis generated by the Storegga slides. Mar Geol
105: 1–21.
Heezen BC and Ewing WM (1952) Turbidity currents and submarine slumps, and the 1929 Grand
Banks (Newfoundland) earthquake. Am J Sci 250: 849–873.
Hornbach MJ, Lavier LL and Ruppel CD (2007) Triggering mechanism and tsunamogenic potential of the Cape Fear Slide complex, U.S. Atlantic margin. Geoch Geoph Geosystems 8.
Imamura F and Hashi K (2003) Re-examination of the source mechanism of the 1998 Papua New
Guinea earthquake and tsunami. Pure Appl Geoph 160: 2071–2086.
Jiang LC and LeBlond PH (1992) The coupling of a submarine slide and the surface waves which it generates. J Geophys Res 97: 12731.
Moscardelli L and Wood L (2008) New classification system for mass transport complexes in offshore Trinidad. Basin Res 20: 73–98.
Mosher DC, Moran K and Hiscott RN (1994) Late Quaternary sediment, sediment mass flow processes and slope stability on the Scotian Slope, Canada. Sedimentology 41: 1039–1061.

744 L. Moscardelli et al.
Mosher DC, Piper DJW, Calvin CD and Jenner KA (2004) Near-surface geology and sedimentfailure geohazards of the central Scotian Slope. Am Assoc Pet Geol Bull 88: 703–723.
Murty TS (1979) Submarine slide-generated water waves in Kitimat Inlet, British Columbia.
J Geophys Res 84: 7777–7779.
Okal EA and Synolakis CE (2004) Source discriminants for near-field tsunamis. Geoph J Internat
158: 899–912.
Pelinovsky E and Poplavsky A (1996) Simplified model of tsunami generation by submarine landslides. Phys Chem Earth 21: 13–17.
Skvortsov A and Bornhold B (2007) Numerical simulation of the landslide-generated tsunami in
Kitimat Arm, British Columbia, Canada, 27 April 1975. J Geophys Res 112.
Sullivan S, Wood LJ and Mann P (2004) Distribution, nature and origin of mobile mud features offshore Trinidad. Program and Abstracts – GCSSEPM 24: 36.
Tappin DR and Suppl (1999) Submarine slump generation of the 1998 Papua New Guinea tsunami; the evidence so far. In Watts P, Borrero JC, Okal E, Bardet JP, Grilli S, Matsumoto T and Synolakis CE (eds) Eos, Transactions, Am Geophys Union 80: 750.
Watts P (1997) Water waves generated by underwater landslides. Pasadena, California Institute of
Technology, CA.
Wei G and Kirby JT (1995) A time-dependent numerical code for extended Boussinesq equations.
J Waterway Port Coast Ocean Eng 120: 251–261.