Giant submarine landslides on the Colombian margin and tsunami risk in the Caribbean Sea

Earth and Planetary Science Letters 449 (2016) 382–394

Contents lists available at ScienceDirect

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

Giant submarine landslides on the Colombian margin and tsunami risk in the Caribbean Sea Stephen C. Leslie ∗ , Paul Mann Department of Earth and Atmospheric Sciences, University of Houston, Houston, TX 77204, USA

a r t i c l e

i n f o

Article history: Received 2 February 2016 Received in revised form 23 May 2016 Accepted 24 May 2016 Available online 13 June 2016 Editor: P. Shearer Keywords: mass transport deposit submarine slide seismic reflection Colombian Caribbean margin tsunami Caribbean Sea

a b s t r a c t A series of three giant, previously unrecognized submarine landslides are defined on a 16,000 line km grid of multi-channel 2D seismic reflection profiles along the active margin of northern Colombia in the western Caribbean Sea. These deposits record the collapse and mobilization of immense segments (thousands of cubic kilometers) of the submarine slope and are comparable in scale to the largest known landslides on Earth. We show that the breakaway zone for these events corresponds to the tectonically over-steepened slopes of the Magdalena Fan, an extensive submarine fan composed of sediments sourced from the northern Andes and deposited by the Magdalena River. An over-pressured zone of weakness at the base of the gas-hydrate stability layer within the fan likely facilitates slope failure. Timing of these massive slope failures is constrained by well control and occurred from the mid-to-late Pliocene to mid-Pleistocene. To understand the tsunamigenic hazards posed by the recurrence of such an event today, we model the potential tsunami source created by a submarine landslide of comparable thickness (400 m) and lateral extent (1700 km2 ) derived from the over-steepened upper slopes of the present day Magdalena Fan. Our modeling indicates the recurrence of an analogous slope failure would result in a major tsunami that would impact population centers along the Caribbean coastlines of Colombia, Central America, and the Greater Antilles with little advance warning. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Submarine landslides have the potential to cause unexpected or “surprise” tsunamis in regions not typically thought of as tsunami prone (Ward, 2001; Hornbach et al., 2010; Lo Iacono et al., 2012). When they occur, these tsunamis waves arrive in coastal areas with little advance warning and can devastate coastal communities. For example, in 1998 a 7.1 Mw earthquake in Papua New Guinea triggered a submarine landslide of ∼4 km3 which caused a catastrophic tsunami with local run-up heights of up to 15 m that killed more than 2200 coastal inhabitants (Heinrich et al., 2001; Tappin et al., 2008). While only 7% of tsunamis worldwide are thought to originate by submarine landslide or a combination of earthquake and landslide (National Geophysical Data Center, 2016), the destructive potential of these events is highest when they affect areas with large coastal populations uninformed and unprepared for tsunami hazards, such as the Caribbean Sea (Fig. 1). Hundreds of thousands of residents, tourists, and touristindustry workers are potentially at risk from tsunamis around the densely populated urban and tourist areas of the Caribbean

*

Corresponding author. E-mail address: [email protected] (S.C. Leslie).

http://dx.doi.org/10.1016/j.epsl.2016.05.040 0012-821X/© 2016 Elsevier B.V. All rights reserved.

(von Hillebrandt-Andrade, 2013). Unfortunately, unlike earthquake generated tsunamis in seismically active coastal regions with long and meticulously maintained historical records (e.g. Japan), the infrequent nature of landslide tsunamis and the relatively brief historical records (∼500 yr) in the Caribbean region make understanding the likelihood of these events in this area difficult (Ward, 2001; Harbitz et al., 2012). In the absence of historical records of tsunamis occurring along Colombian Caribbean margin it is necessary to investigate the history and magnitude of past tsunamis using the geologic record. In this study, a series of giant (>1000 km3 ), previously unrecognized, Plio-Pleistocene age submarine landslide deposits are identified on seismic reflection records from the submarine Magdalena fan along the Caribbean margin of northern Colombia (Fig. 1). Our observations include the frequency, location, lateral extent, and thickness of the largest slope failures in the area. We use these observations as parameters for modeling the effects of a future tsunami produced by a similar event (Lo Iacono et al., 2012; Harbitz et al., 2014). Sediments resulting from submarine slope failures range in scale from tens of cubic meters to thousands of cubic kilometers and are classified as mass transport deposits (MTDs) (Weimer, 1989; Mulder and Cochonat, 1996). MTD deposition occurs by a variety of matrix-supported, mass movement

S.C. Leslie, P. Mann / Earth and Planetary Science Letters 449 (2016) 382–394

383

Fig. 1. a. Map of the active subduction margin along the northern coast of Colombia including its accretionary prism, the South Caribbean Deformed Belt (SCDB), and the submarine Magdalena Fan system (shaded outline). Three submarine landslides, or mass transport deposits (MTDs) are recognized from SW to NE along the proximal to distal lower slopes of the Magdalena Fan: the Cartagena MTD, Barranquilla MTD, and Santa Marta MTD. Dashed white lines indicate grid of 2D seismic reflection profiles used to map the slides including one seismic line that ties to DSDP site 153, providing age control. Additional lithologic correlation is provided by ODP site 999, located on a basement high to the west of the distal Magdalena Fan. Major, active strike-slip faults in the region shown are potential triggers for submarine slope failure evens. Large Colombian cities susceptible to tsunami hazard are shown as yellow circles. b. Location of larger map (dashed box), Magdalena Fan (green outline), and MTDs identified in this study (yellow outlines) in the tectonic context of northwestern Colombia and its offshore margin. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

processes including creep, slide, slump, and debris flow (Weimer, 1989). A single deposit can cover thousands of square kilometers and can travel up to hundreds of kilometers from the source region prior to deposition (Talling et al., 2007). The volume of MTDs on a scale resolvable by seismic reflection data and reviewed in the literature follows a lognormal distribution, with a median MTD size of ∼40 km3 (Moscardelli and Wood, 2015). Located on the distal flanks of the Magdalena Fan, the three giant MTDs recognized by our study are orders of magnitude larger than those described previously in the area (Vinnels et al., 2010; Cadena and Slatt, 2013; Alfaro and Holz, 2014; Ortiz-Karpf et al., 2015). Each of the three MTDs falls within the top 5–10% of the global distribution of all deposits as compiled by Moscardelli and Wood (2015). The largest deposit, the Santa Marta MTD, covers an estimated area of 34,700 km2 and has a volume of 5255 km3 – recording the mobilization, in a single event, of enough material to cover an area equivalent to the U.S. state of California with 12.4 m of sediment. A comparable MTD and paleo-tsunami analog to the three Colombian MTDs identified in this study is the 1300 km3 Holocene Storegga slide and its well-documented paleo-tsunami with up to 25 m of observed sediment run-up off the Atlantic coast of Norway (Bugge, 1983; Smith et al., 2004; Moscardelli and Wood, 2015). The timing of the Colombian MTDs associated with the Magdalena Fan is likely to have been strongly influenced by climate induced

changes in sea level during the late Pliocene to Pleistocene, as previously documented for similar extensive MTDs recognized from the Pleistocene section of the Amazon Fan (Maslin et al., 1998). The initial slope failures were likely facilitated by pre-existing planes of weakness along the base of gas-hydrate stability zones, which are seismically imaged as prominent Bottom Simulating Reflectors (BSRs) that are observed throughout the Magdalena Fan. 2. Geologic setting The Caribbean Margin of Colombia extends ∼850 km from the Isthmus of Panama to the Guajira Peninsula and is actively deforming by oblique subduction of the Caribbean Plate and Colombian basin beneath the overriding continental South America plate at a rate of ∼2 cm/yr (Krause, 1971; Ladd et al., 1984; Bernal-Olaya et al., 2015). The active subduction process has formed a 65 to 165 km-wide accretionary prism, the Southern Caribbean Deformed Belt (SCDB), along the margin (Krause, 1971; Ladd et al., 1984; Bernal-Olaya et al., 2015) (Fig. 1). The Magdalena Fan is a 325 km by 390 km submarine fan that spills across the SCDB onto the subducting Caribbean Plate. The Magdalena Fan is composed primarily of late Miocene to recent sediments transported from an extensive, 260,000 km2 watershed draining the Andean Mountains of northwestern South Amer-

384

S.C. Leslie, P. Mann / Earth and Planetary Science Letters 449 (2016) 382–394

ica (Kolla and Buffler, 1984; Kolla et al., 1984; Hoorn et al., 1995; Ercilla et al., 2002; Flinch, 2003) (Fig. 1a). The slopes of the Magdalena Fan along the Colombian margin are locally over-steepened, exceeding 10◦ in proximal areas of the fan. Slope steepness is controlled by the interplay of rapid sedimentation and tectonic uplift related to active shortening on buried, blind thrusts of the SCDB (Breen, 1989). The Magdalena fan is folded and faulted within the accretionary prism of the SCDB on its proximal margins. To the south-west of our study area, the SCDB is known as the Sinu– San Jacinto fold belt; the SCDB is known as the Tayrona fold belt to the northwest (Ercilla et al., 2002; Alfaro and Holz, 2014; Ortiz-Karpf et al., 2015). The central and distal portions of the fan are relatively undisturbed at shallow levels possibly as the result of either simple burial by rapid fan sedimentation or, alternatively, because of the inability of thrusts to continue slipping under rapid burial conditions (Breen, 1989) (Fig. 2a). Thickness of the Magdalena fan sedimentary section varies from up to 12 km in its proximal area to <1 km along the distal lower slopes. Quaternary sediment deposition rates in proximal areas of the Magdalena Fan are approximately 1.5 m/kyr (Rincón et al., 2007) while hemi-pelagic sediment contribution from the Magdalena fan at the distal ODP 999 site (located atop a 1000 m basement high) is ∼0.03 m/kyr (Shipboard Scientific Party, 1997). A number of linear slope-failure scars have been previously mapped along the proximal slopes of the Magdalena fan, especially in the zone between the shelf break (∼100 mbsl) and the distal lower slopes (∼2000–3000 mbsl) (Ercilla et al., 2002; Romero et al., 2015) (Fig. 2b). Numerous elevated channel-levee complexes are present between the slide scars. In some cases, the channel-levee complexes are truncated by the slope failures, indicating that some of the scars post-date formation of the channellevee complexes (Ercilla et al., 2002). In addition, the largest of the slide scars exhibit high backscatter on GLORIA side scan sonar surveys (labeled 1–4 in Fig. 2b), indicating that they are relatively fresh features that are not draped by hemi-pelagic sediments (Romero et al., 2015). Interpretation of a 1970’s vintage seismic reflection profile across the proximal to distal slopes of the Magdalena Fan reveals a number of lenticular, chaotic packages of seismic reflections concentrated in the proximal lower slope area of the central Magdalena Fan and interpreted as stacked sets of MTDs (Fig. 2c, 2d). The shallowest of these MTDs is exposed at the seafloor and exhibits a rugose top that corresponds to a group of compressional seafloor ridges, typical of MTD deposition as imaged on high-resolution bathymetric surveys (Fig. 2b) (Mulder and Cochonat, 1996). Individual thicknesses for these MTDs range from 40 to 150 m and they extend laterally from 20 to over 100 km downslope. The prevalence of these features demonstrates MTDs are recurring at regular intervals along the slopes of the Magdalena fan and exhibit a wide range of scale from a few cubic kilometers to thousands of cubic kilometers (Ercilla et al., 2002; Alfaro and Holz, 2014; Ortiz-Karpf et al., 2015; Romero et al., 2015). Methane hydrates are solid, crystalline, water-based molecules that trap methane molecules and are common in deepwater, deltaic and fan settings where significant gas-prone, organic material is trapped within the offshore, sedimentary section (Kvenvolden, 1993). The occurrence of methane hydrates within the shallow sediment section of offshore Colombia were first recognized based upon the presence of bottom simulating reflectors (BSRs) (Shipley et al., 1979). Beneath such gas hydrate zones, free gas is generated by decay of organic material in the sediment column and accumulates at the base of the gas hydrate layer, acting to increase pore pressure and decrease sediment strength (Kvenvolden, 1993). The base of the gas hydrate stability zone along the mid-to-upper slopes is clearly imaged on seismic re-

flection profiles crossing the Magdalena Fan (Fig. 2c, 2d) and is proposed to act as a potential failure plane for the large MTDs identified in this study. Potential triggering events for these slope failures could include infrequent earthquakes along major strikeslip faults along the margin or on underlying thrust faults associated with the SCDB (Tappin et al., 2014) (Fig. 1a). In northern Colombia, large earthquakes of either type are comparatively rare and the seismic risk for this area of Colombia is considered low based upon the 500 yr long historical record (Sarria et al., 1996) and the sparseness of earthquake events in the SCDB and its Benioff zone (Bernal-Olaya et al., 2015). 3. Seismic reflection profiles and age control A number of multi-channel seismic reflection surveys have been conducted along the Caribbean coast of Colombia over the last 40 years for the purpose of either academic research or offshore hydrocarbon exploration (Edgar et al., 1973; Ladd et al., 1984) (Fig. 1). Interpretation of 16,000 km of 2D seismic reflection data from multiple surveys covering an area of ∼250,000 km2 reveals for the first time the broad areal extent of a series of massive, long runout MTDs. The primary seismic reflection profiles presented here (Fig. 5, distal portion of Fig. 6, Fig. 8, and Fig. 9) were collected in 2006 with 10-km, 804-channel receiver array and a 34-airgun array (86.85 l volume) as the source. The additional seismic reflection profile presented in Fig. 2 was collected in 1975 with a 2.2-km, 24 channel receiver array, using a 73.7 l source by the R / V Ida Green (Kolla et al., 1984). The seismic line shown in the proximal (landward) portion of Fig. 6 was collected in 1977 with a 2.4-km, 24 channel receiver array, using a 30.5 l source by the R / V Robert Conrad (Ladd et al., 1984). Age correlation between the seismic reflection profiles and the subsurface, sedimentary section is provided by a tie to DSDP well 153, located to the northeast of the Magdalena Fan on the abyssal sea floor at a water depth of 4046 m, along the southern edge of the Beata Ridge (Edgar et al., 1973) (Fig. 1). The DSDP well at site 153 encountered a 755-mthick succession of Late Cretaceous to Recent age sediments lying conformably on basalts of the Late Cretaceous Caribbean oceanic plateau. Seismic velocity from analysis of the DSDP 153 well data in the sedimentary section ranges from 1575 to 5635 m/s and provide the basis for converting stratigraphic horizons interpreted in two-way travel-time to surfaces in depth (Moore and Fahlquist, 1976) (Fig. 3a, 3b). Present day bathymetric data includes the continuous terrain model General Bathymetric Chart of the Oceans (British Oceanographic Data Centre, 2008) and a high resolution bathymetric dataset kindly provided by the Institut de Ciències del Mar, Barcelona, Spain. 4. Recurring giant submarine landslides Three previously undocumented giant MTDs are identified and described by this study along the distal marine slopes of the Colombian Caribbean margin, including their seismic reflection character and stratigraphic/structural position (Weimer, 1989; Moscardelli et al., 2006). From southwest to northeast they are described and named here as: 1) the Cartagena MTD (1610 km3 ), 2) the Barranquilla MTD (1470 km3 ), and 3) the Santa Marta MTD (5255 km3 ) (Fig. 1). We interpret the source area for these MTDs as the over-steepened and actively deforming submarine slopes of the Magdalena Fan, based upon the morphology of the present-day sea floor directly upslope from the MTDs, paleo-flow indications from the seismic reflection data, and the stratigraphic and structural positions of the three MTD’s (Fig. 2, 5b, 6b). The timing of these failures, constrained from offset well and seismic data, is interpreted as Pliocene (5.3 Ma) through early to mid-Pleistocene (0.8 Ma). Relative position within the stratigraphic section indicates that the

S.C. Leslie, P. Mann / Earth and Planetary Science Letters 449 (2016) 382–394

385

Fig. 2. a. High resolution bathymetry along Colombian Caribbean margin showing the proximal Magdalena fan and the SCDB (bathymetric basemap from Romero et al., 2015 and shallow structures from Ercilla et al., 2002 and Romero et al., 2015). b. Interpreted bathymetry showing recent slide scars or areas of slope excavation (orange) interpreted from bathymetry. Orange scars labeled 1–4 correspond to areas of high reflectance identified on GLORIA side-scan sonar surveys (Romero et al., 2015). Sediment routes (channels) are labeled in blue (Ercilla et al., 2002). A series of compressional seafloor ridges are prevalent along the proximal lower slope areas and are down dip of major slide scars associated with areas of MTD excavation. Clusters of seafloor pockmarks present along the distal lower slope areas are likely associated with fluid escape processes from sediments associated with MTD deposition. Fold axes of the SCDB predominately trend SW–NE and are buried beneath the proximal slopes of the Magdalena Fan. Subsurface outlines of the three paleo MTDs of Pliocene–Pleistocene age identified in this study: the Cartagena MTD (CMTD), Barranquilla MTD (BMTD), and Santa Marta MTD (SMMTD) as shown on the map. Position of seismic profile A–A extends from the mid-lower to the distal slopes of the Magdalena Fan. c. Seismic profile extending 190 km from the mid-to-lower slope to the distal lower slope of the Magdalena Fan. d. Interpretation of seismic line showing numerous stacked MTDs along the proximal lower slope. Note rugose sea-floor on seismic inset that corresponds with area of seafloor ridges shown in (b). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

386

S.C. Leslie, P. Mann / Earth and Planetary Science Letters 449 (2016) 382–394

Fig. 3. a. Age control of seismic interpretations provided by DSDP 153 well tie. b. Block diagram showing subsurface structure and stratigraphy of the Colombian Caribbean Margin through the Magdalena Fan and SCDB. Submarine landslide deposits identified in this study are shown in dark red as lenticular deposits within Pliocene to Recent sediments, labeled from left to right as the Santa Marta MTD, Barranquilla MTD, and Cartagena MTD. Note that the Cartagena MTD is deformed by the frontal folds of the accretionary prism of the SCDB, while the Santa Marta MTD and Barranquilla MTD are emplaced on the more tectonically-stable, down-going Caribbean plate. Inferred MTD source areas are outlined along flanks of Magdalena Fan with arrows indicating direction of sediment transport. Several coastal population centers along the Colombian coastline would be impacted by tsunami waves generated by a recurrence of a massive submarine slope failure along the Colombian margin today. These cities include Cartagena (pop. 1.24 M), Barranquilla (pop. 2.37 M), Santa Marta (pop. 0.45 M), and Riohacha (pop. 0.15 M). c. Inset map showing position of 3D block diagram and position of MTDs identified in this study with yellow outlines. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 4. Regional correlation of major geologic events, eustatic sea level changes (Haq et al., 1987), ODP 999 bulk sediment composition (Shipboard Scientific Party, 1997), relative stratigraphic position of MTDs identified in this study and lithologic and age constraints provided by DSDP 153 (Edgar et al., 1973). Major regional geologic and tectonic events include the onset of late Pliocene to Pleistocene glaciation (Bartoli et al., 2005), formation of the proto-Magdalena River (Hoorn et al., 1995), collision of the Panama Arc with western Colombia (Coates et al., 2004), and the initial uplift of the northern Andean Santander Massif (Hoorn et al., 1995).

S.C. Leslie, P. Mann / Earth and Planetary Science Letters 449 (2016) 382–394

387

Fig. 5. a. Regional seismic reflection section B–B , in two-way travel time, extending from the distal Magdalena fan in the northwest to the toe of the SCDB in the northeast. b. Expanded view of a part of line B–B , showing the internal structure of Santa Marta MTD. c. Interpretation of regional line B–B , Santa Marta MTD, in yellow, extends ∼150 km from southwest to northeast. Barranquilla MTD, in green, is deeper in the section and thins to the northeast. Both MTDs have a relatively continuous, rugose top reflection and a less uniform basal reflector that sporadically cuts into underlying sediments and is characteristic of MTDs. Base of gas hydrate layer shown as red dashed line. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Cartagena MTD and Barranquilla MTD are roughly contemporaneous, with the Barranquilla MTD slightly older (deeper in the section) (Fig. 4). The Santa Marta MTD is the youngest of the three MTDs and sits ∼400 m shallower in the section, above the Barranquilla MTD. Seismic line B–B extends for 173 km from the lower slopes of the Magdalena Fan to the frontal thrusts of the SCDB and images both the Santa Marta MTD and deeper Barranquilla MTD (Fig. 5). Both features are easily recognized as lenses of chaotic, disorganized, low-amplitude and steeply-dipping seismic reflections which differ markedly from the flat-lying, continuous, and more coherent reflections of the overlying sedimentary section. The top boundaries of both MTDs are rugose and represent the uneven upper surface of the submarine landslide immediately after

the flow stopped moving. Localized depressions along these rugose top surfaces are in-filled with smooth, laterally continuous reflections more typical of deepwater turbiditic or hemi-pelagic sedimentation (Fig. 5b, 6b). The basal surface of the Santa Marta MTD varies from relatively smooth, continuous, and easily identifiable along the northeastern half of line B–B to more irregular and rugose towards the southwestern half of the seismic reflection profile. The basal reflector of the Barranquilla MTD varies from smooth to rugose in a similar manner to the Santa Marta MTD, but this reflector cuts down more prominently into the underlying sedimentary section in a series of steps or scours (Fig. 5c). These basal steps and scours are a diagnostic characteristic of MTDs recognized from detailed studies using 3D seismic reflection data (Moscardelli et al., 2006).

388

S.C. Leslie, P. Mann / Earth and Planetary Science Letters 449 (2016) 382–394

Fig. 6. a. Regional arbitrary seismic reflection section C–C , in two-way travel time, through part of the SCDB in the south and extending to north across the abyssal plain of the Colombian basin. b. Inset showing the Santa Marta MTD on two intersecting seismic lines of 1977 and 2006 vintage (left) and the distal termination of the Santa Marta MTD (right). C. Interpretation of arbitrary seismic line showing lateral extent and thickness of Santa Marta MTD and Barranquilla MTD. Base of gas hydrates shown as red dashed line. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Located along the distal, northeastern slopes of the Magdalena Fan, the Santa Marta MTD is 260 km long by 160 km wide, with its long axis trending northwest to southeast, parallel to the trend of the modern shoreline of northern Colombia (Fig. 6). The appearance and thickness of the MTD remains remarkably consistent along the distal lower slope and basin floor, but terminates abruptly at its distal margins (Fig. 6b). We estimate the area covered by this MTD to be 34,700 km2 , slightly larger than the U.S. state of Massachusetts. The actual areal extent of the MTD may be greater, as the distal, northern edge of the feature is not covered by our grid of seismic lines. The maximum thickness of the Santa Marta MTD is 235 m, with an average thickness of 150 m. The volume of the Santa Marta MTD is calculated to be ∼5255 km3 , placing this newly described submarine landslide within the top

5% of the known occurrences of MTDs compiled from the geologic record by Moscardelli and Wood (2015) (Fig. 7). The Santa Marta MTD is several orders of magnitude larger than MTDs previously recognized on the slopes of the Magdalena Fan or in the nearby SCDB (Vinnels et al., 2010; Cadena and Slatt, 2013; Alfaro and Holz, 2014; Ortiz-Karpf et al., 2015). The deeper, older Barranquilla MTD lies ∼400 m beneath the Santa Marta MTD within the same, Pliocene age sedimentary section (Fig. 5, 6). The Barranquilla MTD is 150 km long by 90 km wide and is estimated to cover an area of 9220 km2 . The maximum thickness of the Barranquilla MTD is 396 m, with an average thickness of 159 m and a volume calculated to be ∼1470 km3 . Locating the proximal source area for these long run-out submarine landslides of the Colombian margin is challenging as the

S.C. Leslie, P. Mann / Earth and Planetary Science Letters 449 (2016) 382–394

389

Fig. 7. Cumulative distribution of MTD volume worldwide based on compilation by Moscardelli and Wood (2015). Santa Marta MTD falls in the top 5% of the distribution. Cartagena MTD and Barranquilla MTD are in the top 10% and top 11% of the distribution, respectively.

primary failure scarps have been infilled and covered by later sediments. In addition, the upslope source or excavation area for these types of MTDs is likely to be smaller in both area and volume than the final deposit, as the moving slide scoured and incorporated additional material from the seafloor as it spread outward and downward during the failure event (Mulder and Cochonat, 1996). However, we interpret the most likely source areas for both the Santa Marta and Barranquilla MTDs along the NE flank of the Magdalena Fan based upon 1) the anomalous ∼1700 km2 indentation of the sea floor contours in this area (Fig. 1, 2b), 2) the internal geometry of each MTD indicating paleo-flow direction was to the N–NE (Fig. 6), and 3) the structural/stratigraphic position of the deposits along the distal slopes of the Magdalena Fan (Fig. 3, 4, 5, 6). The Cartagena MTD, located along the southwestern flanks of the Magdalena fan is visible on seismic line D–D as a layer of discontinuous, chaotic reflections (Fig. 8). The long axis of the Cartagena MTD trends 155 km to the SW–NE, following the trend of the frontal thrust of the SCDB. The deposit is 60 km wide and is estimated to cover an area of 6730 km2 . The maximum thickness of the Cartagena MTD is 565 m, with an average thickness of 240 m and an estimated volume of ∼1610 km3 . The source area of the Cartagena MTD is inferred to be along the SW flank of the Magdalena fan, as indicated by the incised nature of the sea floor contours in the area immediately upslope from the shallowest portion of the deposit (Vinnels et al., 2010) and the overall morphology of the MTD in map view (Fig. 1a, 2b). 5. Causative factors for the Colombian MTDs Extreme slope instability along the flanks of the Magdalena Fan results from a ‘perfect storm’ of environmental conditions that allows for the formation of precariously over-steepened slopes with the potential for calving into massive submarine landslides as illus-

trated on seismic line E–E , located along the axis of the Magdalena Fan (Fig. 9). Sea floor slopes in the area are high, averaging 4–5◦ on the upper slope and locally reaching values over 10◦ (Fig. 9c). These anomalously over-steepened slopes are produced by a combination of two key factors, 1) high rates of sediment input from the Magdalena River, which drains a significant portion of the northern Andean Mountains, and 2) tectonic uplift produced by thrust faults of the underlying and buried accretionary prism of this subduction setting along the SCDB (Fig. 9b). Within the sediment column, a potential zone of instability exists at the base of the gas-hydrate layer due to pore-space overpressure (Kvenvolden, 1993). This zone is easily recognized as a high-amplitude Bottom Simulating Reflector (BSR) that cuts across structures and stratigraphy in the sediment column (Figs. 5c, 6c, 8b, 9b) with a depth controlled by the temperature of the sediment column (Kvenvolden, 1993). Along the upper slope of the Magdalena Fan on line E–E , the BSR is imaged as strong reflector ∼435 m beneath the seafloor dipping to the northwest (Fig. 9b). This BSR delineates the top of an over-pressured zone of weakness (failure plane) within the sediment column that facilitates unstable conditions across broad sectors of the Magdalena Fan. The dip of these potential failure planes is greatest beneath steep submarine slopes of the proximal Magdalena Fan, due to high rates of sediment input and tectonic uplift. In addition, rapid oscillations in sea level and changing rates of sediment discharge related to deglaciation during the late Pliocene to Pleistocene (Maslin et al., 1998) may have further destabilized the margin. In this tectonicallyactive setting, a relatively small earthquake from thrusts within the accretionary wedge or from the nearby Oca or Santa Marta/Bucaramanga strike-slip faults (Flinch, 2003) could trigger a major collapse of the submarine slope, resulting in some the largest volume MTDs recognized in the geologic record and possibly generating a series of paleo-tsunamis that affected many coastal areas of the Caribbean Sea.

390

S.C. Leslie, P. Mann / Earth and Planetary Science Letters 449 (2016) 382–394

Fig. 8. a. Seismic reflection section D–D through the proximal portion of the Cartagena MTD. b. Outline of Cartagena MTD in yellow. Note rapid thinning of the MTD to the SE and truncation by the frontal thrust of the SCDB. Base of gas hydrates shown as red dashed line. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

6. Tsunamigenic implications of giant MTD’s along the Circum-Caribbean coastline Historical compilation of tsunami events documented since 1498 revealed that 85 tsunamis have affected the coastlines of the Caribbean Sea (Harbitz et al., 2012). Of these 85 tsunamis, 17 have resulted in a combined loss of ∼15,000 lives (Harbitz et al., 2012; NGDC, 2016). No tsunami events have been recorded along the Colombian coastline over the last 500 years, consistent with the observed low rate of modern seismicity along this slowly subducting margin (Bernal-Olaya et al., 2015). However, the existence of paleo-submarine landslide deposits described in this paper suggests that tsunamis generated by submarine slope-failures on the Colombian margin pose a significant but infrequent risk to the coastal population of Colombia and the greater Caribbean (Fig. 10). In particular, the nearby locations of the Colombian coastal cities of Cartagena, Barranquilla and Santa Marta relative to the giant MTDs of the Magdalena Fan would provide a warning time of less than 1 h between a potential slope failure and the onset of a tsunami wave.

To understand the tsunamigenic potential of a recurrence of a giant submarine landslide along the slopes of the Magdalena Fan, we model an event similar in size to the Barranquilla MTD using GeoWave tsunami source and propagation code (Watts and Tappin, 2012). GeoWave simulates tsunami generation, propagation, and inundation using a Boussinesq wave model and can model various tsunami source generation mechanisms, including submarine landslides. Initial wave height in the GeoWave model is most influenced by water depth and slide density (Rahiman and Pettinga, 2006; Moscardelli et al., 2010). Slope failure events occurring in shallower water will produce stronger perturbations of the sea surface, while failures occurring in deeper water will have less expression on the sea surface. The scenario tested in our model is based upon our best estimate of the initial conditions that include: 1) a translational submarine landslide that is 46 km long and 37 km wide; and 2) the geographic center of the slide located ∼57 km W/NW from the mouth of the present day Magdalena River in water depths of 1500 m; 3) the thickness for the initial failure estimated at 400 m, resulting in a conservative initial slide volume

S.C. Leslie, P. Mann / Earth and Planetary Science Letters 449 (2016) 382–394

391

Fig. 9. a. Seismic reflection section E–E along axis of the Magdalena Fan from shelf to lower slope. b. Interpretation of potential submarine landslide break-away zone in yellow. Factors contributing to likelihood of catastrophic slope failures include high rate of sediment input from the Magdalena River, tectonic uplift from underlying thrust faults related to the SCDB, and potential planes of weakness related to overpressure directly beneath gas hydrate zone (dashed red line). c. Plot of average seafloor slope over 500 m long window along seismic line. Note steep slopes from shelf edge to mid-upper slope, locally greater than 10◦ in the area overlying active, blind thrusts of the SCDB. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

of ∼680 km3 ; 4) the inclination of the main failure plane beneath the landslide for the simulation at 2.5◦ ; and 5) the direction of motion towards N 5◦ E (Fig. 10a). The maximum tsunami wave heights predicted in the model for such an event would occur along the proximal Colombia coastline, but would also devastate extensive and more distant coastal areas of the Caribbean Sea in other parts of northern South America, Central America, and the Greater Antilles (Fig. 10a). The initial tsunami wave generated by the slope failure has a maximum trough amplitude of −65.8 m and a maximum peak amplitude of 19.2 m. Maximum wave height at the shoreline is most pronounced close to the source area and around bathymetric features that focus or impede wave energy such as the prominent Bahoruco Peninsula of the southernmost Dominican Republic and the various atolls and shallow banks of the western Caribbean between Jamaica and Nicaragua (Fig. 10a). The maximum height radiates in a ‘butterfly’ pattern away from the initial source area offshore northern Colombia due to interference effects between individual

waves within the tsunami wave train in the direction of failure (Fig. 10a). A time series of the tsunami propagation illustrates how quickly the train of tsunami waves reaches population centers in coastal areas of the Caribbean Sea (Fig. 10b–10d). The effects of the tsunami on large cities in the region would include a 21 m high tsunami wave arriving 56 min after the failure event at Cartagena, Colombia (pop. 1.24 M), a 7.5 m high tsunami wave in Kingston, Jamaica (pop. 1.04 M) 82 min after the failure event, and a 6 m high tsunami wave in Santo Domingo, Dominican Republic (pop. 2.91 M) 102 min after the failure event. The results of such tsunami wave-heights would be devastating to affected coastlines, with inundation possible several kilometers inland from the shoreline in areas of relatively flat topography, such as the Ciénega Grande area in Colombia, to the east of the city of Barranquilla (Fig. 1). While such a large tsunami has not occurred in the area during historic times, evidence does exist in the Western Caribbean for

392

S.C. Leslie, P. Mann / Earth and Planetary Science Letters 449 (2016) 382–394

Fig. 10. a. Maximum tsunami wave height predicted after a submarine landslide offshore Northern Coast of Colombia with dimensions based upon the Plio-Pleistocene MTDs described in this paper. Coastal areas affected by tsunamis with wave heights of 6+ m marked with heavy red line outlined in black and include a large region of Colombia, Central America, and the Greater Antilles. Coastal cities affected by tsunami shown as yellow circles. b. Tsunami wave height predicted 30 min after failure event. c. Tsunami wave height predicted 50 min after failure event. d. Tsunami wave height predicted 80 min after failure. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

potential tsunamigenic deposits on the island of Bonaire (Scheffers, 2004) and the Yucatán Peninsula (Shaw and Benson, 2015), although other workers have interpreted these as deposits related to episodic hurricanes and large storms (Morton et al., 2006). If these preserved, terrestrial deposits are indeed tsunami deposits, they are likely related to younger, Holocene tsunamis and it is presently unclear if these more recent tsunamis were generated by an earthquake, submarine landslide, or volcanic flank collapse. In the uplifted and exposed Pliocene sediments along the northern Colombia shoreline, there are possible candidates for paleotsunami deposits within the Tubará formation (Molinares et al., 2012), although further field work is necessary to determine if deposition of these units were related to storm events or tsunamis. As our knowledge and recognition of submarine landslides and associated tsunami hazards in the Caribbean continues to evolve,

data from the Colombian margin should be incorporated into the construction of a Caribbean-wide tsunami hazard, monitoring plan, and public outreach effort, especially as the travel time to major coastal cities across the Caribbean is less than 90 min from areas of likely slope failures on the Colombian Margin. 7. Conclusions Seismic reflection data and seafloor morphology clearly demonstrate that three massive submarine landslides on a scale that falls within the top 5–10% of the largest known submarine slides on Earth, have taken place along the active subduction margin of the Caribbean coast of northern Colombia. These giant slides – although imaged on individual seismic reflection surveys dating back to the 1970’s – were not previously recognized as region-

S.C. Leslie, P. Mann / Earth and Planetary Science Letters 449 (2016) 382–394

ally and globally significant submarine landslides. The implications of a modern submarine slide event of comparable size to the pre-historic examples described in this paper would be significant damage to infrastructure and loss of life along the Colombian coastline and the Caribbean margins of Central America and the Greater Antilles. Acknowledgements We thank Spectrum ASA for providing the seismic reflection data presented in this paper and the Institut de Ciències del Mar, Barcelona, Spain for kindly sharing high resolution bathymetric data covering a part of the Magdalena Fan. We gratefully acknowledge the assistance of Dr. Jean Roger of G-MER Marine Studies for his assistance with the tsunami modeling presented here. We thank Drs. Lesli Wood and David Tappin for their helpful reviews of an earlier version of this paper. Finally, we thank the industry sponsors of the Caribbean Basins, Tectonics, and Hydrocarbons Project (CBTH) for their continued support of our studies at the University of Houston. References Alfaro, E., Holz, M., 2014. Seismic geomorphological analysis of deepwater gravitydriven deposits on a slope system of the southern Colombian Caribbean margin. Mar. Pet. Geol. 57, 294–311. Bartoli, G., Sarnthein, M., Weinelt, M., Erlenkeuser, H., Garbe-Schönberg, D., Lea, D.W., 2005. Final closure of Panama and the onset of northern hemisphere glaciation. Earth Planet. Sci. Lett. 237, 33–44. Bernal-Olaya, R., Mann, P., Vargas, C.A., 2015. Earthquake, tomographic, seismic reflection, and gravity evidence for a shallowly dipping subduction zone beneath the Caribbean Margin of Northwestern Colombia. In: Bartolini, C., Mann, P. (Eds.), Petroleum Geology and Potential of the Colombian Caribbean Margin. American Association of Petroleum Geologists, pp. 247–270. Breen, N., 1989. Structural effect of Magdalena fan deposition on the northern Colombia convergent margin. Geology 17, 34–37. British Oceanographic Data Centre, 2008. General Bathymetric Chart of the Oceans, version 20100927. http://www.gebco.net. Bugge, T., 1983. Submarine slides on the Norwegian continental margin with special emphasis on the Storegga Slide. IKU Report, pp. 1–152. Cadena, A.F., Slatt, R.M., 2013. Seismic and sequence stratigraphic interpretation of the area of influence of the Magdalena submarine fan, offshore northern Colombia. Interpretation 1, SA53–SA74. Coates, A.G., Collins, L.S., Aubry, M.-P., Berggren, W.A., 2004. The geology of the Darien, Panama, and the late Miocene–Pliocene collision of the Panama arc with northwestern South America. Geol. Soc. Am. Bull. 116, 1327–1344. Edgar, N.T., Saunders, J.B., Bolli, H.M., Donnelly, T.W., Hay, W.W., Maurrasse, F., Pérez, H., Premoli, I., Riedel, W.R., Schneidermann, N., 1973. Site 153. Initial reports of the Deep Sea Drilling Project, pp. 367–406. Ercilla, G., Alonso, B., Estrada, F., Chiocci, F.L., Baraza, J., Farran, M.l., 2002. The Magdalena Turbidite System (Caribbean Sea): Present-day morphology and architecture model. Mar. Geol. 185, 303–318. Flinch, J.F., 2003. Structural evolution of the Sinu–Lower Magdalena area (Northern Colombia). In: Bartolini, C., Buffler, R., Blickwede, J. (Eds.), The CircumGulf of Mexico and the Caribbean: Hydrocarbon Habitats, Basin Formation, and Plate Tectonics. The American Association of Petroleum Geologists, Tulsa, pp. 776–796. Haq, B.U., Hardenbol, J., Vail, P.R., 1987. Chronology of fluctuating sea levels since the Triassic. Science 235, 1156–1167. Harbitz, C.B., Glimsdal, S., Bazin, S., Zamora, N., Løvholt, F., Bungum, H., Smebye, H., Gauer, P., Kjekstad, O., 2012. Tsunami hazard in the Caribbean: Regional exposure derived from credible worst case scenarios. Cont. Shelf Res. 38, 1–23. Harbitz, C.B., Lovholt, F., Bungum, H., 2014. Submarine landslide tsunamis: how extreme and how likely? Nat. Hazards 72, 1341–1374. Heinrich, P.H., Piatanesi, A., Hébert, H., 2001. Numerical modelling of tsunami generation and propagation from submarine slumps: the 1998 Papua New Guinea event. Geophys. J. Int. 145, 97–111. Hoorn, C., Guerrero, J., Sarmiento, G.A., Lorente, M.A., 1995. Andean tectonics as a cause for changing drainage patterns in Miocene northern South America. Geology 23, 237–240. Hornbach, M.J., Braudy, N., Briggs, R.W., Cormier, M.-H., Davis, M.B., Diebold, J.B., Dieudonne, N., Douilly, R., Frohlich, C., Gulick, S.P.S., Johnson III, H.E., Mann, P., McHugh, C., Ryan-Mishkin, K., Prentice, C.S., Seeber, L., Sorlien, C.C., Steckler, M.S., Symithe, S.J., Taylor, F.W., Templeton, J., 2010. High tsunami frequency as

393

a result of combined strike-slip faulting and coastal landslides. Nat. Geosci. 3, 783–788. Kolla, V., Buffler, R.T., 1984. Morphologic, acoustic, and sedimentologic characteristics of the Magdalena Fan. Geo Mar. Lett. 3, 85–91. Kolla, V., Buffler, R.T., Ladd, J.W., 1984. Seismic stratigraphy and sedimentation of the Magdalena Fan, Southern Colombian Basin, Caribbean Sea. Am. Assoc. Pet. Geol. Bull. 68, 316–332. Krause, D.C., 1971. Bathymetry, geomagnetism, and tectonics of the Caribbean Sea north of Colombia. Mem. Geol. Soc. Amer. 130, 35–54. Kvenvolden, K.A., 1993. Gas hydrates—geological perspective and global change. Rev. Geophys. 31, 173–187. Ladd, J.W., Truchan, M., Talwani, M., Stoffa, P.L., Buhl, P., Houtz, R., Mauffret, A., Westbrook, G., 1984. Seismic reflection profiles across the southern margin of the Caribbean. In: Bonina, W.E., Hargraves, R.B., Shagan, R. (Eds.), The Caribbean– South American Plate Boundary and Regional Tectonics. In: Mem. Geol. Soc. Amer., vol. 162. Geological Society of America, Boulder, CO, pp. 153–160. Lo Iacono, C., Gràcia, E., Zaniboni, F., Pagnoni, G., Tinti, S., Bartolomé, R., Masson, D.G., Wynn, R.B., Lourenço, N., Pinto de Abreu, M., Dañobeitia, J.J., Zitellini, N., 2012. Large, deepwater slope failures: implications for landslide-generated tsunamis. Geology 40, 931–934. Maslin, M., Mikkelsen, N., Vilela, C., Haq, B., 1998. Sea-level- and gas-hydratecontrolled catastrophic sediment failures of the Amazon Fan. Geology 26, 1107–1110. Molinares, C.E., Martinez, J.I., Fiorini, F., Escobar, J., Jaramillo, C., 2012. Paleoenvironmental reconstruction for the lower Pliocene Arroyo Piedras section (Tubará – Colombia): implications for the Magdalena River – paleodelta’s dynamic. J. South Am. Earth Sci. 39, 170–183. Moore, G.T., Fahlquist, D.A., 1976. Seismic profile tying Caribbean DSDP Sites 153, 151, and 152. Geol. Soc. Am. Bull. 87, 1609–1614. Morton, R.A., Richmond, B.M., Jaffe, B.E., Gelfenbaum, G., 2006. Reconnaissance Investigation of Caribbean Extreme Wave Deposits; Preliminary Observations, Interpretations, and Research Directions. U.S. Department of the Interior, U.S. Geological Survey. Moscardelli, L., Hornbach, M., Wood, L., 2010. Tsunamigenic risks associated with mass transport complexes in offshore Trinidad and Venezuela. In: Mosher, D.C., Shipp, R.C., Moscardelli, L., Chaytor, J.D., Baxter, C.D.P., Lee, H.J., Urgeles, R. (Eds.), Submarine Mass Movements and Their Consequences. Springer Netherlands, Dordrecht, pp. 733–744. Moscardelli, L., Wood, L., 2015. Morphometry of mass-transport deposits as a predictive tool. Geol. Soc. Am. Bull. http://dx.doi.org/10.1130/B31221.1. Moscardelli, L., Wood, L., Mann, P., 2006. Mass-transport complexes and associated processes in the offshore area of Trinidad and Venezuela. Am. Assoc. Pet. Geol. Bull. 90, 1059–1088. Mulder, T., Cochonat, P., 1996. Classification of offshore mass movements. J. Sediment. Res. 66, 43–57. National Geophysical Data Center, 2016. National Geophysical Data Center/World Data Service (NGDC/WDS): Global Historical Tsunami Database. Ortiz-Karpf, A., Hodgson, D.M., McCaffrey, W.D., 2015. The role of mass-transport complexes in controlling channel avulsion and the subsequent sediment dispersal patterns on an active margin: the Magdalena Fan, offshore Colombia. Mar. Pet. Geol. 64, 58–75. Rahiman, T.I.H., Pettinga, J.R., 2006. The offshore morpho-structure and tsunami sources of the Viti Levu Seismic Zone, southeast Viti Levu, Fiji. Mar. Geol. 232, 203–225. Rincón, D.A., Arenas, J.E., Cuartas, C.H., Cárdenas, A.L., Molinares, C.E., Caicedo, C., Jaramillo, C., 2007. Eocene–Pliocene planktonic foraminifera biostratigraphy from the continental margin of the southwest Caribbean. Stratigraphy 4, 261–311. Romero Otero, G.A., Slatt, R., Pirmez, C., 2015. Evolution of the Magdalena deepwater fan in a tectonically active setting, offshore Colombia. In: Bartolini, C., Mann, P. (Eds.), Petroleum Geology and Potential of the Colombian Caribbean Margin. AAPG, pp. 675–708. Sarria, A., Garcia, L.E., Espinosa, A., Yamin, L.E., 1996. Seismic risk maps for the update of the Colombian Seismic Code. In: Eleventh World Conference on Earthquake Engineering. Acapulco, Mexico, pp. 1–8. Scheffers, A., 2004. Tsunami imprints on the Leeward Netherlands Antilles (Aruba, Curaçao, Bonaire) and their relation to other coastal problems. Quat. Int. 120, 163–172. Shaw, C.E., Benson, L., 2015. Possible tsunami deposits on the Caribbean coast of the Yucatán Peninsula. J. Coast. Res., 1306–1316. Shipboard Scientific Party, 1997. Site 999. In: Sigurdsson, H., Leckie, R.M., Acton, G.D., et al. (Eds.), Proceedings of the Ocean Drilling Program Initial Reports Leg 165. Ocean Drilling Program, College Station, TX, pp. 131–230. Shipley, T.H., Houston, M.H., Buffler, R.T., Shaub, F.J., McMillen, K.J., Ladd, J.W., Worzel, J.L., 1979. Seismic evidence for widespread possible gas hydrate horizons on continental slopes and rises. Am. Assoc. Pet. Geol. Bull. 63, 2204–2213. Smith, D.E., Shi, S., Cullingford, R.A., Dawson, A.G., Dawson, S., Firth, C.R., Foster, I.D.L., Fretwell, P.T., Haggart, B.A., Holloway, L.K., Long, D., 2004. The Holocene Storegga Slide tsunami in the United Kingdom. Quat. Sci. Rev. 23, 2291–2321. Talling, P.J., Wynn, R.B., Masson, D.G., Frenz, M., Cronin, B.T., Schiebel, R., Akhmetzhanov, A.M., Dallmeier-Tiessen, S., Benetti, S., Weaver, P.P.E., Georgiopoulou, A.,

394

S.C. Leslie, P. Mann / Earth and Planetary Science Letters 449 (2016) 382–394

Zuhlsdorff, C., Amy, L.A., 2007. Onset of submarine debris flow deposition far from original giant landslide. Nature 450, 541–544. Tappin, D.R., Grilli, S.T., Harris, J.C., Geller, R.J., Masterlark, T., Kirby, J.T., Shi, F., Ma, G., Thingbaijam, K.K.S., Mai, P.M., 2014. Did a submarine landslide contribute to the 2011 Tohoku tsunami? Mar. Geol. 357, 344–361. Tappin, D.R., Watts, P., Grilli, S.T., 2008. The Papua New Guinea tsunami of 17 July 1998: anatomy of a catastrophic event. Nat. Hazards Earth Syst. Sci. 8, 243–266. Vinnels, J.S., Butler, R.W.H., McCaffrey, W.D., Paton, D.A., 2010. Depositional processes across the Sinú Accretionary Prism, offshore Colombia. Mar. Pet. Geol. 27, 794–809.

von Hillebrandt-Andrade, C., 2013. Minimizing Caribbean tsunami risk. Science 341, 966–968. Ward, S.N., 2001. Landslide tsunami. J. Geophys. Res., Solid Earth 106, 11201–11215. Watts, P., Tappin, D.R., 2012. GeoWave validation with case studies: accurate geology reproduces observations. In: Yamada, Y., Kawamura, K., Ikehara, K., Ogawa, Y., Urgeles, R., Mosher, D., Chaytor, J., Strasser, M. (Eds.), Submarine Mass Movements and Their Consequences: 5th International Symposium. Springer Netherlands, Dordrecht, pp. 517–524. Weimer, P., 1989. Sequence stratigraphy of the Mississippi fan (Plio-Pleistocene), Gulf of Mexico. Geo Mar. Lett. 9, 185–272.