OS21E-01 INVITED 08:00h
Physical Property Correlations from Cascadia Great Earthquakes: What Are They Telling Us About The Triggering Events?
We have found that it is possible to correlate the physical property signatures of probable earthquake-generated turbidites from locale to locale down individual channels. This indicates that the details of the turbid flow that are relevant to deposition of the turbidite, apparently maintain their integrity for long distances within channels. This in itself is somewhat surprising, but we have been able to correlate event signatures not only down individual channels, but between channel systems, some of which never meet. We see a general correspondence of turbidite size that is reflected in these separate channels, as well as correlatable details such as the number of coarse pulses (density and magnetic peaks). For example, Cascadia events T5, T10, and T12 are small events in all cores at all sites. T6, T8 and T16 are large triplet events in all cores, and most other events follow similar size patterns across the margin. We observe similar patterns in our SAF cores thus far. This suggests that there may be some fundamental relative size relationship to the underlying earthquakes, or alternatively perhaps to sediment supply events such as major storms or volcanic eruptions. The correlation we see between events also suggests that there may be some persistent signature of individual events that is recorded in the cores. Why should this be the case? One might expect that such correlation could be due to details of how the turbid flow initiated in the canyon's upper reaches. An earthquake, unlike other triggers for submarine landslides, is likely to trigger multiple failures within a canyon. Thus the turbid flow should contain multiple inputs, each perhaps containing a coarse fraction pulse, which coalesce down-channel. In fact we do see this effect in some cores, however we commonly see correlation between channels that never meet, where persistent multiple input signature cannot be the cause. We suggest that the only plausible commonality in such cases is the original earthquake itself, in effect they the signatures may be crude paleoseismograms.
OS21E-02 08:15h
Propagation of Short, Dispersive Tsunami Waves in Ocean Basins
Tsunami waves generated by mass failure events oscillate with relatively higher frequencies than waves generated in response to seismic faulting. As a result, these waves can be significantly modified by frequency dispersion effects while propagating over distances typically found in coastal seas or open ocean basins. Recent investigations have indicated that large scale underwater slide events are capable of generating significant transoceanic tsunami waves, as (for example) is hypothesized in the case of the 1946 Ugamak slide in the Aleutian Islands. In order to provide a comprehensive framework for modeling weakly dispersive waves from source to eventual landfall, we have developed a Boussinesq-type model for water surface waves in an ocean-scale basin. The basic model is formulated in spherical polar coordinates. Scaling analysis is utilized to simplify the model and to distinguish between the regime of long waves (with horizontal length scales approaching the basin dimensions) and shorter waves with horizontal scales commensurate with the short wave packet dimensions. The second scaling produces a model in which the usual weakly-dispersive Boussinesq model corrections are retained. The model retains a fully nonlinear formulation in its treatment of surface boundary conditions, allowing for accurate modeling of runup and propagation in the nearfield of shorelines. The model also retains the specification of an arbitrary (but for now externally prescribed) bottom motion, allowing for direct simulation of the response at the source. Finally, we are presently developing a formulation in curvilinear coordinates to allow for mapping computational grids to the ocean basin geometry. The model is used to examine tsunamis generated by several idealized sources in idealized ocean basins, in order to characterize farfield response to hypothetical slide properties. We hope to examine the 1946 Aleutian Islands event in detail by the time of the presentation.
OS21E-03 INVITED 08:30h
Landslide Basal Friction as Measured by Seismic Waves
Dynamical predictions of landslide runout require measurements of the basal friction. We present the first seismically determined bounds on the frictional coefficients for three large volcanic landslides. A landslide generates seismic waves by both shearing and loading the surface as the mass moves from a steep to a shallow slope. The effective force system is a horizontal single force. The amplitude of the seismic waves is proportional to the force drop during the landslide, just as during an earthquake the seismic wave amplitude is proportional to the seismic moment, i.e., the force drop multiplied by the source dimension. For landslides we know an additional variable that is unknown for the earthquake case. We know the gravitational driving force of the landslide while the magnitude of the tectonic forces that drive earthquakes are generally unknown. Therefore, we can find the absolute value of the frictional force for landslides whereas we are unable to perform this calculation for earthquakes. We studied three landslides (Bezymianny, Russia 1956, Sheveluch, Russia 1964 and Mount St. Helens, USA 1980) that were all followed immediately by eruptions. The landslide masses vary by a factor of 5. We test the data against the hypothesis of a constant value of apparent friction. The apparent friction $\mu_{app}$ is defined as the ratio of the amplitude of the horizontal single force to the weight of the landslide. Since the Mount St. Helens seismic source is very well-constrained, we use the amplitude of this landslide force drop as a starting point. We calculated the value of $\mu_{app}$ for Mount St. Helens using previous seismic results and the geological data. We then test whether or not the other two landslides are consistent with the same value of $\mu_{app}$. We use teleseismic and regional seismic data to show that all three landslides are consistent with an apparent coefficient of friction of 0.2 which corresponds to an actual areally-averaged frictional coefficient of 0.2--0.6. We find that the apparent friction is independent of the quantity of hot gas subsequently released, i.e., all the landslides are consistent with $\mu_{app}$=0.2 even though some had directed blasts and others did not. In addition, the data rule out viscous flow as the constitutive model for basal shear. The seismic data are consistent with a model where the amplitude of the shear force scales with landslide mass, but they are inconsistent with a model where the force scales with landslide area.
OS21E-04 08:45h
Dynamical Simulation of Subaerial and Submarine Landslides Detaching From the Volcanic Island of Stromboli, Italy: a Sensitivity Analysis
Stromboli is an active volcano that is found in the southern Tyrrhenian sea, Italy, and is known to be unstable. In addition to large lateral collapses in the order of 1 cubic kilometre that were ascertained to occur several times in the last 13000 years, the volcanic cone is affected by mass failures involving volumes of millions of cubic metres, that may generate tsunamis. These slides are usually concomitant with the paroxysmal eruptions of the volcano that repeated on average every 15-20 years in the last century. The last documented case is the double tsunami that took place on December 30, 2002, and was induced by two main episodes of failure, occurring only 7 minutes apart in the north-west flank of Stromboli, known as "Sciara del Fuoco" (SdF). The tsunamis were very damaging and hit the northern coast of Stromboli with wave heights larger than 10 meters in several places. In this work we perform numerical simulations of slides that are supposed to detach from the subaerial and submarine portion of the SdF. The simulations are carried out by applying an enhanced Lagrangian model that was originally implemented by Tinti et al. (1997). It is a centre-of-mass model, and consists in dividing the mass in a chain of blocks (1D) or in a matrix of blocks (2D) that are allowed to deform: the motion of each centre of mass is computed, depending on the geometrical characteristics of the sliding body and on the environmental conditions (basal friction, drag and shear resistance exerted by the water, etc.). The sensitivity analysis we have performed enables us to determine the influence of the initial volume, geometry, position, density, etc. of the body on the dynamical evolution of the sliding (trajectory, velocity, etc.) and on the final run-out distances, and to make some inferences on the main characteristics of the generated tsunami.
OS21E-05 INVITED 09:00h
The Silent Earthquakes of Kilauea's South Flank and their Ramifications for Catastrophic Failure
The south flank of Kilauea Volcano is one of the most actively deforming regions on the planet. GPS measurements taken there since the early 1990s show a persistent seaward migration at rates exceeding 5 cm/yr. Large earthquakes occur on Kilauea with alarming frequency. The largest in historical times, a M7.2 event in 1975, caused more than 6 m of seaward displacement in addition to 3.5 m of coastal subsidence. In the last few years, an intermediate form of south flank deformation has been observed. With slip rates of about 10 cm/day, these so-called ``silent earthquake'' are much slower than the nearly instantaneous brittle failure of normal earthquakes, but vastly faster than the creep that carries the south flank relentlessly toward the sea. Kilauea is the latest subaerial volcano in an archipelago of volcanic islands stretching back to the Kamchatka Peninsula. In just the youngest part of this chain, the modern Hawaiian islands, there are more than 70 submarine debris fields, each thought to represent an ancestral flank collapse. If these collapses occurred catastrophically, they almost certainly created tsunami large enough to inundate nearby islands and perhaps long-lived enough to threaten the entire Pacific basin. Evidence for prehistoric inundation on the Hawaiian island consists of anomalously elevated detrital coral deposits, found at several locales and interpreted as the diaspora of passing tsunami. As the youngest and most active of the subaerial Hawaiian volcanoes, it is natural to wonder if Kilauea--in particular the south flank--is the most likely place for future catastrophic collapses. Since the creation of the continuous GPS network on Kilauea in the mid-1990s, as many as four silent earthquakes have been detected. Two of these, occurring in November 2000 and July 2003, have resulted in elastic deformation fields large enough to model. The modeling shows that these events occurred on a shallow ($\sim 5$ km depth) landward dipping fault, possibly the down-dip extension of the Hilina normal fault complex. Distinctive bathymetric features offshore from the south flank are highly suggestive of a buttressing toe. Thus, the entire structure of surficial normal faults, inferred down-dip extension, and the observed bathymetric morphology, can be plausibly interpreted as a listric fault that accommodates gravity-driven block rotation. If this structure also accounts for the steady creep and occasional large earthquakes, then the current active deformation within Kilauea's south flank probably promotes stability rather than catastrophe. It is difficult to see how a rotational block can ``run away'' along a listrically curved fault that is blocked at its distal end by a buttressing toe. So, Kilauea, at least in its current stage of growth, is likely not a locus for catastrophic or even large flank collapse.
OS21E-06 09:15h
Reinterpretation of a Stage-11 21-m Sea Level High Stand Deposit on Bermuda: Megatsunami from a Giant Submarine Landslide?
A deposit of coral-bearing, marine shell conglomerate found at an elevation 21 m above sea level (asl) in a Bermuda cave has previously been interpreted as evidence of a towering sea level high stand during marine isotope stage 11. Hearty et al. (1999) U-series dated this deposit at 420 +/- 30 ka and correlated it with carbonate sand deposits at up to 21 m asl in the Bahamas; they attributed this putative sea level high stand to a 20% decrease in Antarctic ice volume and suggested direct geologic evidence for ice sheet instability. Although stage 11 was an extended interglacial period similar to the Holocene, few marine isotope sea level curves support a high stand greater than that of stage 5e over the past million years, and those that do suggest a maximum height of no more than about 9 m asl. The sea level evidence is primarily from Bermuda, where no subsidence has been identified, and Oahu, Hawaii, where the rate of uplift is generally known and the Kaena high stand, containing corals U-series dated at 400 to 550 ka , is exposed at up to 30 m asl. On Oahu, the uplift for the island is recorded by a nearly linear (r = 0.999; n = 7) emergence rate of 0.069 +/- 0.003 mm/kyr, using isotopically-dated, uplifted marine terraces back to nearly 500 ka BP. This uplift rate is supported by identical stage 7 comparisons with Bermuda, Papua New Guinea and Barbados (Jones, 1993). When this rate is applied to the "best" U-series age for stage 11 (430 ka) at the maximum elevation of 30 m asl, we arrive at a maximum sea level height for stage 11 that is no greater than the present sea level. This evidence, together with: (1) a lack of layering in archived samples of the 21-m Bermuda conglomerate; (2) its reported field relations; and (3) the occurrence of stage 11 deposits at only 5 +/- 3 m asl elsewhere on Bermuda (Hearty et al., 1992), suggest an alternative deposition mechanism for this deposit. We hypothesize that this conglomerate was rapidly deposited by a "megatsunami" during stage 11, an interpretation that is consistent with the climate correlation of McMurtry et al. (2004). Likely causes for such a megatsunami are a flank collapse of an Atlantic oceanic island volcano, such as the Canary or Azores Islands, or a giant submarine landslide on the Atlantic continental margin.
OS21E-07 09:30h
Slope Instability and Gas Hydrates in the Hudson Canyon Region, U.S. Atlantic Continental Margin
The continental slope and the upper rise centered on Hudson Canyon offshore New York and New Jersey lie within a major gas-hydrate province. This region exhibits evidence of gravitational mass movements and possible methane expulsion, as inferred from our bathymetric and water-column surveys conducted in 2002 with support from NOAA/OE, and prior data. The bathymetric data cover our study area (200 km by 110 km; 37\deg40'N to 39\deg50'N, 70\deg00'W to 72\deg30'W) from the inner edge of the continental slope (depth 200 m) seaward to the middle rise (c.3500 m). The world's largest hub of submarine telecommunications cables partially passes through this area. Evidence of gravitational mass movements and of probable gas release is extensive. Examples of the former include: (1) blocks of landward-dipping strata up to 2-km wide and 150-m high that lie at the base of the continental slope (water depth 2100-2200 m) seaward of an over-pressured zone beneath the continental slope (639 mbsf in ODP Hole 1073A; water depth 650 m; Dugan and Flemings, 2000); (2) boulders of Eocene chalk that litter the lower slope and upper rise; (3) a semicircular, tabular glide block, about 20 km in diameter, which thickens to about 150 m at its seaward margin; the block is centered at 39\deg23.5'N, 71\deg10.0'W between 2450 and 2600 m depth on the upper rise, about 15 km downslope from a congruent scarp at 2200 m on the lower slope; (4) apparent penecontemporaneous faulting and gliding in strata inclined sub-parallel to the seafloor along the upper rise; 5) apparent clogging of Hudson Canyon with hummocky sediment at a right-angle turn of the axis (depth 3368 m; 38\deg39.6'N, 71\deg01.8'W); 6) changes in stratification from the upper to middle rise; uneven layering beneath the upper rise (seafloor mean inclination 0.75\deg down to 2700 m) is inferred to reflect disturbance by gravitational mass movements; even layering parallel to the seafloor beneath the middle rise (inclination increase seaward from 0.25\deg to 0.76\deg) may reflect less disrupted hemipelagic sedimentation. Evidence of gas release includes: (1) a zone of irregular pits, each up to 500-m in diameter and spaced kilometers apart, that extends along the upper rise (2600-2650 m); (2) a line of eight depressions each 50-70 m in diameter with several meters relief and spaced 100-120 m apart that trends NNW (2150 m; near 39\degN, 71\deg52'W)and that may indicate gas or porewater escape along a local fault; 3) a plume of sediment suspended in fluid that is discharging through the seafloor and rising about 1 m at 2625 m depth on the SW margin of Hudson Canyon (38\deg52.4'N, 71\deg31.0'W) recorded on deep-towed video; and 4) three zones of regional, water-column, methane anomalies, which exceed background values, deeper than 2500-m depth, centered at 1200-m depth, and near 200-m depth. These features indicate past and present dissociation of gas hydrates and/or venting of free gas. They also indicate that coherent gravitational mass movements over low seafloor slopes (< 1\deg) have been facilitated by excess fluid (water and/or gas) pressures. The continental slope and rise in the Hudson Canyon region comprise a natural laboratory in which to study slope instability in a gas hydrate province and to assess the hazards to telecommunications cables. Dugan, B. and P.B. Flemings (2000), Overpressure and fluid flow in the New Jersey continental slope: implications for slope failure and cold seeps, Science, 289, 288-291.
OS21E-08 09:45h
Mapping and Modelling of the PNG Slump - 3-D Evidence to demonstrate a Tsunami Source?
The original offshore data set for the slump that is now generally accepted as the source of the 1998 PNG tsunami was originally presented as 2-D bathymetry images, seismic sections and as seabed photographs. The dataset images a cohesive rotational failure offshore off the north coast of Papua New Guinea, mathematical modelling of which provides appropriate run-ups comparable with onshore measurements from field surveys. The regional bathymetry and seismic data acquired off the north coast of PNG images a deeply incised, sediment-starved convergent margin subsiding along the New Guinea Trench. An arcuate shaped feature off the Sissano Lagoon, termed the `amphitheatre', is identified as the source location of the failure that is located in the east of this feature. The presence of a slump in this eastern area is confirmed by seismic data and observations from Remote and Manned Submersibles, that show seabed features, such as fissures and fractured limestone, on the surface of the slump and interpreted as due to sediment movement. Absolute dating of slump failure is not possible with the present data set, but the relative, recent, age of failure is based on the fresher appearance of fissures in the slump area as well as a greater concentration of a chemosynthetic cold-water biota together with active fluid expulsion on the slump surface. The chemosynthetic biotas comprise mussels and tubeworms and bacterial mats. Laminar bedded chemosynthetic limestone was only observed on the slump surface and represents a low volume background sulphide and methane rich fluid seepage. The concentration of living cold-water faunas on the slump surface is interpreted as the result of an increased fluid expulsion rate associated with the slumping. This presentation uses new interactive software, Fledermaus, to image the northern PNG offshore area, including the amphitheatre, to show the seabed morphology in 3-D and the relationships between the regional geology and the slump area. Use of this software allows the integration of the surface morphology and the subsurface seismic data together with the seabed images, allowing a better understanding of the region and its tsunamigenic potential.