DOI 10.1007/s00367-002-0113-y

Original

An experiment demonstrating that marine slumping is a mechanism to transfer methane from seafloor gas-hydrate deposits into the upper ocean and atmosphere

C. K. Paull^1,, P. G. Brewer¹, W. Ussler III¹, E. T. Peltzer¹, G. Rehder² and D. Clague¹

(1)	Monterey Bay Aquarium Research Institute, 7700 Sandholdt Road, Moss Landing, California 95039, USA
(2)	GEOMAR, 24148 Kiel, Germany

E-mail: paull@mbari.org

Received: 13 June 2002 / Accepted: 27 September 2002 / Published online: 13 November 2002

Abstract. The scientific community is engaged in a lively debate over whether and how venting from the gas-hydrate reservoir and the Earth's climate is connected. The various scenarios which have been proposed are based on the following assumptions: the inventory of methane gas-hydrate deposits is locally enormous, the stability of marine gas-hydrate deposits can easily be perturbed by temperature and pressure changes, enough methane can be released from these deposits to contribute adequate volumes of this isotopically distinct greenhouse gas to alter the composition of oceanic or atmospheric methane reservoirs, and the mechanisms exist for the transfer of methane from deeper geologic reservoirs to the ocean and/or atmosphere. However, some potential transfer mechanisms have been difficult to evaluate. Here, we consider the possibility of marine slumping as a mechanism to transfer methane carbon from gas hydrates within the seafloor into the ocean and atmosphere. Our analyses and field experiments indicate that large slumps could release volumetrically significant quantities of solid gas hydrates which would float upwards in the water column. Large pieces of gas hydrate would reach the upper layers of the ocean before decomposing, and some of the methane would be directly injected into the atmosphere.

Introduction

Many authors (e.g., Paull et al. 1991; Dickens et al. 1995; Haq 1998; Mienert et al. 2001) have postulated that the release of large volumes of methane into the Earth's oceans from gas hydrate stored within the shallow geologic reservoirs along the continental margins would alter the composition of either the dissolved carbon in the ocean or the methane concentration within the atmosphere. Such conjecture is based on both the size of the gas-hydrate reservoir and the distinct isotopic composition of methane.

Currently, the amount of methane carbon which is stored as gas hydrate in the shallow sedimentary sections along the world's continental margins is estimated to be 25% of the amount of dissolved inorganic carbon in the ocean, and ~10⁴ times the amount of methane carbon in the atmosphere (Kvenolden 1988). The isotopic composition of the methane contained in marine gas hydrates (<-60 permil PDB) is distinct from the isotopic composition of other oceanic and atmospheric carbon reservoirs (0±7 permil PDB). Thus, if a fraction of the methane carbon which is currently in the shallow subsurface along continental margins were transferred into the ocean, where it would be oxidized (Scranton and Brewer 1978; Rehder et al. 1999) to dissolved inorganic carbon (DIC), this event could alter the isotopic budget of the ocean's DIC reservoir (Dickens et al. 1995, 1997).

Abrupt excursions in the stable isotopic composition of marine carbonate contained in stratigraphic sections from several areas and time periods suggest that large releases of isotopically light carbon have occurred in the past, which are interpreted to have come from methane gas hydrate (Dickens et al. 1995, 1997; Hesselbo et al. 2000; Kennett and Frackler-Adams 2000; Kennett et al. 2000). Moreover, a rapid and direct transfer of this methane into the atmosphere could significantly change the Earth's atmospheric methane composition. Because methane is an effective greenhouse gas, this transfer could potentially affect the Earth's radiant heat budget (Lelieveld and Crutzen 1992).

Mechanisms to transfer gas from gas hydrates out of the sediments and upwards into the ocean and atmosphere, associated with the thermal decomposition of gas hydrate, have been evaluated by, amongst others, Xu et al. (2001). However, the consequence of turbulence associated with submarine landslides destroying the sediment fabric and allowing the release of massive pieces of gas hydrate from the sediment matrix has not been extensively considered. Once freed, the solid gas-hydrate pieces would float upwards because of their lower density (~0.91 g/cm³) relative to seawater (>1.03 g/cm³). The effectiveness of this process depends on the amount of gas hydrate which is released, the efficiency of its transfer, and the level which the gas hydrate reaches before completely decomposing.

Continental margins contain abundant scars left by submarine landslides (Schwab et al. 1993). The size and shape of these scars suggest that the volume of material removed in these slumping events may involve tens and even thousands km³ of material. Many authors have tentatively related some of these major slumps on continental margins to instability associated with the breakdown of gas hydrate (see references in Paull et al. 2000). This inference is usually based on the observation that bottom simulating reflectors (BSR), the only commonly available remote detection indicator for the presence of gas hydrate, occur in the sediments which are immediately adjacent to the slide scar. Although a causal relationship between slope failures and gas-hydrate decomposition has not been proven, this observation clearly implies that the sediments involved in many slumping events initially contained gas hydrates. Whether or not gas hydrates play a role in generating weakness in sediments along continental margins, the fate of the gas-hydrate-bound methane contained in the affected slump mass needs to be considered.

Gravity-driven slope instabilities can take various forms (Hampton et al. 1996). Many slides, and especially large slides, are commonly associated with enough turbulence to destroy the cohesion between sediment grains. As a result, the original fabric of the sediments will be destroyed as the sediment disintegrates into its constituent grains and a dense turbid water mass is produced. This turbid water mass may move energetically downslope as a gravity-driven flow. Such flows can travel great distances before depositing thick, graded sedimentary deposits which are common in deep-sea basins. The location of slide scars indicates that the sources for materials in the gravity flows frequently are continental margin sediments which are likely to have contained gas hydrate (Booth et al. 1994). Thus, the likely presence of gas hydrate in near-seafloor sediments and the disruption of these sediments during slides may provide opportunity for escape of methane to the ocean/atmosphere system.

In this study, the fate of gas-hydrate solids which are released from the sediment matrix are considered using data and observations collected from Hydrate Ridge, offshore Oregon. These data include (1) multi-beam seafloor reflectivity and remote operated vehicle (ROV) observations which document the area over which evidence of gas hydrates occurs, (2) ROV-collected pressure core samples which document how much gas hydrate occurs in the surface sediments, and (3) ROV observations of what happens when gas-hydrate-bearing sediments are disturbed. Taken together, these observations provide constraints which should allow us to assess the potential for transfer of methane from seafloor sediments into the ocean and atmosphere.

Materials and methods

Multi-beam bathymetry data over Hydrate Ridge off Oregon (Fig. 1) were collected in 1998 using an EM-300 system (Clague et al. 2001). This system collects both bathymetry and acoustic backscatter data.

Fig. 1. A Illuminated bathymetry and B backscatter amplitude (light shading is high backscatter) data from Hydrate Ridge, Oregon. Contour interval is 50 m. Also shown are ODP site 892C on the northern ridge, and the sites near the southern ridge proposed for drilling during ODP leg 204 (HR-01, HR-02, and HR-03). Arrow indicates location of the ROV Tiburon dive sites

Monterey Bay Aquarium Research Institute's (MBARI) ROV Tiburon made seven dives on the southern part of Hydrate Ridge in 2000. During these dives, areas of seafloor which were associated with the highest reflectivity in the EM-300 data on the southern crest of Hydrate Ridge were visited to provide ground truth for the causes of higher reflectivity.

To determine how much gas is contained in these sediments, four push cores were taken with the ROV from areas of high seafloor reflectivity and placed into the MBARI clathrate bucket (Erickson et al. 2001). This is a chamber which can be sealed on the seafloor so that short push cores (0.6-l volume) are recovered at in-situ pressures. After the vehicle was recovered, pressure from the clathrate bucket was progressively released by venting gas through a series of valves into an overturned, graduated cylinder placed in a salt-saturated water bath (Paull et al. 1996) which allows the total volume of evolved gas to be measured.

The ROV's robotic arm (Kraft Telerobotic) was used to probe the near-seafloor sediments. The wrist on the ROV's arm is on a ball joint which makes it capable of continuous rotation. Thus, the surface sediments (0-20 cm) could be churned by rotating the wrist while the arm was inserted into the sediment. Individual pieces of material which floated upwards into the water column were followed by the ROV up to 180 m above the seafloor before visual contact was lost. Observations of the rise rate and changes in the character of the floating material were made as a function of depth.

Another experiment involved capturing buoyant clumps of gas hydrate within an imaging box (Fig. 2), which consists of a rectangular plastic box (25 cm deep, 30 cm wide and 95 cm tall) with an open bottom and a porous, meshed top. The 30 times 95 cm side of the imaging box faces the ROV's camera and is transparent, whereas the other side is composed of white backlit plastic. Buoyant clumps of gas-hydrate-bearing sediments were captured under the mesh on top of the imaging box while the ROV's arm churned the sediment. The ROV then moved upwards at a rate which retained most of the buoyant pieces within the box, thus simulating the undisturbed, free ascent of fragments of gas hydrate and bound sediment while allowing continuous observation. Changes in this material, as it ascended from depth through the top of the gas-hydrate stability zone to the surface, were observed and documented with the ROV's video cameras. Both experiments to track gas-hydrate solids within the water column are more fully documented in Brewer et al. (2002).

Fig. 2. Video frame grab taken from the Tiburon in 779-m water depth, showing a carbonate- and bacterial mat-covered seafloor where the ROV's mechanical arm (to the right) is about to disturb the sediments underneath the imaging box (to the left)

Results

The detailed bathymetry of Hydrate Ridge shows two broad summits with crests at 588- and 773-m water depths (Fig. 1). The local conditions are such that the top of the gas-hydrate stability field is in the water column at ~400 m water depth, well above the summits of Hydrate Ridge (Peltzer and Brewer 2000; Brewer et al. 2002). Previous sampling in this area (Suess et al. 2001) documents that gas hydrate and other related phenomena, such as authigenic carbonates and chemosynthetic communities, occur on the seafloor and in the subsurface associated with both ridge crests. The crests of Hydrate Ridge are associated with higher acoustic backscatter than the surrounding seafloor. These higher backscatter areas are believed to be produced by chemosynthetic biological communities, authigenic carbonates and/or gas-hydrate occurrences in the near-seafloor environment (Clague et al. 2001).

The Tiburon dives focused on the higher backscatter areas on the southern crest of Hydrate Ridge. Extensive patches of bacterial mats, massive outcrops of authigenic carbonate and some clam shell beds (Fig. 2) were common near the ridge crest. In three locations, small streams of bubbles were seen emanating from the seafloor, sometimes associated with bacterial mats but in one case from otherwise unremarkable sediment. The locations of areas in which authigenic carbonates, chemosynthetic communities, and venting gas bubbles occurred matched the areas associated with high backscatter in the EM-300 data.

Under most areas associated with chemosynthetic macrofaunal communities and bacterial mats, the ROV could only insert push cores or its arm ~20 cm into the sediment before encountering a hard substrate.

The total volume of gas (free and gas hydrate) contained within the material overlying the hard substrate in the four clathrate bucket push core samples varied considerably (Table 1). Core T-171 CB-1 was a full core and contained typical hemipelagic sediment. Core T-172 CB-1 recovered no sediment. Thus, it is ambiguous as to whether the recovered gas was from the decomposition of gas hydrate or from gas bubbles captured within the corer. Cores T-168 CB-1 and T-169 CB-1 were hard to insert and contained black sediment and authigenic carbonates, indicative of active anaerobic diagenesis (Suess et al. 2001). These two cores suggest that there is ~5 volumes of gas at Earth surface conditions per volume of in-situ sediment above the hard substrate.

Table 1. Volume of gas released from core samples collected in the MBARI clathrate bucket from the southern summit of Hydrate Ridge

Sample	Volume gas (l)	Volume gas per volume core
T-168 CB-1	2.67	>4.44
T-169 CB-1	3.72	5.36
T-171 CB-1	0.13	0.21
T-172 CB-1	0.52	-

The release of gas bubbles was stimulated by probing the near-surface sediment with the ROV's mechanical arm. The rate of gas bubble escape increased when the wrist of the ROV's arm was rotated within the sediments for a few minutes. Clumps of material floated upwards off the seafloor out of the excavated hole. The first pieces were small (~1-2 cm in maximum dimension) but, after a few minutes of churning, some larger pieces (up to 15 cm in maximum dimension) floated out (Fig. 3).

Fig. 3. Video frame grab taken from the Tiburon during an ascent (647-m water depth), showing buoyant sediment clumps within the imaging box. The clumps contained enough gas hydrate to float off the bottom when the ROV's manipulator disturbed the sediments on the seafloor at 779-m (cf. Fig. 2)

The clumps of sediments released were pursued by the ROV as they ascended in the water column (more fully documented in Brewer et al. 2002). A free ascent rate of 20 m/min was determined by following individual pieces upwards. The dark color of the pieces suggested that they consisted of mixtures of sediment and gas hydrate.

With the aid of the imaging box, which served as a corral for the floating clumps of sediment, buoyant pieces of gas hydrate were observed as they ascended from the seafloor (779 m) through the water column to the shallowest depth at which the Tiburon can operate (100 m). During ascent, the clumps broke up into smaller pieces, and the sediments initially adhered to the gas hydrate washed off and settled out through the open bottom of the imaging box. Large (>10 cm) transparent pieces which appeared to be individual crystals of gas hydrate were thereby exposed.

Above 400-m water depth, bubbles formed on the surfaces of the gas-hydrate crystals, suggesting that the gas-hydrate crystals were decomposing. The ascent rate increased upon bubble formation and exceeded the ROV's ability to keep up. However, gas-hydrate crystals were still trapped under the mesh at the top of the imaging box at 100-m depth. The remaining gas hydrates were observed at 100-m depth for several minutes (Fig. 4). During this period a continuous stream of bubbles was observed to rise from the top of the imaging box. Based on the associated ascent rate versus decomposition measurements (Brewer et al. 2002), this was an adequate period of time for some of the gas hydrates to have reached the surface intact (if they were free of the mesh top of the imaging box).

Fig. 4. Video frame grab taken from the Tiburon in 99.4-m water depths, showing streams of bubbles rising through the mesh at the top of the imaging box. The bubbles are coming from decomposing gas hydrates which have traveled from the seafloor at 779-m on the southern crest of Hydrate Ridge to this depth, and remained stationary at this depth for ~4 min

Discussion and conclusions

Sedimentary deposits associated with gravity-driven density flows commonly show the effects of hydraulic sorting of grains. Grains composed of heavy minerals become concentrated in the basal layers whereas grains of less dense minerals become concentrated in the upper layers because they settle more slowly. Seeing that gas-hydrate crystals are less dense than seawater, the effects of sediment sorting which occur within a slump mass containing gas hydrate will be even more extreme because gas-hydrate-bearing sediments will be buoyant, and isolated gas-hydrate crystals will actually float upwards in the water column when released from their sediment matrix. Thus, a slumping event which results in disintegration of the sediments into constituent particles will release the gas-hydrate grains into the water column (Fig. 5).

Fig. 5. Schematic representation showing hydraulic sorting of gas-hydrate-bearing sediment clusters in a sediment gravity flow. Gas hydrate floats upwards but remains within the P-T conditions for stability until it reaches the top of the gas-hydrate stability zone (GHSZ)

The sedimentary fabric of the gas-hydrate-bearing sediments on the crest of southern Hydrate Ridge was disaggregated into discrete clumps by the ROV's mechanical arm. Most of the disaggregated sediment remained on the bottom but some sediment clumps, which contained gas hydrate, were observed to float upwards.

The proportion of gas hydrate (0.91 g/cm³) which is needed to float sediments in seawater (1.03 g/cm³) can be estimated (assuming that water-saturated clays have a density of 1.6 g/cm³). Thus, the floating clumps should contain >83% gas hydrate.

The volumes of gas recovered in the clathrate bucket samples were much larger than those contained in traditional cores which freely degas in route to the surface. However, the samples contained much less gas hydrate (<6%) than the floating clumps are calculated to have had. Assuming full gas-hydrate cage occupancy, a structure I hydrate could contain up to 164 times the gas volume under standard surface temperature and pressure conditions as the gas-hydrate volume under in-situ conditions. The volume of gas hydrate which was in the clathrate bucket samples could have occupied ~5% of the core volume. Thus, the distribution of gas hydrate within the shallow subsurface is not uniform.

ROV observations of the sediment clumps along their ascent path toward the surface show that the dark tan sediment progressively washes off, revealing a core of transparent crystals or crystal aggregates, some being due over 10 cm long. The clarity of the exposed crystals shows that they lack sediment inclusions. Thus, the crystals we observed were not pore-filling cements. Rather, the clarity of the crystals shows that they either grew in open voids or, more likely, excluded sediment as they grew.

The existence of nodular gas hydrates within 20 cm of the bottom requires that they be trapped within the bottom by the sediments. Otherwise their buoyancy would have released them from the sediment. The ROV arm manipulation showed that some disaggregation of the sediments is needed before gas-hydrate-bearing sediment clumps will float upwards off the seafloor.

The disaggregation of the sediments by the ROV's mechanical arm is analogous to what occurs within a sediment failure which results in turbulent gravity flow. The sediment failures which produce gravity flows must also release gas-hydrate-bearing sediment clumps. Gas-hydrate-bearing sediment clumps have densities which are lower than similarly sized, gas-hydrate-free sediment clumps. Thus, the gas-hydrate-bearing sediment clumps will be preferentially kept in suspension. Moreover, as the clumps are exposed to turbulence in the water column, they are continuously being washed free of the adhering sediment, which makes the remaining material even more gas-hydrate-rich and thus less dense. When the mass of the clumps is mostly gas hydrate, the clumps will buoyantly rise in the water column.

Our observations indicate that large crystals of gas hydrate rise through the water column with only minor loss while within the gas-hydrate stability zone. Bubble formation on the surface of the gas-hydrate crystals reveals the rapid decomposition of the gas hydrate which occurs after the material has risen out of the depth where it is stable (given the local temperature conditions, this is ~400 m) at Hydrate Ridge. Thus, the gas which was within the gas hydrates will primarily be released into the near-surface ocean. Some gas will be released as bubbles within the surface mixed layer, and some gas bubbles and gas hydrate will be transported to the ocean surface and released directly into the atmosphere (Fig. 5).

The volume of gas which could be released from a sediment failure involving sediments like those which occur on Hydrate Ridge is difficult to estimate. The clathrate bucket cores taken from the southern crest of Hydrate Ridge suggest that ~5 times the volume of gas at surface conditions was initially contained in the near-seafloor sediment. The occurrence of floating pieces of sediment suggests that much higher concentrations are common. If the clathrate bucket values are characteristic of the region, a relatively small slide may cut into the bottom to a depth of 10 m and involve 1 km² of seafloor area. This would provide 2.2 times 10⁹ mol of gas (or 2.6 times 10¹⁰ g methane carbon) to be released primarily into the upper water column and atmosphere. This would have only a local effect on the water-column methane concentrations. Larger slides involving similar material (removing 100 m of section over an area of 1,000 km²) could release 10⁴ times this amount of material. Additional gas may be tapped from the free gas which exists underneath the gas-hydrate stability zone (Paull et al. 1991).

Buoyancy following the disintegration of the sediment fabric associated with submarine landslides which contain nodular gas hydrates needs to be considered as one mechanism to transport methane from continental margin sediments into the upper ocean and atmosphere.

Acknowledgements. Support for this work was provided from the David and Lucile Packard Foundation.