Abstract

High-resolution multibeam bathymetry data collected with an autonomous underwater vehicle (AUV) complemented by compressed high-intensity radar pulse (Chirp) profiles and remotely operated vehicle (ROV) observations and sediment sampling reveal a distinctive rough topography associated with seafloor gas venting and/or near-subsurface gas hydrate accumulations. The surveys provide 1 m bathymetric grids of deep-water gas venting sites along the best-known gas venting areas along the Pacific margin of North America, which is an unprecedented level of resolution. Patches of conspicuously rough seafloor that are tens of meters to hundreds of meters across and occur on larger seafloor topographic highs characterize seepage areas. Some patches are composed of multiple depressions that range from 1 to 100 m in diameter and are commonly up to 10 m deeper than the adjacent seafloor. Elevated mounds with relief of >10 m and fractured surfaces suggest that seafloor expansion also occurs. Ground truth observations show that these areas contain broken pavements of methane-derived authigenic carbonates with intervening topographic lows. Patterns seen in Chirp profiles, ROV observations, and core data suggest that the rough topography is produced by a combination of diagenetic alteration, focused erosion, and inflation of the seafloor. This characteristic texture allows previously unknown gas venting areas to be identified within these surveys. A conceptual model for the evolution of these features suggests that these morphologies develop slowly over protracted periods of slow seepage and shows the impact of gas venting and gas hydrate development on the seafloor morphology.

INTRODUCTION

Here, we report on seafloor morphologies observed where fluids are venting along the Pacific margin of North America. Areas where hydrocarbon-bearing fluids are seeping onto the seafloor are among the most dynamic deep-sea environments. Chemosynthetic biological communities surround these sites, supported by energy from hydrocarbon oxidization (e.g., Sibuet and Olu, 1998; Levin, 2005). Diagenetic reactions are enhanced in seep environments, notably those that result in the precipitation of methane-derived carbonates (e.g., Ritger et al., 1987; Kulm and Suess, 1990; Paull et al., 1992; Bohrmann et al., 1998; Peckmann et al., 2001). Gas hydrate formation and decomposition occur within the near-seafloor sediments, altering seafloor sediment properties (Kvenvolden, 1999; Sloan and Koh, 2008). As a consequence, seafloor seepage areas have become a focus of the research community. Seafloor seepage sites also pose special geohazard issues, in part because of the potential for unstable seafloor conditions and the possible existence of overpressured gas in the near subsurface (Chiocci et al., 2011). Thus, the hydrocarbon industry avoids installing seafloor structures near seeps (Hough et al., 2011).

Most of the known seafloor seepage sites were initially identified by regional side-scan sonar and surface vessel multibeam bathymetry surveys, because exposures of methane-derived carbonates and chemosynthetic biological communities result in high seafloor reflectivity (e.g., Carson et al., 1994; Suess et al., 2001; Naudts et al., 2008; Jones, et al., 2010). These areas became targets for human occupied vehicle (HOV) and remotely operated vehicle (ROV) dives to conduct detailed observations and sampling programs. Water-column acoustic anomalies have helped to identify other sites (e.g., Merewether et al., 1985; Gardner et al., 2009).

The best grid resolution of ship-mounted multibeam bathymetric data is ∼10 m (e.g., Hughes Clarke et al., 1998). While this level of resolution nicely reveals the general shape of the seafloor, an appreciable gap exists between the morphologies that are resolvable in surface ship multibeam data and what can be observed using the field of view provided by HOVs and ROVs.

Recently, it has become possible to use autonomous underwater vehicles (AUV) to map selected areas of the seafloor at 1 m grid resolution. The AUV surveys presented here provide a resolution that bridges the gap between visual observations and the best resolution obtainable from surface ship mapping (≥10 m grids). Because these surveys were collected with the same vehicle, meaningful comparisons can be made between sites, and the recurring characteristics of seafloor fluid venting sites can be delineated for the first time.

METHODS

AUV Surveys

Surveys were conducted at seven areas (Fig. 1; Table 1) where gas venting occurs using an AUV developed at the Monterey Bay Aquarium Research Institute (MBARI) specifically for seafloor mapping (Caress et al., 2008). The vehicle carried a Reson 7100, 200 kHz multibeam sonar, and an Edgetech 2 to 16 kHz Chirp subbottom profiler until the end of 2011, when it was replaced with an Edgetech 1 to 6 kHz Chirp subbottom profiler. The AUV was preprogrammed to proceed to >200 waypoints during each dive. Missions lasted up to 18 h and were designed for the vehicle to run at a speed of 3 knots while maintaining an altitude of 50 m off the seafloor. Track lines were spaced ∼150 m apart. In this mode, the AUV obtains overlapping multibeam bathymetric coverage at a vertical resolution of 0.15 m and a horizontal footprint of 0.7 m, and Chirp seismic-reflection profiles with a vertical resolution of 0.11 m. Initial navigation fixes were obtained from global positioning system coordinates when the AUV was at the sea surface, and subsequently updated with a Kearfott inertial navigation system (INS) and a Doppler velocity log (DVL). The AUV was launched from the mother ship R/V Zephyr and spun down to the seafloor through the water column with only INS navigation available until the seafloor was within ∼100 m. Data processing was done using MB-system (Caress and Chayes, 1996; Caress et al., 2008). Subbottom depths to reflectors imaged in the Chirp profiles are reported in meters below seafloor (mbsf) assuming a sound velocity of 750 m per second two-way traveltime.

ROV Observations, Sampling, and Sample Processing

The ROV Doc Ricketts was used to explore the seafloor associated with distinctive seafloor morphologies identified in the AUV surveys (Fig. 1; Table 1). Doc Ricketts dives provided ground truth verification through the collection of high-definition video and samples, including lithified substrates, push cores (≤22 cm long), and vibracores (≤170 cm long). Sediment samples obtained in these dives and from the International Ocean Discovery Program (IODP) core repository (Table 2) were processed in a similar fashion.

To determine sediment ages, ∼25 cm3 sediment samples from selected vibracores were wet-sieved through a 63 µm screen. Tests of ∼1000 planktonic (mixed species) or benthic foraminifera were handpicked from the >63 μm fraction. Where possible, two samples from vibracores were measured to constrain sediment accumulation rates. Radiocarbon measurements were made at National Ocean Sciences Accelerator Mass Spectrometry (NOSAMS) facility at the Woods Hole Oceanographic Institution (Table 2).

RESULTS

The seafloor morphology of areas with fluid seepage is presented here as bathymetric grids of 1 m resolution. An overview of previously published evidence and/or new ROV observations documenting the occurrence of gas venting in the survey areas is also presented. Sites are presented from north to south (Fig. 1).

Neptune Transect

The existence of gas hydrate along the Cascadia margin is well documented (e.g., Hyndman and Davis, 1992; Westbrook et al., 1994; Hyndman et al., 2001; Novosel et al., 2005; Riedel et al., 2002, 2005, 2006a). Three contiguous AUV dives mapped the surface of an elongate plateau in the vicinity of the Bullseye Vent node on the Neptune cable (Fig. 2). Five subareas of distinctive rough topography identified within these three dives are discussed here: Bullseye Vent, Bubbly Gulch, Spinnaker Vent, and two ridge crests.

Bullseye Vent

Bullseye Vent is associated with a distinct 305-m-long, 65–75-m-wide, NE-SW–oriented depression, developed on the surface of a broad plateau (Figs. 2, 3A, and 4A). The maximum relief from the rim to the floor of the depression is 6 m (Fig. 3, B–B′). This distinct depression was not recognized in previous bathymetric surveys because it is below the resolution of surface vessel multibeam sonar. The sides of the depression drop off abruptly along ∼5-m-high, ∼25°-sloping scarps. Within the floor of the depression, there are ∼5 roughly circular subdepressions that coalesce to form the overall feature. The seafloor outside the depression is smooth but gently shoals to form an elevated rim that is ∼2 m higher than the surrounding seafloor.

Observations during ROV dives show that the seafloor inside the depression is very hummocky, with scattered pieces of carbonate rubble, shells, and a few small clusters of living Vesicomya clams. Slabs of authigenic carbonate (∼1–4 cm in thickness) sticking out of the scarp wall (Fig. 5A) appear to be the edges of truncated sedimentary beds. Outside the depression, the seafloor is smooth and lacks rubble.

AUV Chirp profiles resolve distinct subhorizontal reflectors down to ∼50 mbsf across most of the plateau away from Bullseye Vent. However, the continuity of these reflectors progressively diminishes with proximity to the Bullseye Vent depression, and a seismic blanking zone occurs underneath the depression. Although a slight upturn in the reflectors within 200 m of the structure indicates a subtle doming centered underneath Bullseye Vent, the reflector geometry suggests intact strata at one time continued undisrupted through what is now the blanking zone (Figs. 3, section B–B′, and 6).

Bubbly Gulch

Persistent water-column acoustic anomalies (Riedel et al., 2010) have been identified above a small WNW-ESE–oriented embayment called Bubbly Gulch, located on the flank of the plateau, northeast of Bullseye Vent. AUV surveys partially mapped this structure, which extends laterally ∼850 m into the adjacent basin (Figs. 2 and 3). The head of Bubbly Gulch consists of an ∼120-m-radius amphitheater with 30 m relief and a smooth slope, except for a subtle bench 90 m NE of its headwall. Reflectors from the Chirp profile, which can be traced underneath the width of the plateau, crop out along a subtle terrace rimming the sides of this re-entrant (Fig. 3, B–B′).

Observations during ROV dives identified small (≤1 m high, and 5–10 m wide) mounds (Figs. 5B and 5C) rimming the sidewall of the embayment, at depths between 1270 and 1265 mbsf, coincident with the bench resolved in the AUV bathymetry. The mounds are covered with white patches of bacterial mats (Fig. 5B), which occasionally exhibit 5-cm-wide open cracks in hexagonal grid patterns suggestive of blistering and expansion of the seafloor surface (Fig. 5C). Vibracore collection on the mounds along this bench elicited vigorous release of gas (Fig. 5D). Spontaneous periodic gas bubbling was also seen emanating from the cracks. However, no indication of authigenic carbonate, tubeworms, or living Vesicomya clams was seen.

Ridge Crest Sites

Patches of rough topography that contrast with the smooth seafloor surrounding them were identified on the crest of some of the ridges jutting out of the plateau. One of them is at the western end of the large ridge on the NE edge of the AUV survey (Fig. 2). This patch of rough topography consists of a series of coalescing depressions ranging in width from 10 to 70 m and in depth from 1 to 5 m (Fig. 4B). Characteristically, the smooth seafloor turns up at the rim of the patch of rough topography. ROV observations (Fig. 5E) indicate that the seafloor has a nearly continuous pavement of carbonate, rising toward the rim of the depressions, cut by occasional cracks and open joints, with few scattered clamshells and few isolated clusters of living Vesicomya clams. The walls of the depressions expose the truncated beds of the pavement, which are rough and heavily cemented, producing more than 1 m of overhang in places. The floors of the depressions have variable amounts of scattered rubble, occasional larger isolated slabs of carbonate, and some apparently recent sediment fill. Angular slabs of lithified mudstone, presumably from the underlying formation, are exposed in the deeper depressions. Another patch with ∼12 similar depressions was also identified on the crest of a smaller ridge (Figs. 2 and 4C).

Spinnaker Vent

Spinnaker Vent is another site of persistent water column acoustic anomalies, 6.3 km NW of Bullseye Vent (Figs. 2 and 7). Despite the absence of identifiable features in surface vessel multibeam data, 1-m-resolution AUV surveys show the presence of an ∼500-m-long, ∼100-m-wide, NE-SW–oriented patch of distinctive rough topography. The patch is dominated by a 50–70-m-diameter and 2-m-deep depression, with smaller 5–12-m-diameter and 0.5-m-deep depressions to the NE and SW. Two topographic highs or mounds up to 2 m tall also occur on the SW side of the patch (Figs. 5F and 5G). Chirp profiles from a broad region around Spinnaker Vent resolve ∼10 m of horizontal strata around Spinnaker Vent, but in proximity to the rough topography, the reflectors dome upward. Seismic blanking is present, and ≤2 m section of layered strata is resolved under the area of rough topography (Fig. 7, F–F′).

ROV observations document that the walls of the central depression consists of steep and occasionally overhanging, 1–2-m-tall scarps built from the truncated edges of rough authigenic carbonates beds of considerable permeability (Fig. 5H). The floor of the depression was resistant to coring and contained scattered carbonate rubble, shell hash, and small beds of living Vesicomya clams, especially at the base of the small scarps on the sides of the depressions.

The two ∼2-m-high mounds to the SW of the depression exhibited surface cracking and had a small surface crater (Fig. 5F) from which intermittent gas bubbles emanated. Horizontal layers of up to 10-cm-thick, solid white material were exposed in the cracks (Fig. 5G), which were inferred to be gas hydrate because they floated away when broken loose. The process of collecting push cores from the flanks of these mounds stimulated vigorous gas releases.

Age Constraints

Along the Neptune plateau, 14C content was measured on 41 foraminifera samples from the vibracores and 10 samples from IODP Leg 311, Sites 1327C and 1328B (Fig. 2; Table 2). All cores with two or more measurements show increasing age with depth (Fig. 8). The cores from Bubbly Gulch came from the flank of the plateau and sampled the truncated edge of units that can be traced in Chirp profiles underneath the plateau into the Bullseye Vent depression blanking zone (Fig. 3).

Sediment accumulation rates calculated from the age versus depth regressions from Bullseye Vent, the flank of Bubbly Gulch, and IODP Sites 1327C and 1328B indicate that glacial marine silts and clays (Clague and Ward, 2011) were accumulating from at least 50,000 up to 13,100 14C yr B.P. at rates between 0.4 m/k.y. and 1.1 m/k.y. (Fig. 8D). Holocene age sediments are noticeably absent across the plateau, and the intersection of the seafloor age versus depth regressions suggests sediment accumulation ceased in this area ca. 13,100 14C yr B.P. Similarly, the ages of the sediment near the seafloor at Spinnaker Vent are ≥12,200 14C yr B.P., with the exception of an 8620 14C yr B.P. measurement on sediment from within 18 cm of the modern seafloor. Again, significant accumulations of Holocene sediment are noticeably absent.

The ages of sediment samples from vibracores show that the near-surface sediments within the Bullseye Vent depression are systematically older (≥15,100 14C yr B.P.) than the sediments on its rim, but outside the depression. The age verses depth progression seen in these cores is consistent with the removal of overburden equivalent to the depth of the depression (Figs. 8B and 8D).

Barkley Canyon

The occurrence of gas hydrate on the flank of Barkley Canyon was originally discovered by a commercial fishing trawler (Spence et al., 2001; Fig. 9). Subsequently, ROV dives in ∼860 m water depths showed lenses of pure gas hydrate exposed on the seafloor in cracks on the flanks of small topographic mounds on a terrace on the western flank of the canyon (e.g., Chapman et al., 2004; Pohlman et al., 2005; Hester et al., 2007).

Bathymetry collected during an AUV survey in 2009 shows a line of ∼10 circular mounds, up to 2 m high and 10 m in diameter, on the terrace associated with the gas hydrate occurrences (Fig. 9, inset). No internal reflections are resolved in the Chirp data underneath this terrace near where the gas hydrate mounds are found. However, up to 50-m-thick sections of finely layered sediments occur in places on the floor and flanks of the canyon (Fig. 9, G–G′). ROV observations confirm the absence of sediment drape on the mounds. Although the rock samples from the mounds are mudstones barren of microfossils, their firm brittle texture suggests these mounds developed within pre-Quaternary strata exposed on the side of the canyon.

Northern and Southern Hydrate Ridge

Extensive areas of authigenic carbonates, living chemosynthetic organisms, and gas hydrate in the near subsurface (e.g., Kulm et al., 1986; Westbrook et al., 1994; Bohrmann et al., 1998; Suess et al., 1999, 2001; Tréhu et al., 2003) were documented on Hydrate Ridge by HOV and ROV dives and Ocean Drilling Program (ODP) and IODP drilling (Figs. 10, 11, and 12).

The surface of northern Hydrate Ridge has an abundance of distinctive rough topography in multiple patches along the crest of the ridge (Figs. 10 and 12). This site is distinguished from the others by the large size of individual patches (e.g., up to 1400 m long), and by the extent of relief (commonly 10 m or more) within these patches. One of these patches contains a roughly circular crater-like feature with a raised rim surrounding a 40-m-deep depression that is 300 m in diameter (Figs. 10 and 12A). A ridge extends over 600 m to the SW from this crater-like feature. This ridge has a trough near its crest interpreted to be a tension fracture formed during inflation of the seafloor.

The AUV data show that the surface of southern Hydrate Ridge is generally smooth, except for two contiguous patches of distinctive hummocky topography with maximum diameters of 350 m and 500 m (Figs. 11 and 12C). These areas are ringed with small, apparently erosional scarps, and thus the strata exposed within these patches are stratigraphically lower than the surrounding smooth seafloor. The fine-scale topography within these circular rough patches is characterized by small, sometimes circular, ∼0.5-m-deep pits, and local highs and lows separated by ∼0.5-m-high ledges, which appear to be irregularly eroded bedding surfaces. Similar shapes also occur at larger scales. The topographic high previously called the Pinnacle (Suess et al., 2001; Tréhu et al., 2004; Paull and Ussler, 2008) is ∼100 m across with ≥35 m of vertical relief (Figs. 11 and 12C) and occurs within the center of a ∼500-m-wide depression in the western patch of distinctive rough topography.

Chirp data from Hydrate Ridge are characterized by a strong seafloor reflection with only a few wispy sub-bottom reflectors identified during these surveys. These reflectors outline irregular surfaces that are up to 15 mbsf (Fig. 10, section H–H′; Fig. 11, section I–I′) and that consistently pinch out around the areas of rough topography. No subbottom reflectors can be identified under the areas of rough topography.

Foraminifera from 2 to 71 cmbsf at Sites 1249 and 1250, which are both within the areas of distinctive rough topography on southern Hydrate Ridge, yielded ages of ≥43,800 14C yr B.P. (Table 2). U/Th measurements on samples from northern Hydrate Ridge indicate that the timing of cement formation on the seafloor was 68.7–71.7 k.y. B.P. (Teichert et al., 2003).

Eel Slump

Water-column plumes were discovered off the coast of northern California (Fig. 13; Gardner et al., 2009). Some plumes emanate from a topographic high that is ∼650 m long, 350 m wide, and stands nearly 60 m higher than the surrounding seafloor at 1850 m depth. This topographic high occurs within a 3.5-km-wide slide scar informally known as Eel Slump.

The surface of this topographic high is characterized by a hummocky topography consisting of small, approximately circular, ∼0.5-m-deep pits, and local highs and lows separated by ∼0.5-m-high ledges that could have been formed by irregular erosion of bedding surfaces (Fig. 13). A semicircular crater-like depression that is ∼80 m across and more than 10 m deep occurs on the NW flank of this feature.

ROV observations show scattered living and dead Vesicomya clams and patches of white bacterial mat on the flanks and crest of this topographic high. Nearly continuous streams of gas bubbles were observed emanating from a few places on the crest of this topographic feature. Analysis of this gas indicates that it is primarily methane, mixed with other thermogenic hydrocarbon gases (Gwiazda et al., 2011). Small drops of oil were also released from the near-surface sediment when a push core was taken.

Chirp profiles show that a drape of layered sediment that is at least 40 m thick covers the bottom and sidewalls of Eel Slump, suggesting that the slope failure that created the large scarp was not recent. No subbottom reflectors are traceable underneath the ovoid mound in the center of Eel Slump (Fig. 13, J–J′).

NE Flank of the Guaymas Basin

After water-column acoustic anomalies were discovered along the transform margin on the NE flank of the Guaymas Basin, Gulf of California (Lonsdale, 1985; Merewether et al., 1985), subsequent investigations in this area showed extensive areas of methane-derived authigenic carbonate, seafloor chemosynthetic communities, vigorous gas venting, and gas hydrate exposures (Paull et al., 2007). Two AUV dives mapped a ∼3.5 km section of a NW-SE–oriented ridge with a crest in 1540–1600 m water depths (Fig. 14). The orientation of this ridge is coincident with the trend of the plate margin in this area. The ridge crest is associated with a blanking zone, but strata extending from underneath the Guaymas Basin thin and/or onlap on the SW side of this ridge (Fig. 14, L–L′). A gas vent known as Pinkie’s Vent occurs on the steeper NE side of this ridge (Fig. 14; Paull et al., 2007). The ∼1.4-km-wide basin to the NE of the main ridge has smaller NW-SE–oriented ridges that are up to 1.9 km long and have up to 70 m of relief.

Patches of distinctive rough topography containing circular mounds and depressions occur along the crests of the main ridge and subsidiary ridges (Figs. 14A and 14B). The largest mound has 9 m of local elevation within its 130 m diameter (Fig. 14B), but it occurs on a 750-m-long, ∼10-m-high swell on the main ridge. The largest depression is 60 m in diameter and 4 m deep, and is flanked by elevated topography on both sides (Fig. 14B). ROV observations confirm that the topography consists of irregular, broken pavements of methane-derived authigenic carbonates and contains both living and dead chemosynthetic communities.

Chirp data show that the main ridge is composed of a wedge of upturned sediments that dip to the WSW and thin to the NNE, where they are truncated on the flank of the ridge (Fig. 14, L–L′), The continuity of the reflectors is lost underneath the areas of distinctive rough topography (Fig. 14, M–M′).

SW Flank of the Guaymas Basin

A previously unexplored area, 3.6 km long and 3.2 km wide, on the SW side of Guaymas Basin was mapped, revealing two elongated topographic ridges that straddle what is believed to be the main trace of the transform fault (Fig. 15). A distinct NW-SE–oriented, 2.8-km-long trough, only ∼100 m wide, separates these ridges. From this trough, the ridge to the SW rises over 140 m. The ridge on the NW side is smaller, with only 30 m of relief with respect to the trough. Ovoid patches of a distinctive rough topography occur along the crests of both of these ridges (Fig. 15), which are 50–150 m across and contain 1–5-m-deep depressions and roughly circular topographic highs.

ROV observations show that the patches of distinctive rough topography on both ridge crests are associated with exposures of methane-derived carbonates. Beds of living Vesicomya clams and tubeworms also occur on the crests of both ridges within the areas of distinctive rough topography. On the larger ridge, living clam beds occur discontinuously along a 1-km-long stretch of the ridge where turbidite beds containing sandy horizons outcrop. Open cracks between blocks of carbonates and depressions in the seafloor that are ∼1 m deep were observed in the ovoid patches of high relief on the NW flank (Fig. 15, inset).

DISCUSSION

Seafloor surveys using AUVs provide bathymetric grids with 1 m resolution and show areas associated with distinctive rough seafloor topography at multiple sites along the west coast of North America. The topography is composed of mixtures of circular depressions and local topographic highs. The distinctive topography occurs within approximately ovoid patches that range in size from a few tens of meters to hundreds of meters long and commonly have between 2–10 m of local relief (Figs. 4, 7, and 9–16).

This distinctive rough topography occurs in areas where the near-seafloor sediments have been exposed to methane-rich pore waters. The evidence for hydrocarbon seepage includes water-column acoustic anomalies, widespread occurrence of methane-derived authigenic carbonates in the upper layers of the seafloor, living and dead chemosynthetic biological communities, occasional direct observations of gas bubbling and of gas hydrate exposures on the seafloor, samples of gas hydrate recovered in boreholes, and blanking zones in seismic profiles. All the areas with this distinctive topography are at water depths (i.e., pressures) and temperatures where methane hydrates are stable at the seafloor and within the near subsurface (Sloan and Koh, 2008).

Here, we argue that the distinctive seafloor morphologies are produced by a combination of widespread chemical changes, including diagenesis associated with anaerobic methane oxidation, and physical changes. The physical changes include seafloor inflation and collapse associated with formation and decomposition of gas hydrate within sediments, sediment erosion that is specifically focused at the seepage sites and associated with increased biological activity, and sediment rafting due to the inherent buoyancy of gas hydrate.

Authigenic Carbonate Formation

ROV observations show that the distinctive topography is consistently associated with exposures of methane-derived authigenic carbonate (e.g., Ritger et al., 1987; Paull et al., 1992; Sakai et al., 1992; Bohrmann et al., 1998; Stakes et al., 1999; Roberts, 2001). Methane-derived authigenic carbonates are a by-product of the anaerobic oxidation of methane (AOM; Caldwell et al., 2008). AOM occurs within marine sediments along the sulfate methane transition zone where the concentration of both methane and sulfate are nearing zero (e.g., Borowski et al., 1999; Ussler and Paull, 2008; Bhatnagar et al., 2008). The dominant fine-grained lithology of these authigenic carbonates indicates they were formed slowly within the host strata as a by-product of AOM (e.g., Ussler and Paull, 2008).

The addition of authigenic carbonate cement changes the mechanical properties of hemipelagic sediments and increases their resistance to mechanical erosion. Without the increased strength provided by the authigenic carbonate cement, the rugose bottom observed in the AUV multibeam data and ROV images could not persist. In areas experiencing net erosion, variations in cement content result in differential erosion that leaves more densely cemented areas standing above the surrounding seafloor. Authigenically cemented sediments are subject to brittle failure (e.g., Bjørlykke and Hoeg, 1997), and the presence of angular breccia (e.g., Bohrmann et al., 1998) confirms that brittle failures do take place. Cement addition also reduces the sediment porosity and permeability, forms barriers that inhibit the upward migration of hydrocarbons, sharpens near-seafloor pore-water chemical gradients, and allows accumulation of gas underneath carbonate pavements.

Seismic Blanking Halos

Chirp profiles characteristically show the presence of layered sediments away from areas of distinctive rough topography and exposed authigenic carbonates (Figs. 3, 6, 7, 9, 10, 11, and 13–15), but the lateral continuity of the reflecting horizons is consistently lost within seismic blanking zones beneath the rough topography. In some places, the loss of reflectivity is because the layered sediments pinch out at the seafloor (Figs. 9, 10, and 11); in other places, no change in orientation is seen in the reflecting horizons near the areas of rough topography and seismic blanking zones (Figs. 3, 6, 7, 13, and 15).

In the absence of physical disruptions of the beds, seismic blanking can be attributed to the presence of interstitial gas bubbles (e.g., Judd and Hovland, 1992; Roberts, 1999), dispersed gas hydrate (e.g., Lee and Dillon, 2001; Riedel et al., 2006b), or authigenic carbonate (Riedel et al., 2002). Distinguishing between the potential causes of the blanking is not possible, and all three may be contributing to varying extents. However, the concentration of methane in pore waters that is needed to form either gas hydrate or gas bubbles is much higher than that needed to stimulate the formation of methane-derived authigenic carbonate (e.g., Ussler and Paull, 2008; Bhatnagar et al., 2008). For the blanking in these Chirp profiles (Figs. 3, 6, 7, 13, and 15) to be primarily attributed to the presence of gas hydrate and/or gas bubbles requires that these phases were present underneath the entire blanking zone when the surveys were conducted. This would require ongoing fluxes of methane capable of maintaining relatively high pore-water methane concentrations (i.e., tens of millimoles) over widespread areas. In contrast, the microbial communities that stimulate the formation of authigenic carbonate require considerably lower methane concentrations (i.e., <1 mM) and thus a considerably lower methane flux. Moreover, because the carbonate cements remain after the methane supply stops, conditions appropriate for carbonate formation do not have to occur over more than a small fraction of the entire area at any point in time. The blanking zones within these high-resolution profiles may primarily reflect the integrated impact of carbonate formation associated with a modest amount of methane seepage along active conduits that migrated around or through these seepage areas over protracted periods of time.

Deformation Associated with Gas Hydrate Formation

Gas hydrate has been observed to occur within sediments as solid lenses that are up to 1 m in thickness (e.g., Kvenvolden and McDonald, 1985; MacDonald et al., 1994). Several-centimeter-thick layers of pure hydrate have been observed exposed on the seafloor at Spinnaker Vent (Fig. 5G), Barkley Canyon (Chapman et al., 2004; Hester et al., 2007), Hydrate Ridge (Bohrmann et al., 1998; Suess et al., 2001), and in the Gulf of California (Paull et al., 2007). Drilling at Bullseye Vent (Riedel et al., 2006a) and Hydrate Ridge (Carson et al., 1994; Tréhu et al., 2003) recovered lenses of pure hydrate near the seafloor as well.

Development of subsurface layers of pure gas hydrate requires that sediment be excluded during gas hydrate formation. Analysis of pressure core samples confirms the existence of significant grain displacements (Holland et al., 2008). This process may be similar to the formation of segregated ice in permafrost, where the formation of ice lenses in the subsurface commonly results in the expansion of the sediment column by 50% or more (Murton et al., 2006). By analogy, development of lenses of pure hydrate near the seafloor will result in surface blistering (MacDonald et al., 1994; Hovland and Svensen, 2006; Vardaro et al., 2006; Paull et al., 2008; Serié et al., 2012), the gas hydrate equivalent of pingos in permafrost areas (e.g., Mackay, 1998). The amount of uplift will be equivalent to the total thickness of the pure hydrate layers formed in the near subsurface.

The smallest seafloor inflation features, or mounds, observed in these surveys on steep slopes at Bubbly Gulch and Pinkies Vent (Figs. 5A and 14) were less than 1 m high, i.e., too small to be detected in the AUV surveys but easily seen within the field of view of an ROV. Seafloor mounds at Bubbly Gulch have surface cracks consistent with inflation of the seafloor (Fig. 3). Bubbly Gulch is also the only area where, despite the methane seepage, carbonates were not observed exposed on the seafloor. These mounds may be too young to have accumulated extensive deposits of slowly growing authigenic carbonate, or, if present, erosion has not exposed them yet on the seafloor.

Similarly, small circular mounds also occur at both Barkley Canyon and Spinnaker Vent (Figs. 5F, 7, and 9). In both of these areas, open cracks and lenses of apparently pure gas hydrate are exposed on the sides of these mounds (Fig. 5G). The seafloor inflation and deformation associated with the formation of hydrate lenses explain the formation of these mounds.

The detailed morphologic surveys show similar-sized mounds occurring in different areas, which are all interpreted to be caused by inflation related to gas hydrate growth in the near-surface sediments. For example, mounds with elevations more than 10 m high occur on the NE side of the Guaymas Basin (Figs. 14, 16A, and 16C), the Santa Monica Basin (Fig. 16B; Paull et al., 2008), and on southern Hydrate Ridge (Figs. 11, 12C, and 16D). Topographic features of comparable size off West Africa have been attributed to seafloor inflation as well (Serié et al., 2012).

While the general shapes of these features are similar (Fig. 16), the extent of seafloor roughness on their flanks can be explained by variations in the amount of erosion. This is consistent with what is known about the relative ages of these features. The Santa Monica mound (Fig. 16B) is a late Holocene feature (Paull et al., 2008) and has relatively smooth flanks, while the Pinnacle on southern Hydrate Ridge (Fig. 16D), of early Holocene age or older (Suess et al., 2001; Teichert et al., 2003), exhibits a much more eroded topography.

Still larger topographic highs, which are either circular or ovoid, occur at Eel Slump (Fig. 13) and northern Hydrate Ridge (Figs. 10 and 12A). As the topographic highs increase in size, the extent to which inflation is the cause for the increase in topography becomes more uncertain. Nevertheless, the >2-km-long ridge on northern Hydrate Ridge has extensive open depressions or cracks that suggest expansion (Figs. 10 and 12A).

Hydrate Dissolution and Seafloor Collapse

While the seafloor in all the surveys presented here is at pressures and temperatures appropriate for methane hydrate stability, pore waters must have adequate concentrations of methane for methane hydrate to form (Zhang and Xu, 2003; Bhatnagar et al., 2008; Sloan and Koh, 2008), yet methane concentrations in normal seawater are significantly lower than this. Thus, methane hydrate occurrence on or very near the seafloor requires a sustained advective flux of methane (Torres et al., 2004). Conversely, when methane hydrate is exposed to seawater or pore waters with low methane concentration, it dissolves (Brewer et al., 2003; Zhang and Xu, 2003; Rehder et al., 2004; Lapham et al., 2014).

Over time, the subsurface plumbing associated with seafloor vents changes as a consequence of authigenic carbonate and gas hydrate formation, which alter the sediment porosity and permeability and cause internal deformation within the sediment. Shifts in fluid pathways will result in some conduits being cut off (Solomon et al., 2008). Without a sustained methane flux, diffusion and mixing with overlying seawater will decrease the methane concentration within the sediment below the level needed for methane hydrate stability, and the hydrate lenses will start to dissolve (Brewer et al., 2003; Zhang and Xu, 2003; Rehder et al., 2004; Lapham et al., 2014).

When gas hydrate lenses dissolve, sediment overburden that previously was supported by gas hydrate collapses, refilling the void previously occupied by gas hydrate. However, irreversible changes associated with cement addition, internal deformation, and brittle failures prevent the overburden from simply collapsing back into its original position. This process contributes to the formation of the distinctive topography that surrounds seafloor methane seepage areas.

The distinctive rough surface morphologies revealed in these surveys have some morphologic similarities to subaerial thermokarst landscapes (Washburn, 1980; Williams and Smith, 1989). Thermokarst landscape evolves by both formation and decomposition of ice. Expansion during ice formation forces sediment apart and results in surface inflation and ice wedging, which break up rigid surfaces into slabs. Some of the textures associated with this deformation remain imprinted in the sediments after the ice thaws. This analogy may provide some explanation as to the way in which the distinctive rough topography develops in areas that have experienced gas hydrate formation and decomposition.

Seafloor Erosion

The distinctive rough topography of the seafloor suggests that hemipelagic sediment was exhumed, thereby exposing carbonates on the seafloor, widening cracks, undercutting the edges between carbonate blocks, and producing the depressions. Clearly, these are areas that have experienced erosion. Various mechanisms to explain how fluid seepage erodes the seafloor should be considered.

Patches of distinctive rough topography characteristically occur on the surfaces of plateaus and crests of ridges, sites that are preferentially exposed to winnowing by bottom currents. The rates of current-induced erosion decrease significantly as deep-sea sediments become cemented by carbonates (Young and Southard, 1978). Thus, in areas experiencing net erosion, the more cemented patches stand elevated compared to the adjacent less-cemented areas.

Within the patches of distinctive rough topography, there are numerous small depressions (Figs. 4, 7, 10, 12A, 15, and 16). Similar depressions occur at other active gas venting sites (e.g., Hovland and Judd, 1988; Paull et al., 1995; Roberts, 2001; Gay et al., 2007; Freire et al., 2011). However, physical models of how gas venting results in the excavation of seafloor to form the observed depressions are still lacking.

The possibility that seafloor depressions within these gas-rich environments are generated by eruptive events that eject fluids and solids into the water column has been suggested (e.g., Hovland, 1989; Prior et al., 1989; Hovland et al., 2002; Bangs et al., 2010). Near-seafloor gas pockets and overpressured conditions may occur when impermeable lenses of gas hydrate trap gas in the near subsurface (Flemings et al., 2003; Hornbach et al., 2004). However, the dynamics of the proposed eruptive events are unclear, and questions remain regarding the required pressures and gas volumes needed to generate such eruptions.

A physical model has been proposed to explain how capillary forces in the pore spaces within gas chimneys build slowly with time until they abruptly fail, liquefying sediment and erupting gas and presumably some liquefied sediment (Cathles et al., 2010). This model predicts physical distortion of the underlying sediment column, which is observed at some gas chimneys elsewhere, but which is not observed at these sites. Moreover, none of these seepage areas shows a recognizable halo of debris, as indicated by the eruptive model. These morphologies may be the result of the cumulative effects of less dramatic but observable processes that occur slowly.

Rafting away of pieces of gas hydrate along with some adhering sediments has been observed at seepage sites (Paull et al., 1995, 1999, 2002; Suess et al., 2001). When clumps of gas hydrate and adhering sediment occupy >83% of the volume, the sediment clump will be buoyant (given densities of 0.91 g/cm3, 1.03 g/cm3, and 1.6 g/cm3, for hydrate, seawater, and water-saturated clays, respectively). Once hydrate pieces break free from the bottom, the rate of hydrate dissolution is comparatively slow relative to the rate of lateral transport in bottom currents (Brewer et al., 2003); thus, the adhering sediments may be moved a considerable distance before falling back to the seafloor. Thus, sediment excavation by hydrate flotation may not leave an identifiable halo of debris surrounding the seepage area. These rafts can be small pieces of hydrate forming under an overhang, growing in size, and breaking off periodically. The occurrence of larger rafts has also been inferred (Bangs et al., 2010).

The erosional impact of the greatly enhanced biological activity supported by hydrocarbon seepage (Sibuet and Olu, 1998; van Dover et al., 2002; Levin, 2005) is unquantified. Bioerosion is a significant process in the reduction of hard bottom substrates in shallow water (e.g., Neumann, 1966; Glynn, 1997), especially where high abundances of organisms and hard bottom substrates occur. By analogy, it should be a factor in seep settings that also have high abundances of organisms and hard bottom substrates. Abundant snails, urchins, crabs, and other grazing organisms (Fig. 5H) rasp and scrape substrate while feeding on microbial communities (Sibuet and Olu, 1998; van Dover et al., 2002; Levin, 2005). The burrowing of Vesicomya clams directly erodes firm substrates (Paull et al., 2005). Enhanced bioturbation by infauna irrigates the sediment, enhancing chemical dissolution, and physically mobilizes and excretes fine sediment, which in part becomes resuspended (Ziebis and Haese, 2005; Meadow et al., 2011). Microbial mats thrive on and within seep carbonates (Caldwell et al., 2008), and such endolithic communities inevitably contribute to substrate reduction (e.g., Golubic et al., 1975; Edwards et al., 2003). The effects of bioerosion are preferentially focused where fluid seepage is active, often in the bottoms of the cracks and around the base of the carbonate blocks where the chemosynthetic organisms occur. Bioerosion is inferred to be a significant process contributing to progressive deepening of cracks, the undercutting of carbonate blocks, and excavation of depressions. The integrated effect of these processes will have major impact on the local seafloor morphology if the seepage is persistent.

Time Required to Form the Distinctive Rough Topography

Some constraint on the amount of time required to generate the observed areas of distinctive rough topography is provided by geologic context. Sediment-hosted methane-derived authigenic carbonate exposed on the seafloor within these areas of distinctive rough topography originally formed within hemipelagic sediments. They grow when and where zones in the subsurface are subjected to focused diagenesis in the sulfate-methane interface. The exposure of these carbonates on the seafloor inherently implies erosion.

Inevitably, the positions of the active seepage conduits shift as carbonates form and gas hydrate grows and dissolves. These shifts will alter the position of the sulfate-methane interface and thus where authigenic carbonates are forming at any particular time. The observed patches of rough topography reflect an amalgamation of carbonates formed as the position of the seepage conduits shifted over time. As individual authigenic carbonate nodules can take hundreds if not thousands of years to form (Ussler and Paull, 2008), the time needed to generate the large patches of distinctive rough topography may require thousands if not tens of thousands of years.

The age of sediment with known stratigraphic relationships to the patches of distinctive rough topography helps constrain the time required for these morphologies to develop. The extreme example of this distinctive rough topography occurs on Northern Hydrate Ridge, where some local topographic highs have ∼15 m of relief. A 20-m-deep, ∼100-m-wide, and 350-m-long open crack was mapped on the crest of a transverse ridge (Figs. 10 and 12A). The open cracks suggest both expansion of the seafloor surface and removal of material. The >49,400 14C yr B.P. age of the underlying sediments and exposures of carbonate on the surface dating to 267.6 ka (Teichert et al., 2003) indicate that the patches of distinctive rough topography on Northern Hydrate Ridge may have developed over tens of thousands of years. Similarly, the sediments within the upper 11 cm of the seafloor on southern Hydrate Ridge are also approaching 14C depletion (Table 2), indicating that the younger sediments have been removed or were never deposited. The considerable areas of distinctive rough topography on both crests of Hydrate Ridge are examples of comparatively mature seepage morphologies.

The best age control to assess the duration of exposure to seepage conditions comes from the areas surrounding Bullseye Vent (Fig. 8; Table 2). No sediment occurs either inside or outside Bullseye Vent that is younger than 13,100 14C yr B.P. The profound change from rapid deposition prior to 13,100 14C yr B.P. to being sediment-starved subsequently (Fig. 8D) is consistent with the glacial history of British Columbia (Porter and Swanson, 1998; Clague and James, 2002). The ice sheets that extended to the shelf edge and calved into the open ocean reached their maximum extent ca. 14,000 14C yr B.P. but retreated rapidly and were gone from most of what is now southern Vancouver Island by 13,100 14C yr B.P. (Dallimore et al., 2008). Thus, Bullseye Vent depression is in comparison a young feature having developed within the last 13,100 yr, since the rain of glacially delivered sediment stopped. The depth of the Bullseye Vent depression only requires an average removal rate of 0.4 mm of sediment per year, which is a rather modest rate. Similar rates of excavation applied over longer time intervals may be adequate to generate the more extensive areas of distinctive rough topography observed elsewhere.

Progressive Maturity of Seepage-Related Geomorphologies

Considerable variation exists in the extent of seepage-related geomorphic modification between the surveyed sites. These variations are interpreted to reflect varying stages of development associated with exposure to methane-rich fluids. The size of the patches of distinctive rough topography is consistent with what is known about their exposure history. For example, Bullseye Vent is among the youngest datable features, and it is associated with a relatively small area (Table 3). Hydrate Ridge has the largest patches of this distinctive rough topography, the greatest seafloor relief, the most extensive blanking zones, and appears to have the longest exposure history. The observations documented in the seven surveyed areas of distinctive rough topography are interpreted to represent varying stages in the geomorphic evolution of the seafloor in the presence of methane venting and gas hydrate formation within near-seafloor sediments.

CONCLUSIONS

AUV bathymetric surveys of the best-known methane seepage sites on the Pacific margin of North America show these sites in unprecedented detail and reveal the distinctive rough topography that characterizes methane seeps. This distinctive topography includes a progression of seafloor depressions that range from <1 m to over 100 m in depth, and topographic highs that exceed 15 m of vertical relief. The origin of this morphology is attributed to multiple processes that are stimulated by methane seepage. The increased biological activity associated with chemosynthetic biological communities enhances bioerosion rates. Formation of lenses of pure hydrate in near-subsurface sediments causes expansion, which in turn results in seafloor blistering and generation of open cracks and raised topographic features. Buoyant clumps of hydrate periodically break loose from the seafloor, rafting adhering sediment away. The dissolution of hydrate lenses when a sustained methane flux ceases leaves the overlying sediment unsupported and results in sediment collapse to refill voids. The variations in the morphologies shown in these seven surveys collectively provide a mosaic of observations that illustrate the increasing geomorphologic maturity associated with increasing duration of exposure to methane seepage.

The David and Lucile Packard Foundation provided support. Special thanks are given to the crews of the R/V Zephyr, R/V Western Flyer, the Monterey Bay Aquarium Research Institute (MBARI) autonomous underwater vehicle (AUV) group, and the MBARI remotely operated vehicle pilots. Reviews from Tom Lorenson, Pete Dartnell, and two anonymous reviewers greatly improved the manuscript.