Active methane venting observed at giant pockmarks along the
U.S. mid-Atlantic shelf break
Kori R. Newmana,b,⁎, Marie-Helene Cormiera,1, Jeffrey K. Weissela, Neal W. Driscollc,
Miriam Kastnerc, Evan A. Solomonc, Gretchen Robertsonc, Jenna C. Hillc,2,
Hanumant Singhd, Richard Camillid, Ryan Eusticed,3
aLamont-Doherty Earth Observatory of Columbia University, Palisades, NY 10964, United States
bDepartment of Earth and Environmental Sciences, Columbia University, New York, NY 10027, United States
cScripps Institution of Oceanography, La Jolla, CA 92093, United States
dWoods Hole Oceanographic Institution, Woods Hole, MA 02543, United States
Received 27 June 2007; received in revised form 22 November 2007; accepted 24 November 2007
Available online 14 December 2007
Editor: M.L. Delaney
Detailed near-bottom investigation of a series of giant, kilometer scale, elongate pockmarks along the edge of the mid-Atlantic continental shelf
confirms that methane is actively venting at the site. Dissolved methane concentrations, which were measured with a commercially available
methane sensor (METS) designed by Franatech GmbH mounted on an Autonomous Underwater Vehicle (AUV), are as high as 100 nM. These
values are well above expected background levels (1–4 nM) for the open ocean. Sediment pore water geochemistry gives further evidence of
methane advection through the seafloor. Isotopically light carbon in the dissolved methane samples indicates a primarily biogenic source. The
spatial distribution of the near-bottom methane anomalies (concentrations above open ocean background), combined with water column salinity
and temperature vertical profiles, indicate that methane-rich water is not present across the entire width of the pockmarks, but is laterally restricted
to their edges. We suggest that venting is primarily along the top of the pockmark walls with some advection and dispersion due to local currents.
The highest methane concentrations observed with the METS sensor occur at a small, circular pockmark at the southern end of the study area. This
observation is compatible with a scenario where the larger, elongate pockmarks evolve through coalescing smaller pockmarks.
© 2007 Elsevier B.V. All rights reserved.
Keywords: pockmarks; seafloor venting; methane; AUV
It is estimated that 6.6–19.5 Tg of methane per year are
released from the marine environment into the atmosphere,
making natural gas seeps an important part of the global meth-
ane budget (Judd et al., 2002). Methane seeps can occur in most
marine environments (Judd, 2003) with seep characteristics
ranging from diffuse seafloor venting to more focused escape
(Lonke et al., 2004). In addition to the environmental signifi-
cance, gas in marine sediments might hold possible geohazard
and resource significance (e.g. Sills and Wheeler, 1992).
Pockmarks associated with the venting of gas-rich fluids
have become widely observed seafloor features since their first
Available online at www.sciencedirect.com
Earth and Planetary Science Letters 267 (2008) 341–352
⁎Corresponding author. Lamont-Doherty Earth Observatory, 304B Oceano-
graphy, P.O. Box 1000, Palisades, NY 10964, United States. Tel.: +1 845 365
8461; fax: +1 845 365 8156.
E-mail address: firstname.lastname@example.org (K.R. Newman).
1Present address: Department of Geological Sciences, University of Missouri,
Columbia, MO 65211, United States.
2Present address: Center for Marine and Wetland Studies, Coastal Carolina
University, Conway, SC 29526, United States.
3Present address: Department of Naval Architecture and Marine Engineering,
University of Michigan, Ann Arbor, MI 48109, United States.
0012-821X/$ - see front matter © 2007 Elsevier B.V. All rights reserved.
discovery by King and McLean (1970) offshore Nova Scotia
(Hovland and Judd, 1988). The cross-sectional shape of these
features varies from U-shaped and V-shaped seafloor depres-
sions to truncated cones with steep, low angled or asymmetric
walls. Some are circular in plan view while others are elongate
(Dimitrov and Woodside, 2002; Hovland et al., 2002). While
most agree that pockmarks are the result of focused fluid flow
(Hovland et al., 2002), the exact nature of venting remains
poorly understood (Paull et al., 2002). Kelley et al. (1994)
suggest two models for pockmark formation: 1) organic matter
deposited above an erosional surface decomposes, releasing gas
that excavates the pockmark; once the excavation extends to
the erosional surface, the pockmark spreads out laterally along
the erosional surface, and 2) a catastrophic event such as an
earthquake or tsunami reduces the confining pressure in the
area, allowing gas and fluids to suddenly escape. The first
model can explain why U-shaped, V-shaped and flat-floored
pockmarks are observed, and the latter model why pockmark
formation and increased methane venting have been documen-
ted to occur in response to earthquakes (Hovland et al., 2002;
Christodoulou et al., 2003).
Using newly released bathymetry from NOAA, several large,
elongate, en echelon pockmarks were discovered at the edge of
the Virginia/North Carolina continental shelf (Fig. 1) by Driscoll
et al. (2000). While the usual scale of pockmarks ranges from a
few meters to ~300 m in diameter and up to 25 m in relief
(Dimitrov and Woodside, 2002; Christodoulou et al., 2003; Çifçi
uptoa kilometeracrossand50minrelief.Untilthese pockmarks
were discovered, pockmarks exceeding 350 m in diameter and
35 m in relief were classified as “giant” (Kelley et al., 1994).
normal faults diagnostic of some incipient slope failure (Driscoll
et al., 2000). However, further investigation in 2000 using
sidescan sonar and high-resolution sub-bottom profiling (chirp)
showed that these features are produced by gas seepage because
abundant gas is imaged in the sedimentary section housing the
giant pockmarks (Fig. 2) (Hill et al., 2004). Those authors
proposed a mechanism for formation of the pockmarks in which
methane migrates upslope beneath an impermeable shelf-edge
delta, creating an overpressure which, combined with downslope
creep, eventually leads to failure during which gas is expelled.
Hill et al. (2004) thus describe them as “gas blowouts.” This
scenario implies a pockmark age younger than the last glacial
maximum when the shelf-edge delta presumably formed.
Based on the existing shipboard data, it was not clear
whether gas continues to vent through the expulsion features
since their development, and whether the gas is thermogenic or
biogenic in origin. For example, pockmark fields recently
mapped in Belfast Bay, ME and off-shore Big Sur, CA show no
sign of active venting (Paull et al., 2002; Ussler et al., 2003).
The grid of chirp profiles collected during the 2000 study
clearly document gas withinthe shallow sediment at the walls of
the pockmarks (Fig. 2) (Hill et al., 2004). However, unlike what
has been reported in some other regions (e.g. Christodoulou et
al., 2003; Judd and Hovland, 2007), gas bubble plumes have not
been acoustically imaged in the water column. No sampling
was performed during the 2000 survey that would verify if gas
venting is presently occurring.
In July 2004, we carried out a detailed survey of the giant,
shelf-edge pockmarks with the R/V Cape Hatteras to determine
Fig. 1. Bathymetric map of the survey area produced with the ELAC-1180
multibeam sonar during the July 2004 survey. Core locations are plotted as
circles, black for cores squeezed for pore water geochemistry and red for those
saved for stratigraphy. Green stars show the locations of hydrocast sampling.
Cores 10P, 23P, 25P, 30P and 31P and hydrocasts 5 and 7 are identified. Red
boxes show the areas displayed in Fig. 7. Inset is an overview map of the area
with the red star showingthe location of the survey area. Visible coastlines in the
inset map are, from north to south, the southern tip of New Jersey, the Delmarva
Peninsula, and the barrier islands offshore North Carolina.
342 K.R. Newman et al. / Earth and Planetary Science Letters 267 (2008) 341–352
the source of gas and its fate in the water column. We made
in situ, near-bottom measurements of dissolved methane con-
centration in the pockmarks and surrounding areas using two
emerging technologies, an Autonomous Underwater Vehicle
(AUV) and a vehicle-mounted, underway METS methane sen-
sor. We collected cores, pore fluids and water column samples
for geochemical analysis to document the presence and nature
of gas discharge.
2. Data acquisition
2.1. Multibeam bathymetric survey
We acquired new high-resolution multibeam swath bathyme-
try data to better define and locate sediment and water column
sampling sites, as well as provide improved navigation for the
AUV missions (Figs. 1 and 7). The area was previously mapped
html) using a 36 kHz Hydrochart II multibeam bathymetric
of greater accuracy than the NOAA data, so a pole-mounted
SeaBeam/ELAC multibeam system was leased to produce a
new, higher resolution map of the pockmarks. This multibeam
system operates at 180 kHz with a swath width of 153° and
126 beams per ping. Velocity profiles used for processing the
multibeam data were calculated from daily casts of expendable
bathythermographs (XBTs). Ship tracks, aligned parallel to the
shelf-edge, were spaced 150 m apart to ensure ~100% swath
overlap in constructing the final bathymetric map (Fig. 1).
2.2. SeaBED AUV missions
AUVs are now sophisticated enough that they can per-
form accurately geo-referenced, detailed, near-bottom surveys
that were previously considered too expensive using remotely
operated vehicles (ROVs) or manned submersibles (Whitcomb
et al., 2000; Singh et al., 2004a). The AUV SeaBED (Fig. 3)
was designed at the Woods Hole Oceanographic Institution
for easy transport to remote locations as well as to be de-
ployable from small ships of opportunity, both of which re-
duce operational costs and ease survey logistics (Singh et al.,
The SeaBED AUV completed 16 successful dives (out of 18
deployments) across the pockmarks during which the vehicle
followed a pre-programmed track at a speed of approximately
0.5 m/s overground while maintaininganaltitude of 3m±0.1 m
above the seafloor. The AUV made continuous, in situ mea-
surements of the water properties in the pockmarks. A METS
methane sensor, manufactured by Franatech GmbH, Germany,
to measure in situ dissolved methane concentration, and a Sea-
bird Fastcat CTD were mounted on the AUV. Water was si-
multaneously pumped into both instruments so that each was
analyzing the same water sample. Microbathymetric data and
color photographs of the seafloor were also acquired contin-
uously along track. Water property measurements were made
approximately every second and photographs taken every 3 s.
The interval between photographs was selected to ensure some
overlap between frames so that continuous photomosaics could
The shipboard Acoustic Doppler Current Profiler (ADCP)
was used to determine the overall trend and strength of currents
in the area, as needed for planning the AUV dives. Underway
AUV navigation was based on compass readings on the vehicle
and a vehicle-mounted ADCP. Track navigation was adjusted
after the completion of each dive using ship-to-vehicle so-
nar ranging in conjunction with shipboard differential GPS.
expose the instrumentation.
Fig. 2. Chirp seismic profile across a shelf-edge pockmark, modified from Hill et al. (2004). Gas-charged sedimentsare visiblealong the western wall of the pockmark,
extending westward under the shelf. Gas-charged sediments are identified as a high amplitude reflector that obscures underlying reflectors.
343 K.R. Newman et al. / Earth and Planetary Science Letters 267 (2008) 341–352
Further re-navigation, comparing AUV measured seafloor
depth with the shipboard, GPS-navigated multibeam bathy-
metry, was needed because the vehicle's bottom-track velocity
measurements included a component of the strong SSW shelf-
edge current (Eustice et al., 2005). This current, which is clearly
expressed in the shipboard ADCP data (Fig. 4), flows south-
ward along the US east coast shelf-edge (Bumpus, 1973).
2.3. Water and sediment sampling program
Hydrocasts, located on Fig. 1, were deployed to collect wa-
ter column samples. The Conductivity, Temperature and Depth
(CTD) profile from each descent of the hydrocast were used to
identify optimal water sample collection depths during ascent.
Collected water samples were immediately stored in nitrogen-
purged 125 mL serum bottles, poisoned with mercuric chloride
to halt any methane production and oxidation, and were later
δ13C. Methane concentrations were measured with a gas chro-
matograph equipped with a flame ionization detector (GC 8A,
Shimadzu Corp.) and the methane δ13C isotopic analyses were
performed on a Finnigan MAT252 mass spectrometer with a
GC1 interface at the University of Hawaii following the tech-
nique of Popp et al. (1995). The average percent precision of
the methane concentration and methane δ13C analyses are b3%
and ±0.6‰, respectively. Other water aliquots were analyzed
aboard the ship for salinity and alkalinity and the rest were
preserved for shore-based analyses.
Sediment cores were collected throughout the survey area
(Fig. 1). Approximately half of the cores were saved for sedi-
mentology and stratigraphy. Selected cores were sub-sampled
under anaerobic conditions and the pore fluid was extracted
onboard using titanium squeezers. All pore fluids were passed
through 0.45 µm Gelman polysulfate filters to remove the re-
maining suspendedsiltandwereimmediately sub-sampledunder
anaerobic conditions for various shore-based analyses. Aliquots
polypropylene centrifuge tubes and acidified with Optima ni-
tric acid. Dissolved inorganic carbon (DIC) samples were poi-
soned with a saturated mercuric chloride solution and stored in
vacutainers. Sample aliquots for sulfate analyses were added to
acid-cleaned polypropylene centrifuge tubes containing a 50%
CdNO3solution to precipitate out the sulfide, thus leaving on-
ly sulfate in solution. Alkalinity and pH were measured im-
mediately onboard by Gran titration, and chloride concentrations
were determined by titration with AgNO3. Sulfate concentrations
were determined via ion chromatography (precision b0.6%),
stable isotope ratio mass spectrometer, with an average percent
precision b1.8%. Sediment sub-samples for pore fluid methane
concentration and δ13C isotope ratios were immediately taken
after core recovery, stored in nitrogen-purged serum bottles, and
preserved with a saturated mercuric chloride solution. Methane
concentrations and carbon isotopic ratios were determined using
the same techniques as described for the water column samples.
3.1. New multibeam bathymetric map
The new bathymetric map (Figs. 1 and 7) is of higher
resolution than the existing NOAA bathymetry. The higher
sonar frequency and the dense across track spatial sampling
allowed us to produce a new map based on a grid spacing
of 8 m. Although the giant pockmarks are well imaged with
the NOAA 3″ grid, the higher resolution map yields more
accurate information in steeper terrain. In particular, it high-
lights the striking linearity of the landward pockmark walls
(Figs. 1 and 7).
The amount of material excavated the pockmarks can be
estimated from the new high-resolution bathymetric map. The
average volume of the pockmarks is on the order of 107m3with
the smallest pockmark having a volume of 3·106m3and the
largest having a volume of 4·107m3. The total volume of
the pockmarks within our study area is ~108m3. Assuming a
Fig. 4. Effect of near-bottom currents measured by the ship's ADCP in the
survey area. The bottom three bins (each bin is 8 m) above the seafloor have
been averaged. The plot shows total displacement of a parcel of water over a
12 hour period with ticks every hour. A southerly current of 0.2 km/h domi-
nates, which is consistent with the previously documented shelf-edge current
(e.g. Bumpus, 1973). Tidal effects are expressed as east/west excursions.
344 K.R. Newman et al. / Earth and Planetary Science Letters 267 (2008) 341–352
typical porosity of 60–70%, this indicates that about 3–
4·107m3of sediment have been removed from the pockmarks
since their formation.
3.2. AUV along track data
The METS sensor, mounted on the SeaBED AUV, routinely
exhibit a systematic negative correlation of dissolved methane
concentration with temperature and salinity (Fig. 5). We found,
however, that the METS sensor response is characterized by a
significant time lag to changes in dissolved methane concentra-
tion. At places where sharp gradients in salinity and tempera-
concentrations grow or decay over a ~15 minute period and
plateau until a similar sharp salinity and temperature gradient
is encountered. This pattern suggests that the methane con-
centration values are affected by instrument performance and
we have devised a method to correct for the instrument re-
sponse (see Appendix A). After correction, the resulting dis-
solved methane concentration data mirror that of salinity and
temperature, increasing when salinity and temperature decrease
(Figs. 5 and 6). Additionally, we observe a correlation between
methane concentrations and bathymetry where methane anoma-
lies are all located at depths shallower than 130 m, which cor-
responds to the upper walls of the pockmarks and the adjacent
shelf-edge (Figs. 6 and 7).
Dissolved methane concentrations also vary temporally. The
data from the first three AUV dives show generally high meth-
ane concentrations throughout the deployments. Two storms
occurred during the early part of the cruise, before dive 4 and
before dive 9. Immediately after the second storm, methane
concentrations dropped to very low levels for the next three
dives. During the remainder of the AUV dives the average
background methane concentration increased slowly with time
after the storms. This is demonstrated by cross-over errors
in relative methane concentration between the data collected
recently after the storm and those collected near the end of the
survey (Fig. 7b).
Inspection of the 44,000 photos collected by the AUV does
not show the faunal or bacterial communities typical of cold
seeps (e.g. Hovland and Judd, 1988; Sibuet and Olu, 1998; Judd
and Hovland, 2007). Instead, seafloor sediment texture varies
from mud to gravel, and the commonly observed fauna include
fish (such as skate and chain dogfish), starfish and anemones.
Hence, if cold seeps are present within the study area, their
lateral extent must be less than the average spacing of the AUV
tracks (50 m to a few hundred meters).
Fig. 5. Near-bottom water properties collected by the SeaBED AUV during dive 16 (located in Fig. 7a). The black line is the raw dissolved methane data generated by
the METS sensor. The red line is the data filtered using the first principle component from empirical orthogonal function analysis (see Appendix A). The green line is
the corrected dissolved methane. The blue line is salinity, measured by the AUV mounted SeaBird CTD. Temperature data are not plotted, but they follow a similar
pattern as the salinity data, as illustrated in Fig. 6.
Fig. 6. Along track data from dive 16 (located in Fig. 7a) displaying the categorization of the methane anomaly. White areas are background concentration, darkly
shaded are high methane concentration and lightly shaded are intermediate methane concentration. A correlation is typically observed in all dives between salinity,
temperature and dissolved methane concentration.
345 K.R. Newman et al. / Earth and Planetary Science Letters 267 (2008) 341–352
3.3. Spatial distribution of methane anomalies
Instead of interpreting the temporally varying absolute meth-
ane concentrations, we grouped methane anomalies into three
categories: background concentration, high concentration and
intermediate concentration (Fig. 6). Background concentration
is defined as the average low concentration for each dive (usu-
anomalies well above background (N50 nM) and are char-
acterized by a steep along-track gradient at either end. Areas
identified as intermediate concentration have dissolved methane
concentrations that are higher than background, but are either
lower than areas identified as high concentration, are slowly
increasing or decreasing along-track, or are fluctuating.
Fig. 7 displays the spatial distribution of methane anomalies
according to the above three categories. It includes 13 SeaBED
dives that we evaluated to be the most consistent and reliable of
the successful deployments on the basis of cross-over errors at
track intersections. The northern part of the survey area (Fig. 7a)
shows high methane on the landward walls of the pockmarks
and on the shelf landward of the pockmarks with no meth-
ane over the floors of the pockmarks. The southern pockmarks
(Fig. 7c) show a similar pattern. However, the small, circular
pockmark in the south appears to have little or no methane
to the north and west, but shows a streak of elevated meth-
ane concentration extending from its southern wall toward the
southeast. The middle section of the survey (Fig. 7b) is the
most complex and the most densely sampled. High methane is
observed along the western walls of the pockmarks and
continues along the shelf to the west. Elevated methane is
observed at the bathymetric highs along the eastern edges of the
pockmarks. All dives, except dive 4, show little to no methane
venting at the floors of the pockmarks. Dive 4, one of the
earliest dives, is anomalous and displays high methane through-
out most of the deployment, even along the floor of the pock-
mark. Low methane concentration is only observed on Dive 4
outside of the pockmark, down the continental slope from the
shelf break at depths greater than 160 m (Fig. 7b). Dives 5 and 8
were not plotted because the corrected methane concentra-
tions appear too internally inconsistent within the deployments:
methane concentrations change only when the vehicle begins to
travel down slope suggesting that other factors, such as the
vehicle's response to the changing bathymetry, may also affect
measurements. Dive 1 displays higher dissolved methane con-
centrations compared to all the other dives and was omitted
due to inability to separate the data into the three previously
mentioned classifications and because the dissolved methane
concentration, salinity and temperature correlation is less ro-
bust. This pattern might be due to initial calibration issues.
3.4. Dissolved hydrocarbons in water column samples
We measured dissolved methane concentrations as high as
40 nM in water column samples collected with the hydrocasts
(Supplementary Table 1). The measured concentrations are
significantly higher than those in average seawater (1–4 nM
Fig. 7. Bathymetry maps of the pockmarks showing the spatial distribution of the methane anomaly. Black is background methane concentration, red is high methane
concentration and green is intermediate methane concentration (see text for details). AUV dive numbers are given at the beginning of the dive track.
346 K.R. Newman et al. / Earth and Planetary Science Letters 267 (2008) 341–352
(Holmes et al., 2000; Sansone et al., 2001; Reeburgh, 2007)). In
all the hydrocast profiles, low methane occurs near the floors
of the pockmarks and higher concentrations occur at depths
corresponding to the top of the pockmark walls (100–130 m),
consistent with the near-bottom measurements made from the
AUV. In most profiles, the correlation between salinity, tem-
perature and methane concentration, observed for near-bottom
AUV measurements (Fig. 6), breaks down with altitude above
the seafloor (Fig. 8). In addition to the elevated methane con-
centrations, trace amounts of higher molecular weight hy-
drocarbons, mainly ethane and propane, were measured in the
water column. Dissolved ethane, propane and butane concen-
trations are usually negligible, but in some samples ethane and
propane concentrations are more prominent. Isotopic analysis
of the dissolved methane in the hydrocast samples shows that
the methane δ13C values range from −65 to −45‰, with most
measurements less than −60‰ (Supplementary Table 1).
3.5. Pore fluid geochemistry
Depth profiles of DIC δ13C values and sulfate concentration
in sediment pore fluids yield information about the nature of the
microbiological reactions, on organic matter diagenesis,
methane flux, and anaerobic methane oxidation (AMO). Sulfate
profiles have been obtained for pore fluids squeezed from piston
cores 10P, 23P, 25P, 30P, 31P and 34P and DIC δ13C profiles
from 10P, 23P and 25P (Fig. 9; Table 1; Supplementary Table
2). Sulfate concentration decreases with depth below the
seafloor, reaching zero concentration at depths as shallow as
50–65 cm below the seafloor in core 30P and 34P. DIC δ13C
also decreases with depth, reaching a minimum of −34.4‰ at
115–138 cm below the seafloor in core 10P (Table 1). The
shallowness of the sulfate reduction zone and isotopically light
DIC δ13C suggest active methane advection and AMO at the
pockmarks (Borowski et al., 1999).
4.1. Methane venting at the pockmarks
The observed methane anomaly concentrations of 50–
150 nM and ~30 nM observed in the METS sensor data
and in the hydrocast samples, respectively, are significantly
higher than average seawater dissolved methane concentra-
tions (1–4 nM), confirming that methane is actively venting in
the pockmarks. While these values are lower than the 200–
1500 nM concentrations measured at some other pockmarks
(e.g. Bohrmann et al., 2002; Christodoulou et al., 2003), they
Fig. 8. Top panel: CTD and dissolved methane concentration profiles from
hydrocast 5, which sampled the eastern wall of a large pockmark (located in
Fig. 1). Salinity is plotted in black, temperature in gray and the laboratory
measured dissolved methane concentration as points. Some correlation between
dissolved methane concentration, salinity and temperature is seen in the lower
part of the profile, but it begins to break down at depths shallower than the peak
methane concentration. A step is visible in salinity and temperature at 110 m, the
depth at which methane begins to increase in the profile. Bottom panel: CTD
profile from AUV dive 9 ascent, which sampled the center of the same
pockmark (Fig. 7b). No steps, except for noisy excursions, are visible in this
profile, suggesting that the methane-rich water mass that is slightly colder and
fresher than the bottom water is not present at this location.
Fig. 9. Sulfate and DIC δ13C profiles for core 23P. Sulfate is plotted in black,
DIC δ13C in gray. Sulfate concentration is nearly zero at 150–159 cm below
seafloor, marking the depth of the sulfate–methane interface.
347K.R. Newman et al. / Earth and Planetary Science Letters 267 (2008) 341–352
are significantly higher than 1–3 nM concentrations observed in
pockmarks with no evidence of venting (Paull et al., 2002).
The shallowness of the sulfate-methane interface (SMI)
and the isotopic signature of the DIC also support methane
advection and AMO (Table 1). Two processes contribute to
sulfate reduction (Claypool and Kaplan, 1974; Borowski et al.,
1999): Organic Matter Oxidation (OMO)
2ðCH2OÞ þ SO2−
and Anaerobic Methane Oxidation (AMO)
The DIC is expressed in the bicarbonate ion (HCO3
reactions. OMO, which is the more common of the two re-
actions, occurs in all sulfate-reducing environments when or-
ganic matter is present. In the presence of only OMO, the
minimum δ13C of the resulting DIC is equal to that of the
organic carbon involved in the reaction. Organic carbon has
a typical δ13C range of −20‰ to −22‰ for marine carbon
and −26‰ to −32‰ for terrigenous carbon (Hedges, 1992).
However, east coast U.S. rivers deliver particulate organic
carbon that can be isotopically lighter (−33.7% for the Parker
River (Raymond and Bauer, 2001)). Conversely, DIC δ13C
values resulting from AMO can be less than that of OMO
because methane is isotopically lighter. We measured meth-
ane δ13C less than −60‰ in our water column samples. The
minimum δ13C of the DIC in the pore fluids ranges from
−30.9‰ to −34.4‰ (Table 1). Given that the isotopic com-
position of the pore fluid DIC represents a mixture of all
end member sources of carbon involved in sulfate reduction
(e.g. Blair et al., 2004), the minimum value of the DIC δ13C
found in the pore fluids suggests that AMO is the dominant
sulfate-reducing process in the shallow sedimentary section.
This is consistent with the shallow depth to the SMI observed
in the core pore fluids (Table 1). The pore fluid geochemistry
results strongly suggest that methane is actively being advected
toward the seafloor.
In the pockmarks, elevated methane concentrations coin-
cide with T–S anomalies, indicating that the vented fluids are
cooler and less saline than the background water. A possible
source of this anomalous water is water emplaced as ground-
water during the last glacial maximum when sea level was
~130 m below current sea level, leaving the shelf-edge exposed.
Groundwater would have been present beneath the shelf-edge
and would be fresh by nature and cold because of the cooler
3þ HS−þ H2O:
−) in these
ambient temperature. Accordingly, ODP sites 1071 and 1072 on
the New Jersey margin sampled interstitial water with salinities
as low as 25 (Austin et al., 1998), showing that fresher water
is present in the subsurface.
4.2. Methane source
The δ13C of the methane is sufficiently light to indicate
a primarily biogenic source for the vented methane. Methane
δ13C lower than −60‰ is generally attributed to biogenic
methane, whereas methane δ13C heavier than −45‰ is con-
sidered thermogenic in origin (Whiticar et al., 1986; Ussler
et al., 2003). Since most of the water column methane δ13C is
less than −60‰, the vented methane has a primarily biogenic
Isotopically light methane can also be produced abiogen-
ically under serpentinizing conditions through reactions cata-
lyzed by metallic minerals found in igneous rocks (Horita and
Berndt, 1999). However, in our region 10 km of sediment
overlie the basement within the study area (Holbrook et al.,
1994) and faults are not observed that would allow fluids to
migrate through the sediment carapace. Thus, it seems unlikely
that serpentinization processes contribute to the vented methane
4.3. Diffuse vs. focused venting
The spatial distribution of the elevated methane concentra-
tions (Fig. 7) and the CTD profiles (Fig. 8) show that methane-
rich fluids are not present along the floors of the pockmarks.
This is consistent with previous observations based on high-
resolution (chirp) seismic profiling (Hill et al., 2004), where
gas-charged sediments are seismically imaged along the walls
of the pockmarks, but in most cases gas is not seismically
imaged beneath the floors of the pockmarks (e.g. Fig. 2).
The pattern of methane venting illustrated by the AUV data
itself is not sufficient to determine the overall venting pattern in
the area. Since the AUV maintained an altitude of 3 m above
seafloor during the dives, its depth below the sea surface
constantly changed. Thus, the observed variations in methane
concentration might be due to horizontal stratification, such as
those observed by Berner et al. (2003), rather than to focused
venting of methane-rich fluids. To determine if a horizontally
extensive methane-rich layer exists, the near-systematic cor-
relation between salinity, temperature and dissolved methane
concentration was exploited by examining CTD profiles ac-
quired by the AUV during its descent to and ascent from the
seafloor. The observed near seafloor relationship between dis-
solved methane concentration, salinity and temperature tends to
break down in the water column (Fig. 8), likely due to mixing
occurring in a water column with existing vertical structure.
Vertical profiling may still be useful to evaluate the horizontal
extent of a well defined methane-rich layer. Near the seafloor,
methane anomalies occur in areas where the salinity and tem-
perature are lower than the surrounding water, indicating that
the vented methane-rich fluid is less saline and cooler than
the bottom water. These observed salinity and temperature
Pore fluid geochemistry data
Core number Depth to SMI
Min. δ13C DIC
SMI not reached
348 K.R. Newman et al. / Earth and Planetary Science Letters 267 (2008) 341–352
gradients are not due to the interfingering of shelf and slope
water resulting from the summer weakening of the shelf/slope
front as described by Flagg et al. (1994), Burrage and Garvine
(1982) and Gordon and Aikmann (1981). Structure related to
that feature is visible farther up in the water column, at water
depths shallower than 80 m. The overall trend in the 100–150 m
water depths of the study area is that both temperature and
salinity increase significantly with depth (Figs. 8 and 10).
Therefore, as altitude increases, the affect on temperature
and salinity due to mixing will become less pronounced as the
physical characteristics of the water approach those of the
vented fluids. This water column structure enables us to locate
the base of the methane-rich layer by a sharp gradient in tem-
perature and salinity, but above that step-like variation in phys-
ical properties, the previously observed correlation between
dissolved methane concentration, salinity and temperature will
become less robust and the top of the methane-rich layer may
not be sharply defined in the temperature and salinity data.
We observe an evidence of a methane-rich layer in most
of the CTD profiles generated during the AUV descents and
ascents that extend deeper than 100 m and are sited near the
landward (west) walls of the pockmarks. However, profiles
collected during the ascents of dives 9 and 18 (Fig. 8) do not
show steps in salinity and temperature, suggesting that the
observed methane-rich water mass is not laterally continuous.
These two profiles are far enough away from the pockmark
walls that advection due to tidal forcing (Fig. 4) would not have
carried the methane-rich fluids to the profile location.
Based on these observations, we hypothesize that the pattern
in the spatial distribution of the methane anomaly mostly re-
flects an area of venting along the walls of the pockmarks with
some transport by local currents contributing to the observed
distribution of the methane (Fig. 10). We have considered other
venting scenarios, including localized venting at specific sites,
either along the pockmark walls or through the floor of the
pockmarks, or a methane-rich water mass transported from
elsewhere. However, these are unlikely because they require the
presence of a laterally extensive methane-rich water layer that is
not supported by the data.
4.4. Temporal variations in methane venting
During this survey we observed temporal variations in the
methane concentration with background methane concentra-
tions near the seafloor dropping to insignificant levels after two
storms and then gradually increasing. Examination of XBT and
CTD profiles collected throughout the cruise shows that the
structure of the water column also changed in response to the
two storms. Thus, it seems likely that the storms either shifted
the water masses in the area, clearing the methane-rich water
that had accumulated in the pockmarks, or induced water col-
umn mixing so that methane concentrations were greatly re-
duced near the seafloor. In effect, these storms seem to have
“reset the system,” and provided an unexpected opportunity to
track the build up of the methane anomalies.
Changes in seepage rates that coincide with tidal variations
are commonly observed (e.g. Mikolaj and Ampaya, 1973;
Orange et al., 1997; Boles et al., 2001; Torres et al., 2002;
Forrest et al., 2005). Christodoulou et al. (2003) observe both
seasonal changes in methane concentration after sampling wa-
ter above pockmarks on a monthly basis as well as increased
methane concentrations following an earthquake. Boles et al.
Fig. 10. Proposed methane venting scenario. Left panel data are from hydrocast 7 and plotting conventions are the same as in Fig. 8. The right panel shows a
representative chirp profile across one of the pockmarks (Hill et al., 2004). White arrows denote the location of methane venting. The shaded areas indicate where
methane-rich water is found, suggesting spreading due to diffusion and advection due to currents. The bold line simulates the AUV track across the pockmark and the
dashed line simulates the location of the hydrocast data presented in the left panel.
349 K.R. Newman et al. / Earth and Planetary Science Letters 267 (2008) 341–352
(2001) document tidal forcing resulting in 4–7% variations
about the mean in methane concentrations measured in situ by
tent-shaped traps placed on the seafloor. Within the study area,
the tidal effect at the seafloor is apparent in the shifting current
directions (Fig. 4). However, because our measurements were
not taken in a stationary position over a tidal cycle, the mag-
nitude of the tidal forcing on these pockmarks cannot be quan-
tified. Nonetheless, based on reported variations in methane
concentration in other nearshore settings and cross-over errors
between subsequent AUV dives, we estimate that tidal forcing
results in less than a 20% variation in methane concentration
about the mean local value.
4.5. Evolution of elongate pockmarks
Most of the pockmarks in our survey area are kilometer
scale, elongate features, but a smaller, circular pockmark is
present at the southernmost part of the survey area (Figs. 1
and 7c). This pockmark is ~300 m in diameter and has slightly
less relief (b40 m) than the other pockmarks in the area. The
largest methane anomaly was recorded along the southwestern
wall of this pockmark during AUV dive 12, possibly indicating
that venting is more vigorous in this pockmark. Pockmark fields
with no evidence of gas or fluid venting have been identified
(e.g. Paull et al., 2002). Judd and Hovland (2007) suggest that
features like these might be relict, indicating earlier gas or fluid
venting. Ivanov et al. (2007) similarly conclude that the remains
of chemosynthetic communities found in pockmarks on the
Vøring Plateau off western Norway give evidence of previous
fluid venting that has since ceased. We further hypothesize that
as the pockmarks age and the reservoir of vent material becomes
depleted, venting rates might be reduced. According to this
scenario, the southernmost pockmark might be the youngest.
Our data are spatially limited, so it is possible that high methane
concentrations exist in other parts of the survey area and were
not sampled. However, if the smallest pockmarks are indeed the
youngest, it would support Kelley et al.'s (1994) model of
pockmark formation in which the pockmarks are gradually
enlarged due to gas or fluid escape.
Hill et al. (2004) attribute the elongate nature of the pock-
marks to stress changes resulting from downslope creep with-
in the shelf-edge delta. Çifçi et al. (2003) hypothesized that
elongate pockmarks on the Turkish shelf of the Black Sea
formed by the merging of smaller pockmarks. There, the round
pockmarks were 1/4–1/2 the size of the larger, elongate pock-
marks. These two mechanisms are not mutually exclusive and
their combined effect might be responsible for the formation
of the pockmarks in our study area. The smaller pockmarks
could be younger and thus have not yet had sufficient time
to grow and merge with adjacent pockmarks. The irregularly
shaped pockmark where cores 10P, 23P and 25P were collected
(Fig. 7b) might get its shape from the coalescing of three or
more smaller pockmarks (the thin northward extension and
the easternmost section of the pockmark could each have been
separate features before joining with what is now the central
portion of the pockmark). Downslope creep also likely con-
tributes to the shape of the features by influencing where the
pockmarks form and by causing them to preferentially align and
spread parallel to the shelf break. The western walls of the
pockmarks are very linear, systematically oriented parallel to
the continental slope, and are arranged in an en echelon, left-
stepping pattern, all of which suggest their formation is partially
controlled by local stress at the shelf-edge. Hence, the tensional
regime induced by creep processes in the deltaic sediments
draped over the shelf-edge controls the locations of the pock-
marks, likely resulting in linear array of circular pockmarks that
eventually coalesce into en echelon, elongate features.
Near seafloor dissolved methane concentration measure-
ments from the SeaBED AUV, combined with CTD profiles
document the distribution of active venting in the pockmarks
along the U.S. mid-Atlantic shelf-edge. Methane venting is
concentrated along the upper parts of the pockmark walls and
adjacent shelf area and is not occurring through the floors of the
pockmarks. A correlation is observed, both in the AUVand the
lower sections of hydrocast data, between increased methane
concentration and decreased salinity and temperature. This
correlation allows the use of CTD casts to determine that the
methane-rich water mass is not laterally extensive across the
pockmarks. The formation and linear arrangement of these
pockmarks is likely related to linearly trending tension due to
downslope creep at the shelf break. Their elongate shape may be
related to the progressive merging of smaller, initially more
circular pockmarks, consistent with apparently more vigorous
venting at a smaller circular pockmark at the southern end of the
We thank the captain and crew of the R/V Cape Hatteras for
their cooperation and Kevin Tomanka of Seafloor System, Inc.
for his assistance in acquiring the multibeam bathymetric data.
We thank Robert Houghton and Douglas Martinson for their
helpful insights into ocean mixing processes. This work was
supported by NSF grants OCE-0242426, OCE-0242804 and
OCDE-0242449 and ONR grant N00014-02-1-0691. This is
LDEO contribution number 7114.
Appendix A. Response and correction of the METS sensor
The METS sensor allows for near real time measurement
of dissolved methane concentration from a moving platform.
Most users report that the METS sensor reacts as expected
with the ability to detect subtle changes in methane concentra-
tion (Bussel et al., 1999). However, other studies (Lamontagne
et al., 2001; Paull et al., 2002) show a time lag in its response
and a delay in returning to “normal” values after reading high
methane concentrations. Occasionally, concentrations measured
by the METS sensor are significantly lower than those measured
analytically. Conventional methods for determining dissolved
methane concentration involve retrieving water samples from
depth for later analysis (e.g. Clarke et al., 2000; Christodoulou
350 K.R. Newman et al. / Earth and Planetary Science Letters 267 (2008) 341–352
et al., 2003). The METS sensor employs a semiconductor whose
resistance varies with the amount of methane present in the
detection chamber. As methane molecules in the water diffuse
across a silicon membrane into the chamber, they participate in
an electron exchange with oxygen and modify the resistance
across the semiconductor. The resulting change in the measured
voltage is directly related to the dissolved methane concentration
Visual inspection of the raw dissolved methane concentra-
tion, salinity and temperature time series data (Fig. 5) shows a
correlation between the three constituents: elevated dissolved
methane concentrations are observed in areas of decreased
salinity and temperature. However, it appears that the variations
in dissolved methane concentration lag the corresponding sa-
linity and temperature variations. When a square-shaped signal
is observed in the salinity and temperature data, the dissolved
methane concentration, recorded by the METS sensor, begins to
increase at the start of the excursion and continues to increase
until the end of the salinity/temperature anomaly, at which time
it decays back to background levels. This is the expected re-
sponse for diffusion across a membrane (Newman et al., 2005;
Fukasawa et al., 2006). The theoretical response of this pro-
cess is that concentrations should increase as a function of
1−exp(−t/τ) and decay as a function of exp(−t/τ), where τ
is the time constant of the system. Fukasawa et al. (2006) give
τ as the function (VL)/(RTAPT), where V is the volume of
the detector room, L is the membrane thickness, R is the gas
constant, Tis the water temperature, Ais membrane permeation
area and PTis boundary layer resistance. The response of the
sensor can be expressed as the finite difference function
ð Þ ¼ y tn?1
ð Þ þ x tn?1
ð Þ ? y tn?1
ðÞ½? 1 ? e?Dt=s
where x(t) is the input function and y(t) is the output. The
actual signal can then be retrieved as
ð Þ ¼ y tn
ð Þ þy tnþ1
ð Þ ? y tn
1 ? e?Dt=s
All dive data were corrected using the above algorithm
(Fig. 5) with the time constant for the system of approximately
11 min giving the best visual fit to the data. Although a low
signal to noise ratio exists in the data recorded by the METS
sensor, the data had to be low pass filtered prior to applying
the correction because the algorithm amplifies high frequency
noise. The noise was removed through empirical orthogonal
function analysis. Each time series was analyzed separately. In
all cases the first principle component was used as the filtered
form of the data because it represented over 93%, and in most
cases, over 97% of the variance in the data.
Corrected methane concentrations are significantly larger
than those measured in the hydrocast samples. This may be
accounted for by the hydrocast samples being taken at a greater
altitude above the seafloor than the AUV measurements. This is
also consistent with the results of Chirstodoulou et al. (2003)
where they observe an order of magnitude difference between
near seafloor and upper water column measurements.
This instrumental response can also explain the differences
in METS measured and analytically measured methane con-
centrations by Lamontagne et al. (2001). Since the amount of
time spent in some methane-rich areas is considerably shorter
than the time constant of the instrument, the concentration
measured by the instrument would not have had enough time
to ramp up to the true value.
Appendix B. Supplementary data
Supplementary data associated with this article can be found,
in the online version, at doi:10.1016/j.epsl.2007.11.053.
Austin Jr., J.A., Christie-Blick, N., Malone, M.J., et al., 1998. Proc. ODP. Init.
Repts., 174A. Ocean Drilling Program, College Station, TX.
Berner, U., Poggenburg, J., Faber, E., Quadfasel, E., Frische, A., 2003. Methane
in the ocean waters of the Bay of Bengal: its sources and exchange with the
atmosphere. Deep-sea Res., Part 2, Top. Stud. Oceanogr. 50, 925–9250.
of particulate organic carbon across coupled watershed–continental margin
systems. Mar. Chem. 92, 141–156.
Bohrmann, G., Heeschen, K., Jung, C., Weinrebe, W., Baranov, B., Cailleau, B.,
Heath, R., Hühnerbach, V., Hort, M., Masson, D., Trummer, I., 2002. Wide-
spread fluid expulsion along the seafloor of the Costa Rica convergent
margin. Terra Nova 14, 69–79.
Boles, J.R., Clark, J.F., Leifer, I., Washburn, L., 2001. Temporal variation in
natural methane seep rate due to tides, Coal Oil Point area, California.
J. Geophys. Res. 106 (C11), 27077–27086.
Borowski, W.S., Paull, C.K., Ussler, W., 1999. Global and local variations
of interstitial sulfate gradients in deep-water, continental margin sediments:
Sensitivityto underlyingmethane and gas hydrates. Mar. Geol.159, 131–154.
Bumpus, D.F., 1973. A description of the circulation on the continental shelf of
the east coast of the United States. Prog. Oceanogr. 6, 111–157.
Burrage, D.M., Garvine, R.W., 1982. Summertime hydrography at the shelf-
break front in the Middle Atlantic Bight. J. Phys. Oceanogr. 18, 1309–1319.
Bussel, J., Klinkhammer, G., Collier, R., Linke, P., Appel, F., Heeschen, K.,
Çifçi, G., Dondurur, D., Ergün, M., 2003. Deep and shallow structures of
large pockmarks in the Turkish shelf, Eastern Black Sea. Geo Mar. Lett. 23,
seepage in two contrasting pockmark fields in the Patras and Corinth gulfs,
Greece. Geo Mar. Lett. 23, 194–199. doi:10.1007/s00367-0063-0151-0.
Clarke, J.F., Washburn, L., Hornafus, J.S., Luyendyk, B.P., 2000. Dissolved
hydrocarbon flux from natural marine seeps to the southernCalifornia Bight.
J. Geophys. Res. 105, 11509–11522.
Claypool, G.E., Kaplan, I.R., 1974. The Origin and Distribution of Methane in
Marine Sediments. In: Kaplan, I.R. (Ed.), Natural Gases in Marine Sedi-
ments. Plenum Press, New York, pp. 94–139.
Dimitrov, L., Woodside, J., 2002. Deep sea pockmark environments in the
eastern Mediterranean. Mar. Geol. 195, 263–276.
Driscoll, N.W., Weissel, J.K., Goff, J.A., 2000. Potential for large-scale slope failure
Eustice, R., Camilli, R., Singh, H., 2005. Towards bathymetry-optimized
Doppler re-navigation for AUVs. Proc. IEEE/MTS OCEANS Conf. Exhib.,
Washington, DC, pp. 1430–1436.
Flagg, C.N., Houghton, R.W., Pietrafesa, L.J., 1994. Summertime thermocline
Top. Stud. Oceanogr. 41 (2/3), 325–340.
Forrest, M.J., Ledesma-Vásquez, J., Ussler, W., Kulongoski, J.T., Hilton, D.R.,
Greene,H.G., 2005.Gas geochemistryof a shallowsubmarinehydrothermal
351 K.R. Newman et al. / Earth and Planetary Science Letters 267 (2008) 341–352
vent associated with the El Requesón fault zone, Bahía Concepción, Baja Download full-text
California Sur, México. Chem. Geol. 224, 82–95.
Fukasawa, T., Hozumi, S., Morita, M., Oketani, T., Masson, M., 2006. Dis-
solved methane sensor for methane leakage monitoring in methane hydrate
production. Proc. IEEE OCEANS Conf. Exhib., Boston, MA, pp. 1–6.
Gordon, A.L., Aikmann III, F., 1981. Salinity maximum in the pycnocline of the
Middle Atlantic Bight. Limnol. Oceanogr. 26 (1), 123–130.
Hedges, J.I., 1992. Global biogeochemical cycles: progress and problems. Mar.
Chem. 39, 67–93.
Hill, J.C., Driscoll, N.W., Weissel, J.K., Goff, J.A., 2004. Large-scale elongate
gas blowouts along the U.S. Atlantic margin. J. Geophys. Res. 109, B09101.
Holbrook, S.W., Purdy, G.M., Sheridan, R.E., Glover III, L., Talwani, M.,
Ewing, J., Hutchinson, D., 1994. Seismic structure of the U.S. mid-Atlantic
continental margin. J. Geophys. Res. 99 (B9), 17871–17891.
Holmes, M.E., Sansone, F.J., Rustt, T.M., Popp, B.N., 2000. Methane production,
consumption, and air–sea exchange in the open ocean: an evaluation based on
carbon isotopic ratios. Glob. Biogeochem. Cycles 14 (1), 1–10.
Horita, J., Berndt, M.E., 1999. Abiogenic methane formation and isotopic frac-
tionation under hydrothermal conditions. Science 285, 1055–1057.
Hovland, M., Judd, A.G., 1988. Seabed Pockmarks and Seepages, Impact on
Geology, Biology and the Marine Environment. Graham and Trotman,
Hovland, M., Gardner, J.V., Judd, A.G., 2002. The significance of pockmarks to
understanding fluid flow processes and geohazards. Geofluids 2, 127–136.
Ivanov, M., Westbrook, G.K., Blinova, V., Kozlova, E., Mazzini, A., Nouzé, H.,
Eos Trans. AGU 88 (19), 209 & 212.
Judd, A.G., 2003. The global importance and context of methane escape from
the seabed. Geo Mar. Lett. 23, 147–154. doi:10.1007/s00367-003-0136-z.
Judd, A.G., Hovland, M., 2007. Seabed Fluid Flow: the Impact on Geology,
Biology and the Marine Environment. Cambridge University Press, New
York. 475 pp.
Judd, A.G., Hovland, M., Dimitrov, L.I., Carcía Gil, S., Jukes, V., 2002. The
geological methane budget at Continental Margins and its influence on
climate change. Geofluids 2, 109–126.
Kelley, J.T., Dickson, S.M., Belknap, D.F., Barnhardt, W.A., Henderson, M.,
1994. Giant sea-bed pockmarks: evidence for gas escape from Belfast Bay,
Maine. Geol. 22, 59–62.
King, L.H., McLean, B., 1970.Pockmarkson the ScotianShelf. Bull. Geol. Soc.
Am. 81, 3141–3148.
Lamontagne, R.A., Rose-Pehrsson, S.L., Grabowski, K.E., Knies, D.L.,
2001. Response of METS sensor to methane concentrations found on the
Texas–Louisiana Shelf in the Gulf of Mexico. Naval Research Laboratory
Memorandum Report NRL-MR-6110-01-8584, Washington, DC. 15 October.
pockmarks and mounds in the Nile deep-sea fan (Eastern Mediterranean):
geophysical evidences. Mar. Pet. Geol. 21, 669–689.
Mikolaj, P.G., Ampaya, J.P., 1973. Tidal effects on the activity of natural sub-
marine oil seeps. Mar. Technol. Soc. J. 7, 25–28.
2005. Geophysical and geochemical evidence for methane venting at large
gas blowouts along the US mid-Atlantic Shelf edge. Eos Trans. AGU 86 (52)
Fall Meet. Suppl., Abstract OS32A-05.
Orange, D.L., Saffer, D., Maher, N., 1997. Tidally mediated fluid expulsion at
a cold seep. Eos Trans. AGU 78 (52), Fall Meet. Suppl. Abstract T31E-11.
2002. Pockmarks off Big Sur, California. Mar. Geol. 181 (4), 323–355.
Popp, B.N., Sansone, F.J., Rust, T.M., Merritt, D.A., 1995. Determination of
concentration and carbon isotopic composition of dissolved methane in
sediments and nearshore waters. Anal. Chem. 67, 405–411.
Raymond, P.A., Bauer, J.E., 2001. Riverine export of aged terrestrial organic
matter to the North Atlantic Ocean. Nature 409, 497–500.
Reeburgh, W.S., 2007. Oceanic methane biogeochemistry. Chem. Rev. 107,
Sansone, F.J., Popp, B.N., Gasc, A., Graham, A.W., Rust, T.M., 2001. Highly
elevated methane in the eastern tropical North Pacific and associated iso-
topically enriched fluxes to the atmosphere. Geophys. Res. Lett. 28 (24),
Sibuet, M., Olu, K., 1998. Biogeography, biodiversity and fluid dependence of
deep-sea cold-seep communities at active and passive margins. Deep-sea
Res., Part 2, Top. Stud. Oceanogr. 45, 517–567.
Sills, G.C., Wheeler, S.J., 1992. The significance of gas for offshore operations.
Cont. Shelf Res. 12 (10), 1239–1250.
Sensing Technol. Appl. 5 (1), 498–514.
Singh, H., Can, A., Eustice, R., Lerner, S., McPhee, N., Pizzaro, O., Roman, C.,
2004b SeaBED AUV Provides New Platform for High Resolution Imaging.
Eos Trans., AGU 85 (31), 289, 294–295.
Torres, M.E., McManus, J., Hammond, D., Angelis, M., Heeschen, K.U.,
Colbert, S.L., Tryon, M.D., Brown, K.M., Suess, E., 2002. Fluid and
chemical fluxes in and out of sediments hosting methane hydrate deposits on
Hydrate Ridge, OR, I. Hydrological provinces. Earth Planet. Sci. Lett. 201,
Ussler III, W., Paull, C.K., Boucher, J., Friederich, G.E., Thomas, D.J., 2003.
Submarine pockmarks: a case study from Belfast Bay, Maine. Mar. Geol.
202 (3–4), 175–192.
Whitcomb, L., Yoerger, D., Singh, H., Howland, J., 2000. Advances in un-
derwater robot vehicles for deep ocean exploration: navigation, control and
survey operations. Robotics Research ― The Ninth International Sym-
posium. Springer-Verlag Publications, London, pp. 439–448.
Whiticar, M.J., Faber, E., Schoell, M., 1986. Biogenic methane formation in
isotope evidence. Geochim. Cosmochim. Acta 50, 693–709.
352K.R. Newman et al. / Earth and Planetary Science Letters 267 (2008) 341–352