Sonar surveys provide an effective mechanism for mapping seabed methane flux
emissions, with Arctic submerged permafrost seepage having great potential to
significantly affect climate. We created in situ engineered bubble plumes
from 40 m depth with fluxes spanning 0.019 to 1.1 L s
On a century timescale, methane (CH
Arctic continental shelf sediment accumulates five times faster than in other world oceans. Sedimentation for the Siberian Arctic Shelf where the six Great Siberian Rivers outflow has deposited organic carbon that approximately equals accumulations over the entire pelagic area of the world's oceans. This leads to the thickest (up to 20 km) and most extensive sedimentary basin in the world (Gramberg et al., 1983).
Terrestrial Arctic permafrost CH
Sonar is the most common survey approach and has been used on concentrated
seep areas covering
Sonar has also mapped significantly larger and stronger seepage in the Coal
Oil Point (COP) marine hydrocarbon seep field offshore California. The COP
seep field covers
ESAS seepage is on a dramatically larger scale with
Given the ESAS seepage extent there is a critical need for new approaches to effectively, rapidly, and quantitatively survey seepage areas. Video is inadequate to survey extensive or widely dispersed seepage, a task for which sonar (active acoustics) excels. This study demonstrates an improved approach to quantify seabed seepage using in situ calibrated sonar-derived bubble fluxes. Bubble plumes were observed in the ESAS and offshore California. In combination, the in situ studies covered a broad range of flows and included fine-depth resolution of near-source (growth) plume processes (California) and coarser resolution of plume processes to tens of meters where the plume is self-similar.
Both multibeam echosounder (MBES) and single-beam echosounder (SBES) data were collected in the ESAS, just MBES data of rising engineered bubble plumes were collected in California. These data were collected both as a depth-dependent calibration and to investigate the effect of multiple acoustic scattering in bubble plumes.
The calibration was applied to quantify in situ sonar observations of three natural seepage areas in the ESAS. Because the calibration bubble plumes and seep bubble plumes were different gases and from different source depths, bubble dissolution rates are different – i.e., for the same seabed mean volume flux, the depth-window-averaged volume fluxes are different. We demonstrate a first correction attempt based on a numerical bubble plume model between the calibration and natural seepage bubble flows.
The Siberian Arctic Shelf contains vast CH
The ESAS is the world's largest and shallowest shelf (covering
ESAS subsea permafrost degradation allows the release of sequestered
CH
Onshore and offshore Arctic permafrost can thaw from the top downward, with
the active layer expanding downward, creating taliks (bodies of thawed
permafrost). They also degrade bottom-up from geothermal heat flux, thawing
frozen sediments from below (Osterkamp, 2010; Shakhova and Semiletov, 2009).
The latter only has a significant effect for submerged offshore permafrost
(Romanovskii et al., 2005). Recent observations of offshore permafrost
(Shakhova et al., 2014) show that the ESAS permafrost lid is perforated, with
year-round CH
There are important geologic controls on the subsea permafrost's thermal state. On millennia timescales, increasing temperatures of the overlying bottom seawater affect subsea permafrost through heat transfer and salinization (Shakhova et al., 2014, 2015; Soloviev et al., 1987). Geologic control also arises from heat transport by large Siberian rivers, which drives bottom water warming and is proposed to control the distribution of open taliks in coastal ESAS waters (Shakhova et al., 2014). Global warming enhances terrestrial riverine heat including from ecosystem responses, degradation of terrestrial permafrost, and increased river runoff. Warm riverine runoff drives a downward heat flux to shelf sediments and subsea permafrost (Shakhova and Semiletov, 2007; Shakhova et al., 2014).
Heat flow in rift zones also provides geologic control (Drachev et al., 2003; Nicolsky et al., 2012). High heat flow areas include relic thaw lakes and river valleys that were submerged during the Holocene inundation. These still drive modern permafrost degradation (Nicolsky and Shakhova, 2010; Nicolsky et al., 2012; Shakhova and Semiletov, 2009).
Subsea permafrost degradation is greatest in the outer shelf waters (deeper than 50 m) where submergence occurred first, such as the outer Laptev Sea where models predict mostly degraded permafrost (Bauch et al., 2001). Riverine input to the Laptev Sea also supports the formation and growth of subsea thaw lakes and taliks, which are effective gas migration pathways to the seabed (Hölemann et al., 2011; Nicolsky and Shakhova, 2010; Shakhova and Semiletov, 2007; Shakhova et al., 2014).
Active seafloor spreading in the Laptev Sea, which is undergoing continental
rifting, leads to strong geologic heat flow (85–117 m W m
Marine seepage is a global phenomenon where CH
The fate of dissolved seep
Bubble size is important with most seep bubbles in a narrow range. Based on a
review of 39 bubble plume size distributions (the most comprehensive
published dataset to date), Leifer (2010) found that the vast majority of
reported seep bubble plumes could be classified in two categories termed
major and minor, with minor being the most common – see also the studies
reviewed in Leifer (2010). Bubble plume size distributions (
Seep
Sonar interpretation is highly challenging, even for qualitative assessments of relative emission strength. For SBESs, there is geometric uncertainty (Leifer et al., 2010) and the plume's angular location is unknown; this problem is resolved by MBESs. Additionally, sonar (SBES or MBES) loses fidelity from multiple plumes in close proximity (Schneider von Deimling et al., 2011; Wilson et al., 2015) where the sonar returns along multiple pathways, creating ghosts, shadow noise, off-beam returns, scattering loss, and other artifacts (Wilson et al., 2015). Note that if bubble spatial densities are sufficiently high for artifacts to occur between plumes, then the bubbles inside the plumes will produce artifacts inside the plumes. The vessel acoustic environment can be noisy and signal loss from scattering can also occur from suspended sediment and biota, often in layers.
There are many challenges to the quantitative derivation of bubble emission
flux from sonar return, which at its basis relates to the interaction of
sound with a bubble. For a single spherical bubble, the relationship has long
been known with resonance given by the Minnaert (1933) equation:
However, bubbles are often clustered in close proximity in seep bubble
plumes, which allows multiple scattering between bubbles that decreases
This study reports on the use of in situ engineered plumes for
calibration of
A precursor study was conducted in the COP seep field (Fig. 1) prior to the
Arctic field experiment to demonstrate time-resolved, 3-D seep
monitoring by a scanning MBES (Fig. S1 in the Supplement). The
rotator–lander was deployed
Field data were obtained during an expedition onboard the R/V
Map for the R/V
The calibration experiments were conducted in a region of no natural seepage
and almost flat seafloor in the
Kara Sea (Fig. 3) to reduce or eliminate off-beam acoustic seabed
scattering. Water depth was 45 m and weather was favorable: calm sea with a
wind speed of 1–3 m s
Locations of oceanographic stations for the RV
The vessel was anchored during the engineered bubble plume experiments.
Engineered bubble plumes were made from nitrogen supplied by a pressure tank
on the vessel foredeck. A 70 m long, 12 mm diameter, 6 mm wall thickness
gas supply tube was attached by a
Kevlar rope to a heavy metal weight (
Gas flow was controlled using standard flow meters. One port was connected to
a PVC tube and a second port was connected to a two-way valve. The third port
was connected to the gas tank through the gas manifold. The manifold
consisted of a high-pressure sensor for the tank pressure and a low-pressure
sensor for the outgoing pressure (5.5 bar). We used temperature-compensated
differential-pressure sensors with a manufacturer-specified range of
The same MBES was used in the ESAS and COP seep field. The SBES was a SIMRAD
EK15 SW 1.0.0 echosounder (
Bubbles have high density contrast with water and thus are strong sonar
targets that can be distinguished easily from the background (Fig. 4b). For
the engineered bubble plume experiments, the wave-mixed layer (WML) extended
to
Sonar data analysis and visualization was performed with custom MATLAB
routines (MathWorks, MA) that first georectified each ping and then assembled
the data for each experimental run into a three-dimensional array of depth
(
A numerical bubble propagation model was used to explore the relative dissolution rates for seep versus calibration bubble plumes and to calculate a volumetric correction factor that accounted for the difference. The bubble model is described elsewhere (Leifer et al., 2006; Leifer and Patro, 2002; Leifer et al., 2015b; Rehder et al., 2009). The model solves the coupled differential equations describing bubble molar content (Eq. 3), size (Eq. 4), pressure, and rise for each bubble size class in a bubble plume. These equations describe how sonar observations of bubble volume (size) relate to bubble mass (molar content).
Bubble dissolution or gas flux (
Unfortunately, bubble size distributions were not measured, and thus a typical minor bubble size distribution from the literature was used. Implications of these simplifying assumptions are discussed in Sect. 4.4.
The model was initialized with a typical (Leifer, 2010) minor
Sonar return for the two (high and low) calibration plumes (Fig. S2) was
thresholded above background (bubble-free water) and integrated for each beam
during rotation across each calibration plume. The thresholded
Field sonar data from the Coal Oil Point seep field for air bubbles
in 22 m deep water. Sonar return counts integrated across the plume (
There is significant geometric uncertainty in SBES data, which is evident in the overlap in time of sonar returns for the calibration bubble plume (Fig. S4). This overlap results from current advection of the plume orthogonal to the page. MBES addresses this SBES deficiency. For example, the SBES loses the bubble plumes once they have risen into the WML, where currents often shift, but the MBES continues to observe them to 13 m depth, slightly below the vessel's draft.
The most common sonar return ping element is noise, which was isolated from
the bubble plume signal by setting a threshold from the sonar return
probability distribution function (
Plume-integrated sonar return (
For the engineered bubble plume experiments, plumes with volume flux (
Sonar return (
For low- versus high-flow plumes,
Sonar return (
These are point source plumes that disperse as they rise, thus bubble–bubble
multiple scattering should decrease with height. With the exception of the
strongest plume, plume rise decreases
The depth-dependent calibration curves (
These in situ
These differences cause different bubble plume evolution and thus different volume height profiles. A volumetric correction factor was developed based on the ratio of the volume height profiles between a calibration and a seep bubble plume (same bubble size distribution) based on numerical bubble propagation model simulations (Fig. 9).
The numerical simulations show that for the first three 5 m depth windows,
the depth-averaged total bubble plume volume (
The size distribution of a minor seep bubble plume changes dramatically as it
rises from a 70 m depth, with the smallest bubbles dissolving and the
largest bubbles growing (Fig. 9d). Overall, air uptake and decreasing
hydrostatic pressure largely balance dissolution for the first 50 m of
bubble rise and
Combining the volumes from the two simulations provides the volume correction
factors, 0.948, 0.868, and 0.775 for the 65–70, 60–65, and 55–60 m depth
windows, respectively. Thus, the calibration plume
Seep mass flux (
The calibration function (
Seep area 2 was stronger than the other seep areas by an order of magnitude and clearly showed a northeast–southwest trend, which is apparent in all seep areas. Some of the striation patterns, primarily of the weaker returns, are consistent with the very strong currents detraining small bubbles out of the plume in the direction of the sonar beam fan. On a second, east–west leg, Seep area 1 was surveyed with currents not aligned with the sonar beam fan and does not exhibit striation. Further evidence of the effect of currents is shown in the sonar ping data (Fig. 10b vs. 10c and d), where Seep area 1 does not show the extreme tilt across beams as sonar data for Seep areas 2 and 3. Thus, the linear seep trends reflect geological control.
Seepage spatial structure showed numerous seeps clustered around the
strongest seep with an apparent modulation at distances of
Seep mass flux (
The mass flux (
Seep area 2 exhibits both greater fluxes and a shallower power law than other
seep areas (Fig. 12c). Furthermore, all seep areas exhibited positive
anomalies or peaks in
Integrated depth-windowed methane flux estimates.
Fit parameters for the seep area flux probability distribution function.
Fit from
Total flux in each seep area was determined by area integration and was 5.56,
42.73, and 4.88 mmol s
We presented results of an in situ engineered bubble plume experiment to
investigate the evolution of bubble plume sonar return for flows spanning two
orders of magnitude. This range was comparable from typical low-flow minor plumes
to very strong high-flow major plumes (Leifer, 2010).
Calibration plume sonar return increased strongly and nonlinearly with flux,
As a high-flow bubble plume rises, the weak
As low-flow calibration plumes rise and disperse,
The artifact striations in the natural seep sonar data from currents are
consistent with a non-negligible bubble–bubble acoustic interaction
(Fig. 11). Specifically, seep bubble plumes were imaged for high currents
that advected small bubbles out of the plumes into the down-current water.
When detrained bubbles were in the beam fan orientation, they were observed,
but not when the beam fan was perpendicular to the currents. For
co-orientation, scattered acoustic energy interacts with nearby down-current
bubbles, which remain in the beam. This arises because the cross-track beam
dimension is very broad (120
Bubble size distributions have been reported for other ESAS seep sites (Shakhova et al., 2015), but the equipment to make bubble measurements was unavailable for this study. Bubble modeling was used to address the effect of the evolving bubble size distribution with flow in the application of calibration air or nitrogen (preferred for safety reasons over methane) bubble plumes to seep bubble plumes (Fig. 9). Thus, we applied a first approximation using a typical minor bubble plume size distribution. Clearly initializing the model with measured plumes would improve the accuracy of the volume correction factor and hence the sonar-derived flux. Still, the primary goal in our study is to demonstrate with a simple approximation that bubble size evolution matters and should not be neglected.
Although the simulations were conducted to correct between a nitrogen
calibration plume and pure methane seep bubbles, if the seep bubbles
contained other gases at non-trace levels, their outgassing could
significantly impact bubble size evolution. In particular, CO
The MBES and SBES systems were calibrated with the same nitrogen gas bubble
plumes; thus, the two systems should agree in terms of flux observations.
Calibration flows spanned very weak flows (
Field observations showed far better agreement between systems for Seep area 2 than the other seep areas (Table 2). This most likely relates to the greater relative importance of stronger seeps that are well above the noise level relative to the other seep areas. The calibration flows (Fig. 8) showed weaker sonar return for the SBES than for the MBES for the same flow. Geometric uncertainty likely played a role in the SBES negative bias.
The seepage spatial map in the ESAS (Fig. 11) shares similarities with spatial patterns in the COP seep field (Fig. 1). Subsurface geologic structures control the seepage spatial flux distribution by creating the pathways through which seepage migrates to the seabed and ocean; seepage areas must occur where geologic structures allow. In the COP seep field, strong seepage areas are located at intersecting non-compressional faults and fractures (Leifer et al., 2010). Furthermore, these faults and/or fractures themselves are preferred migration pathways that connect subsurface reservoirs to the seabed, with seepage tending to manifest along their trend.
Two spatial trends were manifested in the ESAS seepage map (Fig. 11): one northeast–southwest of individual vents and the second a north–south elongation in Seep area 2. Both trends were aligned with the two weaker seepage areas. Furthermore, the northeast–southwest trend is apparent within Seep area 2. Here, fractures in submerged permafrost could play a similar role to the role of fault intersections in the COP seep field; however, more extensive seep area mapping is needed for validation and/or penetration sonar data that can image near-surface rock strata. On smaller length scales, there is an evident striation pattern in vent locations suggesting a subsurface linear geological control on meter-length scales.
High-flow seepage requires high permeability migration pathways, while low-flow seepage occurs along low permeability migration pathways if the driving pressure between the deeper reservoir and the seabed is constant across the active seepage area (Leifer and Boles, 2005). Thus, the stronger, more numerous, and extensive seepage emissions from Seep area 2 indicate higher subsurface permeability and subsurface connectivity with more numerous migration pathways than the other seep areas (Fig. 11). Seepage connectivity can be envisioned topologically as an inverted branched structure (Leifer and Boles, 2005) where central stronger seepage is surrounded (generally) by weaker seepage (Fig. S7). Given that permeability is inversely related to resistance in the migration pathways, stronger seepage is fed by migration along pathways with lower resistance (higher permeability), while weaker seepage is fed by migration along pathways with stronger resistance (lower permeability). The balance between seepage emissions for different migration pathways with a range of permeability underlies the flux probability distribution function (Fig. 12).
The seepage emissions map demonstrates similar geologic spatio-flux control.
Specifically, weak seepage exhibited a
This power law does not extend to the largest seep fluxes, which manifest as peaks in the flux probability distribution function. Thus, higher-flow plumes could represent normal seabed structure failure (that governs the weak seepage) from stresses and/or talik melting, leading to focused high-flow migration pathways that help define where the seep areas lie.
In the Arctic, subsea permafrost degradation from heating both below (geologic, strongest in faulted zones) and above (riverine inputs and overall Arctic Ocean warming) creates migration pathways that manifest as seep spatio-flux distributions. The presence of active seepage in this region likely relates to these heat flows, with the hotspots likely related to taliks and/or subsea thaw lakes whose locations are controlled by linear geologic structures. In the ESAS, grabens are often linear structures, which often are correlated with paleo-river valleys, and could cause co-aligned fractures controlling seepage along linear trends. The similarity in the emission probability distribution power laws between seep areas indicates that subsurface permeability exhibits similar fractal distribution between the three areas. This argues for a similar formation mechanism, i.e., taliks. In this case, at the intersection of the two linear trends, fluid migration and thus heat flow are likely higher, leading to more rapid talik development providing high permeability migration pathways.
There are enormous carbon stores sequestered in marine-permafrost in the
Arctic, which are of particular concern for release as the warming Arctic
Ocean transfers heat to the atmosphere faster than it is transferred from the
atmosphere to terrestrial permafrost. Migration from this submerged
permafrost reservoir to the ocean has created a vast marine seep field that
lies entirely in shallow waters with emissions contributing directly to the
atmospheric budget (Shakhova et al., 2014). Widespread ESAS seabed bubble
emissions have been documented (Shakhova et al., 2014, 2015) demonstrating
permafrost integrity failure that makes CH
These observations support the hypothesis that the subsea permafrost is a
controlling factor for spatial variability in seabed CH
In this study, bubble plumes spanning an almost two orders of magnitude flow
(0.019 to 1.1 L s
This study featured the novel use of a numerical bubble plume model to correct for different size evolution between calibration gas bubble plumes and seep bubble plumes. Uncertainty arises from the bubble size distribution, which needs to be measured for the calibration and seep bubble plumes at multiple flow rates. Our approach was a simplified first effort with room for improvement.
In this study, we present a methodology for using an in situ
plume calibration approach to derive quantitative sonar methane emissions. We
created in situ engineered bubble plumes from a 40 m depth spanning an
almost two orders of magnitude flow (0.019 to 1.1 L s
The in situ calibration curve was applied to natural seepage from
70 m depth in the Laptev Sea
outer shelf where subsea permafrost is predicted to be degraded in modeling
studies. A correction then was made for the different volume evolution of the
nitrogen calibration plume and the methane seep bubble plume through the use
of a numerical bubble plume model. The model was initialized with a typical
(assumed) minor bubble plume size distribution and suggested
The seepage occurrence probability distribution function was bimodal, with weak seepage well described by a power law. This was interpreted as suggesting primarily small minor bubble plumes. The seepage-mapped spatial patterns suggested subsurface geologic control along linear trends. The analysis showed that a probability distribution could provide insights into geologic control.
The underlying data are proprietary.
The authors declare that they have no conflict of interest.
We thank the crew and personnel of the expedition onboard the research vessel