Volume 47, Issue 18 e2020GL089256
Research Letter
Free Access

Dynamics of Gas Bubbles From a Submarine Hydrocarbon Seep Within the Hydrate Stability Zone

Binbin Wang

Corresponding Author

Binbin Wang

Department of Civil and Environmental Engineering, University of Missouri, Columbia, MO, USA

Correspondence to:

B. Wang,

[email protected]

Search for more papers by this author
Inok Jun

Inok Jun

Zachry Department of Civil and Environmental Engineering, Texas A&M University, College Station, TX, USA

Search for more papers by this author
Scott A. Socolofsky

Scott A. Socolofsky

Zachry Department of Civil and Environmental Engineering, Texas A&M University, College Station, TX, USA

Search for more papers by this author
Steven F. DiMarco

Steven F. DiMarco

Department of Oceanography, Texas A&M University, College Station, TX, USA

Geochemical and Environmental Research Group, Texas A&M University, College Station, TX, USA

Department of Ocean Engineering, Texas A&M University, College Station, TX, USA

Search for more papers by this author
John D. Kessler

John D. Kessler

Department of Earth and Environmental Sciences, University of Rochester, Rochester, NY, USA

Search for more papers by this author
First published: 30 August 2020
Citations: 15

Abstract

We validate a new model for mass transfer and bubble transport for natural seeps on the continental margins using an integrated observation of a seep at 883 m in the Northern Gulf of Mexico. In the model, mass transfer is assumed to transition from clean to dirty bubble mass transfer rates following a characteristic hydrate formation time that depends on the initial bubble surface area and the hydrate subcooling. We show that buoyancy-induced upwelling is negligible for the bubble stream. We initialize the model using precise data observed at the seafloor and validate the model results to optical and acoustic measurements within the bubble stream. The model accurately predicts the acoustic attenuation of the bubble flare by lateral spreading and bubble shrinkage throughout the water column. The model shows that up to 99.4% by mass of the released gases from this seep dissolves into the ocean within the hydrate stability zone.

Key Points

  • Hydrocarbon dissolution at seeps slows from clean to dirty bubble mass transfer coefficients after a characteristic hydrate formation time
  • Seep-induced upwelling at natural seeps can be negligible, evidenced by both measurement and modeling
  • The model predicts nearly full dissolution of released hydrocarbons within the hydrate stability zone for a seep originating at 883 m depth

Plain Language Summary

Natural bubble seepage in the deep ocean is increasingly observed by shipboard surveys using sonars. These acoustic signals of these bubble flares are observed high into the ocean water column, raising the question of what the vertical distribution of natural seep gases, including methane, originating at natural seeps may be in the oceans. Present numerical models of natural seeps cannot consistently predict the observed rise heights of seep bubbles. Here, we validate a new model for natural seep bubbles to an integrated in situ and multibeam survey of a seep site in the Northern Gulf of Mexico. Our model assumes that bubbles initially dissolve following convective mass transfer until a hydrate coating slows the mass transfer to rates consistent with solid particles. The characteristic transition time for hydrate formation depends on the initial bubble surface area and a characteristic hydrate formation rate. The model shows that the bulk of the released gases (up to 99.4% by mass) dissolves in the deep ocean, within the hydrate stability zone, and that the model may justifiably neglect other aspects of gas hydrates or buoyancy effects of the rising bubbles. This work is important to understand the role of natural seeps in the oceanic biogeochemical cycling.

1 Introduction

Submarine hydrocarbon seeps are ubiquitous on the continental margins and are important to the global methane inventory (Brothers et al., 2013; Skarke et al., 2014; Talukder et al., 2013; von Deimling et al., 2010; Westbrook et al., 2009). Gas bubbles emitted at natural seeps are commonly observed in data from shipboard sonars, appearing as flares or columns of bubbles extending high into the water column, and raise the question of how natural gas bubbles persist over sometimes more than 1,000 m of vertical ascent (Brewer et al., 2002; McGinnis et al., 2006; Rehder et al., 2002; Römer et al., 2012). In many cases, seep sites originate within the hydrate stability zone (HSZ) of the released gases, and it is expected that hydrate shells forming on bubbles affect their rates of dissolution (Rehder et al., 2009; Warzinski et al., 2014). An upwelling velocity associated with the buoyancy of the bubble column may also help preserve bubbles over great distances (Dsouza et al., 2016; Leifer & Patro, 2002). While focused laboratory (Warzinski et al., 2014) and field investigations (Rehder et al., 20022009) have sought to quantify these processes, numerical simulations based on current understanding of dissolution and transport dynamics fail to consistently predict the observed flare heights of natural seeps (McGinnis et al., 2006; Römer et al., 2012). Because bubble persistence depends on bubble shrinkage rate and rise velocity, the failure of existing models must stem from errors in the mass transfer rates of bubbles, their ascent speeds, or both. Thus, there is a fundamental need to develop reliable models for mass transfer and rise speed of natural gas bubbles in the deep ocean to predict bubble flare dynamics from natural seeps and to evaluate the distribution of dissolved gases released on the continental margins and entering the ocean water column.

Previous observations of bubble shrinkage rate in the oceans and within the HSZ have reported a range of values. Rehder et al. (2002) followed by Rehder et al. (2009) released pure methane bubbles at various depths and found that shrinkage rates for these bubbles immediately after release were consistent with clean bubbles having internal circulation, followed by slower mass loss at later times, consistent with dirty bubbles having immobilized bubble-water interfaces. McGinnis et al. (2006) correlated the transition from clean to dirty bubble behavior with the bubble diameter using the (Rehder et al., 2002) data set and found 3.5 mm as the threshold of clean-to-dirty transition. This criterion, however, does not appear to be universal in the Rehder et al. (2009) data set, and the model developed by Rehder et al. (2009) relies on the measured value of the transition point to explain their data. Römer et al. (2012) applied the McGinnis et al. (2006) model to echosounder surveys of seep flares offshore Pakistan. They found that model simulations for flare height were strongly dependent on the initial bubble size and that the model performance was inconsistent among seep sources originating at different depths. Because the literature lacks an accurate understanding of how natural gas bubbles within the HSZ transition from clean to dirty mass transfer rates, dissolution rates for natural seep bubbles cannot be accurately predicted.

The rise velocities of bubbles in natural seep flares are also uncertain. In buoyancy-driven, single-phase plumes, the collective buoyancy of the release provides the driving force, and the entrainment of ambient fluid results in the plume width increasing linearly with height above the source (Ditmars & Cederwall, 1974; Lee & Chu, 2003; Socolofsky et al., 2008). Bubble plumes can also behave as buoyant plumes, with the effect of the entrained fluid adding a vertical upwelling velocity to the individual bubble slip velocity. This entrainment theory has been widely used to predict deepwater oil spill plumes (Dissanayake et al., 2018; Johansen, 2000; Yapa et al., 1999). However, for submarine hydrocarbon seeps, with much lower gas fluxes and void fraction, bubbles may behave as isolated particles shortly after they leave the source (Wang et al., 2016). In this case, there is minimal or near-zero upwelling (Leitch & Baines, 1989; Wang et al., 2019). Whether or not to include a buoyancy-induced upwelling velocity in calculations of bubbles rising from natural seeps remains an open question and is critical to the fate of natural gas bubbles in the ocean water column.

Much of the uncertainty in determining field-scale dissolution rates and rise velocities for natural seep bubbles stems from the difficulty of measuring the full-scale seep flare and the bubble-scale shrinkage rates simultaneously. To help understand these processes, we carried out a comprehensive field campaign at a natural seep site in the Northern Gulf of Mexico in 883 m water depth (nearly 500 m beneath the hydrate stability line; see later Figure 3). Measurements at the seafloor included gas composition, bubble size distribution, and source discharge observed by a high-speed, stereoscopic camera system mounted on a remotely operated vehicle (ROV). Measurements within the water column included discrete measurements of bubble sizes using the stereo camera system and observations of the acoustic anomaly throughout the bubble streams using two different multibeam sonars, one deployed in situ from the ROV and another haul-mounted on an exploration vessel. We utilize these data to validate a numerical model of the seep bubbles and their acoustic properties. The model uses clean and dirty bubble mass transfer rates coupled by a transition time that depends on the initial surface area of the bubbles and the hydrate subcooling. We further show from the field data that a buoyant upwelling velocity is absent from typical natural seeps. By validating our seep bubble model to these detailed observations, we elucidate the dominant mechanisms controlling the fate of bubbles released from natural seeps.

2 Methods

2.1 Field Site and Observations

Here, we analyze observations collected during the Gulf Integrated Spill Research (GISR) G08 cruise from 9–21 April 2015 in the Northern Gulf of Mexico on the E/V Nautilus (Nautilus cruise NA056) and using the ROV Hercules. Our main field site was in Mississippi Canyon, lease block 118 (MC 118), where active seeps are the result of a crustal fault network near a carbonate-hydrate mound complex called Woolsey Mound (located at 28.85216903°N, 88.49182788°W, and 883 m depth; see Figure S1 in the supporting information and Macelloni et al., 2016).

Several in situ observations were made from the ROV. To find and track the bubble stream, we used a forward looking M3 multibeam sonar (500 kHz, Kongsberg Socolofsky, 2015a). These data provided a profile of the acoustic cross section of the bubble flare, which we followed from the seafloor to a depth of 442 m, where the acoustic signal diminished to background levels. Qualitative properties of the seep source and bubble stream in the water column were documented using the ROV's high definition (HD) camera (Socolofsky, 2015b). Quantitative measurements of the gas bubble size distribution, rise velocity, and flow rate at the seafloor were made using a stereoscopic, high-speed camera system (Socolofsky, 2015c), following the same procedure as described in Wang and Socolofsky (2015) and Wang et al. (2016). Because the bubble stream is spread out by the ocean currents, stereo camera observations were also made at altitudes of 50, 90, 200, 300, and 400 m above the source. The composition of the seep bubbles was obtained from gas samples collected at the seafloor and later analyzed at the Woods Hole Isotope Laboratories (Leonte et al., 2018).

Other measurements were made from the E/V Nautilus (Socolofsky, 2015a) and nearby observation platforms. We used a haul-mounted EM 302 multibeam sonar (30 kHz, Kongsberg) to survey the local bathymetry and the acoustic anomaly of the seep flares in the water column. Temperature profiles were obtained from expendable bathythermograph (XBT) casts made to support the EM 302 data processing; temperature and salinity were also collected by a SeaBird SBE 49 FastCat mounted on the ROV. We obtained ocean currents from an acoustic Doppler current profiler mounted on a drill ship located 1.98 km from the seep source and available from the National Data Buoy Center (NDBC, Station ID 42883).

2.2 Lagrangian Particle Model for Natural Seep Bubbles

We use the single bubble model (SBM) module of the Texas A&M Oil spill/Outfall Calculator (TAMOC Dissanayake et al., 2018) to simulate the dynamic behavior of seep bubbles rising through the ocean water column. TAMOC uses real-fluid equations of state for complex hydrocarbon mixtures to compute thermodynamic and chemical properties (Gros et al., 2016, 2017) and employs standard empirical correlations for mass transfer coefficients and terminal velocity of bubbles (Clift et al., 1978). The SBM solves for the trajectory of a dissolving bubble in a Lagrangian reference frame, forced by the measured ambient currents and by a random-walk model for turbulent diffusion. Boundary conditions, including local salinity, temperature, and pressure, are imposed from the measured data (Wang & Socolofsky, 2018).

For the present simulations, we employ a new criterion for selecting between clean and dirty mass transfer coefficients for these natural seep bubbles. This kinetics-based model assumes the rate of hydrate growth (area per time) scales with the hydrate subcooling ΔT at the release depth. Then, the time required to transition a bubble of initial spherical diameter de from clean to dirty behavior is
urn:x-wiley:grl:media:grl61161:grl61161-math-0001(1)
where α and β are empirical coefficients. Clean bubble mass transfer coefficients are assumed to apply from the release until the time ttrans, after which dirty bubble mass transfer coefficients are used. Jun (2018) calibrated α and β to the data in Rehder et al. (2009) and obtained the values urn:x-wiley:grl:media:grl61161:grl61161-math-0002 m2/(sK1/3) and urn:x-wiley:grl:media:grl61161:grl61161-math-0003.

2.3 Calculation of Acoustic Response of Modeled Seep Bubbles

We use the seep model results for bubble size, location, and flow rate for 1,000 bubbles released at the seafloor to compute the acoustic response of these bubbles in the water column so that these data can be compared to the multibeam sonar data from the M3 on the ROV and EM 302 on the vessel. To compute an acoustic response from the model that is comparable to the sonar measurement, we calculate the total backscatter contributed by all bubbles within the sample volume of each sonar to obtain a target strength (Medwin & Clay, 1997; Weber et al., 2014; Weidner et al., 2019). Details for each of these calculations are provided in Text S1.

3 Results and Discussion

3.1 Seep Flare Initial Conditions Observed at the Sea Floor

During the GISR G08 cruise, the seep source at Woolsey Mound was quite dynamic, with individual streams of bubbles occurring at multiple locations that moved around within a seafloor region of approximately 10 m2 (see Movies S1–S8). The average bubble size distribution in volume at the seafloor fits well to a log-normal distribution with mean of 1.63 mm and standard deviation of 0.35 mm (Figure S2). Similarly, bubble rise velocities measured by the camera system agree with empirical correlations in Clift et al. (1978) (Figure S3). Based on the in situ gas samples, methane was the main gas component in the bubbles (68.7 mol%), followed by 19.21 mol% nitrogen, 2.46 mol% ethane, and the remainder other gases (see Table S1 and Leonte et al., 2018). We estimated the overall gas discharge rate during our surveys by combining detailed measurements from the stereoscopic cameras with full-field observations from the HD cameras (Text S2 and Table S2). From these data, we derive an average emission rate of 1.0 ± 0.2 L/min of gas at in situ conditions of temperature and pressure, or equivalently, (2.0 ± 0.4) · 10−3 kg/s natural gas. Previous studies confirmed that hydrocarbon emissions are thermogenic at Woolsey Mound (Lutken et al., 2011); hence, oil might be expected to be present and affect the mass transfer rates from bubbles (Leifer & MacDonald, 2003). During our experiments, some oil droplets were observed in the ROV videos when the sediment was disturbed. However, for the quantitative imaging, the gas bubbles appeared free of oil coatings or attached droplets.

3.2 Evolution of Bubble Size Distribution Through the Water Column

As the seep bubbles rise through the water column, they lose mass by dissolution, are advected by ambient currents, spread out by turbulent diffusion, and experience continuous decompression. As a result, the bubble size distribution is continuously evolving with height. Figures 1a1c show the volume size distributions of the modeled bubbles at three heights (see Movie S9 for an animation of the size distribution following the trajectory of the flare). The total volume of gas at each cross section (area under the histograms in Figure 1) reduces with height. This occurs because bubbles shrink by dissolution faster than they expand by decompression and some bubbles become small enough that their rise velocity becomes negligible, removing them from the flare. These effects are also illustrated in Movie S10, which animates the trajectories of 200 simulated bubbles released from the seep. These factors (dissolution, decompression, and sorting by the currents) also act to narrow the bubble size distribution and gradually reduce the median bubble size d50.

Details are in the caption following the image
(a–c) Evolution of the bubble size distribution at three different altitudes in the water column z = 0, 200, and 400 m; the entire evolution of the bubble size distribution can be found in Movie S9. (d) Evolution of the statistics of the volume size distribution of the bubbles—equivalent spherical diameter at the median value and the standard deviation in the volume CDF, plotted with height for the measured and modeled data; see Figure S4 for full distributions.

Figure 1d compares the statistics of the modeled volume size distribution with those of observed by the stereoscopic cameras. To sample the model results in a similar way to the camera data, we analyze a subsample of all modeled bubbles passing through a disc of 2 m diameter, centered on the strongest modeled acoustic return, that is, using the same sampling strategy and volume as was used in the field. Because the number of bubbles in the bubble flare decreases with height (see Figures 1a1c), some of the field samples at the higher altitudes may not have converged statistics. Overall, the model-measurement comparison in Figure 1d shows good agreement, with a root mean square (RMS) error for the median bubble size of 0.76 mm (i.e., about 20% of the average value of 4 mm), with a bias of 0.38 mm, and an RMS error in the standard deviation of 0.22 mm (see also Figure S4).

3.3 Lateral Spreading of Bubbles in the Seep Flare

From the M3 acoustic images of the bubble flare cross section, the bubble stream is elongated in the dominate direction of the currents and also spreads laterally relative to the currents (e.g., Figure S5, left column). There are two potential explanations for the lateral spreading. First, if the gas flow rate is large enough, a buoyant plume may develop, and the plume would spread by lateral entrainment, following a linear growth rate of flare half width of 0.1 m/m (Fischer et al., 1979). This process is critical to the fate of bubbles since a plume would also generate an upwelling velocity. Second, if the gas flow rate is sufficiently low and a coherent plume does not form, lateral turbulent diffusion may be responsible for the bubble flare growth, resulting in spread of the half width proportional to z1/2.

We analyzed the lateral spreading by computing the half width of the bubble flare at different heights from the M3 data (Figure 2). In Figure 2a, we show the measured data for ROV Dive H1402 together with linear spreading for a coherent, buoyant plume (dashed line) and results of our seep model with constant lateral turbulent diffusivities (solid lines). This plot clearly demonstrates that the bubble flare does not follow the plume spreading rate. This conclusion is also consistent with Wang et al. (2019), who show through laboratory experiments that natural seep plumes normally lose their coherent plume behavior starting from z/D > 5, where
urn:x-wiley:grl:media:grl61161:grl61161-math-0004(2)
Details are in the caption following the image
Lateral diffusivity of bubble flare spreading rate: (a) comparison between the measured bubble flare half width during Dive H1402 and model results with different lateral diffusion coefficients ( urn:x-wiley:grl:media:grl61161:grl61161-math-0005 m2/s; the linear spreading rate for a coherent bubble plume is also shown as the dashed black line); (b) comparison between measured and modeled results for all dives with urn:x-wiley:grl:media:grl61161:grl61161-math-0006 m2/s.

B is the buoyancy flux of the plume ((ρ − ρb)gQb/ρ), urn:x-wiley:grl:media:grl61161:grl61161-math-0007 is the pure plume entrainment coefficient, and us is the terminal slip velocity of the bubbles; ρ is the ambient water density, ρb is the in situ density of the bubbles, g is the acceleration of gravity, and Qb is the gas flow rate. Using the properties of this seep, urn:x-wiley:grl:media:grl61161:grl61161-math-0008 m, and buoyant plume behavior is not expected above 1 m of ascent.

In contrast, the seep model, using random-walk diffusion, tracks the half-width growth well. We use the seep model to select the best fit turbulent diffusivity for three main reasons. First, the seep flare contains bubbles of a broad size distribution, each with different rise velocities, so that the age of each bubble is different at each measurement height. Second, the differences in bubble rise velocities together with the lateral currents result in smaller bubbles being advected further downstream than larger bubbles in a process called fractionation (Socolofsky & Adams, 2002). Third, the ambient currents change magnitude and direction with height, resulting in a differing level of fractionation in all directions. Hence, the distribution of flare bubbles at any height results from these processes together with random-walk diffusion.

We calibrate the seep model, which naturally accounts for differing rise velocities and fractionation through Lagrangian advection within the measured currents, for Dive H1402, yielding the best fit value of urn:x-wiley:grl:media:grl61161:grl61161-math-0009 m2/s (shown as the solid black line in Figure 2a). This value is in agreement with expected turbulent diffusivities in the open ocean (e.g., Fischer et al., 1979; Okubo, 1971; see Figure S6). Using this value of Ka, Figure 2b shows the half width computed by the seep model compared to the measured data for all ROV dives. The agreement between the model and data is good, with an overall RMS error in modeled flare half width of 0.57 m (i.e., about 20% of the average value of 2.5 m). This finding is significant because it confirms that there is no plume-generated upwelling velocity in seep flares with these seepage rates and that the vertical bubble transport rate, hence the time-scale available for dissolution, is given by the terminal velocity of each individual bubble.

3.4 Acoustic Response of the Observed and Modeled Seep Bubbles

We further test the seep model's ability to predict dissolution by comparing the acoustic response of the simulated bubbles to that observed by the M3 and EM 302. By using the measured ambient currents and calibrated turbulent diffusivity, the spatial pattern of the modeled bubbles agrees well with the M3 observations (see Figure S5). The model prediction of target strength integrated over the cross section of the bubble flare as a function of height also agrees well with the calculated values from the M3 intensity data after an adjustment of 22.5 dB to offset the uncalibrated sonar intensity (see Figure S7). Because the target strength is a function of bubble void fraction and size, these data confirm that the model captures the evolution of these bubble parameters well. The reduction of target strength with height predicted by the model (0.031 dB/m) agrees within 50% of the measured values (0.022 and 0.021 dB/m) for ROV Dives H1402 and H1403.

For the EM 302 data, the water column anomaly captures an instantaneous snapshot of the whole bubble flare in the ocean water column (Figure 3). To identify the bubble flare, we follow Weber et al. (2014) to distinguish the bubbles, using a threshold value for the acoustic background level above the average value at each depth. Figure 3a shows the EM 302 water column anomalies during the survey following ROV Dive H1403; Figure 3b shows the integrated target strength from the EM 302 and simulated by the seep model. The computed target strength from the seep model is adjusted to match the uncalibrated backscatter intensity of the EM 302 at the seafloor, resulting in a shift of 13 dB for the model target strength in Figure 3b (Weber et al., 2014). See Figures S8 and S9 for model-measurement comparisons following Dives H1404 and H1407. With this adjustment, the trend in the modeled acoustic response matches the measured data, with the simulation results plotting within the measurement error for nearly all depth bins.

Details are in the caption following the image
Comparison between the measured and modeled acoustic data from the EM 302 multibeam sonar: (a) acoustic anomaly in the water column measured by the EM 302 after Dive H1403 (the color bar shows the backscatter intensity, dB); (b) comparison of the measured backscatter intensity from the EM 302 (blue circles) and the modeled target strength (red line, offset by 13 dB); the solid black line reports the acoustic background used as a threshold to identify the EM 302 water column anomalies of the bubble flare; (c) the measured acoustic anomalies in the water column by the EM 302 in the context of the surrounding bathymetry; the measured temperature profiles and the calculated hydrate stability curve are superposed in subplot (c), showing the Hydrate Stability Zone (HSZ) from the seafloor to 385 m water depth. (d) The percentage of remaining mass of released natural gas and methane as a function of water depth computed by the model. The horizontal dashed line shows the hydrate stability line, where the remaining mass of total gas is 0.6% and the remaining mass of methane is 0.1%.

Over the height of the seep flare, significant changes in the seep bubbles and acoustic response occur. Lower in the water column, the bubbles are larger, above their threshold sizes for resonance in the 30 kHz acoustic beam of the EM 302, and they remain continuously observable in the EM 302 data to a depth of about 400 m, similar to the maximum observable height from the M3 (442 m depth). Above this depth, there are fewer bubbles in the flare, and they successively shrink below their resonant size (yielding a spike in the modeled data as they pass through resonance) and then rapidly become acoustically transparent as they continue to shrink. In this region, the flare bubbles occasionally exceed the background threshold value, yielding EM 302 observations in agreement with the model as high as 680 m above the seafloor (200 m depth) in Figure 3. At the same time, the target strength decreases by nearly 15 dB, equivalent to a 30 times reduction in signal power, as a result of the bubble dissolution and spreading.

Figure 3c shows all of the EM 302 observations of the Woolsey Mound bubble flare overlaid with the measured bathymetry; the insert also shows the measured ocean temperature profile and a hydrate stability curve for the measured gas composition. The depth of the HSZ for this gas composition and ocean temperature profile is 385 m, 57 m above the highest acoustic observations from the M3 during the G08 cruise and above the continuous observations of the EM 302 (see data gaps in subplot b). Moreover, the seep model predicts less than 0.6% by mass of the released gases and 0.1% by mass of the released methane to reach a depth of 385 m, carried there by the largest bubbles released from the sea floor (Figure 3d). Hence, though it would appear in Figure 3c that the hydrate stability line may be a causation of the bubble flare disappearance, this careful survey and model validation confirm that nearly full absorption of the released seep gases by the ocean water column occurs below this level. Moreover, only a few of the larger bubbles released from the seafloor reach the top of the HSZ, the seep flare remains intermittently observable in the acoustic data above the HSZ, and dissociation of hydrate above the HSZ is not required in the seep model to explain the acoustic observations throughout the water column.

4 Conclusions

In this letter, we compared results of a natural seep model to in situ and acoustic observations of a seep flare in the Northern Gulf of Mexico. The model uses measured boundary and initial conditions and accounts for the effects of gas hydrates on dissolution using dirty bubble mass transfer coefficients that replace clean bubble mass transfer coefficients after a characteristic transition time that depends on the initial bubble surface area and the hydrate subcooling. The model was validated with water column data for the evolving bubble size distribution, the lateral advection and spreading of the seep bubbles, and the acoustic properties of the seep flare. Based on the plume spreading, this study confirms that there is a negligible upwelling velocity resulting from the bubble flare so that the rise time of bubbles is given by their terminal velocity.

The validated model further predicts that 99.4% of the released natural gases by mass and 99.9% of the released methane are absorbed through dissolution by the ocean water column below the HSZ at this site and that no dissociation of hydrate near the HSZ is needed in the model to account for the acoustic properties of the bubble flare throughout the water column. The vertical distribution of dissolved methane quantified in this study may inform the location of the hydrocarbon source for microbial consumption and associated biogeochemical processes. The physical processes of mass transfer and bubble kinematics revealed by this study are expected to be universal, though the actual scales of transport will vary for different seep initial conditions from different source depths. Hence, these results are critical for interpreting observations of natural seep flares in acoustic data on the continental margins and for estimating the vertical distribution in the ocean water column of gases originating from natural seeps.

Acknowledgments

This research was made possible in part by a grant from the Gulf of Mexico Research Initiative to the Gulf Integrated Spill Research (GISR) consortium, the Center for the Integrated Modeling and Analysis of the Gulf Ecosystem (C-IMAGE-II and C-IMAGE-III), and the project Synthesis of the Physical Processes in Subsea Bubble Plume to Connect Natural Seeps and Oil Spills (Syn-bubbles). This work was also supported in part by the U.S. Department of Energy, Methane Hydrates Program, under Award Number DE-FE0028895.

    Data Availability Statement

    Data are publicly available through the Gulf of Mexico Research Initiative Information and Data Cooperative (GRIIDC) (at doi: 10.7266/N79C6VB3, doi: 10.7266/N7GF0RF5, doi: 10.7266/N78P5XKG, and doi: 10.7266/N72Z145Z Socolofsky, 2015a2015b2015c; Wang & Socolofsky, 2015b, 2018).