Geochemical and geophysical evidence of methane release over the East Siberian Arctic Shelf



[1] The East Siberian Arctic Shelf (ESAS), which includes the Laptev Sea, the East Siberian Sea, and the Russian part of the Chukchi Sea, has not been considered to be a methane (CH4) source to hydrosphere or atmosphere because subsea permafrost, which underlies most of the ESAS, was believed, first, not to be conducive to methanogenesis and, second, to act as an impermeable lid, preventing CH4 escape through the seabed. Here recent observational data obtained during summer (2005–2006) and winter (2007) expeditions indicate the ubiquitous presence of elevated dissolved CH4 and an elevated atmospheric CH4 mixing ratio. The CH4 data were also analyzed together with high resolution seismic (HRS) data obtained by means of a “Sonic M-141” system consisting of a high-resolution profiler and side-scan sonar mounted in a towed fish during the Transdrift-X Expedition (2004) onboard the R/V Yakov Smirnitskiy. Results show anomalously high concentrations of dissolved CH4 (up to 5 μM) and an episodically (nongradually) increasing atmospheric mixing ratio of CH4 (up to 8.2 ppm) in some areas of the ESAS. A most likely source is year-round CH4 release through taliks (columns of thawed sediments within permafrost) from seabed CH4 reservoirs such as shallow hydrates and geological sources. Such releases occur not only within the areas underlain by fault zones but also outside of them. This points to permafrost's failure to further preserve CH4 deposits in the ESAS. The total amount of carbon preserved within the ESAS as organic matter and ready to release CH4 from seabed deposits is predicted to be ∼1400 Gt. Release of only a small fraction of this reservoir, which was sealed with impermeable permafrost for thousands of years, would significantly alter the annual CH4 budget and have global implications, because the shallowness of the ESAS allows the majority of CH4 to pass through the water column and escape to the atmosphere.

1. Introduction

1.1. Arctic Overview

[2] The Arctic climate is warming dramatically, threatening catastrophic climate change through rapid mobilization of the vulnerable reservoirs of carbon sequestered by permafrost [IPCC, 2007]. Increasingly, feedbacks in the Arctic are recognized as contributing to climate change, including cycles associated with the powerful greenhouse gas methane CH4, whose atmospheric concentration has more than doubled since the preindustrial epoch [Houghton et al., 2001]. Sustained CO2 and CH4 release to the atmosphere from thawing Arctic permafrost is a positive and likely highly significant feedback to climate warming [Oechel et al., 1993; ACIA, 2005; Zimov et al., 2006]. Other possible feedbacks include destabilization of CH4 hydrate deposits underlying the Arctic seabed, very conservatively estimated to contain 2 × 103 Gt of CH4 [Makogon et al., 2007]. Because of the large uncertainties in Arctic budgets and poor understanding of complex feedback processes, predictions of future emission trends, while critical, are highly unreliable [McGuire et al., 2009].

1.2. Geohistory of the East Siberian Arctic Shelf

[3] The Arctic Ocean comprises ∼1.5% of the global ocean by volume and <5% of its surface area but receives ∼10% of all global river runoff [Aagaard et al., 1981]. The majority (87%) of particulate material delivered by riverine waters and from coastal erosion accumulates in the Arctic continental shelf [Stein and Fahl, 2004; Vetrov and Romankevich, 2004]. Annually, the East Siberian Arctic Shelf (ESAS) accumulates organic carbon equal to that accumulated over the entire pelagic area of the world oceans [Vetrov and Romankevich, 2004]. The sedimentary basins of the ESAS provide favorable conditions for CH4 origination and are predicted to contain a giant natural pool of hydrocarbons stored as oil, natural gas, and CH4 hydrates [Gramberg et al., 1983; Ginsburg and Soloviev, 1994].

[4] The ESAS is the largest (2.1 × 106 km2) and shallowest (mean depth < 50 m) continental shelf among the world oceans [Jakobsson, 2002]. Because of its shallowness, the ESAS has a unique geological history. During cold climate periods, when sea level was more than 100 m lower, the coastline was up to 1000 km further north, exposing the continental shelf; at that time, the entire area of the Siberian coastal accumulative plain was larger than today by a factor of 5 [Romanovskii and Hubberten, 2001; Romanovskii et al., 2005]. Freezing to as deep as hundreds of meters led to permafrost formation (sediments with a 2 year mean subzero temperature), which restricted upward CH4 migration. Trapped CH4 was then transformed into permafrost-associated hydrates [Soloviev et al., 1987; Kvenvolden, 1988]. Replacement of the cold epoch by the current warm Holocene epoch was accompanied by a sea level rise that led to permafrost inundation 5–12 kyr ago [Fleming et al., 1998]. This inundation caused the environmental thermal regime to warm by as much as 12°C; in response, the subsea permafrost started warming to achieve a new quasi-stationary equilibrium [Soloviev et al., 1987; Kvenvolden, 1988]. Numerical models predict destabilization about 5–10 kyr after inundation, depending on the duration of inundation relative to the duration of previous freezing [Soloviev et al., 1987]. However, observational data [Shakhova et al., 2005; Rachold et al., 2006; Shakhova and Semiletov, 2007] suggest that subsea permafrost is prone to more rapid destabilization than has been predicted based on numerical modeling [Romanovskii et al., 2005].

[5] After inundation, terrestrial permafrost-related hydrate deposits became Arctic shallow hydrate deposits. It has been recognized that destabilization of these deposits can lead to a large-scale enhancement of aqueous CH4 [Kvenvolden et al., 1993]. Kvenvolden et al. [1993] conducted seasonal (summer and winter) measurements in the Alaskan Beaufort Sea Shelf over 2 years and found significant CH4 supersaturation in some samples (to 118 nM, n = 250, where n is the number of samples). Although consistent with the destabilization hypothesis, Kvenvolden et al. [1993] estimated that the Arctic shelf contribution to the global CH4 budget was small: ∼0.1 Tg yr−1.

[6] The ESAS comprises ∼25% of the Arctic continental shelf and contains over 80% of existing subsea permafrost and shallow hydrate deposits (Figure 1), which are estimated to contain a total ∼1400 Gt carbon. This reservoir consists of ESAS hydrate deposits estimated to hold ∼540 Gt of CH4 with an additional 2/3 (∼360 Gt) trapped below as free gas [Gramberg et al., 1983; Soloviev et al., 1987]. Because the ESAS is an overlooked sibling to Siberian on-land permafrost that was submerged, terrestrial permafrost estimates are expected to apply to the ESAS. Thus, subsea permafrost is estimated to contain a further 500 Gt within a 25 m thick permafrost layer [Zimov et al., 2006]. Yet despite its reservoir's significance, the role of the ESAS in the modern carbon cycle has received little attention because the ESAS has not been explored until recently [Shakhova et al., 2005; Shakhova and Semiletov, 2007].

Figure 1.

(a) Predicted deposits of Arctic shallow hydrates over the Arctic Ocean (shown in purple, after Soloviev, 2002). Red line marks position of 50 m isobath. (b) Distribution of subsea permafrost over the Arctic Ocean (shown in red, after ACIA, 2005).

[7] In this study, we test the hypothesis that submerged permafrost destabilization is occurring through analysis of data collected from the ESAS in summer 2005 and 2006 and winter 2007 with respect to the ESAS geological history. Elevated aqueous CH4 supersaturation and atmospheric concentrations were found, which were best explained by CH4 release from seabed deposits. Data demonstrate that the CH4 that accumulated over millennia during ages with colder climate under and within largely impermeable subsea permafrost is, to some extent, imperfectly sequestered. On the basis of these findings, we discuss the global climate implications.

2. Regional Setting

2.1. Setting

[8] Several expeditions were conducted to test the hypothesis that migration pathways exist through submerged permafrost, allowing CH4 to escape to the water column and then to the atmosphere. These included annual summer field campaigns (August to September 2005 and 2006) and one over-ice winter expedition (April 2007). We deployed our field laboratories on a small vessel (R/V TB-12, 27 m) and a river boat (Neptun, 14 m) suitable for operation in shallow ESAS waters. Summer 2005 and 2006 cruises surveyed the eastern Laptev and East Siberian seas (Figure 2a). In winter 2007, we studied the Lena Delta adjacent to the Laptev Sea, working on fast ice (Figure 2b).

Figure 2.

Position of oceanographic stations within the study area. (a) Summer cruise (September 2005). Position of transects numbered 1–3 are shown by red dotted arrows, positions of two geophysical surveys are shown in black symbols marked A and B; (b) winter expedition, April 2007.

2.2. Sedimentation

[9] ESAS sedimentation is unique because of the intense supply from riverine runoff and coastal erosion [Vetrov and Romankevich, 2004; Dudarev et al., 2006; Rachold et al., 2006]. As a result, an enormous amount of organic-rich sediments arrive at the shelf and accumulate within its sedimentary basins [Kleiber and Niessen, 2000; Bauch and Kassens, 2005]. The distribution pattern of the surface sediments varies significantly within the ESAS. In areas favorable for accumulation Holocene sediments generally vary from 1 to <5 m, but within limited areas they can be 10 m thick [Kleiber and Niessen, 2000]. In the nearshore, where processes of bottom erosion and sediment resuspension and redistribution prevail, the thickness of the Holocene sediments varies from 0 to 20–25 cm with sands and silt dominating [Vetrov and Romankevich, 2004]. In regions distant from the coast, as well as in areas with monotonously even topography, fine-grained sediments (clayey muds) prevail because near-bottom current velocities are lower [Vetrov and Romankevich, 2004].

[10] Organic carbon (Corg) is the principal component of carbon cycling in bottom sediments [Vetrov and Romankevich, 2004]. Due to limited light, shallowness, and a short trophic web, a large fraction of Corg (∼50% of the initial flux) arrives at the seabed, where 88% is mineralized [Vetrov and Romankevich, 2004]. An estimation of the terrigenous component of organic matter (OM) in the sediments of the Laptev Sea and the western part of the East Siberian Sea showed that the upper sediment layer is up to 100% terrigenous OM [Stein and Fahl, 2004; Semiletov et al., 2005].

2.3. Hydrology

[11] Hydrological parameters of the ESAS reflect influences of winds, Siberian river outflows, ice melt, and eastward flows from the Pacific Ocean [Nikiforov and Shpaikher, 1980]. The western ESAS is strongly influenced by freshwater transport from the Lena River, which is overlapped by runoff from other rivers (Yana, Indigirka, Kolyma). A major plume of riverine freshwater usually transits the East Siberian Sea eastward within the coastal zone [Weingartner and Danielson, 1999]. Hydrological data show that over the shallow western ESAS the water column is strongly stratified; the pycnocline deepens from 3–5 m east of the Lena Delta to 10–20 m in the eastern ESAS [Nikiforov and Shpaikher, 1980]. In the western part of the East Siberian Sea (roughly between 140°E and 160°E) wind-induced mixing extends from the top to the bottom of the water column [Semiletov et al., 2005].

2.4. Geology

[12] The extensive offshore sedimentary basins are believed to have originated initially by rifting and postrift thermal subsidence during the opening of the Arctic oceanic basins. The Laptev Sea rifts are strongly linked to a seafloor spreading axis (Gakkel Ridge) in the Eurasian Basin [Drachev et al., 2003]. The poorly studied rifts of the East Siberian Sea and the northwestern part of the Chukchi Sea probably originated under similar tectonic conditions. The Laptev rift system consists of several deep, subsided rift basins; from the west to the east, there are the West Laptev and South Laptev rift basins, the Ust' Lena (UL) Rift, the Stolbovoi Rift, the East Laptev Horst (ELH) Rift, and the Bel'kov-Svyatoi Nos (BSN) Rift (Figure 3b).

Figure 3.

Distribution of dissolved CH4 in the study area (September 2005). (a) Surface water. Plume areas are shown: black square, east of the Lena Delta; red square, the Dmitry Laptev Strait. (b) Bottom water. Positions of fault zones are shown in black lines (the dashed line shows where position of fault zones is uncertain).

3. Methods

3.1. Water Column CH4 Measurements

[13] Niskin bottle water samples were collected during upcasts at each conductivity/temperature/depth (CTD) station (Figure 2) for 127 stations during the summer cruise 2005 and 57 stations during the winter survey 2007. Because the water depth at most stations was <15 m, water samples were taken from surface and bottom layers only. In deep water, samples also were taken from an intermediate layer. For CH4 measurements, Niskin water samples were immediately poured into replicate 500 mL glass bottles, overfilling 1.5–2 times. Careful subsampling avoided introducing air bubbles. Silicon stoppers sealed with tin lids were used. Gas was extracted from the water with helium by the headspace method [Johnson et al., 1990]. Water samples were placed in a thermostatic water bath shaker and equilibrated for 30 min. Replicate CH4 samples were kept at ambient laboratory temperature and analyzed within a few hours.

[14] All samples were analyzed for CH4 with a MicroTech-8160 gas chromatograph (GC) equipped with a flame ionization detector (FID). The GC oven was operated isothermally at 40°C with helium as the carrier gas. Calibration was performed with a certified 1.96 ppmv CH4 gas standard with the air (Air Liquide, USA). The standard deviation of duplicate analyses (three to five replicates) was <2%. Reproducibility was ∼1% based on multiple standard injections during daily calibrations.

[15] The concentration of dissolved CH4 in the water samples was calculated with the Bunsen solubility coefficient for CH4 [Wiesenburg and Guinasso, 1979] for the appropriate equilibration temperature. Data were analyzed statistically and presented graphically using standard scientific programs such as Statistics 7.0 (StatSoft, Inc., USA), Matlab 7.0 (MathWorks, Inc., USA), and Grapher 6.0 (Golden Software, Inc., USA). Irregularly spaced data (temperature, salinity, aqueous CH4) were interpolated onto uniform grids using a minimum curvature algorithm (Surfer 8.0, Golden Software, Inc., USA), allowing calculation of area-weighted means for topographic plots.

3.2. Atmospheric CH4

[16] Air samples were taken off the vessel's bow at an elevation of ∼3 m through a stationary mounted polypropylene hose, 1 cm diameter and 5 m long. The air inlet was ∼1 cm diameter and 10 m long. Samples were continuously pumped through the GC sample loop and injected every 15 min. Simultaneously, continuous measurements of carbon dioxide (CO2) were conducted using a Li-Cor-820, which was installed in the same location as the CH4 sampler. Atmospheric CO2 concentrations caused by turbulence of the ship's exhaust in response to changes in wind direction (equation image390 μatm) were used to filter CH4 data.

3.3. Sampling CH4 Sea Ice Bubbles

[17] Gas samples were collected from visually detected bubbles in fast ice using a 2 m ice auger with extension to drill through the ice to water in close proximity to the bubble. Then, an underwater hole was made in the side of the borehole with a submerged hand-held steel awl, releasing the bubble to rise 20–30 cm to the sea surface, where it was caught in a hand funnel attached to a 1 cm diameter polyurethane tube. The tubing conveyed the gas to an upside-down, submerged, and water-filled 250 mL sample bottle. After accumulating sufficient gas (equation image5 cm3), the sample bottle was hermetically sealed and analyzed in the field laboratory for CH4, as was done for water samples.

3.4. Hydrological and Other Parameters

[18] Salinity and water temperature T were recorded with a Seabird 19+ CTD. Steady wind speeds were measured with a portable Li-Cor-1400 meteostation ( Differential global positioning system (GPS) signals were recorded every few seconds by a GPS-12 (Furuno, USA).

3.5. Estimation of CH4 Flux

[19] The sea-air gas exchange rate or flux F was calculated from measurements of the concentration difference (ΔC) between the surface aqueous CH4 concentration and equivalent atmospheric CH4 based on Henry's law, specific gas properties (Schmidt number, Sc, T, and short-term records of the 10 m wind speed υ, according to F = 0.31υ2(Sc/660)−0.5ΔC [Wanninkhof, 1992]). Sc for CH4 in seawater was evaluated using the measured T and salinity [Jahne et al., 1987]. F was calculated based on daily records of υ, T, and surface water and atmospheric CH4 concentrations.

3.6. Geophysical Methods Used to Determine Permafrost State

[20] High-resolution seismic (HRS) data were obtained in 2004 during the Transdrift-X Expedition on board the R/V Yakov Smirnitskiy using a “Sonic M-141” system consisting of a high-resolution profiler with a frequency of 4.5 kHz, leading to a vertical resolution of 0.3–0.5 m and a maximum penetration of 30 m. Side-scan sonar at a signal frequency of 100/330 Hz with a scattering angle of 2 × 50° observed a 300 m wide seafloor swath for each seismic line. Data were interpreted using The Kingdom Suite (TKS) software version 7.6 (by Seismic Micro-Technology, Inc., USA). In general, the frozen-unfrozen sediment interface is characterized by a significant change in sound velocity that causes a strong seismic signal reflection [Rogers and Morack, 1980; Rekant et al., 2004].

4. Results

4.1. Dissolved Methane (CH4) in Seawater

4.1.1. Spatial Distribution of CH4

[21] Dissolved CH4 concentrations in September 2005 varied significantly between stations, creating very sharp spatial gradients at several locations (Figure 3). Surface water CH4 concentrations ranged from 1.9 to 651 nM, with a mean of 47.6 ± 11.5 nM (P = 95%, n = 98), where P is the confidence level. Most values were far greater than atmospheric equilibrium (∼3.5–4.0 nM), with saturation levels as high as 16,000% compared to the latitude-specific monthly mean (LSMM) of 1.85 ppm at Barrow, AK [NOAA, 2009]. Over 93% of surface water concentrations were >10 nM, 47% were >30 nM, 31% were >50 nM, and 9% were >100 nM. Bottom water CH4 concentrations were less wide ranging, from 2.1 to 298.5 nM, with a mean of 50.4 ± 3.7 nM (P = 95%, n = 168) (Figures 3a and 3b) with more than 88% (>10 nM), 44% (>30 nM), 23% (>50 nM), and 12% (>100 nM). The highest dissolved CH4 concentrations were measured east of the Lena Delta (Figures 3a and 3b).

[22] These CH4 concentrations are well above values typical of coastal water [Reeburgh, 2007], although similarly high values are reported for areas of active hydrocarbon seepage [Hovland et al., 1993]. Other CH4 seawater sources are microbial reduction of OM in sediments and upward transport by bubbles or diffusion [Hovland et al., 1993], water column methanogenesis [Damm et al., 2005], and riverine input [Shakhova and Semiletov, 2007]. We evaluated the potential contribution from these sources using several approaches. These included analysis of the CH4 vertical profiles, evaluation of seasonal and potential sedimentary and water column methanogenic production, comparison of the spatial distributions of CH4 and sediment OM, and analysis of seismic data.

4.1.2. Vertical Distribution of CH4 in the Water Column

[23] Three distinguishable water column CH4 profile patterns were observed: (1) bottom maxima with concentrations decreasing upward (Figure 4a); (2) an inverse distribution, with a surface maximum and concentrations decreasing downward (Figure 4b); and (3) a uniform concentration distribution across the entire range of concentrations (Figure 4c). We propose that, for these data, the first pattern is consistent with a seabed source and diffusive transport and the second with a persistent seabed source and bubble transport, and the third could be generated by either mechanism.

Figure 4.

Different types of dissolved CH4 profiles within the water column: (a) concentrations increase toward the surface; (b) concentrations decrease toward the surface; and (c) concentrations are uniformly distributed.

4.1.3. Dissolved CH4 Versus Hydrological Parameters

[24] Surface and bottom water salinities varied significantly from 4.9‰ to 33.9‰ and from 7.8‰ to 32.7‰, respectively, reflecting the strong influence of riverine water on the entire water column. Surface and bottom T ranged from 0.5°C to 8.5°C and from −1.6°C to 4.5°C, respectively. To distinguish between lateral CH4 transport and marine sources, we profiled CH4 concentrations and water temperature and salinity along three transects that were largely unaffected by river waters. Concentrations along transect 1 (Figure 5a) were characterized by a pronounced influence of fresher and warmer water and did not exceed 40 nM, while CH4 concentrations along transects 2 and 3, which were less affected by freshwater, were up to 3–4 times higher (Figures 5b and 5c).

Figure 5.

Vertical distribution of dissolved CH4, water temperature (T, C) and salinity (Sal, psu) along transects: (a) transect 1: (b) transect 2; (c) transect 3. Position of transects are shown in Figure 2a in red dotted arrows.

[25] In September 2006, we studied dissolved CH4 along the Bykovskaya Channel of the Lena River from a river boat (Figure 6). Dissolved CH4 concentrations clearly decreased in the downstream direction. This supports the conclusion that the Lena River Delta and the ESAS shelf waters have two distinct CH4 sources.

Figure 6.

Concentrations of dissolved CH4 downstream from the Bykovskii Channel of the Lena Delta in September 2006: (a) surface water and (b) bottom water.

4.1.4. Seasonal Variability of Dissolved CH4

[26] Because the rate of microbial methanogenesis is strongly temperature dependent [Claypool and Kaplan, 1974; Rivkina et al., 2006], production rates should show strong seasonal trends. Thus, we proposed that the microbial contribution to water column CH4 would show temporal trends that would be most distinctive in areas of the ESAS with the largest annual temperature variations of bottom water. To explore this, dissolved winter CH4 concentrations were measured in the water column beneath the ice east of the Lena Delta, where in summer we measured mean CH4 concentrations of 47 nM. Although winter bottom water temperatures were 5°C–7°C cooler than in summer, dissolved CH4 concentrations were up to 10 times higher (Figures 7a and 7b). At several stations, dissolved CH4 concentrations reached 5000 nM or three orders of magnitude higher than atmospheric equilibrium. Here large bubbles (up to 0.3 m diameter) were frozen in the ice (Figure 7c). CH4 in sampled bubbles was as high as 1.14%.

Figure 7.

Dissolved CH4 in the water beneath the sea ice: (a) surface water, (b) bottom water (position of oceanographic stations shown in Figure 2b), and (c) gas bubbles entrapped within the annual ice. Diameter of bubbles is up to 30 cm.

4.1.5. Spatial Distribution of Dissolved CH4 Versus Positions of Fault Zones

[27] Further, microbial CH4 production should be greater in areas like natural depressions that are favorable for sediment accumulation, such as fault zones, deltas and river paleovalleys. Thus, we compared CH4 measurements along transects 1, 2, and 3 (Figure 3b) with geologic and geographic features. Transect 1 crossed the UL fault zone and coincided with the direction of sediment migration toward the eastern Lena subsea valley. Because of the combined downward warming effect of sediments and a greater upward geothermal heat flux from beneath the fault zone than in other areas, this transect would be predicted to have among the highest CH4 concentrations. Nevertheless, dissolved CH4 concentrations along this transect were the lowest among the three transects (equation image40 nM). Transect 2 also intersected the BSN Rift but lay outside the predominant sediment migration pathways. Aqueous CH4 concentrations along this transect were highest (up to 400 nM) at the transect beginning, close to Bolshoi Lyakhovskii Island, and decreased along the transect to background. Transect 3 was outside fault zones and under minimal freshwater influence, but dissolved CH4 concentrations were as high as those measured along transect 2.

[28] On the basis of the transect data, we conclude that the lowest CH4 occurred where the highest microbial production was expected, strongly suggesting that CH4 emissions are likely driven by other mechanisms besides diffusion of dissolved CH4 predominantly originating in sediments.

4.2. Methane Sea-Air Flux

[29] Measurements of atmospheric CH4 in the ESAS were extremely variable, ranging from below the latitude-specific monthly mean of 1.85 ppm to as high as 8.2 ppm (Figure 8).

Figure 8.

Mixing ratios of CH4 in the atmosphere. Mixing ratio was measured ∼5–10 m above the sea surface measured along transects 1, 2, and 3 (the position of transects is shown in Figure 2a by red dotted lines). The latitude-specific monthly mean (LSMM) equals 1.85 ppmv ( and is shown by the blue line.

[30] The highest atmospheric CH4 mixing ratios occurred along transect 1, with a mean of 3.8 ± 0.05 (P = 95%, n = 310) and a range of 1.9–8.2 ppm. As the ship moved coastward and toward the bubble plume area north of the Lena Delta (Figure 3a), aqueous CH4 concentrations decreased (Figure 5a) while atmospheric CH4 mixing ratios increased. Transect speeds were moderate (5–8 m s−1) under northeast winds. These data are consistent with atmospheric CH4 transport from the bubble plume area. The atmospheric CH4 mixing ratio also increased upwind (Figure 9), suggesting that significant CH4 from the plume area, toward which the ship was moving, was primarily advected coastward of the boat's path.

Figure 9.

Spatial distribution of mixing ratios of methane in the atmosphere. Measured values are shown in red scaled picks; actual wind directions and wind speeds are shown in blue scaled arrows. Data are obtained during the summer cruise (September 2005).

[31] Observed decreasing atmospheric CH4 concentrations along transect 2 corresponded with the decreasing trend in aqueous CH4 concentration along the transect. However, while dissolved CH4 concentrations along transect 2 were significantly (3–4 times) higher than those measured along transect 1, atmospheric CH4 concentrations generally were lower (1.97–2.9 ppm, mean = 2.69 ± 0.08 ppm, P = 95%, n = 250). Atmospheric CH4 was 3.9 ppm at the very beginning of transect 3, while dissolved CH4 concentrations varied from 30 to 50 nM. Atmospheric CH4 was measured for westerly winds and thus may reflect another easterly source, which could be more significant than evasion from the underlying water column. The mean atmospheric CH4 mixing ratio along transect 3 was 2.4 ± 0.02 ppm (P = 95%, n = 127) and ranged from 2.2 to 3.9 ppm.

4.3. Summer and Winter Potential Methane Flux in Bhuor-Khaya Bay

[32] We used a simple model [Shakhova et al., 2005] to calculate and compare the late summer to the late spring potential CH4 emission Ep from the water column in the plume area based on the area's CH4 budget A, which is the maximum, area-adjusted dissolved CH4 available for release to the atmosphere and was calculated as

equation image

where s = s(x, y), z are the horizontal and vertical coordinates, S is the area, H(s) is the local depth, and A(s, z) is the spatial distribution of the dissolved CH4 concentration. A(s, z) was obtained by vertical and horizontal linear interpolation between available data. A = 7.6 × 107 and 60.1 × 107 g CH4 in summer and winter, respectively, for the Bhuor-Khaya plume area (∼103 km2, mean depth of 10 m, Figure 3a, marked with a black square). CH4 excess or Ep was defined as Ep = AAe, where Ae is CH4 storage if the area is at equilibrium with the atmosphere. Using an atmospheric concentration of 1.85 ppm, Ae = 56 × 104 g CH4, orders of magnitude less than A. The calculated summer Ep was more than 8 times higher than that of the winter, demonstrating that winter ice leads to significant accumulation of dissolved CH4 in the surface water.

4.4. Geophysical Data

[33] The high-resolution seismic profiles identified a large number of gas migration pathways in the studied area, appearing as semiblanked seismic signatures with sharp vertical side boundaries (Figure 10).

Figure 10.

High-resolution seismic data obtained in the study area (Transdrift-X, 2004). (a and b) Two typical vertical blanked zones, interpreted as gas migration pathways. Positions of sites A and B are shown in Figure 2a.

[34] Two potential gas migration patterns were identified within the study area. The first type is characterized by relatively wide (up to several hundred meters) homogeneous areas of weak-amplitude acoustic signal (see Figure 10a, position of the site shown in Figure 2a as a black square marked A). This type could be a zone of diffusive gas release. The top of these features has an unevenly degraded surface mainly at 5–10 m below the seafloor. We propose that these features may indicate high-permeability zones within the permafrost, possibly associated with talik development. Within these zones, the seismic data clearly show gas extending up to an impenetrable sediment stratum, possibly a clay horizon. Farther upward movement can occur through discontinuities if the overlying strata are disturbed by sediment settlement and adjustment, thermal contraction, seismic events, or further talik development.

[35] The second gas migration pattern was narrower (Figure 10b, position of the site shown in Figure 2a as a black square marked B) and suggested focused gas release through discontinuities within the permafrost. The tops of such structures are often situated just below the seafloor. We propose these anomalies occur within more coarse-grained sediments (e.g., sandy silt, silt), and are associated with high-permeability pathways through which gas and/or gas-bearing fluids can seep to the seafloor and water column.

5. Discussion

5.1. ESAS Methane Sources

5.1.1. Microbial Production in Sediments Production in Modern Sediments

[36] The data show that the ESAS is a strong CH4 source in winter and summer. Indeed, >90% of bottom and >60% of surface water samples were supersaturated in dissolved CH4 by factors of up to 160 and 1200 in summer and winter, respectively. Moreover, the spatial pattern of dissolved CH4 in bottom waters did not correspond to the distribution of Corg in surface sediments [Dudarev et al., 2006]. Instead, areas with the highest concentrations of dissolved CH4 correlated with fault zones (Figure 3b). Thus, the spatial patterns of dissolved CH4 are distinct from those of surface sediments, strong evidence that the dominant water column CH4 source is not microbial sediment production.

[37] It is well known that biogenic CH4 is produced in organic-rich seabed sediments, such as estuaries, river deltas, and continental shelves and slopes, as long as sufficient OM is available [Claypool and Kaplan, 1974; Hovland et al., 1993; Judd 2004; Reeburgh, 2007]. In the ESAS, such production is limited to the Holocene sediment layer accumulated above the permafrost, which caps deeper sediments. This layer is significantly deep (>1 m) only in areas of preferential sediment accumulation, primarily in water depths greater than 50 m, which were outside the study area. Because the thickness of the layer above the permafrost cap in nearshore waters is generally uniform, the spatial distribution of production rates and, thus, dissolved CH4 fluxes should relate to the Corg fraction in the sediments. Further, although dissolved CH4 concentrations varied by several orders of magnitude, from 2.0 to 5000 nM (Figures 3 and 7), Corg varied by just a factor of ∼4.

[38] For the Arctic Ocean, an initial CH4 content in sediments of ∼0.5 mM g−1 was thought to be sufficient to create a bottom CH4 concentration of ∼50 nM [Lupton et al., 1985; Damm et al., 2007]. Thus, bottom water concentrations of 600 nM would require CH4 content in sediments of equation image6 mM g−1. However, it was shown by Lein et al. [2007] that CH4 production in surface sediments was extremely low even in the Chukchi Sea, where primary production is highest: production rates vary from 9 × 10−5 to 0.029 nM g−1 d−1, while rates of oxidation vary from 0.001 to 0.009 nM g−1 d−1; in 30% of stations, rates of oxidation exceed rates of methanogenesis. Finally, it is difficult to invoke methanogenesis to explain significantly enhanced winter, as compared to summer, aqueous CH4 concentrations, even though methanotropic metabolic reactions occur at subzero temperatures; microorganisms in permafrost remain viable to −20°C [Rivkina et al., 2004]. Specifically, because of the strong temperature dependency of methanogenesis, a temperature increase of 3°C corresponds to an order of magnitude increase in productivity of methanogenic bacteria in permafrost [Rivkina et al., 2006]. Thus, bottom water CH4 concentrations should not be orders of magnitude lower in the summer than they are in the winter. Production in Taliks

[39] Taliks are a proposed source of microbial CH4 to the water column [Zimov et al., 1997]. Permafrost can thaw locally under the warming effect of riverine waters and/or geothermal heat flux, allowing unfrozen sediments to develop [Delisle, 2000; Kasimskaya 2005; Romanovskii et al., 2005]. Bottom-up or top-down developing taliks (bulbs or columns of thaw sediments) allow CH4 production from biogenic substrates previously sequestered by permafrost [Shakhova and Semiletov, 2007]. It has been proposed that the greatest methanogenic potential occurs in buried ice-rich permafrost deposits including peat and loesslike silty sand, termed edomas, which are highly enriched with Corg by as much as 12% [Zimov et al., 2006]. Edomas were proposed to cause northern lake ebullition [Zimov et al., 1997]. Because edomas are widely spread over the Siberian coastal plain, they should be present in the ESAS seabed. However, recent analysis of dozens of coastal Siberian cores into edomas showed the majority contained little or no CH4 and limited CH4 productivity [Rivkina et al., 2006]. Microbial production in submerged ESAS edomas was shown by Koch et al. [2008] to vary from 0 to 284 nM g−1 CH4, with negligible concentrations of CH4 found in overlying sediments. Thus, net CH4 production was less than anaerobic oxidation in the upper sediment layers.

5.1.2. Microbial Production in the Water Column

[40] Water column CH4 production is proposed to explain the elevated CH4 in the Laptev Sea [Cramer and Franke, 2005]. Indeed, it is known that microbial oxidation of marine OM, primarily from primary production, produces CH4. Specifically, in oxygen-saturated stratified waters, CH4 production occurs within limited anaerobic lenses at the base of the pycnocline >100 m below the sea surface. There, enough OM accumulates to maintain subsurface concentrations of up to 9 nM [Ward et al., 1987; Damm et al., 2005; Sasakawa et al., 2008]. For several reasons, ESAS water column methanogenesis should be significantly less. ESAS primary productivity is suppressed by the lack of sunlight; the 1% light penetration depth is at most 12–15 m, decreasing to 5 m on the shallow shelf [Tuschling et al., 2000]. Even where photosynthesis is highest (up to 0.1–0.3 g C m −2 d−1), it is two to three orders of magnitude lower than in typical ocean waters and decreases toward the shelf edge [Sorokyn and Sorokyn, 1996].

[41] An alternative potential substrate for water column methanogenesis could be terrestrial-sourced particulate material (PM), which is common in the ESAS [Semiletov et al., 2005]. However, only the water extractable fraction of organic carbon (WEOC) could potentially serve as a suitable substrate. This fraction comprises just ∼1% of the Corg present in PM [Wagner et al., 2005], and whether it provides a suitable substrate for seawater methanogenic bacteria is unknown. Thus, we conclude that water column methanogenesis is unlikely to explain more than a minor fraction of the observed dissolved CH4.

5.1.3. Lateral Input of Microbially Produced CH4

[42] Russian Arctic rivers are another potential CH4 source to the ESAS [Shakhova and Semiletov, 2007]. However, the river CH4 transect showed that CH4 decreased with downriver distance in the Lena River (Figure 6), demonstrating that the Lena River was, at best, a minor CH4 source to the ESAS. We propose that this is because the majority of CH4 entering the rivers from terrestrial sources oxidizes in aerobic waters and escapes to the atmosphere before river waters reach the ocean. Thus, we conclude that modern biogenic sources could only be minor contributors to the dissolved CH4 inventory in the ESAS.

5.1.4. Seabed Methane Sources Geological Sources

[43] A clear signal of C1–C5 hydrocarbons was detected in all sediments; the lack of any obvious regional variations suggests a thermogenic source within the ESAS [Ilyin et al., 1998; Cramer and Franke, 2005]. Concentrations of CH4equation image 37.5 nM were measured at a water depth of 100 or 150 m directly above the seabed sediment surface in an area that is also underlain by a deep-reaching strike-slip fault called the Northern Fracture, which may provide a preferential pathway for gas migrating upward. Arctic Shallow Hydrate Deposits

[44] Arctic shallow hydrate deposits represent an extensive and massive potential CH4 source, which can occur at depths as shallow as 20 m [Chuvilin et al., 2000, 2005]. These deposits have specific features, which differ from non-Arctic hydrate deposits. Specifically, because they originated under very low subzero temperatures, destabilization can occur with an energy input 3 times less than typical [Makogon et al., 2007]. Also, they are very highly spatially concentrated, have extremely high pore saturation (20%–100% of pore space, in contrast to oceanic hydrates which occupy only 1%–2% of pore space), can exist in thick layers of >100 m, and occur 3 times more frequently offshore than onshore in the Arctic environment [Hyndman and Dallimore, 2001; Soloviev, 2002].

[45] In the ESAS, the effect of bottom-up warming from geothermal heat flux transforms CH4 from ice-like hydrates into free gas, which accumulates beneath still-impermeable upper permafrost layers. Because of this permafrost cap, the pressure increases [Romanovskii et al., 2005; Makogon et al., 2007], implying that when migration pathways develop, fluxes can be high. Also, this high pressure acts as a geological force enabling migration pathways to develop even through continuous permafrost [Paull et al., 2007; Lorenson et al., 2008].

5.2. Migration Pathways and Transport Mechanisms

5.2.1. Permafrost Migration Pathways

[46] Clearly, migration pathways are key determinants of whether CH4 from seabed deposits can escape to the ocean. At the seabed, pathways manifest as pockmarks, mud volcanoes, funnels, chimneys, and pingo-like structures, or they may be morphologically unspecified [Hovland et al., 1993; Judd, 2004]. Additional pathways could be provided by completely submerged thaw lakes, underlain by taliks, which formed on the Siberian coastal plain prior to inundation. A number of such lakes have been transformed into sea lagoons [Romanovskii et al., 2005] or have left seabed depressions in the ESAS, which are interpreted as a typical thermokarst terrain landscape similar to the terrain of the Siberian Lowland [Rekant et al., 2009].

[47] Evidence of the existence of migration pathways through and within the permafrost is provided by seismic data, specifically as low-amplitude anomalies sometimes referred to as a “washed-out” or “semiblanked” zone. In the marine environment, widespread washed-out zones have often been attributed to gas hydrates [Holbrook, 2001]. In permafrost, low seismic amplitude may also result from variations of physical properties or property changes associated with development of deep taliks. The underlying mechanism is the significant variation in the velocity of sound waves as a function of the fraction of unfrozen pore water [Zimmermann and King, 1986]. Thus, taliks, which include unfrozen or partially unfrozen areas, attenuate seismic waves more severely than do fully frozen sediments, producing low-amplitude areas (Figure 10). Such anomalies are found beneath lakes and drainage features in permafrost, where they are interpreted as providing possible migration pathways [Wright et al., 2008].

5.2.2. CH4 Transport Within the Water Column Water Column Signature of Ebullition

[48] CH4 transport from the seabed to the atmosphere occurs by diffusion and by bubble ebullition where fluxes are sufficiently high [Hovland et al., 1993; Judd 2004]. Diffusion is slow, allowing significant water column CH4 oxidation [Claypool and Kaplan, 1974]. Water column microbial oxidation time scales are estimated to be 5–100 years [Tilbrook and Karl, 1995]. Bubble-mediated CH4 transport is far more efficient than diffusion. Ebullition occurs from sediment microbial sources [Claypool and Kaplan, 1974; Best et al., 2006] and, through migration pathways, connects trapped CH4 to the seabed [Hovland et al., 1993]. Transit times are on the order of minutes for upwelling fluid and less for bubbles; thus, negligible CH4 oxidation occurs. As a bubble rises, CH4 flows out of the bubble, while other gases flow in [Leifer and Patro, 2002]. Thus, bubble plume transport is consistent with a concave-up profile of the dissolved CH4 distribution (Figure 4b). This is in contrast to vertical profiles that are concave-down (Figure 4a), typical of diffusive CH4 transport [Reeburgh, 2007].

[49] Bubble plumes can also lead to more uniform depth profiles (Figure 4a). This can be partially explained by hydrological processes [Damm et al., 2005]. In some areas of the ESAS, where the water column is mixed from the top to the bottom (like the Dmitry Laptev Strait area; see Figure 3a, red square), ebullition occurs within nonstratified water. However, uniform profiles with very high dissolved CH4 concentrations were also observed year-round in strongly stratified waters such as east of the Lena Delta (Figure 3a, area marked with a black square). The following is one possible explanation for the summer gas profile. Leifer et al. [2010] observed that upwelled water had significantly lower density than surrounding surface waters and sank after outwelling. Rising bubbles drive an upward flow that transports this enriched fluid vertically toward the sea surface [Leifer et al., 2010]. Significant loss from the upwelling flow occurs where the bubble plume traverses density stratification, leading to plume detrainment into intrusion layers. This plume detrainment process is well studied for lake destratification [McGuiness et al., 2006] and has been identified for marine seepage [Leifer and Judd, 2002; Leifer et al., 2010]. This overturning of the water column could lead to mixing and a more uniform CH4 profile.

[50] We propose that winter water column CH4 profiles could be governed by unique plume dynamics. During the fall, many intense storms occur over the ESAS [Nikiforov and Shpaikher, 1980], leading to intense vertical mixing that extends to the bottom in shallow areas [Kulakov et al., 2003]. When the freezing period starts, ice formation blocks CH4 bubbles from reaching the atmosphere; thus, bubble contents are deposited in the surface layer where some bubbles and water become entrapped in the fast ice for the winter. The rising bubble plume flow also lifts CH4-rich water and deposits it at the ice base. These processes create a profile with a surface maximum that could persist through the summer until deep fall convection occurs. Ebullition and Atmospheric Emissions

[51] Calculated CH4 fluxes ranged from 116 to 5240 μM m−2 d−1 (mean = 456 ± 89 μM m−2 d−1). Kourtidis et al. [2006] used a simple box model to show that diffusive CH4 fluxes of equation image480 μM m−2 d−1 would result in an increased atmospheric CH4 mixing ratio of a few tens of parts per trillion, contributing negligibly to regional atmospheric CH4 concentrations. The average mixing ratio of atmospheric CH4 (2.4 ppm) was significantly increased compared with typical Arctic values, implying a strong source. Atmospheric CH4 concentration may not correlate with near-surface aqueous CH4 concentration, in part because both atmospheric and aqueous concentrations are cumulative; they depend on transport and the location of heterogeneous sources [Leifer et al., 2006]. The temporal history of marine CH4 also may be important. Thus, to explain the atmospheric levels of CH4 found in our study area, fluxes must be many orders of magnitude greater than calculated ones.

5.3. Implications for the Global Methane Cycle and Climate

5.3.1. Vulnerability of Arctic Shallow Hydrates

[52] Arctic shallow hydrates, because of their inundation, have been exposed to temperatures about 5°C–10°C warmer than temperatures of terrestrial Arctic shallow hydrates for the past 5–10 kyr. On the basis of the heat transfer downward from relatively warmer ocean waters and upward from below, numerical models predict destabilization after ∼5–10 kyr of inundation [Romanovskii et al., 2005]. As a result, it is probable that large-scale hydrate destabilization will occur first in the ESAS and other areas of submerged shallow permafrost. In fact, it is feasible that hydrate destabilization in the Arctic is currently creating free gas reservoirs trapped below the largely impermeable permafrost layer. In contrast to other areas of the Arctic Ocean, the ESAS water column provides a very short conduit for releasing CH4 to the atmosphere. This makes the ESAS a primary, important region for CH4 release, compared to other areas of the Arctic Ocean where the majority of CH4 passing through the water column is oxidized [Westbrook et al., 2009].

[53] Continued hydrate destabilization will lead to increasing pressure in these shallow reservoirs. In such case, fracturing and thawing of the permafrost will create pathways for deeper, hydrate-derived CH4 deposits to escape to the sea surface, a process that we propose is currently occurring in the ESAS and is consistent with our data. Further, the shallowness of the ESAS implies hydrate-derived CH4 here will affect atmospheric budgets much more than will most of the CH4 from deep-sea hydrate deposits at lower latitudes [Kvenvolden, 1988, 2002].

5.3.2. Potential Fluxes

[54] CH4 flux from the ESAS has enormous implications for the Arctic ecosystem and global climate. The potential CH4 emission from the ESAS during the short Arctic summer (100 days) by diffusive flux alone has been estimated, based on a limited data set, to be as much as 1.5 Tg [Shakhova et al., 2005]. However, the ESAS budget consists of many components. First, because fall convection mixes the entire water column [Kulakov et al., 2003], significant release could occur during the stormy fall period and during freezeup [Mastepanov et al., 2008]. Second, winter emissions may be nonnegligible because of deep convection in flaw polynyas (band-like ice-free areas). Flaw polynyas can be tens of kilometers wide, and they migrate with fast ice up to hundreds of kilometers [Smolyanitsky et al., 2003]. Sea-air CH4 fluxes from flaw polynyas can be 20–200 times higher than the ocean average [Damm et al., 2007]. Third, significant CH4 emissions likely occur during ice breakup from areas unaffected by polynyas, which would allow CH4 that has accumulated over the winter in the water column and within the sea ice to escape. Additionally, the contribution of ebullition, which represents the predominant mechanism of CH4 transport in the ESAS water column, still remains to be apportioned. If one hypothesizes that the seabed CH4 flux to the water column is largely seasonal (based on heat flux processes, which operate on time scales longer than a year), then the current, annual CH4 flux from the ESAS may be significantly greater than the current estimate of summer flux.

[55] Current estimates of CH4 emission are based on the current Arctic thermal regime. However, the largest global warming trend, from 4°C to 7°C, is forecasted for the Eurasian Arctic by the end of the 21st century [ACIA, 2005]. Warmer thermal regimes will progressively decrease the integrity of the permafrost cap, leading to the release of vast quantities of CH4. Further, CH4 release occurs because of permafrost integrity failure; thus, the process is unlikely to scale linearly with temperatures. Given that the CH4 reservoir is enormous, estimated at 1400 Gt, the release of only a very small fraction has the potential to dramatically alter modern biogeochemical cycles. Thus, the implications for global climate depend critically on whether time scales are geologic or human.

6. Conclusions

[56] Data presented herein demonstrate significant marine seabed CH4 fluxes into the Arctic Ocean and atmosphere from the ESAS, causing CH4 concentrations far above typical atmospheric and oceanic values. This study supports the hypothesis that these fluxes are driven by increasing degradation of subsea permafrost due to the long-lasting warming that was initiated by permafrost submergence. Nonpermafrost sources were evaluated and considered to contribute only minimally to the extreme CH4 enhancements observed. Instead, observations demonstrated that CH4 seabed fluxes were primarily due to permafrost integrity failure; flux occurred through identified migration pathways as well as via hypothesized structures such as taliks. Further efforts to derive more accurate regional emission fluxes and determine seasonal variations are needed to understand the current role of the ESAS in carbon cycles and to begin to predict the future impacts of warming Arctic climate.


[57] We thank Valentin Sergienko, Larry Hinzman, and Gueorgui Golitsyn for encouraging this study. We thank anonymous reviewers for comments on an earlier draft of this manuscript. This work was supported by the Far Eastern Branch of Russian Academy of Sciences (06-05-96064p), NOAA (Cooperative Agreement NA17RJ1224), the International Arctic Research Center of the University of Alaska Fairbanks, and the Russian Foundation for Basic Research (07-05-00050а, 08-05-00191а).