Methane fluxes from the sea to the atmosphere across the Siberian shelf seas

1 Introduction

Seafloor CH₄ seeps have attracted increased attention following the suggestion that massive releases of subsea CH₄ may have led to rapid climate shifts in the distant past [Kennett et al., 2000; Nisbet, 1990]. Seeps of CH₄ are found in shallow shelf seas worldwide [Hu et al., 2012; Kodovska et al., 2016; Skarke et al., 2014] and are estimated to contribute 18–48 Tg yr⁻¹ [Hornafius et al., 1999] globally to the water column. The amount of CH₄ which reaches the atmosphere from subsea seeps is limited by dissolution into the water column and methanotrophic microorganisms in the water column [Leifer and Patro, 2002; McGinnis et al., 2006; Osudar et al., 2015], at the seafloor [Marlow et al., 2014], or below the seafloor [Overduin et al., 2015].

Nonetheless, reports of high levels of CH₄ in the atmosphere above the Laptev Sea (LS) and East Siberian Sea (ESS) with average concentrations of 2.97 and 2.66 ppm respectively, far above then-mean latitudinal values of approximately 1.85 ppm [Shakhova et al., 2010a, 2010b], suggested a surprisingly large CH₄ source from these seas. The Laptev and East Siberian Seas are wide and shallow, with continental shelf areas averaging 48 and 58 m average depths, respectively (collectively the East Siberian Arctic Shelf or ESAS) [Jakobsson, 2002], and partly underlain by thawing remnant permafrost from the Last Glacial Maximum [Nicolsky et al., 2012; Romanovskii and Hubberten, 2001]. The changing state of this system [Overduin et al., 2007] kindled interest in the possibility of present-day or near-future large-scale release of CH₄ from the ESAS. Such CH₄ could originate from long-locked reservoirs, either deep thermogenic sources [Cramer and Franke, 2005], degradation of Pleistocene and Holocene organic material released and transported from thawing permafrost on land [Charkin et al., 2011] or coastal erosion [Günther et al., 2013], increased primary production due to less ice coverage in recent years [Arrigo et al., 2008], or CH₄ produced by modern subsea permafrost degradation [Dmitrenko et al., 2011]. Such enhanced degradation could be driven by seawater warming via Atlantic intrusion [Westbrook et al., 2009] or warming due to seasonal ice loss [Hölemann et al., 2011].

CH₄ in surface waters enters the atmosphere via diffusion during ice-free conditions, with the flux rate controlled by temperature, relative sea-air concentrations, and wind [Wanninkhof, 1992, 2014]. It has also been suggested that shallow seawaters such as the ESAS are easily transited by sediment-released CH₄-containing bubbles [Leifer and Patro, 2002]; these bubbles released from the seafloor represent a short circuit, a rapid transport process for CH₄ from seabed sources to the atmosphere which may avoid the water column loss processes mentioned above. Such bubbling resulting in CH₄ release to the atmosphere has been suggested to occur over 10% of the ESAS [Shakhova et al., 2010a]. Determining annual sea-air CH₄ flux is complicated by ice cover, which impedes and collects CH₄ emissions for much of the year and may contribute to enhanced flux at ice-out. These processes have been observed in lakes [Greene et al., 2014; Jammet et al., 2015; Lindgren et al., 2016].

Here we combine unique data sets of spatial distributions of CH₄ from continuous surface water and atmospheric measurements from the Swedish icebreaker Oden in the LS and ESS in July and August 2014 during the SWERUS-C3 project. We report on the magnitude of CH₄ transport to the atmosphere from any ESAS sources. We do not speculate about any putative CH₄ source because CH₄ released to the atmosphere must transit the water column and the sea-air interface. Our near-surface in situ approach integrates all local CH₄ sources.

2 Methods

We report here measurements from several overlapping segments of the cruise track divided into regions for comparisons: (1) mostly ice-covered Arctic Ocean (AO) north of the Kara and LS, (2) the shelf break and upper continental slopes to 500 m depth of the LS and ESS, (3) the ice-free segments of the LS and ESS, (4) the entire LS (ice-free during this study), (5) the LS segments well removed from subsea gas seeps, (6) the ice-covered segments of the ESS, and (7) the ice-free segments of the ESS.

Air was continuously pumped from four heights (averaging 9, 15, 20, and 35 m above sea level) during SWERUS-C3. The exact heights varied with ship bunker load; when full at the beginning of the cruise, heights were 70 cm lower, by late August, 60 cm higher than the listed heights. A 10 m meteorological tower provided mounts for the 15 and 20 m inlets. The 9 m inlet was mounted on a triangular bowsprit, 2.5 m forward of the front deck rail. The 35 m inlet was located forward of the exhaust stacks on a cargo container atop Oden's bridge. At stations where the ship was anchored, the 15 m inlet was lowered to 4 m above the sea surface. Air was pulled to the analysis system inside a lab; measurements were determined with a cavity ring-down laser spectrometer (Model 0010, FGGA 24EP, Los Gatos Research (LGR), Mountain View, California, USA). Raw measurements were collected at 1 Hz. Each inlet was monitored for 2 min at a time and then switched to the next inlet in the sequence. The sequence for inlet sampling was 9, 20, 15 (or 4), then 35 m. The last 70 s of the 2 min dwell times after valve rotation are averaged for the values reported here. CH₄ precision is 0.5 ppb; accuracy is 1.5 ppb. CO₂ precision is 0.25 ppm; accuracy is 0.25 ppm. CH₄ precision is determined by target gas measurement variability and is similar to SD of 70 s averages. Target gases were introduced every 2 h. Details are provided in the supporting information.

Wind speed and direction were used to filter the data to avoid air contaminated by the ship. We include data collected when the relative wind speed was >2 m s⁻¹, and the wind direction was >280° and <80° relative to the ship's bow. Any remaining observations with CO₂ concentrations above 450 ppm were also removed. After filtering, atmospheric CO₂ and CH₄ averaged 391.51 ± 0.58 ppm and 1.8834 ± 0.0026 ppm (both 2σ).

CH₄ and CO₂ in the surface water were continuously measured using the Stockholm University Water Equilibration Gas Analyzer System (WEGAS; see supporting information). Seawater was collected from an inlet at 8 m on the hull of Oden. After extraction with a showerhead equilibrator [Johnson, 1999] and drying, CH₄ and CO₂ were analyzed by two cavity ring-down gas spectrometers (G2301 and G2131-i, Picarro, Inc, Sunnyvale, California, USA). No special provision was made to exclude bubbles from the seawater intake. The CH₄ and CO₂ concentrations measured by the WEGAS system thus include any bubbles in the sample and will bias our flux calculation results toward higher values.

Ice coverage during the cruise was determined by visual observation and by satellite measurements from the Special Sensor Microwave Imager/Sounder (SSMIS) instruments onboard the United States' Defense Meteorological Satellite Program satellites using a 91 GHz microwave radiometer channel for ice coverage measurements. The analysis of these measurements was provided by the University of Bremen [Spreen et al., 2008]. The resolution of these satellite data is approximately 13 × 15 km; ice coverage is reported as a percentage within each grid cell.

CH₄ mixing ratios in air and water CH₄ concentration data, as well as ice coverage data for Oden's position, and other ancillary measurements, were integrated and extrapolated using R (version 3.1.2). The resulting combined data set has a temporal resolution of approximately 10 s, which corresponds to an average spatial resolution of about 0.0002 × 0.0009° (about 120 × 500 m in the LS).

Because the speed of Oden during the cruise varied greatly and included many stops for sampling (especially around known seafloor CH₄ seeps, biasing the raw data toward seep areas), a spatially normalized data set was created. The spatially normalized data set was created by rounding all observations' GPS coordinates to 0.001° then binning and averaging air and water data points which shared the same rounded coordinates.

We calculated the expected diffusive water-air CH₄ flux via the Wanninkhof model, which takes as inputs wind speed, water temperature, and water and air concentrations [Wanninkhof, 2014]. This diffusive flux model is especially suited for regions of long fetches, such as the open sea.

3 Results

A portion of the SWERUS-C3 cruise was designed to sample the middle-to-outer portions of the ESAS which had been relatively undersampled in previous regional studies. Therefore, the cruise spent considerable time in relatively deeper water areas of the continental shelf but still in waters <60 m deep (Table 1). Weather conditions during SWERUS-C3 were relatively calm much of the time, often with a very shallow inversion layer, especially during the ESS portion of the cruise [Tjernström et al., 2015], which would tend to trap gases, including CH₄, in the planetary boundary layer, causing local concentrations to rise.

3.1 Seafloor Gas Seeps and Surface CH₄ Enhancements

Relatively rare enhancements in air and water CH₄ observations above background levels (Figure S1 and Text S3 in the supporting information) were generally spatially correlated and occurred in regions with identified active seafloor gas seeps. Seafloor seeps were located as gas bubble trains in the water column observed with the ship's sonar systems; sonar measurements are otherwise beyond the scope of this paper. By design, the first half of the SWERUS-C3 cruise spent extended time sampling around two regions of known seep activity in the LS and ESS. Methane in air was above background levels in the region around the first-encounted subsea seep area in the LS, in waters ~65 m deep, centered at 76.894°N, 127.798°E (the seep center is defined here by the peak in surface water CH₄ measurements). Mixing ratios as high as 2.172 ppm (Tables 1 and S3) were observed though not specifically collocated with the peak in water CH₄ enhancements of up to ~250 ppm (Figure S2 and Table 1). The second seep region was centered at 74.957°E, 161.091°E in the ESS in ice-covered seas (Figure 1c). Enhanced CH₄ in the water (up to ~200 ppm) was less geographically confined in the ESS than in the LS (Figures 1 and 2), which may be due to the sea ice cover. The ESS seep region in ice-covered waters was confined within a few tens of kilometers of the peak-observed water concentrations but with some apparent spreading of the surface water CH₄ out to ~300 km around the region of the most active ebullition (Figure S3). Waters with sea ice concentrations below 90% were 200 km or more from the studied seep region in the ESS (Figure 1c).

The seep region in the LS had no sea ice cover to impede the passage of CH₄ from seawater to the atmosphere. Within approximately 100 m at the sea surface of the peak-observed water CH₄, our estimated CH₄ flux was 83 mg m² d⁻¹. The spatial extent of the high flux was limited; within 1 km, the flux was 58 mg m² d⁻¹ but within 100 km around the seep area; the average flux was estimated at 14 mg m² d⁻¹.

Generally, there was no significant difference (1σ values overlap) between atmospheric CH₄ measured between the lowest (9 m) and highest (35 m) inlets (Figure S4). At four anchor stations, we were able to lower one air sampling inlet to 2–4 m above the sea surface, a strategy to better collect locally emitted CH₄. Even under these conditions, average gradient enhancements were minimal (Figure S4 and Table S3). At all four anchor stations the ship rode the anchor downwind of the seep areas so the surface flow was across the seep areas to our profile mast air intake array. In the LS on Julian day 199 while anchored near the seep area, CH₄ in air measurements did show an apparent gradient. Values reached 2.172 ppm CH₄ at the temporary lowest (4 m) air inlet, while the highest (35 m) air inlet showed values up to 1.90 ppm during the same 4 h period. The mean concentrations during this same 4 h period were 1.95 ppm and 1.88 ppm at 4 and 35 m above sea level, respectively, consistent with our flux calculations in this active seep area.

4 Discussion

The methodological approach in this paper (high-resolution simultaneous continuous measurements in surface waters and in the air) allows us to test two recently proposed hypotheses about sea-air exchange of CH₄ across the ESAS. The proposed extensive bubble flux from subsea CH₄ seeps to the atmosphere could not be validated for the middle and outer ESAS waters >35 m depth. Neither widespread strong enhancements of ambient CH₄ nor strong near-surface gradients could be identified. Our high-resolution data set rather suggests a spatially limited effect of gas seep bubbles on atmospheric CH₄ mixing ratios. Instead, diffusive fluxes alone can explain observed atmospheric mixing ratios that are slightly elevated in some areas but much less than those reported for shallow inshore areas.

The average spatially normalized CH₄ flux we derived for the entire studied open water ESAS region was 3.8 mg m² d⁻¹, is considerably higher than has been reported for many shelf seas, such as the North Sea (0.04 mg m⁻² d⁻¹) [Bange et al., 1994] and subarctic Alaskan bays (0.008–0.013 mg m⁻² d⁻¹) [Kodovska et al., 2016]. Although this study does not achieve the spatial coverage closer to shore as some previous work [Shakhova et al., 2005, 2010a, 2010b, 2014], the cruise track provides us with a vast data set of synchronized surface water and air measurements across a variety of ESAS environs near the average depth of the ESS and LS; therefore, we have applied our average-calculated fluxes across the continental shelf areas of the Laptev, East Siberian, and Chukchi Seas (2,105,000 km²); this would yield an annual flux of 2.9 Tg CH₄ yr⁻¹. However, important caveats of our limited spatial coverage must be applied to this estimate, and sea ice covers these areas for about 70% of the year [Proshutinsky et al., 1999], as discussed below. Our annual flux, not accounting for impermeable ice cover periods, is in rough agreement with previously reported regional fluxes based on measurements in open waters of the LS and ESS, 1–4.5 Tg yr⁻¹ [Shakhova et al., 2005], though later work noted similar fluxes with additional contribution from subsea CH₄ seeps [Salyuk and Semiletov, 2010]. A more recent study focused on shallower portions of the LS and ESS (6–24 m depth) and found an average flux of 287 mg m⁻² d⁻¹, based on bubble fluxes derived from sonar [Shakhova et al., 2014]. This same study also extrapolated shelf-wide fluxes as well, reporting a flux from subsea CH₄ seeps to the atmosphere of 9 Tg CH₄ yr⁻¹, and a total flux (including diffusive fluxes) of 17 Tg CH₄ yr⁻¹ for the LS and ESS, by estimating that their study accounted for 10% of extant ESAS seeps. Finally, we note that our in situ results (along with all previous in situ results) are higher than a recent model of CH₄ release (bubbling + diffusion) from the Siberian continental shelves of 0.42 Tg yr⁻¹ [Archer, 2015] but are similar to a recent modeling estimate based on Pan-Arctic CH₄ measurements from long-term monitoring stations on land [Berchet et al., 2016].

As noted above, our seawater CH₄ measurements do not exclude bubbles, and over several days of continuous measurements crossing the seep regions, our seawater CH₄ values certainly incorporate some bubble-delivered CH₄ in the surface waters. There is a large difference in transport success of bubble-contained CH₄ in 6 m deep water and 40 m deep water [McGinnis et al., 2006]; rapid exchange and loss of CH₄ to the generally deeper water column in our study alone may account for some of this difference between our and previous studies closer to shore [Sergienko et al., 2012; Shakhova et al., 2014], in shallower depths than studied during SWERUS-C3 (>35 m). Assuming an average CH₄ concentration for the entire water column, we calculated the necessary sustained fluxes to raise the atmospheric concentrations by various amounts. We assumed a very shallow 200 m mixing height (very low inversions were observed during SWERUS-C3) [Tjernström et al., 2015], a 35 m water column, and an 80 ppb atmospheric CH₄ enhancement, raising atmospheric CH₄ from apparent background levels of about 1.87 ppm to 1.95 ppm. This requires a sustained CH₄ flux of about ~12 mg m⁻² d⁻¹, similar to the fluxes we observed during SWERUS in areas near subsea gas seeps on the ESAS (Table 2). An average flux near ESAS subsea seeps of 13 mg m⁻² d⁻¹ was reported previously [Sergienko et al., 2012]. Sustaining higher atmospheric CH₄ concentrations is even more difficult: to sustain 2.1 ppm CH₄ in the atmosphere (35 m water column, 200 m mixing height) would require a flux similar to a subarctic wetland (~36 mg m⁻² d⁻¹) [Bartlett and Harriss, 1993] or an order of magnitude above the average fluxes we observed in the ice-free ESAS. (Various mixing scenarios' effects on atmospheric CH₄ are shown in the supporting information Table S2.)

Surveys conducted in shallower waters, closer inshore, have reported substantially higher atmospheric CH₄, 2.97 and 2.66 ppm average in the LS and ESS, respectively, with spikes to 8.2 ppm [Semiletov et al., 2012; Shakhova et al., 2010a, 2010b, 2007]. Such average atmospheric mixing ratios of CH₄ across large expanses of the LS and ESS require sustained regional CH₄ fluxes of roughly 75–200 mg m⁻² d⁻¹, depending on local winds and atmospheric mixing. Such atmospheric enhancements are not sustainable by local diffusive CH₄ fluxes alone but require a substantial bubble contribution, due to mixing limitations and the strong salinity gradient which acts as formidable barrier to mixing and transport in the ESAS, especially in the LS [Wåhlström et al., 2012]. Using an entrainment velocity model (supporting information), we note that even with sustained gale force winds of 20 m s⁻¹, entrainment velocities in the LS are only 10 cm h⁻¹, suggesting that a week is needed to mix from 15 m depth. Storms with winds >15 m s⁻¹ occur on average only once or twice each summer in the ESAS [Proshutinsky et al., 1999], even less close to shore [Günther et al., 2015], though rare large storms have occurred [Simmonds and Rudeva, 2012]. Thus, during the ice-free sea season, generally, only the upper 10 m of water is subject to surface layer turbulent mixing. This stratification likely contributes to limiting the diffusive transport of CH₄ from deeper waters to the surface.

Winter studies in the Beaufort Sea showed no substantial sea-air CH₄ flux during the ice-covered period [Kvenvolden et al., 1993], except at ice leads [Kort et al., 2012]. If we assume conservatively that our sampled waters are open a third of the year and fluxes are relatively constant in that period, then we would expect 0.87 Tg CH₄ yr⁻¹ to be released during the ice-free season. It is possible, even likely, that larger fluxes exist at ice-out due to overwinter collection of CH₄ in or beneath the ice. Below-ice seawater CH₄ concentrations in the ice-covered portions of the ESS appear to support the idea of storage of CH₄ beneath the ice (we had no measurement of CH₄ within the sea ice during SWERUS-C3). In these ice-covered regions of the AO and ESS, we report hypothetical diffusive fluxes, we report hypothetical diffusive fluxes, which would occur if the sea ice were removed (Table 2). The ESS ice-covered segment became ice-free approximately 1 week after Oden's passage. Diffusive CH₄ fluxes at ice-out would have been substantially higher than those we calculated for the open water portion of the cruise, but such a degassing event could not be sustained for long. If we include these high below-ice CH₄ concentrations in our flux calculations, our whole-ESAS average flux estimate would become 12.2 mg m² d⁻¹ (9.4 Tg yr⁻¹ or 2.8 Tg yr⁻¹ assuming that 30% of the year is ice free); however, because this below-ice CH₄ would be rapidly degassed after ice-out and not result in a sustained flux, we find this estimate to be unrealistically high. It has been suggested that somewhat higher CH₄ emissions may occur during the subsequent month due to convection induced as sea surface temperatures again drop [Damm et al., 2015]. The ESAS begins freezing after the sea ice minimum occurs in September. Subsequent CH₄ releases may then be trapped under or in the sea ice lid until spring [Golden et al., 2007; Shadwick et al., 2011] with some loss through wintertime polynyas [Damm et al., 2007]. Better quantifying CH₄ oxidation rates in ESAS waters [Bussmann, 2013] and CH₄ behavior in ice [Damm et al., 2015] are critical for understanding the efficiency of winter CH₄ storage.

5 Conclusions

Our measurements of CH₄ in the atmosphere and surface water across the middle and outer ESAS during July and August 2014 show an average flux of CH₄ from the Siberian shelf seas to the atmosphere (along the ice-free portions of the cruise track) of 3.8 mg m⁻² d⁻¹. In a region of CH₄ seeps in the LS, fluxes reached 14 mg m⁻² d⁻¹. Enhanced levels of CH₄ were observed in below-ice waters of the ESS; such CH₄ would have to be stored for winter months and released with near-100% efficiency after late summer or early autumn ice-out, providing a short-duration increase to the total flux, to reach annual fluxes of 2.9 Tg yr⁻¹. Such short-duration fluxes at ice-out must be better quantified to constrain the total annual flux. We note that the below-ice CH₄ concentrations in the ESS were considerably below what would be expected if CH₄ was collected over the entire ice-covered months, suggesting overwinter loss processes and/or incorporation into sea ice. This, combined with a lack of knowledge of fluxes through the full ice-free season, causes us to regard our annual estimates of ESAS sea-air CH₄ fluxes (2.9 Tg yr⁻¹) as very likely high rather than low. Unquantified ice-out fluxes could increase the annual ESAS CH₄ flux above the estimated ice-free season flux of 0.87 Tg yr⁻¹. Although our estimated sea-air CH₄ fluxes for the ESAS far exceed fluxes reported for other shelf seas, they are roughly an order of magnitude below ESAS CH₄ flux estimates reported previously. Reconciling these differences requires much better knowledge of the spatial extent of ESAS CH₄ sources (including near-shore areas not accessible during SWERUS-C3 where riverine and terrestrial CH₄ sources may play a greater role), especially the seemingly highly localized bubble sources, as well as quantification of stored CH₄ released at ice-out.