Temporal Constraints on Hydrate-Controlled Methane Seepage off Svalbard

See allHide authors and affiliations

Science  17 Jan 2014:
Vol. 343, Issue 6168, pp. 284-287
DOI: 10.1126/science.1246298

What Does It All Mean?

Strong emissions of methane have recently been observed from shallow sediments in Arctic seas. Berndt et al. (p. 284, published online 2 January) present a record of methane seepage from marine sediments off the coast of Svalbard showing that such emissions have been present for at least 3000 years, the result of normal seasonal fluctuations of bottom waters. Thus, contemporary observations of strong methane venting do not necessarily mean that the clathrates that are the source of the methane are decomposing at a faster rate than in the past.


Methane hydrate is an icelike substance that is stable at high pressure and low temperature in continental margin sediments. Since the discovery of a large number of gas flares at the landward termination of the gas hydrate stability zone off Svalbard, there has been concern that warming bottom waters have started to dissociate large amounts of gas hydrate and that the resulting methane release may possibly accelerate global warming. Here, we corroborate that hydrates play a role in the observed seepage of gas, but we present evidence that seepage off Svalbard has been ongoing for at least 3000 years and that seasonal fluctuations of 1° to 2°C in the bottom-water temperature cause periodic gas hydrate formation and dissociation, which focus seepage at the observed sites.

Large quantities of methane, a powerful greenhouse gas, are present in the continental margin west off Svalbard, where they are stored as marine gas hydrate (13). Because hydrate stability is temperature-dependent, Arctic warming is a potentially major threat to both the environment and the global economy. If even a fraction of the methane contained in Arctic hydrates were released to the atmosphere, the effect on climate could be dramatic (4, 5).

Water-column temperature measurements and mooring data suggest a 1°C bottom-water temperature warming for the past 30 years (6, 7). Numerical modeling of hydrate stability predicts that such warming would result in the dissociation of hydrates in the shallowest sediments (69). Therefore, the discovery of numerous gas flares—that is, trains of gas bubbles in the water column precisely at the water depth where gas hydrate is expected to dissociate—was interpreted as the onset of submarine Arctic gas hydrate dissociation in response to global warming, which may potentially lead to large-scale escape of methane into the water column and ultimately into the atmosphere (6). In order to assess the consequences of methane venting on ocean and atmosphere composition, it is necessary to establish how the rates of methane emissions from hydrate systems change through time (10).

The margin of Svalbard (Fig. 1) can be considered a model system to study a temperature-related gas hydrate destabilization scenario, because water temperature in the Fram Strait oceanographic gateway will be more affected by changes in global atmospheric temperature than elsewhere in the Arctic; therefore, any corresponding changes to a hydrate system should be easier to observe here than elsewhere (11). The continental margin of Svalbard is characterized by abundant contourite deposits (12) that consist of fine-grained sediments with high water content, which cover most of the margin between water depths of 800 and 3000 m. It is likely that these contourites are underlain by Miocene sediments with 3% (by weight) of total organic carbon as found at Ocean Drilling Project site 909 (13) and that the emanating gas was produced by these sediments. Proximally, that is, shallower than 700- to 800-m water depth, Pleistocene and Pliocene highly heterogeneous, terrigeneous glacial deposits overlie the contourites (14, 15). In the glacial deposits, there is only limited evidence for free gas and no clear geophysical evidence for gas hydrate, such as a bottom simulating reflector. Yet, seismic evidence for gas hydrate occurrence is conclusive for the contourite deposits farther west (16), where gas hydrate has also been sampled at a vent site in ~900-m water depth (17).

Fig. 1 The Svalbard gas hydrate province is located on the western margin of the Svalbard archipelago (inset).

At water depths shallower than 398 m, numerous gas flares have been observed in the water column (color-coded dots for different surveys) by using EK60 echo sounders and high-resolution side scan sonar. The gas flares are located between the contour lines at which gas hydrate is stable in the subsurface at 3° (brown) and 2° (blue) C average bottom-water temperature. The black lines show the location of PARASOUND profiles with 40-m separation, that is, complete coverage, for flare mapping. The black arrows point to the location of submarine dives discussed in the text. The red line shows the location of the modeling transect (bold section shown in Fig. 3). The large cluster of seeps at the northern limit of the gas flare line at a water depth of 240 to 260 m can be explained by the presence of an elsewhere-absent glacial debris flow deposit that is deviating gas laterally within the prograding debris flow deposits and cannot have anything to do with gas hydrate dynamics (16, 23).CTD, conductivity, temperature, and depth.

Several oceanographic expeditions were able to corroborate the location of the gas flares discovered in 2008. During the MSM21/4 survey in August 2012, we collected a series of PARASOUND 18-kHz parametric echosounder profiles with 40-m spacing around the site of the MASOX (Monitoring Arctic Seafloor–Ocean Exchange) observatory (Fig. 1). An ~40-m footprint of the PARASOUND system at 390-m water depth allowed us to obtain a complete coverage of the flare locations within the area of this survey, which means these data were no longer biased by selection of ship tracks as in previous surveys. Our results show that the gas flares align between 380- and 400-m water depths, which corresponds to the upper limit of the gas hydrate stability zone (GHSZ) considering present-day bottom-water temperature of around 3°C (16). Geological structures that may focus gas from deeper parts of the plumbing system are absent (16). Thus, we interpret this match of gas flare origination depth and the calculated landward termination of the gas hydrate stability zone in the sediments as strong circumstantial evidence for a link between gas hydrate dynamics and gas seepage. At the gas flares, substantial amounts of methane are liberated into the water column, leading to bottom-water CH4 concentrations of up to 825 nM and a net flux of methane to the atmosphere (18).

One objective of this study was to deduce a minimum age for the onset of marine methane release from the sea floor. For detailed sampling of the gas seeps, we carried out 10 dives with the manned submersible JAGO. Our observations substantiate the presence of more than 5-m-wide and typically more than 20- to 40-cm-thick outcropping carbonate crusts at the Polarstern (246 m) and the HyBIS (385 m) (Fig. 2) sites; small carbonate nodules at the MASOX site (395 m) were found in gravity cores. We analyzed carbonates from the HyBIS and the MASOX site. The mineralogical composition of the carbonates was heterogeneous and admixed with high amounts of detrital silicates. They were characterized by low δ13C isotope values between –27.1 and –41.4 per mil (‰) Vienna pee dee belemnite (V-PDB) (17). Consequently, these carbonates can be regarded as an archive of microbially induced, methane-related authigenic precipitation processes (19). The most reliable single-age data were obtained from aragonite-dominated surface samples. U/Th isotope measurements and resulting minimum seepage age for the MASOX site imply that significant methane-related precipitation was already occurring at 3000 years before the present (yr B.P.) (18). For comparison, the derived ages for the HyBIS site are overlapping or older, for example, sample SV-2 (8200 ± 500 yr B.P.) or sample SV-3 (4600 ± 500 yr B.P.). The youngest isochron-based age of ~500 yr B.P. was deduced from carbonates that were found in sediments at the MASOX site at 40- to 50-cm depth below the sea floor. Because of changes in the path of methane-bearing fluids, inclusion of impurities, and alteration of sample material, it was not possible to decipher potential on/off stages or chemical variation of the seeping fluids beyond the results presented in this paper. Hence, it is possible that seepage strength and transport of methane from the sediment to the water column and atmosphere varied over time.

Fig. 2 Photograph of the massive authigenic carbonate crusts observed at the HyBIS site in 385-m water depth.

For scale, the total length of the larger white sessile ascidia (white stalklike animal on the crest of the uplifted carbonate plate) is about 15 cm. Carbonate crusts such as these take at least several hundred years to develop through anaerobic oxidation of methane.

We propose that carbonate formation in this area continues until today, because surface sediments (0- to 10-cm depth below sea floor) at gas vents at both the HyBIS and the MASOX sites were characterized by high rates of anaerobic oxidation of methane (AOM; maximum of 11.3 μmol CH4 cm−3 day−1), which is the driver for carbonate precipitation at methane seeps (19). AOM correlated with high concentrations of methane (max 14,800 nM), sulfide (maximum of 11,000 nM), and total alkalinity (maximum of 29 meq l−1) in the sediment. Chemosynthetic communities (sulfur bacteria mats and frenulate tubeworms) were present at both sites (18).

Observations of old carbonate crusts imply that seepage must have been ongoing at all three sites for more than 500 years. Detailed paleoceanographic reconstructions for the Svalbard area (11) show a pronounced warming since the end of the 19th century. However, even this 100-year time span seems too short to explain the observed thicknesses. The ages of the recovered carbonate crusts, which are all significantly older than 100 years, support this conclusion. Thus, it is unlikely that an anthropogenic decadal-scale bottom-water temperature rise is the primary reason for the origin of the observed gas flares, although this temperature rise may contribute to keeping gas pathways open longer and further.

During the cruise, we recovered the MASOX observatory, which had been deployed twice for a total of 22 months within a cluster of flares between 390- and 400-m water depth. The observatory contained a bottom-water temperature sensor sampling every 15 min during both deployments. The recorded time series revealed fluctuations of bottom-water temperature between 0.6° and 4.9°C, with lowest temperatures between April and June and highest temperatures between November and March (Fig. 3). In both years, the temperature difference between spring and fall/winter was around 1.5°C, but during the second year the average bottom-water temperature was generally about 0.5° higher than that recorded during the first deployment. The time series implies that there is a strong seasonal change of sea floor temperature.

Fig. 3 Temperature and the GHSZ.

(Top) Daily means of bottom-water temperature recorded by the MASOX observatory. The times when the extent of the GHSZ was at its maximum and minimum are marked by solid red and dashed blue lines, respectively. (Bottom) The seasonal dynamics of the GHSZ. Driven by changes in bottom-water temperature, the GHSZ advances and retreats in the course of the year. The solid red lines and the dashed blue lines indicate maximum and minimum extents of the GHSZ, respectively. The area in which gas hydrates are stable in the long-term is shaded in yellow. The difference between maximum and minimum extents of the hydrate stability zone is shaded in orange and corresponds to the seasonal GHSZ, in which gas hydrate dissociation and formation alternate periodically. The triangles filled in magenta represent the projected locations of all flares detected within 1000 m of the transect line. The green diamond shows the position of the MASOX observatory. An animated illustration of the modeling results is provided in the supplementary materials.

In order to obtain better constraints on the heat exchange between the sediment and the bottom water, we conducted in situ sediment temperature and thermal conductivity measurements by using a 6-m-long heat flow probe along transects down the slope. Between 500- and 360-m water depths, our measurements revealed a landward increase in thermal conductivity from 1.5 to 2.6 W m−1 K−1, with a maximum around the position of the MASOX observatory. High sediment thermal conductivity, large temporal variability in bottom-water temperature, and possibly formation and dissociation of gas hydrates resulted in very irregular sediment temperature profiles, which made it difficult to determine the heat flow along the transect line from our data. On the basis of our measurements at 500-m water depth, we estimate the regional heat flow to be around 0.05 W m−2. Given the comprehensive evidence for seepage, this value is likely modulated by convective heat transport.

On the basis of the recorded bottom-water temperature time series and the acquired thermal conductivity data, we developed a two-dimensional model of the evolution of the GHSZ along the transect line. As illustrated in Fig. 3, the seasonal changes in bottom water temperature are accompanied by large lateral shifts of the GHSZ at least within the top 5 m of surface sediments. During the cycle of a year in which bottom-water temperature varies as observed in 2011 and 2012, the volume of the GHSZ varied between a maximum value in summer and a minimum value in winter. During the time period covered by our measurements, the GHSZ was at its maximum in June 2011, when it extended to a 360-m water depth. Increasing bottom-water temperatures from June until December were accompanied by a retreat of the GHSZ at the seafloor to more than 410-m water depth. In the subsurface, the GHSZ retreated further until it reached its minimum in March 2012.

The modeling shows that persistent supply of dissolved methane from below the GHSZ in this section of the slope would lead to the formation of hydrate from winter until summer. The newly formed hydrate would dissociate again during the second half of the year and thus augment methane emissions from the seabed both by opening pathways to gas ascending from underneath and by releasing gas from the hydrate phase. The total volume of sediment that was affected by seasonal shifts of the GHSZ amounted to between 3000 and 5000 m3 per meter of the margin. Assuming a gas hydrate concentration of 5% of the pore space and a porosity of 50%, the seasonal GHSZ has the potential to periodically store and release between 9 and 15 tons of CH4 per meter of the margin. However, these amounts represent the upper limits of the seasonal buffering capacity, because the latent heat of hydrate kinetics was not included in the simulation. Depending on the concentration and distribution of gas hydrates in the sediment, alternating formation and dissociation would dampen the oscillation of the GHSZ and thus reduce its volume.

Although the modeling shows that seasonal bottom-water temperature variations are capable of modulating the observed gas emissions, we found no direct evidence in the heat flow data that would suggest that the slope sediments experienced decadal-scale warming. The combined data demonstrate that hydrate is playing a fundamental role in modulating gas seeps between 380- and 400-m water depth at the upper limit of the GHSZ, whereas ascending gas would be trapped or deviated up along the base of the GHSZ further seaward. Long-term variations in seepage may exist, but presently available data are insufficient to document annual, decadal, or centennial changes in seepage. Our data suggest that shallow hydrate accumulations are sensitive to bottom-water temperature changes and therefore that significant anthropogenic warming will affect the shallow parts of the hydrate system. This sensitivity demonstrates the need for quantifying the total amount of gas hydrate in the shallowest part of the gas hydrate stability zone if climate feedback mechanisms are to be assessed beyond simple global models (20, 21). Our observations show that methane seepage west off Svalbard has been ongoing for much longer than anthropogenic warming. Therefore, observations of large contemporary emissions reported in other studies cannot be considered proof of accelerating hydrate destabilization, although neither do they prove that catastrophic destabilization is not accelerating.

Supplementary Materials

Materials and Methods

Fig. S1 to S4

Tables S1 and S2

Movie S1

References (2444)

References and Notes

  1. See online supplementary materials.
  2. Acknowledgments: This manuscript is dedicated to the memory of our beloved colleague and friend Victoria Bertics. We are grateful to K. Bergmann and the officers and crew of R/V Maria S. Merian for their help at sea. The German Research Foundation (DFG), the Swiss National Science Foundation, and the Cluster of Excellence “The Future Ocean” supported the project financially. Further support came from the ESONET project (European Seas Observatory NETwork), the PERGAMON project (European Cooperation in Science and Technology), and the Alexander von Humboldt Foundation. Figure 1 was drafted by using Generic Mapping Tools (22). Supplementary data are available at
View Abstract

Navigate This Article