The seepage of methane (CH4) through the seabed of the world's shelves is considered to be ubiquitous. Although numerous observations of methane seepages from shallow marine sources have been reported, there were only a very few observations made over the Arctic shelves and none for the East-Siberian Sea (ESS) or Laptev Sea (LS). We present two years of data obtained during the late summer period (September 2003 and September 2004) for both the ESS and LS shelves. According to our data, the surface layer of shelf water was supersaturated up to 2500% relative to the present average atmospheric methane content of 1.85 ppm. Anomalously high concentrations (up to 154 nM or 4400% supersaturation) of dissolved methane in the bottom layer of shelf water suggest that the bottom layer is somehow affected by near-bottom sources. Considering the possible formation mechanisms of such plumes, we favor thermo-abrasion and the effects of shallow gas or gas hydrates release.
 Methane is one of the most effective greenhouse gases and is subject to considerable global variations; its marine sources and sinks have yet to be fully quantified [Intergovermental Panel on Climate Change (IPCC), 2001]. The atmospheric methane budget presently stands at an estimated overall input flux of 560 Tg yr−1(1 Tg = 1012 g) [IPCC, 2001]. The contribution from all marine sources is assumed to range between 5 and 20 Tg CH4 yr−1.
 Sub-seabed methane on continental shelves is produced primarily by bacterial and thermogenic processes. It is known that seafloor perturbations of temperature and pressure are resulting in sub-sea methane accumulation in the form of methane gas hydrates. The shelf and continental slope methane hydrates reservoir is estimated to be roughly 6 × 1018 g or 6000 Gt [Makagon, 1982]. Recent research points to regional releases of CH4 from gas hydrates in ocean sediments during the last 60,000 years [Kennett et al., 2003]. Doubling of the atmospheric CH4 from present conditions requires release of less than 0.1% of the sub-sea permafrost hydrate reservoir (<4 × 1015 g CH4). That is why the Arctic regions of offshore permafrost seem to be critical to the problem of increasing methane emission from gas hydrates dissociation [Kvenvolden, 1993].
 The other but usually omitted source of methane from continental shelves is thawing submarine permafrost itself. Studies by Rogers and Morack  on sub-sea permafrost and sea level history lead to the inference that offshore permafrost may persist beneath any part of the Arctic shelves inshore from about the 90-m isobath. Permafrost contains a huge amount of ancient organic matter that might be involved in current biogeochemical cycling due thawing of the upper permafrost and restoring the activity of viable methanogens (bacteria that produce CH4 as a metabolic product) preserved in permafrost [Rivkina et al., 1998].
2. Study Area and Methods
 The extensive Russian Arctic shelves play an especially important role because of their large area (more than 3 × 106 km2) and usually shallow sea depth (70% of their total area is less than 40 m deep). The hydrology and sedimentation of the Russian Arctic shallow shelf is determined mainly by Arctic river (Lena, Yana, Indigirka, Kolyma) fluvial sediment discharge, coastal sediment input and sub-sea permafrost ‘bottom thermo-abrasion’ [Semiletov, 1999; Romanovskii et al., 2000; Stein and Macdonald, 2003].
 Our studies were performed in the nearshore open water zone between the coast and drifting ice of the ESS and LS, between 132°E–179°E and 69°N–74.5°N. In total, 44 oceanographic stations were conducted during September 2003 and 118 stations during September 2004. Water samples from different horizons were collected on each station with Niskin bottles. Bottom sediment temperature was obtained using four thermal sensors installed 50 cm apart along the 2 m steel rod. Water samples were analyzed for methane with a MicroTech-8160 gas chromatograph equipped with a flame ionization detector. The headspace technique for equilibrating between the dissolved and gaseous phases was applied [Semiletov et. al., 1996]. The concentration of dissolved methane in the water samples was calculated with the Bunses adsorption coefficient for methane [Wiesenburg and Guinasso, 1979] at the appropriate equilibration temperature. Methane flux estimates were done following Wanninkhof .
3.1. Dissolved Methane Distribution and Depth Profiles
 The concentration of dissolved methane in the surface layer ranged from 2.1 nM to 28.2 nM in 2003 and reached 110 nM in plume areas in 2004. The average surface concentrations for 2003 and 2004 were 10.5 nM and 13.5 nM respectively. This represents supersaturation of the surface layer up to 800% in 2003 and up to 2500% in 2004 relative to the present average atmospheric methane content of 1.85 ppm (http://www.cmdl.noaa.gov/ccgg/insitu.html). Locations of plume areas correspond with the input of the Indigirka and Kolyma rivers (spots 1, 2, 3, 4, 5, Figures 1a and 1b; spots 2, 4, Figures 2a and 2b); these rivers are sources of dissolved methane which comes from Kolyma-Indigirka Lowland permafrost degradation through numerous channels from “taliks” beneath lakes [Semiletov et al., 1996; Zimov et al., 1997]. The bottom concentrations of dissolved methane reached anomalously high values of 87 nM in 2003 and 154 nM in 2004, while averaging 14.4 and 25.5 nM respectively. The bottom plume areas were enriched with dissolved methane up to 4400%, suggesting that the deeper part of the methane plume is somehow affected by near-bottom sources. One of the bottom plumes (spot 3, Figure 2b) was recorded in the vicinity of the Dmitry Laptev Strait (140°E and 73°N); another was located northwest of Chaunskaya Bay (165°E and 175°N) (spot 6, Figure 1b). Intensity of both methane plumes is comparable to reported magnitudes of plumes resulting from shallow gas hydrates decay in the Barents Sea and the Sea of Okhotsk [Lammers et al., 1995; Obzhirov, 2002].
3.2. Methane Flux Estimate
 The methane equilibrium concentration for the LS and the ESS shelves during stable summer conditions is 3.5 nM at 20‰ salinity and 6.2°C and an assumed atmospheric partial pressure of methane of 1.85 ppm. Significant supersaturation of surface layers suggests that this part of the Arctic Ocean is a net source of atmospheric methane. The rate of gas exchange between the ocean surface and the atmosphere (F) is described by Wanninkhof  as a function of the difference between CH4 concentration in the surface water and the concentration of CH4 in the air (ΔC), specific gas properties (Schmidt number, Sc), water temperature (t), and short-term records of the wind velocity (υ): F = 0.31 υ2 · (Sc/660)−0.5 · ΔC. After Wanninkhof , Sc is defined as a function of temperature: Sc = 2039.2 − 120.31 · t + 3.4209 · t2 − 0.040437 · t3.
 The value of the Schmidt number for methane dissolved in seawater at 6°C is 1432. The average regional wind speed for September used in the calculation was 6 m s−1 [Proshutinsky et al., 1994]. The flux from the surface water of the surveyed area, according to the concentration difference of ΔC = 10.0 nM, is about 11.9 × 10−10 g CH4 cm−2 h−1. Taking into account that the period of ice-free water in the study area lasts not more than 90 days (from middle July to middle October), we evaluated the summertime flux to be about 2.6 × 104 g CH4 km−2 on average. For plume areas the summertime flux can reach values of 7.23 × 104 g CH4 km−2 to 39.0 × 104 g CH4 km−2. We can assume that methane flux from the seafloor is at least of the same order of magnitude, and over the year it may reach 1–1.5 g CH4 m−2 yr−1 in plume areas.
3.3. Area Methane Storage and Potential Methane Emission
 Potential methane emission is the area-adjusted maximum of dissolved methane available for release to the atmosphere. The estimates of methane storage (A) in 2003 and 2004 for the comparison area (shown in blue in Figure 3) were calculated as
where s = (x, y), z are the horizontal and vertical coordinates, H(s) is the local depth and A(s, z) is the spatial distribution of the dissolved methane concentration. A(s, z) was obtained by vertical and horizontal linear interpolation between available data points. The final estimate of comparison area storage of dissolved methane was found to be 5.7 × 109 g CH4 in 2003 and 1.57 × 109 g CH4 in 2004. Methane excess or potential emission (Ep) was calculated as: Ep = A − Ae, where Ae is methane storage if the area is at equilibrium with the atmosphere. Ae = 0.58 × 109 g CH4 if we proceed on the assumption that 1.85 ppm is atmospheric CH4 by volume and 3.5 nM is equilibrated dissolved methane concentration.
 The calculated late-summer potential emission decreased more than five times from 5.12 × 109 g CH4 in 2003 to 0.99 × 109 g CH4 in 2004. This increase may be associated with changing hydrological, meteorological and geological conditions from year to year.
3.4. Thermal Sediment Measurements
 No thermal sediment studies have been done over the shallow East-Siberian shelf before. Our first in situ measurements of bottom sediment demonstrated positive temperatures in the Dmitry Laptev Strait of up to 3°C in the top 1 m sediment layer, whereas across the East Siberian shelf the temperature of marine sediments ranged from negative values (down to −1°C) to positive values (up to 2–3°C). Using original and historical measurement of seawater temperature in all seasons (about 2–3°C in summer and near −1°C in winter) we calculated the mean annual sea floor temperature in the Dmitry Laptev Strait area as equal to 1°C. We associate formation of such warm local conditions with the heating plume of the Lena River. We suggest offshore permafrost in this area could begin thawing due to the effect of energy input from the sea floor surface.
 Generation, removal and transport have to be considered as the general controlling parameters for the distribution of methane in the water column. The most probable mechanism for transport of the methane to the surface water is bubbling [Brooks, 1979]. The proportion of methane bubbles that survives passing through the water column to reach the atmosphere generally depends on water depth and bubble size. It was shown by Semiletov et al.  that from a depth of 10 m, bubbles with radius of 0.1 cm reach the air-water interface in 62 s losing only 15.1% of their methane. In our study we analyzed dissolved methane depth profiles for each station. As surface concentration was on average only ≤30% lower than bottom ones, we can conclude that ebullition may play a significant role in enrichment of surface water with dissolved methane and in methane emission to the atmosphere. Accordingly, this would imply that about 30% of methane is recycled within the water column.
 As the rate of gas exchange between the ocean surface and the atmosphere is a function of wind speed, the changing of the relatively stable summer situation on the Artic shelf to fall storm conditions appears to be a possible trigger mechanism to increase the methane flux [Wanninkhof, 1992]. When the wind velocity increases to 15 m/s or more [Proshutinsky et al., 1994], such a velocity can lead to a 6–7 times higher flux between water and air. For the shallow Arctic shelf, fall convection is particularly important in late September – early October (freeze-up period) when the probability of convection penetrating down to the seafloor can reach 40–50% [Kulakov et al., 2003]. Thus, we can assume that a significant part of the stored methane could release into the atmosphere abruptly during the short freeze-up period.
Hovland et al.  estimated the global seepage of methane from continental shelf sediments at values between 8 Tg CH4 yr−1 and 65 Tg CH4 yr−1 applying the mean seepage rates 4.7 g CH4 m−2 yr−1 (low seepage potential) and 38.3 g CH4 m−2 yr−1 (high seepage potential), respectively, to the adjusted seepage areas. It was assumed that the flux from areas of low seepage potential is zero. Our data from the Arctic shelves suggests another point of view on the area of low seepage rates. According to our calculations methane flux from the sea floor in the study area ranged between 0.4 g CH4 m−2 yr−1 and 1.5 g CH4 m−2 yr−1, below the lowest rate classification of Hovland et al. . But applying our low flux numbers to the total Arctic shallow shelf area (3 × 106 km−2) we obtain estimates from 1.0 to 4.5 Tg CH4 yr−1. Following some authors this range almost reaches the lowest estimate of whole World Ocean methane contribution [Ehhalt and Schmidt, 1978] and exceeds up to four times the annual flux estimated for all coastal seas [Cynar and Yayanos, 1993].
 The near-shore system of the LS and ESS is strongly affected by global warming, and exhibits the highest range of coastal erosion compared to other near-shore systems across the world [Stein and Macdonald, 2003]. We assume three main sources of methane from the erosion processes to form its background levels: methane trapped in coastal ice-complexes; methane trapped within the sub-sea permafrost; and modern methane, originating primarily from eroded carbon in sub-marine sediments, as it was shown that eroded carbon is bio-available [Semiletov, 1999; Guo et al., 2004]. Considering the possible formation mechanisms of plumes, we favor bottom thermo-erosion and the effects of shallow gas or gas hydrates. Following Kennett et al.  and their Clathrate Gun Hypothesis, we suggest that instability of the methane hydrates reservoir has played a critical role in Late Quaternary climate change. This instability stems from relatively large and frequent oscillations in temperature of waters on upper continental margins and glacial-eustatic sea regression which has taken place on the shelf of the LS and ESS. For instance, the shallow bottom sediment (and underlying permafrost) in the study area has been warmed about 15°C after flooding before, prior to and during the Holocene optimum (about 6–8 kyr ago [Romanovskii and Hubberten, 2001]). The shoreline moved southward in the LS by 300–400 km and in the ESS by 800–900 km. As a result, during the Holocene the Arctic shelf was covered by numerous thaw lakes which later transformed into “thermokarst lagoons”. Submarine lake taliks, affected by processes of “bottom thermo-erosion” may be completely penetrated allowing methane release from disturbed gas hydrates [Romanovskii et al., 2000].
5. Conclusions and Outlook
 The ESS and LS represent two of the broadest and shallowest shelves in the World Ocean. The observed distribution of dissolved methane and possible mechanisms of methane release suggest that both these shelves are important natural sources of methane to the atmosphere. The extreme methane anomalies in plume areas indicate the presence of bottom methane sources, which might be strongly affected by geological, hydrological and climatic factors. This work points out the great need for data from long-term investigations of this hitherto insufficiently studied region. In comparison to other global sources of methane the Arctic Ocean is thought to be a small source to the global atmosphere, but it does not seem to be too small for not being able to participate in answering the still open question about the atmospheric methane maximum located over the Artic (http://www.noaa.gov).
 We thank Valentin Sergienko, Syun Akasofu, Gueorgui Golitsyn and Candace O'Connor for comments on an earlier draft of this manuscript. This work was supported by the Far-Eastern Branch of the Russian Academy of Sciences, RAS (through the RAS Program #13, Direction #7: “Environmental changes in the East-Siberian region under climate effects and catastrophic processes”), the International Arctic Research Center of the University Alaska Fairbanks, and the Russian Foundation for Basic Research.