By comparison of the methane mixing ratio and the carbon isotope ratio (δ13CCH4) in Arctic air with regional background, the incremental input of CH4 in an air parcel and the source δ13CCH4 signature can be determined. Using this technique the bulk Arctic CH4 source signature of air arriving at Spitsbergen in late summer 2008 and 2009 was found to be −68‰, indicative of the dominance of a biogenic CH4 source. This is close to the source signature of CH4 emissions from boreal wetlands. In spring, when wetland was frozen, the CH4 source signature was more enriched in 13C at −53 ± 6‰ with air mass back trajectories indicating a large influence from gas field emissions in the Ob River region. Emissions of CH4 to the water column from the seabed on the Spitsbergen continental slope are occurring but none has yet been detected reaching the atmosphere. The measurements illustrate the significance of wetland emissions. Potentially, these may respond quickly and powerfully to meteorological variations and to sustained climate warming.
2. Isotopic Signature of Sources of Methane to the Arctic
 Arctic CH4 sources can be identified by isotopic signature (Table 1). In the Eurasian Arctic, emissions from wetlands sampled in ambient air give a characteristic and consistent isotopic signature with δ13CCH4 in the range −69 to −65‰. Emissions occur from May melt to October freeze-up, and may show an exponential (Arrhenius) increase with temperature [Nisbet, 1989; Westermann and Ahring, 1987]. Keeling plot experiments by our group reported elsewhere (S. Sriskantharajah, manuscript in preparation, 2011) find a relatively constant source δ13CCH4 −68.5 ± 0.7‰ for wetland emissions from Northern Finland during the summer. Warming permafrost and thermokarst can emit CH4, made by methanogens from organic matter. The isotopic composition of CH4 from thermokarst lakes in E. Siberia ranges widely, from −83‰ to −58‰ [Walter et al., 2006]. Fire is another CH4 source likely to increase with warming and drought. With warm dry summers, such as in Russia in 2010, tundra fires may become more frequent [Qiu, 2009]. We report here new results of source studies that have been carried out to identify the isotopic source signature of CH4 from forest fires in Canada. Source signatures were identified using Keeling plots of CH4 measured in ambient air collected close to prescribed burns. The mean isotopic composition of CH4 emitted by a boreal jack pine forest fire in Ontario, Canada (46°47′N, 83°20′W) in May 2007 was −27.8 ± 0.3‰. CH4 from a mature pine fire in Northwest Territories, Canada (61°21′N, 117°40′W) in June 2010 had a mean isotopic composition of −28.7 ± 0.7‰. These results fall within the range measured for pine forest fire emissions in the United States (−30 to −21‰) [Chanton et al., 2000]. Fire-sourced CH4, δ13CCH4 around −28‰ for northern pine forests, is isotopically very distinct from wetland CH4. Increased burning, also fingerprinted by elevated CO, should enrich CH4 in 13C.
Table 1. δ13C in Northern Methane Sources
Wetland, N. Finland: summer spring thaw autumn freeze-up
 Industrial gas leaks are a major component of global anthropogenic CH4 emissions, with many of the world's largest gas fields being north of the Arctic circle. The giant gas fields around the Ob estuary supply much of Europe's heat and electric power via long pipelines, and Arctic gas leaks may have contributed in part to the rapid rise in the global CH4 burden in the 1980s [Dlugokencky et al., 1994]. Gas can escape to the atmosphere both from leaks in wells and pipes, and from normal technical operations [Reshetnikov et al., 2000]. CH4 from the W. Siberian fields varies isotopically. W Siberian gas measured by the University of Heidelberg [Nisbet, 2001] and leaked gas in average ambient air in the production region around Korotchaevo tower in the Ob River region measured by our group [Nisbet, 2005] has an isotopic signature of −51 ± 2‰, which is comparable to other measurements of W Siberian production gas [Cramer et al., 1999].
 CH4 emitted by decomposing Arctic gas hydrates is also isotopically variable, depending on the original source of the CH4 trapped in hydrate, or free gas pooled below it, and also on fractionation that may occur on release to the atmosphere. Hydrate is a capacitor, trapping CH4, not a source per se. Thermogenic gas from deep geological carbon stores is relatively rich in 13C. In contrast, gas sourced from biological methanogenesis at shallow levels is richer in 12C. Measurements of CH4 from decomposing CH4 hydrate show a wide isotopic range: hydrate gas with δ13CCH4 −62 to −56‰ has previously been measured in the Norwegian Arctic [Milkov, 2005]; and ∼−72 to −66‰ in the gas in the Nyegga pockmarks [Vaular et al., 2010]. Extensive Arctic CH4 hydrates are vulnerable to decomposition with warming [Nisbet, 1989], if shallow submarine and terrestrial clathrates destabilise. There is evidence for strong CH4 ebullition to the atmosphere in response to the warming of shallow offshore sediments of the East Siberian Arctic Shelf in the far eastern Arctic [Shakhova et al., 2010a, 2010b]. In deeper water, CH4 bubble plumes have recently been identified from the shelf edge west of Spitsbergen [Westbrook et al., 2009].
 We report here (Table 2) δ13CCH4 source signatures of −55 ± 2‰ and −46 ± 3‰ in samples of CH4 hydrate in two sediment cores collected from within the gas hydrate stability zone, during a cruise along the continental slope NW of Spitsbergen in 2008 [Westbrook et al., 2009]. CH4 in any bubbles that escape from the sea-surface to the atmosphere may be further enriched in 13C following partial oxidation of gas in the water column or sediments, because methanotrophs would preferentially consume the lighter carbon isotope. Isotopic fractionation of CH4 in the water column has been observed in the nearby Spitsbergen continental shelf [Damm et al., 2005].
Table 2. Hydrate Samples Collected in the Fram Strait in September 2008a
Seawater Depth (m)
Depth of Hydrate Below Sediment Surface (cm)
Methane δ13C was measured in 3 chips from each sample of hydrate.
Plume field 78°41.07′N, 08°16.36′E
−54.6 ± 1.7
Vestnesa ridge 79°00.39′N, 06°54.26′E
−45.7 ± 2.7
3. Experimental Methods
 To identify major emissions of Arctic CH4 and assess their relative inputs to the Arctic CH4 budget, ambient air samples were collected daily at the Zeppelin station, Spitsbergen, during late summer/autumn 2008, spring 2009 and late summer/autumn 2009. Zeppelin (78°54′N, 11°53′E), at 475 m above sea level is a representative background site normally above the planetary boundary layer with limited influence from local CH4 sources [Pedersen et al., 2005]. Simultaneously with the summer 2008 campaign, in the nearby Fram Strait, off W. Spitsbergen, intensive air sampling was carried out on the bridge of RRS James Clark Ross. Shipboard sampling was repeated during two cruises with RV Jan Mayen over the gas plume field in the Fram Strait in July 2009 and October 2010.
 Air was collected by pumping into 5 litre tedlar bags. Mixing ratio and δ13C of atmospheric CH4 was measured in all air samples in the Dept. of Earth Sciences at Royal Holloway University of London. For air samples collected at Zeppelin in Autumn 2009 and shipboard samples from 2010 CH4 mixing ratios were measured using a Picarro Cavity Ringdown spectrometer (CRDS) with a repeatability of ±0.3 ppb. For all other air samples CH4 mixing ratios were determined using an HP 5890 gas chromatograph (GC) with a flame ionization detector (FID) with a repeatability of ±5 ppb. The GC and CRDS instruments are calibrated for CH4 in the range 1831 to 1965 ppb using NOAA air standards and all CH4 mixing ratio data are given on the NOAA04 scale [Dlugokencky et al., 2005]. CH4δ13C was analysed using a modified gas chromatography isotope ratio mass spectrometry (GC-IRMS) system (Trace Gas and Isoprime mass spectrometer, Isoprime Ltd.) with 0.05‰ repeatability [Fisher et al., 2006]. All measurements were made in triplicate. Isotope ratios are given in δ-notation on the VPDB (Vienna Pee Dee Belemnite) scale.
4. Results and Discussion
 During the summer 2008 cruise, more than 250 plumes of gas bubbles were identified [Westbrook et al., 2009] using a 38 kHz sonar. Plumes were identified in the same area in 2009 (Figure 1). In seawater samples collected in 2008 dissolved CH4 concentrations were up to 20 times greater at the sea bottom than in the surface waters above the plumes. Some bubble plumes rose close to the sea surface. Despite the proximity of the plumes beneath the ship's track, air collected on the ship did not contain elevated CH4 compared with contemporaneous samples from Zeppelin (Figure 2), which is located 70 km NW of the observed plume field. Nor were the shipboard results significantly different isotopically from Zeppelin samples. Probably, complete CH4 dissolution and gas stripping [McGinnis et al., 2006; Rehder et al., 2009] or methanotrophic oxidation occur as the bubbles rise. Oceanic CH4 and oxygen measurements following the 2010 Deepwater Horizon oil spill in the Gulf of Mexico [Kessler et al., 2011] showed that rapid methanotrophic oxidation of released CH4 occurred in the water column. Our results show that seabed CH4 emissions from much shallower waters, though demonstrably present in the sonar records and water column measurements, did not reach the atmosphere.
 Bulk CH4 inputs to Arctic air may be assessed by sampling air masses from a range of directions, using high precision δ13CCH4 measurement, if isotopic signatures of regional sources are known. The y-intercept of a Keeling plot of δ13CCH4 against 1/CH4 mixing ratio in the shipboard and Zeppelin air samples during the period 23 August to 20 September 2008 gives the bulk source input of δ13CCH4 −67.5 ± 1.1‰. This bulk result is strikingly similar to typical wetland emissions, though with the caution that it may record a mix of disparate sources. Additional shipboard campaigns in summer and autumn 2009 and autumn 2010 produced similar results, with CH4 mixing ratios measured in the gas plume area no higher than those measured at the Zeppelin station. The source signature of CH4 in these samples was δ13CCH4 −68.6 ± 4.5‰ in July 2009 and −68.7 ± 4.4‰ in October 2010. CH4 in air sampled daily at the Zeppelin station in September to October 2009 had a δ13CCH4 source signature of −67.4 ± 3.1‰ (Table 3). Destruction of CH4 by OH is small in the Arctic where OH concentrations are low so there is little isotopic enrichment by reaction with OH over the periods considered. The large observed variations in CH4 concentration are related to meteorological dynamics in the surface boundary layer with a timescale much shorter than the lifetime of tropospheric CH4 and the bulk isotopic signature reflects the CH4 source mix.
Table 3. Methane Source Signatures Calculated From Daily or Twice Daily Sampling at the Zeppelin Station (Ny-Ålesund, Spitsbergen) and in the Fram Straita
The source signatures were given by the y-axis intercept of a Keeling plot of the data from each period listed, using a geometric mean regression.
14th August–14th October 2008
−68.7 ± 2.4
6th March–9th May 2009
−52.6 ± 6.4
5th September–4th October 2009
−67.4 ± 3.1
23rd August–20th September 2008
−66.9 ± 1.3
21st July–26th July 2009
−68.6 ± 4.5
9th October–28th October 2010
−68.7 ± 4.4
 The measured bulk input signature at Spitsbergen of ∼−68‰ for 2008–10 shows the late summer CH4 input is dominantly from δ13C depleted sources such as wetland (δ13CCH4 typically −69‰), or thermokarst (−83 to −58‰) (Table 1). If the mean biogenic source signature is −69 or −70‰, then a simple mass balance equation implies no more than 10% of the −68‰ Arctic summer input to Spitsbergen air can be from isotopically heavier inputs such as gas leaks (circa −51‰), fires, or local clathrate identified on the W Spitsbergen slope (−55 to −46‰).
 Any isotopically heavy CH4 input from boreal fires was small as no large CO excursions were observed during the period. CO mixing ratio measured semi-continuously at the Zeppelin station by the Norwegian Institute for Air Research (NILU) using an in situ RGA3 (Reduction Gas Analyser, Trace Analytical) was compared with the CH4 mixing ratio measured in the bags collected at the Zeppelin station at the same time. There is a weak correlation (R2 = 0.49) between excess CO and CH4 at Zeppelin during the period 09 to 20 September 2008, with a CO:CH4 ratio of 0.3 ppb(CO)/ppb(CH4). This is much smaller than the ratio expected from boreal forest fires which would produce CO and CH4 with a CO:CH4 ratio of between 12 and 38 ppb/ppb [Cofer et al., 1998]. Given the proximity of giant Russian Ob River gasfields, which are known to emit at least some leakage in summer [Reshetnikov et al., 2000], some part of the isotopically heavier contribution to the bulk mix must come from gas fields.
 In contributing to the δ13CCH4 −68‰ signature of the bulk CH4 increment measured at Zeppelin, the total local clathrate input during the sampling period is likely to have been very small. Note however this does not exclude inputs from isotopically lighter shallow-sourced CH4 emitted from clathrates further afield such as in the eastern Arctic. Our results from the high Arctic are consistent with the finding in 1999 in mid-latitude boreal wetland along the Trans-Siberian railroad and Ob river, that δ13Csource = −62.9 ± 0.7‰, which suggested CH4 from wetlands dominated the substantial CH4 excess in western Siberia [Tarasova et al., 2006].
 In spring, when Arctic wetland is still frozen, the CH4 in air samples is significantly more enriched in 13C (Figure 3). In March to May, the bulk Arctic source signature, calculated from measurements in air samples collected daily at the Zeppelin station, gives an Arctic springtime source with δ13CCH4 −52.6 ± 6.4‰. This difference from summer is consistent with observations in the Hudson Bay Lowlands where boreal wetland has a sharp seasonal onset of emissions in June and a seasonal switch-off in September [Pickett-Heaps et al., 2010]. Most likely the dominant Arctic sources in winter and spring are gas field emissions. Some (aseasonal) clathrate-derived contribution to this relatively heavy springtime signature is also possible, but given the summer result, local clathrate sources are likely to be small. More remote clathrate sources may contribute if flaw polynas (ice free regions) allow release of CH4 from shallow clathrate to the Arctic atmosphere during the winter. However, in this spring study very few air masses arrived at Zeppelin with trajectories from known emission areas on the East Siberian Arctic Shelf region [Shakhova et al., 2010a, 2010b].
 Footprint emission sensitivities for air arriving at Zeppelin and at the ship were calculated using the Lagrangian particle dispersion model, FLEXPART [Stohl et al., 1998]. The highest CH4 mixing ratios were measured in air masses from Siberia. The atmospheric transport on a twenty-day timescale shows that the Zeppelin observatory is highly sensitive to surface emissions in the Arctic and particularly to emissions in high latitude Eurasia in winter [Hirdman et al., 2010]. Emissions that led to the highest observed CH4 mixing ratios in the summer 2008 study period (04–08 September) were from northern Siberia, crossing a large area from the Ob River to Eastern Siberia, including the East Siberian Arctic Shelf. The mean source signature, δ13CCH4, for these specific days was −65 ± 3‰. Assuming only wetland (−69‰) and gas leaks (−51‰) are responsible for this CH4, then this signature may be apportioned to 78% wetlands and 22% gas leak, but there may also be significant input from East Siberian Arctic Shelf emissions [Shakhova et al., 2010b] with variable δ13C. Air from Canada and Greenland also had high CH4 content (19th to 20th September) with an isotopic source signature of −69‰, indicating Canadian wetland sources dominated [Kuhlmann et al., 1998].
5. Summary and Conclusions
 The results imply the dominant Arctic summer CH4 source in 2008 and 2009 was biogenic, from wetland. This is consistent with evidence from Siberia of the importance of wetland CH4 [Tarasova et al., 2009]. In winter, gas emissions dominate the CH4 input. Submarine emissions along the West Spitsbergen slope currently input negligible CH4 to the air in summer, despite the clear evidence for gas plumes in the water column. However, this could change rapidly if a warming Atlantic warms the West Spitsbergen current [Westbrook et al., 2009]. Gas hydrates are widespread in thick sediments in the Fram Strait between Spitsbergen and Greenland. If the sea bottom warms, the gas hydrate stability zone will move further down the continental slope. Given the steep slopes, earthquakes triggered by ice-melt unloading could produce submarine landslides, triggering further emissions [Berndt et al., 2009].
 Wetland CH4 emissions respond rapidly to warming, such that the warming can feed the warming [Nisbet and Ingham, 1995], as evidenced by their importance in glacial terminations [Nisbet and Chappellaz, 2009]. In particular, Arctic and boreal wetlands are likely to respond immediately to sustained heatwaves and increases in precipitation. Fire CH4 is also more likely with elevated temperatures. There is a strong need for more regular CH4 isotopic measurements in the high Arctic and intercomparison with the isotopic data from flasks currently collected at Barrow [Miller et al., 2002], Alert [Dlugokencky et al., 2009; Nisbet, 2005] and Zeppelin to measure bulk inputs of CH4 to the Arctic. Isotopic data can then be used to constrain emissions in both regional and global inversion models [Bousquet et al., 2006]. High frequency, ideally continuous, monitoring of CH4δ13CCH4 from a number of Arctic sites, onshore and offshore, will be important if future changes in Arctic sources are to be quantified.
 This work was supported by the Natural Environment Research Council both as part of its Arctic initiative for the International Polar Year 2007–2008 and by a separate grant to Royal Holloway. Support was also provided by the EU through the Geomon programme (FP6-2005-Global-4-036677), the Meth-MonitEUr project (FP5), and earlier INTAS projects. Zeppelin sampling was supported by ARCFAC V, grants 02619-2007-67 and 02619-2008-69 to R.E. Fisher and C.M.R. Fowler. J. Greinert thanks COST Action ES0902 “PERGAMON” for travel support. We thank staff at the Norwegian Polar Institute in Ny-Ålesund for sampling at the Zeppelin station.
 The Editor thanks an anonymous reviewer for their assistance in evaluating this paper.