First observation of direct methane emission to the atmosphere from the subglacial domain of the Greenland Ice Sheet

During a 2016 field expedition to the West Greenland Ice Sheet, a striking observation of significantly elevated CH4 concentrations of up to 15 times the background atmospheric concentration were measured directly in the air expelled with meltwater at a subglacial discharge point from the Greenland Ice Sheet. The range of hourly subglacial CH4 flux rate through the discharge point was estimated to be 3.1 to 134 g CH4 hr−1. These measurements are the first observations of direct emissions of CH4 from the subglacial environment under the Greenlandic Ice Sheet to the atmosphere and indicate a novel emission pathway of CH4 that is currently a non-quantified component of the Arctic CH4 budget.


Results
In the period between the 23 rd and 27 th of August 2016, CH 4 and CO 2 concentrations (Fig. 1A-D) were periodically measured in the air streaming out of one subglacial cavity connected to a lateral subglacial discharge point on the southern flank of the Isunnguata Sermia Glacier on the western Greenland Ice Sheet (Fig. 2). The measured subglacial CH 4 concentrations were consistently higher than the atmospheric background concentration (at 1.9 ppm) and varied between 5.4 and 31.7 ppm (Fig. 1A-D). Subglacial CO 2 concentrations varied around the atmospheric CO 2 level (391 ppm) from 380 to 450 ppm. CO 2 concentrations above the ambient level occurred with the highest CH 4 concentrations on August 25 th and 26 th (Fig. 1C,D). High-frequency fluctuations in the subglacial air stream CH 4 (range 0.12-0.75 ppm) and CO 2 (range 0.21-1.49 ppm) concentrations were observed ( Fig. 1A-D) during the measurements in the subglacial cave and the magnitude of the fluctuations increased with the subglacial CH 4 and CO 2 concentrations (Fig. 1A-D). The fluctuations were higher (except for CO 2 on August 23 rd and 24 th ) than the natural fluctuations of the atmospheric background level CH 4 and CO 2 measured on site with a portable CH 4 and CO 2 analyzer (see Materials and Methods). The fluctuations indicate that the subglacial CH 4 and CO 2 is mixed by air with lower concentrations, which likely is atmospheric air entering the subglacial system via surface moulins or other pathways which connects the subglacial system to the atmosphere. A net emission of CH 4 from the subglacial meltwater to atmosphere was also measured directly ( Fig. 1B) showing that the meltwater itself can be a direct source of CH 4 and sink of CO 2 . In the free air immediately above the subglacial meltwater outside the subglacial cavity, CH 4 concentrations fluctuated rapidly from the ambient atmospheric concentration to approximately 5 ppm CH 4 (Fig. 1B). There was no exchange of either CH 4 and CO 2 (not shown) from the sediment at the edge of the glacier (Fig. 1D).
Based on field observations (Supplementary Video Material) and wind tunnel experiments ( Supplementary  Fig. S1), the velocity of the subglacial air stream was estimated to fall in the range of 0.2 m s −1 to 2 m s −1 (Supplementary Fig. S2). From field observations, the cross-sectional area of the subglacial cavity where air was streaming out was estimated to be in the range of 1-2 m 2 . Using a cross sectional area of 1 m 2 , the estimated range in air velocity of 0.2-2 m s −1 and the span of measured CH 4 and CO 2 concentrations, the total hourly cumulative flux rate per square-meter cross-section area were estimated to range from 3.1 to 134 g CH 4 m −2 h −1 and 537 to 6360 g CO 2 m −2 h −1 (Table 1). However, it should be emphasized that these flux estimates represent the total emission of CH 4 and CO 2 from a so far unknown subglacial catchment area and the average emission per square meter catchment remains unknown.
Acknowledging the fact that any conclusions on the overall climatic implication from this type of CH 4 release would be premature due to the limited spatiotemporal observation of the phenomena, the presented measurements are to our knowledge the first showing direct emissions of CH 4 and CO 2 in the air being ventilated from the subglacial environment below the Greenland Ice Sheet to the atmosphere. The only previous reported study  Fig. 3C for details). The red asterisk in (B) represent a closed chamber measurement (Fig. 3B) of the CH 4 exchange between meltwater and the atmosphere (insert graph). The CO 2 concentrations in the air is not shown due to contamination with human breadth resulting in highly fluctuating measurements of CO 2 in the air outside the cave. This was not the case for CH 4 .  which similarly measured CH 4 concentrations in the air mass directly at the glacial edge of the Greenland Ice Sheet was done in a dry subglacial cave in the Kangerlussuaq area at the Greenland Ice Sheet 16 . In this dry setting, CH 4 concentrations were similar to the ambient atmospheric background level of CH 4 at nearby field sites and the natural atmospheric background measured at the summit station on the Greenland Ice Sheet 17 .

Discussion
The highly elevated CH 4 concentrations and calculated flux rates of this study show an apparent correlation with a moving air mass induced by flowing subglacial meltwater. The observation of direct emissions of subglacial CH 4 to the atmosphere supports earlier findings in the area, namely that this region of the Greenland Ice sheet may be a source of CH 4 to the atmosphere 8 . The direct emission of gaseous CH 4 also shows that biologic aerobic oxidation of CH 4 in the subglacial domain may not be able to fully mitigate subglacial CH 4 production 11 . Nonetheless, the process of CH 4 production and location of the source of the measured subglacial CH 4 is still unknown. Possible explanations for the presence of subglacial CH 4 are either release of old (i.e. radiocarbon dead CH 4 ) microbial and/or thermogenic gas from the subglacial domain 1 which could be formed and stored under the Ice Sheet in the presence of a gas-hydrate stability zone 13,15 . Following entrainment in the subglacial water this old CH 4 could be released following degassing of saturated subglacial meltwater in a similar process as described in Dieser et al. 8 . Alternatively, the emitted CH 4 could be a product of more recently in situ produced biological CH 4 from subglacial sediments 10,18 or intermediately aged CH 4 formed in carbon containing sediments capped during neoglacial readvances 3 . Finally, the possibility exist that the emitted CH 4 could be a combined product of mentioned sources, which could be transported to the emission point via either running meltwater at the bedrock-glacier bed interface or as part of a deeper groundwater aquifer below the Isunnguata Sermia Glacier 19 . A plausible mechanism for the observed CO 2 sink is the dissolution of CO 2 in undersaturated meltwater resulting in a net uptake of atmospheric CO 2 20 . Future research efforts could be guided by two hypotheses focusing on the interrelations between meltwater volumes and glaciological dynamics. Thus, the release of subglacial CH 4 and CO 2 could be (1) caused by venting of the subglacial drainage system if atmospheric air entered the subglacial system due to alternating surface meltwater volumes. This could push out any gas that had accumulated in non-deformed englacial cavities through to the subglacial drainage system resulting in a short-lived but intense period of emissions. In this context, the glacial deformation of the meltwater drainage system will likely determine releases of gases to the atmosphere. Alternatively, (2) continuous emissions extending over the entire melt season could be due to a constant degassing of subglacial meltwater at the margin. This implies also that the subglacial emission would be tightly linked to the meltwater volume and the coupling between climate and supraglacial melting. Finally, obtaining data on the composition of C-isotopes in the emitted CH 4 and CO 2 will shed light on the origin of the emitted C-gases.
Despite these uncertainties in the origin of subglacial CH 4 and CO 2, the mechanism of release and the limited extend of the measurement period, our measurements present the first evidence of a new pathway for direct interaction between the subglacial carbon cycle below the Greenland Ice Sheet and the atmosphere through the direct emission of CH 4 and CO 2 . However, caution should be taken before drawing unjustified conclusions about the importance of subglacial CH 4 emission in the Arctic CH 4 budget and potential climate impact. Improving our understanding of the overall importance of subglacial CH 4 emissions for the Arctic CH 4 budget is therefore to expand the documentation of the spatial and temporal occurrences of subglacial emissions along the Greenland Ice Sheet. Also, further investigation into the release mechanism from the source area(s) can help resolve whether expected future increases in meltwater runoff due to increased surface melting and thinning of the ice sheet in a warming climate will affect the magnitude of subglacial CH 4 emissions.

Materials and Methods
Study site. The study site is located at a lateral subglacial meltwater discharge point on the southern flank at the terminus of the Isunnguata Sermia Glacier at the western margin of the Greenland Ice sheet (67°09′16.40′′N 50°04′08.48′′W) ( Fig. 2A-D). The site was at an elevation of 450 meter above sea level. The lateral outlet has been the focus of previous investigations of the geochemical composition of subglacial meltwater 21,22 as well as drilling projects to assess groundwater chemistry below the Isunnguata Sermia Glacier 23 .
The area in front of the meltwater outlet consisted of the abraded granodioritic gneiss bedrock with large boulders and patches of gravel, sand and silt deposited by flowing meltwater (Fig. 2E). In close proximity to the subglacial meltwater outlet were air-filled cavities in the ice through which air constantly streamed from underneath the Ice Sheet to the atmosphere. Direct measurements of the subglacial CH 4  3-dimensional shape of the subglacial cavity was highly irregular, but the cross-section area (shown in Fig. 2E marked by an A and Fig. 3A) at the terminus was estimated to be approximately 1-2 m 2 . The landscape in the Kangerlussuaq area is typical of west Greenland, intersected by numerous long, narrow and up to 600 meter deep valleys in East to West direction. This type of topography extends below the ice sheet, in places reaching hundreds of meters below sea level. Groundwater recharge and discharge of the area is restricted to taliks, e.i. unfrozen zones in the permafrost, beneath large lakes, rivers and fiords 19 .
Deglaciation in the southern West Greenland area started around 12,300 years BP and the ice sheet margin reached its present position between 6,500 and 7,000 years BP. At approximately 6,000 years BP, the ice margin was east of its present position. Neoglacial re-advance may have started as early as 4,800 years BP and culminated about 2,000 years BP. During the the Little Ice Age (LIA) the ice sheet advanced again reaching a maximum position around 1850 AD with the maximum extent of the edge of the Isunnguata Sermia approximately 50-200 meters beyond its current position 19 .
Continuous permafrost extending to 350 meters below the surface has been reported at the study area 23 . The Isunnguata Sermia Glacier and the surrounding glaciers are underlain by granodioritic gneiss with no or only little sediment or organic material 21 . The glacier is warm based with an annual ice flow of 150-200 meters and surface meltwater reaches the base of the glacier 21 . The meltwater at the discharge point has been characterized as acidic (pH between 5.82-6.53) 22 and poorly mineralized and may represent an outlier in terms of the pH of the subglacial meltwater 24 . However, the reported range in pH values of the subglacial meltwater with a seasonal average of 6.8 ± 0.5 at a nearby subglacial meltwater discharge point 8 are within the range of typical pH values for silica-bedded glaciers and ice sheets (5.1 to 7.8) 24 . Borehole geochemistry at the site suggests that glacial meltwater extends deep in bedrock fractures below the Isunnguata Sermia Glacier and that groundwater below the permafrost is flowing away from the glacier margin 23 . In situ measurements of CH 4 and CO 2 . In situ concentrations of CH 4 and CO 2 of the subglacial air were measured using a portable greenhouse gas analyzer based on state-of-the-art cavity ring-down spectroscopy with a measurement precision of 3 ppb and 0.4 ppm for CH 4 and CO 2 , respectively and a sampling frequency of 1 hz (GasScouter G4301, Picarro Inc., CA, USA). The analyzer was connected to a flux chamber via a PVC tube (4 meter) which was deployed under the following ways: (1) the open chamber was lowered into the opening of one subglacial ice cavity (Fig. 3A), (2) the chamber was placed in a pool of subglacial meltwater forming a closed loop where the CH 4 exchange between the meltwater and atmosphere was measured (Fig. 3B), (3) the open chamber was placed directly in front of cracks near the primary discharge point of the subglacial meltwater channel (Fig. 3C) and (4) the closed chamber was placed on top of loose sediments at the edge of the ice (Fig. 3D). Air temperature and relative humidity of the subglacial air stream was continuously measured inside the open chamber using a datalogger unit (TinyTag Ultra 2 -TGU-4500, Gemini Data Loggers Ltd, UK). The temperature and relative humidity of the subglacial air stream was constant at a level of 0.2 °C and 100%, respectively.
Air flow velocity of the subglacial wind. The velocity of the air flow in the subglacial cavity was estimated by combining in situ observation of a smoke fan placed directly in the air filled cavity in the ice with a verification experiment in the laboratory using a small-scale wind tunnel with smoke injection. The general assumption was that movement of smoke when entrained in moving air in the ice cavity would be identical to movement of smoke entrained in moving air in a wind tunnel. The wind tunnel was constructed using a circular array of parallel straws to create laminar flow placed in the inlet end of a cardboard tube (inner diameter 10 cm). At the outlet end of the tube, a 12 V fan generated a laminar air flow that could be varied by controlling the power of the fan with an adjustable power supply. The flow velocity inside the tube was measured with a hot wire anemometer (model Testo 425, Testo SE & Co KGaA, Germany). Generated wind speeds in the tunnel ranged from 0.2 to 1.2 m s −1 . Details on the wind tunnel setup and test is presented in Supplementary Material. Video footage of in situ smoke movement in the air filled cavity in the ice was compared with observations of the smoke fans in the laboratory under wind speeds of 0.2, 0.5 and 1.2 m s −1 (Supplementary Video Material). This approach provided a best estimate of the in situ velocity of air flow in the subglacial cavity, and indicated that air speeds in the subglacial cavity reached at least 1.2 m s −1 and probably exceeded this value. We thus estimated that the likely range of air flow velocities in the subglacial cavity had a flow range between 0.2-2 m s −1 which was used for the estimation of the subglacial CH 4 and CO 2 flux rates.
Flux rate calculation. Cumulative hourly flux rates of subglacial and CO 2 (g m −2 h −1 ) were calculated combining a constant air flow rate through a 1 m 2 cross-section area of the subglacial cavity and the measured CH 4 and CO 2 concentrations over a one hour measurement period (Equation 1). Flux rates are reported on an hourly basis as this was the maximum measurement time on August 23 rd -25 th . Concentrations were converted to mass using the ideal gas law. where F CO CH / 2 4 is the flux rate in g CH 4 /CO 2 m −2 h −1 , C is the measured 1 hz dry mole fraction concentration (µmol mol −1 ) of CO 2 or CH 4 , ῡ is the wind speed (m s −1 ), A is the cross sectional area (m 2 ), M v is the molar volume (m 3 mol −1 ), T a is the air temperature (°C) measured in the cavity, M is the molar mass of CO 2 or CH 4 (g mol −1 ), the multiplier 3600 converts the flux to hourly values and the constant 10 −6 converts the flux from µg to g CO 2 /CH 4 .