Field work was conducted in five different glacier forefields: the Im Griess (IMG), Griessfirn (GRF), and Griessen (GRI) glaciers located in Central Switzerland in the Cantons of Uri (IMG and GRF) and Obwalden (GRI), and Tsanfleuron (TSA) and the previously investigated Wildstrubel (WIL; Zhu et al., 2018) glaciers located in the Canton of Valais (Figs. 1 and S1 in the Supplement). These forefields were selected for two main reasons. Foremost, their sediments are mainly derived from the calcareous bedrock of the Helvetic nappes (green shaded area in Fig. 1), which consist of a series of nappes (sheets of thrusted rocks) largely composed of Mesozoic limestones, shales, and marls of Jurassic to Eocene age (Pfiffner, 2014; Weissert and Stössel, 2015). They were originally deposited on the shallow northern shelf of the ancient Alpine Tethys Ocean (Weissert and Mohr, 1996) and were subsequently deformed, folded, and stacked on top of each other during Alpine orogeny (Herwegh and Pfiffner, 2005). Whereas individual nappes within the Helvetic nappe system therefore share a similar origin, the lithology and tectonic settings between individual nappes can be quite diverse (Weissert and Stössel, 2015). This was suggested to be a dominant factor determining rock CH4 contents in the WIL catchment (Zhu et al., 2018). Consequently, we chose to investigate glacier forefields distant from one another (e.g., the distance from TSA to IMG is ∼ 136 km), for which sediments are derived from different nappes, but also glacier forefields in close proximity to each other (e.g., the distance from IMG to GRF is ∼ 3.8 km, and the distance from TSA to WIL is ∼ 24 km), for which sediments are derived, at least in part, from the same nappe. A second important reason for the selection was that all five glacier forefields are relatively easy to access, facilitating sample collection and transport to the laboratory.

We conducted our field work in two stages. During stage I in summer 2016, we performed a detailed investigation on the spatial distribution of sediment-entrapped CH4 within a designated sampling zone at the GRF glacier forefield, using high spatial resolution sampling to determine variations in entrapped CH4 contents in relation to sediment depth, sediment age, and glacier-forefield landforms. The GRF forefield was chosen for this purpose mainly because it features well-defined sediment-age classes and well-developed, clearly distinguishable landforms within a previously discretized and characterized sampling zone (Chiri et al., 2015, 2017). We also conducted measurements of sediment thickness (distance between the ground surface and the underlying bedrock) to estimate sediment volumes and, thus, the total mass of entrapped CH4 present in these sediments. Results of the GRF field work were then used to adapt our sampling strategies for field work stage II, which was performed in summer 2017 to quantify the contents and total mass of sediment-entrapped CH4 in the IMG, GRI, TSA, and WIL glacier forefields. During both field work stages, selected sediment and rock samples were used to identify the origin of the entrapped CH4 based on CH4 stable carbon-isotope analyses and analyses of entrapped gas composition (see below).

From the 271 sediment samples from the five glacier forefields, we analyzed 256 samples for entrapped CH4 contents. All analyzed sediments contained detectable amounts of CH4 ranging from 0.08 to 73.81 µgCH4g-1 dry weight (d.w.; Fig. 3), with an average of 14.9 (± 17.0) µgCH4g-1d.w. Gas released from 225 samples was analyzed for C2H6 and C3H8 contents, of which 215 contained detectable amounts of C2H6 ranging from 0.002 to 1.67 µgC2H6g-1d.w., with an average of 0.25 (± 0.32) µgC2H6g-1d.w. In addition, 146 out of 225 samples contained detectable amounts of C3H8 ranging from 0.001 to 0.82 µgC3H8g-1d.w., with an average of 0.11 (± 0.15) µgC3H8g-1d.w. (not shown).

The average gas-wetness ratio for all sediment samples was 75.2 (± 48.4), and the average δ13CCH4 was -28.23 (± 3.42) ‰. Plotting δ13CCH4 values vs. gas-wetness ratios in a so-called Bernard diagram (Fig. 4; Bernard et al., 1978) indicated a thermogenic origin for sediment-entrapped CH4, derived from ancient terrestrial or marine organic matter (kerogen types III and II; Fig. 4). Although CH4 extracted from sediments collected in the IMG glacier forefield showed a higher variability in the gas-wetness ratios than CH4 extracted from the sediments of other glacier forefields, it still fell into the same origin type in the Bernard diagram.

All 72 bedrock samples were analyzed for CH4 content, and 64 contained detectable amounts of CH4 ranging from 0.06 to 108.58 µgCH4g-1, with an average of 11.4 (± 20.0) µgCH4g-1 (Fig. 3). The average δ13CCH4 value of -29.21 (± 2.77) ‰ was similar to that of sediment-entrapped CH4. Likewise, the average gas-wetness ratio of gas extracted from rocks was 78.45 (± 121.84), which is similar in value but with higher variability than gas-wetness ratios for sediment-entrapped CH4 (Fig. 4). Together, these data suggest a common, thermogenic origin of entrapped CH4 in sediments and rocks, with little apparent alteration from physical and/or chemical weathering. Moreover, our data suggest that entrapped CH4 is of similar origin in all five glacier forefields.

Methane contents in 52 samples collected at depths from 20 to 70 cm b.s. ranged from 1.19 to 11.24 µgCH4g-1d.w., with one exceptionally high value at 40 cm b.s. (Fig. 5). Based on these data, there was no clear correlation between sediment depth and entrapped CH4 contents (one-way ANOVA, p=0.9). Thus, we subsequently proceeded to collect sediments from 20 cm b.s. only, and we assumed that these samples were representative in terms of entrapped CH4 content for the entire sediment thickness.

The effects of the sediment age and landform proxies on the entrapped CH4 contents were tested using sediments collected from 20 cm b.s. at 99 locations (Fig. 2a). The CH4 contents in these samples ranged from 0.59 to 34.82 µgCH4g-1d.w. (Fig. 3b), with an average of 5.30 (± 4.86) µgCH4g-1d.w. Two-way ANOVA analysis indicated that landform had a significant effect on sediment-entrapped CH4 contents (p=0.03), whereas the effects of sediment age (p=0.19) and the combined effects of sediment age and landform on entrapped CH4 contents (p=0.37) were insignificant. Post hoc comparisons using the Tukey HSD test indicated that mean values for the sediment-entrapped CH4 content (Table 1) were significantly different between floodplain and sandhill landforms (p=0.03) and were weakly different between floodplain and terrace landforms (p=0.10). The difference between terrace and sandhill landforms with respect to mean sediment-entrapped CH4 content was insignificant (p=0.88).

To estimate the mass of sediment-entrapped CH4 stored within the GRF sampling zone, we used Eq. () with mean values of entrapped CH4 contents, sediment thickness, and sediment-covered area determined for each of the three landforms (Table 1). While C‾CH4 varied by a factor < 1.4 between landforms, sediment thickness was highly variable along the five measured ERT profiles (range 1.0–31.5 m; Figs. 6, S3), and T‾sed varied by a factor of ∼ 2 between landforms (Table 1). Sediment-covered area also showed substantial variation between the different landforms. Within the GRF sampling zone, the sandhill landform comprised the largest sediment-covered area with A‾sed≈ 105 m2, which was about 5 times larger than the values for floodplain and terrace landforms. Consequently, the largest sediment mass was contained in the sandhill landform (a factor 2–3 larger than floodplain and terrace; Table 1). All three landforms combined comprised a surface area of ∼ 1.5 × 105 m2 and contained an estimated mass of ∼ 2.4 × 106 t sediment. Summing the masses of sediment-entrapped CH4 for each landform yielded a total m‾CH4 of 9.7 ± 3.0 t CH4. When calculated using average values for entrapped CH4 contents and sediment thickness, as well as combined sediment-covered area, the estimated m‾CH4 within the GRF sampling zone was 12.7 ± 4.0 t CH4 (last row in Table 1). Uncertainties in individual m‾CH4 values up to ∼ 50% mostly arose from uncertainties in T‾sed (dominated by the large variability in sediment thickness across the GRF sampling zone) and to a smaller degree from uncertainties in C‾CH4.

Methane contents varied substantially between different glacier forefields (Table 2), with the distance between the forefields playing an apparently important role. Specifically, the IMG, GRF, and GRI glacier forefields are located in the northeast of the Helvetic nappes relatively close to each other (Fig. 1) and featured similar, low sediment-entrapped CH4 contents. Likewise, the WIL and TSA glacier forefields are located close to each other in the southwest of the Helvetic nappes and featured similar but high sediment-entrapped CH4 contents. Indeed, our ANOVA results indicated that differences in the sediment-entrapped CH4 contents were insignificant between the IMG, GRF, and GRI glacier forefields (p=0.36) and between the WIL and TSA glacier forefields (p=0.18). Conversely, differences in entrapped CH4 contents between the two groups of glacier forefields were highly significant (p<0.0001).

The total mass of CH4 entrapped in sediments of the five glacier forefields was calculated using estimated values for sediment thickness (10.0 ± 3.0 m; the thickness measured in the GRF sampling zone, see above) and sediment-particle density (2.71 ± 0.15 gcm-3; Daly, 1935) that were assumed to be identical for all five forefields, as well as specific data for each glacier forefield on entrapped CH4 contents, sediment-covered area, and sediment porosity (Table 2). Whereas θ‾t,sed values only varied slightly between the five forefields, A‾sed varied by up to a factor of ∼ 3 (IMG vs. WIL), and C‾CH4 varied by up to a factor of ∼ 7 (GRF vs. WIL). This led to substantial differences in the estimated total mass of sediment-entrapped CH4 between the five forefields, which ranged from 200 ± 74 t CH4 for the GRF glacier forefield to 3881 ± 1367 t CH4 for the WIL forefield (Fig. 7a). Estimates of sediment-entrapped CH4 mass for the WIL and TSA glacier forefields were significantly larger than for IMG, GRF, and GRI. For all five forefields, sediment thickness and sediment-covered area contributed most to uncertainties in the quantification (Fig. 7b). Conversely, entrapped CH4 contents, sediment porosity, and sediment-particle density together contributed ≤ 16% to the calculated uncertainties.

The first estimate of the total mass of sediment-entrapped CH4 in all calcareous glacier forefields in Switzerland was based on published data on glacier retreat in the Swiss Alps, an estimation of the fraction of calcareous glacier-forefield surface area, mean values for sediment thickness, sediment-particle density and sediment porosity, as well as a mean value for sediment-entrapped CH4 content obtained from the five investigated glacier forefields (18.5 ± 4.4 µgCH4g-1d.w.; Table 3). Between the end of the Little Ice Age (∼ 1850) and 2010, the glaciated area within the Swiss Alps has decreased by ∼ 676 km2 to less than 60% of its original value (data sources given in Table 3). When multiplied by the fraction of the calcareous bedrock area in the Swiss Alps (54.6 ± 1.7%), this yielded an exposed calcareous glacier-forefield area of ∼ 369 km2. The total sediment mass contained within this exposed calcareous glacier-forefield area was then computed as 5.62 × 109 ± 1.46 × 109 t. From these numbers, we estimated the total mass of sediment-entrapped CH4 in all Swiss glacier forefields derived from calcareous bedrock to be 1.04 × 105 ± 3.7 × 104 t CH4.

We detected substantial quantities of sediment-entrapped CH4 in all sampled glacier forefields. Entrapped CH4 was ubiquitously encountered at different sediment depths as well as in different forefield landforms and sediment-age classes. We also detected entrapped CH4 in most bedrock samples obtained from these glacial catchments. Furthermore, our data indicated that both sediment- and rock-entrapped CH4 are of thermogenic origin. Thus, the results presented here extend our previous studies (Nauer et al., 2012; Zhu et al., 2018) by providing a more detailed survey on entrapped CH4 contained in glacier-forefield sediments across the Helvetic nappes, and they support our hypothesis on its widespread occurrence and thermogenic origin in calcareous, Swiss Alpine glacier forefields. On the other hand, we cannot entirely reject the possibility of the presence of microbial CH4 sources in certain parts of glacier forefields, particularly in water-logged sediments. Methanogenic potential in isolated hot spots of water-logged sediments has previously been confirmed for the WIL glacier forefield, but it was considered to be of minor importance under field conditions (Zhu et al., 2018). In the present study, no attempt was made to specifically identify potential methanogenic hot spots in the sediments of the other four glacier forefields.

Methane is commonly found in organic-rich sedimentary rocks such as shales, marls, and limestones as a product of the thermal maturation of buried organic matter (Etiope, 2017; Horsfield and Rullkötter, 1994). Previous studies on fluid inclusions in quartz and calcite minerals collected from Alpine fissures and veins within the Helvetic nappes have revealed the existence of four fluid zones, including a large thermogenic CH4 zone (Gautschi et al., 1990; Mazurek et al., 1998; Mullis et al., 1994; Tarantola et al., 2007). The five glacier forefields that we sampled in this study were all located within or near the border of this thermogenic CH4 zone (see Fig. 1 in Tarantola et al., 2007). Thus, our results agree with previous findings on the occurrence of thermogenic CH4 in this region, including the occurrence of thermogenic CH4 detected in gas seeps near Giswil, Central Switzerland, which lies on Penninic Flysch underlain by the Helvetic nappes (Etiope et al., 2010). On the other hand, our results also show that CH4 entrapment within the Helvetic nappes is not restricted to fluid inclusions in fissure minerals and that substantial quantities of CH4 are entrapped within the matrix of the sedimentary bedrock and sediment particles themselves, presumably within inter- and intragranular micro- and macropores (Hashim and Kaczmarek, 2019; Moshier, 1989; Léonide et al., 2014; Abrams, 2017).

Our geochemical data further indicate a common origin for CH4 entrapped in bedrock and glacier-forefield sediments, derived from ancient terrestrial and marine organic matter (kerogen types III and II, respectively; Fig. 4). This provides further evidence that CH4 entrapped in the forefield sediments of the Helvetic nappes has its origin in the calcareous parent bedrock. Moreover, terrestrial and marine organic matter as the ultimate source of sediment- and rock-entrapped CH4 agrees with the origin of the Helvetic nappes: their sediments and organic matter were originally deposited under highly variable climatic conditions on the shallow northern shelf of the ancient Alpine Tethys Ocean (Weissert and Mohr, 1996; Weissert and Stössel, 2015).

Sediment-entrapped CH4 contents showed moderate variability within each glacier forefield (Fig. 3a–e). As sediments were largely derived by glacial erosion from the surrounding calcareous bedrock (Chesworth et al., 2008; Fu and Harbor, 2011), the observed variability in the sediment-entrapped CH4 contents reflects the variability in the entrapped CH4 contents of the various geological formations and the associated mineralogy present in each catchment (Fig. 3f). Entrapped CH4 contents in sedimentary bedrock is typically affected by three main factors: the quantity and quality of organic matter buried during sediment deposition, the thermal history during sediment diagenesis and subsequent organic matter catagenesis, and the resulting permeability of the calcareous bedrock, which affects potential gas migration (e.g., Dayal, 2017; Horsfield and Rullkötter, 1994; Mani et al., 2017). Whereas geological formations contained within the same nappe are expected to possess a similar thermal history, the quantity and quality of organic matter buried may vary substantially between individual formations depending on prevailing conditions during the period of sediment deposition (Weissert and Mohr, 1996; Weissert et al., 1985). Thus, variability in rock- and sediment-entrapped CH4 contents is to be expected for glacial catchments featuring geological formations from different time periods, as was observed for all of the glacier forefields sampled in this study (Table 2).

Our study in the GRF forefield sampling zone indicated that sediment-entrapped CH4 content varied little with sediment depth (Fig. 5) and sediment age. However, we cannot exclude the possibility that such variations could be somewhat larger outside of the sampled depth interval, e.g., in top-layer sediments at depths < 5 cm as a result of enhanced chemical, physical, or biological weathering (Bernasconi et al., 2011; van der Meij et al., 2016; Lazzaro et al., 2009). We refrained from collecting top-layer sediments because they were generally much coarser in all five glacier forefields and, thus, did not appear representative of bulk sediments present at greater depth. We assume that sediment fines are continuously removed from the top layer as a result of physical (wind and water) erosion.

On the other hand, we consider the lack of significant variation with sediment age as an indication that CH4 in glacier-forefield sediments is relatively stable in its entrapped state. This hypothesis is supported by the results of our geochemical analyses for all five glacier forefields, which mostly indicated high similarity between sediment- and rock-entrapped CH4 in terms of the range of measured CH4 contents (Fig. 3f), as well as gas-wetness ratios and δ13CCH4 values (Fig. 4). Thus, although sediments have undergone erosion from the parent bedrock and subsequent weathering, changes in entrapped CH4 geochemical characteristics appeared negligible. This indicates that a potential release of entrapped CH4 from sediment particles by molecular diffusion or from the oxidation of CH4 in its entrapped state within sediment particles may be small in magnitude, as these processes would be expected to cause a noticeable change in CH4 geochemical characteristics (Schloemer and Krooss, 2004; Whiticar, 1999; Zhang and Krooss, 2001). Therefore, our findings suggest that CH4 entrapped in bedrock and sediment matrices resides largely in inaccessible, occluded rather than connected pore spaces. However, a potential release of entrapped CH4 from occluded pore spaces may yet occur via sediment erosion processes, in particular by means of physical and/or chemical weathering of calcareous minerals (Emmanuel and Levenson, 2014; Ryb et al., 2014; Trudgill and Viles, 1998). As these processes act on rock surfaces, they are of great importance to sediments with large specific surface areas, with the specific surface area being inversely related to particle size (Michel and Courard, 2014). Although we are aware that similar erosion processes will act upon large bedrock surfaces, e.g., rock walls and other outcrops within glacial catchments, we have refrained from considering CH4 release from these locations (thus far) due to the much smaller specific surface areas involved. Unfortunately, the release of entrapped CH4 as a result of sediment erosion may not be detectable in the sediment's entrapped CH4 contents, as both CH4 and sediment mass is lost as a result of erosion. Hence, our present data set yields no information on the relevance of erosion processes for CH4 release.

In contrast to sediment depth and sediment age, we detected a small but significant difference in the mean sediment-entrapped CH4 content between landforms within the GRF sampling zone. Specifically, mean entrapped CH4 content in floodplain sediments was significantly higher than in terrace and sandhill sediments (Table 1). We can only speculate about the possible reasons for this observation. One reason could be that floodplain sediments, which are intermittently removed and deposited by the glacial stream during and after flooding events, originate from locations outside of our sampling zone, i.e., from different parent bedrock (Fig. S2). Therefore, it is possible that we missed sampling parent bedrock types (e.g., from steep rock walls) with different (in this case higher) entrapped CH4 contents in this or any of the other glacial catchments. An improved mineralogical investigation of sediments in the various landforms would aid in clarifying this issue.

Finally, our data revealed large regional differences in mean sediment-entrapped CH4 contents between glacier forefields (Table 2). This may be explained by the fact that sediments in glacier forefields located in close proximity to one another are, at least in part, derived from the same individual nappes and geological formations contained therein. For example, both the WIL and TSA glacier forefields harbor sediments derived from the Wildhorn Nappe, featuring several identical geological formations. Hence, this result supports our previous hypothesis that differences in lithology, mineralogy, and tectonic settings between individual nappes play an important role in determining bedrock- and, thus, sediment-entrapped CH4 contents (Zhu et al., 2018). Regional differences in the entrapped CH4 contents paired with differences in the sediment-covered area led to significant variation in the estimates for the total mass of CH4 stored in sediments of the five glacier forefields (Fig. 7a). Uncertainties associated with these estimates were reasonably small and arose largely from uncertainties in sediment thickness and sediment-covered area (Fig. 7b). To further reduce these uncertainties, measurements of these parameters across entire glacier forefields would be of help using techniques such as geophysical methods for sediment thickness (such as the ERT method used in the GRF sampling zone) and field mapping of sediment-covered area in combination with GIS-based methods utilizing digital elevation models (e.g., Geilhausen et al., 2012; Smith and Clark, 2005; Zemp et al., 2005). Unfortunately, field measurements in the rugged alpine environment are typically time-consuming, expensive, and challenging to perform.

Our first, rough estimate of the total quantity of CH4 entrapped in sediments of all calcareous Swiss glacier forefields combined yielded a substantial mass of 1.04 × 105 ± 3.7 × 104 t CH4, contained within a solid volume of ∼ 2.1 km3 glacier-forefield sediments. At first glance, this number appears large when compared with an estimate of the annual CH4 emissions to the atmosphere (5.7 × 103 t CH4) from natural and seminatural sources in Switzerland, including emissions from lakes, reservoirs, wetlands, and wild animals (Hiller et al., 2014). However, while the latter data represent annual CH4 fluxes, the fate of sediment-entrapped CH4 still remains elusive (see below). On the other hand, our number is in good agreement with a previous estimate of the CH4 content for Valanginian marl, a geological formation within the Helvetic nappes, containing calcite fracture fill (∼ 0.7 × 105–2.1 × 105 t CH4 km-3 bedrock; Gautschi et al., 1990).

Our estimate for the total sediment-entrapped CH4 mass is subject to substantial uncertainty. The two largest contributors to the calculated uncertainty are the sediment-entrapped CH4 content and sediment thickness. In addition, there is considerable uncertainty in the exposed calcareous glacier-forefield area, as the latter was only roughly estimated based on glacier retreat and the fraction of calcareous bedrock area in the Swiss Alps. As discussed above for individual glacier forefields, field measurements and GIS-based methods may help to reduce uncertainties related to sediment thickness and exposed area. An important way to reduce uncertainty related to the entrapped CH4 contents would be to generate a database of CH4 contents for different geological formations present within the Helvetic nappes, as lithology and tectonic settings appear to control CH4 contents. Determination of the areal extent of different geological formations would likely help to reduce uncertainties in the sediment-entrapped CH4 mass.

Whether or not sediment-entrapped CH4 plays a role as an emission source to the atmosphere will largely depend upon its rate of release from sediment particles and its potential consumption by MOB in aerated sediments. While we produced some evidence that CH4 is stable in its entrapped state (see discussion above), further investigations will be required to specifically elucidate possible mechanisms and fluxes of CH4 release in forefield sediments, in particular during periods of enhanced physical and/or chemical weathering, e.g., during rainstorms or snow melt (Winnick et al., 2017). Notably, atmospheric CH4 oxidation has previously been detected in several glacier forefields including our GRF site (Bárcena et al., 2011; Chiri et al., 2015; Hofmann et al., 2013). These studies indicated that MOB activity in forefield sediments establishes quickly (within the first 10 years after glacier retreat) and that fluxes of CH4 uptake from the atmosphere increase to values comparable with mature soils within a few decades (Chiri et al., 2015). Nonetheless, intermittent CH4 emissions to the atmosphere were also observed in GRF floodplain sediments (Chiri et al., 2017). Hence, we hypothesize that traces of CH4 released from sediment particles may be consumed by MOB, at least under favorable environmental conditions, and serve as an additional source of energy and carbon to this group of microorganisms. This hypothesis, of course, awaits experimental confirmation.