Towards a Better Knowledge of Natural Methane Releases in the French Alps: A Field Approach

We report investigations performed at some hydrocarbon gas seeps located in the French Subalpine Chains in zones of outcropping Jurassic black shales, increasing the reported number of such occurrences in this part of the Alps. We present the characteristics of each of the seeps, based on soil flux measurements and soil gas measurements. Gases emitted are CH4-rich (87–94%) with the exception of one site (78.5% CH4 + 8.2% CO2) where an active landslide may induce dilution by atmospheric air. CO2 is generally measured at low levels (<1.6%). Concentrations in C2H6 are more variable, from less than 1% to more than 2.3%. Gas is emitted over areas of various sizes. The smallest gas emission area measures only 60 × 20 cm, characterized by a strong hydrocarbon flux (release of about 100 kg of CH4 per year). At a second site, hydrocarbon emissions are measured over a surface of 12m. For this site, methane emission is evaluated at 235 kg per year and CO2 emission is 600 kg per year, 210 kg being related to gas seepage. At the third site, hydrocarbons are released over a 60m area but strong gas venting is restricted to localized seeps. Methane emission is evaluated at 5.1 tons per year and CO2 emission at 1.58 tons per year, out of which 0.53 tons are attributed to gas seepage. Several historical locations remain uninvestigated at present, and numerous others may still be unknown. We outline strategies to search for such unrecorded sites. Considering the topography of the potential alpine and perialpine emission areas, the possibilities to detect gas emissions appear of the size recorded so far seem to be restricted to ground-based methods or to methods offering the possibility to point orthogonally to the soil towards the seep maximum. If such sites are to be investigated in the future in the frame of Environmental Baseline Assessment (EBA), even establishing appropriate monitoring protocols will be challenging.


Introduction
The occurrence of natural Earth degassing-and specifically hydrocarbon degassing-is known since very ancient times. In Europe, methane seeps are reported since the Roman period, e.g., in Greece [1], in Turkey [2], and in France [3]. Later, especially during the XIXth century, in conjunction with the hydrocarbon rush, some of these natural gas releases were investigated in more detail to assess their exploitability [3]. Nowadays, the growing interest for natural hydrocarbon seepage areas has two main angles: firstly, to better understand the production mechanisms of the hydrocarbon gas phases [4] and specifically to determine if the gas is produced by thermal or biological processes from organic matter [5,6] or by abiotic reactions involving hydrogen [7]; secondly, to better describe the degassing patterns [8], to identify gas migration pathways, and to quantify the amounts of gases released into the atmosphere [4,9] and the related impacts onto the global atmospheric methane budget [10]. These investigations are in relation with environmental concerns related to the increasing share of unconventional gas exploitation [11] and to natural and man-induced gas emissions from hydrocarbon fields [12,13] with a special focus on the integrity of gas and oil wells [14], including abandoned wells [15]. It is well established that gas seeps not only induce free gas emissions but also increase dissolved gas levels in aquifers [16][17][18] where their quantifications can be quite challenging [19,20]. Here, we focus on the quantification of free gas leaks at the soil/atmosphere interface [10] where the influence of external parameters (temperature, soil properties, humidity, and so on) on the measurements of gas emissions needs to be taken into account [21].
Investigations of the most prominent French alpine gas seep, the Fontaine Ardente du Gua, have been recently reported [3] together with preliminary reconnaissance of a gas seep of lesser importance (Rochasson gas seep; [22]). When comparing the few known occurrences in the French Subalpine arc to the number of gas seeps reported for Italy (HYSED database; [23]) and for Switzerland, Germany, and Austria [5,24,25], mainly situated in the molasses basins, we suspect a significant knowledge gap, even though the lithological and tectonic contexts as well as the thermal history are not directly comparable. Thanks to an in-depth literature review, including "grey" historical literature, we localized several other sites prone to the release of hydrocarbon gases ( Figure 1).
Here, we report field investigations performed at some of those newly localized sites (red symbols in Figure 1) as, from South to North, Jonchiers (JON), Molières-Glandaz (MG), and Châtillon-sur-Cluses (CLU). A fourth site (Rochasson (ROC)), already monitored in 2015, was investigated again because a landslide had modified its surface in 2016. Investigations were focused on the quantification and spatial patterns of gas emissions to the atmosphere as contributions to the greenhouse gas budget. Soil flux measurements using an accumulation chamber system were performed at all the sites with the exception of JON. Soil gas analyses allowed estimating precisely the relative proportions of gas species.

Geological Settings
In their article in 1990, mainly focused on CO 2 emissions in the French southeast basin, Blavoux and Dazy [26] mentioned the occurrence of mofettes (natural gas seeps) among which some are methane-bearing including JON, MG, and the Fontaine Ardente du Gua (FA). For FA, we provided detailed information in Gal et al. [3]. Gas emissions reported at the Col de Cabre could not be investigated due to safety issues as these emissions were encountered during the drilling of a railway tunnel in 1887 [27]. Nonetheless, the probable source rocks for this site are black shale formations similar to those outcropping at Fontaine Ardente du Gua [3].
The Jonchiers site (JON; Figure 2) is located close to the town of Buis-les-Baronnies. This seep is associated with black shale formations of the Oxfordian age [27]. Local inhabitants discovered this methane seep indirectly by observing persistent flames after having practiced fire clearance (pers. comm.). Blavoux and Dazy [26] reported an analysis of the gas phase with dominant CH 4 (71.97%) and high concentrations in CO 2 (8.39%) and H 2 (5.19%), completed by N 2 (9.51%). These authors also reported an isotope analysis of the carbon of the methane (δ 13 C CH4 = −39 7‰ vs. PDB), suggesting a dominant thermogenic origin for this gas phase. The location of this gas seep reported in the subsoil database BSS of the French Geological Survey (site BSS002BPZV; [28]) revealed to be erroneous (the real location is 730 m to the ESE).
The Molières-Glandaz site (MG; Figure 2) is not reported in any geological database. Blavoux and Dazy [26] only mentioned a location called "les Tiogaux" which corresponds to a farm. Nevertheless, they report a historical analysis from 1916: 70.16% of CH 4 , 14.05% of H 2 , 4.8% of C 2 H 6 , 2.89% of N 2 , 1.2% of O 2 , and 0.3% of CO 2 . This alkane-rich gas is also seeping from black shale formations of the Callovian to Oxfordian age. Hints from the inhabitants oriented our investigations to a dry creek where we detected the seep organoleptically, through its characteristic hydrocarbon smell. Inhabitants have known the existence of this gas seep from immemorial times. It was discovered by observing gas bubbling in the creek during wet periods.
The last investigated site is located much further in the north of the French Subalpine Chains, close to the munici-pality of Châtillon-sur-Cluses (CLU; Figure 2). Contrarily to the other seeps, it is not related to obvious Callovo-Oxfordian black shale outcrops. The existence of gas emission is reported since the XIXth century [30], and an analogy was made with the eternal flames occurring in the Caucasus or in Turkey (see references to Pliny the Elder in [2]). The gas was, at that time, collected at 16 m depth in argillaceous and gypsum formations covered by glacial deposits and used to light houses. Some decades later, a more detailed description was given by Omer [31]. Two locations were prone to gas emissions. The first emission zone was located in the backyard of a house, and some of the gas emerged from a crack of the floor inside a room. The second emission zone was located in a meadow and was discovered by a farmer who observed a bare surface of about 100 square meters, where the earth seemed "burned." After having dug a hole in the ground, he heard a hissing gas, which caught fire instantly after approaching a burning match. These two locations are reported in the French BSS geological database (respectively, recordings BSS001RKMM and BSS001RKML; [28]). The gas phase is essentially methane, and production has been attributed in 1925, without proof, to the degassing of Eocene lignite-rich formations [32]. It can be noted that Lower Jurassic marls (Toarcian) outcrop near the site and have been noted as gas-bearing in a 45 m drillhole (BSS001RKML) even though their thickness cannot be compared to the sites further south. We report only the data from the first site, close to the house. The site owner provided instructive information on the zone (M. Dumont, pers. comm.): the area has been recently subject to a landslide which is currently monitored. Consequently, gas emissions to the atmosphere may be different from those described in the past.

Materials and Methods
The monitoring uses two well-established methods, soil gas flux measurements and soil gas concentration measurements. For all sites, soil gas flux measurements at the soil/atmosphere interface were performed. For most sites, these flux measurements were completed by soil gas measurements at 1 m depth. A detailed overview of these two techniques can be found in the literature [3,33]; below, we provide a summary of the methods used.
Soil gas flux measurements were performed [using the accumulation chamber technique, with external recirculation] (Echo Instruments, Slovenia). The […] chamber has a semispherical geometry with a basal area of 0.0289 m 2 and a volume of 0.0018 m 3 (soft ground) to 0.0023 m 3 (hard ground) [and is equipped with a fan with tunable speed for mixing of the headspace gas]. Pressure, temperature and relative humidity inside the chamber are continuously monitored. CO 2 and CH 4 [are measured] by Non-Dispersive Infra-Red (NDIR) detectors (0 to 5000 ppmv ± 2% and 0 to 10000 ppmv ± 5% […] respectively), O 2 and H 2 S by electrochemical cells (0 to 25% vol. ± 2% of the reading and 0 to 10000 ppmv ± 5% of the reading respectively).
[…] The flux measurement was based on the rate of CH 4 and CO 2 accumulation (positive fluxes) and possible opposite O 2 decrease ("negative" fluxes), in the chamber. Usually positive values indicate fluxes directed from the soil to the atmosphere and negative values flow from the atmosphere into the soil. [Most flux measurements were performed directly on the clayey soil surface with little to absent soil litter. Flux measurements at CLU site were performed on a grassy soil surface.] [3] Soil gas concentrations were measured [in a 10 mm hole drilled down to 1 m depth. A copper] sampling tube was inserted […] and an Infra-Red Gas Analyzer (IRGA, LFG20 by ADC Gas Analysis Ltd., UK) was plugged, pumping at low flow rate (200 mL.min -1  Analytical precision for CO 2 and CH 4 was ±0.5% for low concentration range (0.01-10% vol.), ±3% for higher concentrations (10 to 50% vol.) and up to ±5% above 50% vol […]. The precision for oxygen was ±0.4% full scale reading (0-25% vol.). [The most methane-rich samples were collected in Isotubes® for further laboratory characterization of the gas phase, encompassing] permanent gases (CO 2 , O 2 , N 2 , Ar), alkanes (CH 4 , C 2 H 6 , C 3 H 8 ) and 4 He. Detection limits are 10 ppm for permanent gases, 2 ppm for alkanes and 50 ppm for helium. Precision of the measurement is better than 2% at full scale. [3] Flux measurements preceded soil gas measurements to avoid introducing artefacts due to altered soil permeability. The only exception is the ROC site where soil gas was measured during a previous campaign in 2015. This site was subject to a landslide in between the two surveys in 2015 and 2017, so that the ground surface has been totally reshaped and the locations where soil gas concentrations had been measured two years before have disappeared.

Results and Discussion
4.1. Meteorological Conditions. Meteorological conditions are known to have adverse effect on soil gas emissions from the soil under certain conditions, especially when weather is rainy [21,34]. The field survey was performed during one week in early September 2017. During this period, weather conditions were dry and no rainfall event occurred ( Figure 3). The atmospheric pressure was particularly high. If high-pressure conditions are sometimes reported to reduce gas emissions [34], this may not be always the case [33]. During the survey, the pressure conditions were stable. If gas emissions were reduced by the effect of high-pressure conditions, this effect would have concerned the complete monitoring period so that a hypothetical variability of the gas emissivity linked to pressure effects would be very low. Some thunderstorms occurred some days before the survey, but the amounts of rainfall were highly variable from one location to the other (from less than 10 mm up to 30 mm). Thunderstorms are also events that have generally a local influence, and their impact at a distance of some kilometers to tens of kilometers may be low to inexistent. We used the indications of the relative humidity sensor inside the flux chamber to assess local humidity conditions ( Figure 3). Relative humidity in the air pumped from the soil is low, between 31% and 74%, with a mean value of 50.6%. This is in accordance with limited rainfall amounts and a 10-day period without any rainfall event at the end of August. The effect of humidity on gas emissions is thus negligible.

Soil Gas Fluxes and Soil Gas
Concentrations. Data are presented in Figure 4 as boxplots. Datasets are provided as supplementary material (available here). At the JON site, no methane flux measurements could be performed. Soil background CO 2 flux in the surroundings of the seep is generally low (<1 g·m -2 ·h -1 ) and matches with the values reported for biologically produced CO 2 flux [35]. Only one measurement is higher but still falls in the range of biological CO 2 fluxes reported in summer in France [36]. The CO 2 concentrations are between 1 and 2% with slight O 2 depletion compared to the atmosphere. Soil gas emitted at the seep is diluted by atmospheric air as drilling into the clayey formations, and the subsequent measurements were only possible to 30 cm of depth. By correcting this dilution, assuming zero O 2 , the gas composition was recalculated as follows: high CH 4 concentration (87%); abundant N 2 (10%); presence of CO 2 (1.6%), C 2 H 6 (0.87%), and C 3 H 8 (0.235%); and traces of C 4 H 10 (0.071%), C 5 H 12 (0.019%), and 4 He (0.025%).
At the MG site, only soil gas flux was measured ( Figure 4) with the exception of one soil gas measurement sampled at the vent immediately after the flux measurements. CO 2 flux measurements outside the gas seepage area yielded values similar to those measured at the JON site. For about onethird of the measurements, elevated CH 4 fluxes were measured, ranging from less than 30 to 220 g·m -2 ·h -1 . At the most emissive point, three repeated measurements obtained fluxes of 220, 182, and 204 g·m -2 ·h -1 . Intrinsic "pulsations" [5] of the seep can explain this variability as suggested by short-term changes of flame size [37]. Nonetheless, the variations are   Figure 3: Meteorological parameters before and during the survey. (a) Daily mean air temperatures and rainfall amounts taken at the nearest airport weather stations (data from https://freemeteo.fr); distance between gas seeps and airports is indicated in the caption. (b) Hourly pressure and temperature data from Geneva (Genève) airport compared to temperature and pressure data measured using the flux chamber system during the measurements. Bar chart gives the relative humidity inside the chamber during the measurements. 5 Geofluids relatively weak with a standard deviation of less than ±15% [38], showing a good reproducibility of the flux measurements. If the lowest measured CH 4 flux is comparable to the upper range of diffuse seepage measured near the FA gas seep [3], the highest values are similar to some seeps reported by Etiope [2] in Greece or in Romania. In general, a strong depletion in oxygen in the chamber is linked to an increase in methane. The gas phase emitted at the MG site is CH 4 -rich (94.2%), with abundant C 2 H 6 (2.36%), residual N 2 (0.65%), and traces of CO 2 (0.31%), C 3 H 8 (0.40%), C 4 H 10 (0.056%), C 5 H 12 (0.006%), and 4 He (0.012%).
By comparison, the seepage intensity of the ROC site is apparently lower (maximum CH 4 flux of 54 g·m -2 ·h -1 ), but the size of the degassing area is much larger (a CH 4 flux was detected for more than 80% of the points). The O 2 depletion inside the chamber is also lower. The CO 2 flux can punctually reach a relatively high value (4.23 g·m -2 ·h -1 ) suggesting that part of the emitted CO 2 may originate from the seep. Strong CO 2 enrichment in the soil formations has been reported earlier at the ROC site [22].
The present-day methane emission at the CLU site is not as high as it could have been expected from historical literature. The highest value is only 4.56 g·m -2 ·h -1 , and only 4 points from 40 showed methane emission. Nevertheless, O 2 depletion in the chamber was monitored more frequently, suggesting that even in cases where the CH 4 flux is not 6 Geofluids detectable, there is still an emission of deep oxygen-free gas diluting oxygen in the chamber volume [22]. This may be in relation with the active landslide that could have modified degassing patterns. Soil gas monitoring confirmed the existence of high CH 4 concentrations. The highest peak value was 78.5% monitored only during a short period. Lower CH 4 contents were measured in the laboratory (44.5%) accompanied by abundant nitrogen (47.8), CO 2 at 6.6%, and traces of C 2 H 6 (0.60%), C 3 H 8 (0.009%), C 4 H 10 (0.0015%), and C 5 H 12 (0.0002%). In spite of this supposed decline of methane supply, strong O 2 depletion and associated CO 2 enrichment were also monitored in soil.

Relationships between Gas
Species. The binary plots in Figure 5 describe the relations between gas species. The process-based approach [39] can be applied to the CO 2 -O 2 relationship ( Figure 5(a)). As expected, measurements at the JON site (green lozenges in Figure 5(a)) cover a reduced concentration range and plot close to the replacement line that characterizes respiration processes occurring in soils (aerobic respiration consumes 1 mole of O 2 to produce 1 mole of CO 2 ). The situation is different for the single sample taken at the centre of the JON gas seep (grey lozenge in Figure 5(a)), where O 2 is absent and CO 2 is present at ca. 2%. Similarly, only one point is available for the MG site (with on-site and laboratory measurements) so that it is not possible to characterize the CO 2 -O 2 relation close to the MG gas seep. Some of the measurements at the CLU site (small blue triangles in Figure 5(a)) plot on the replacement line 7 Geofluids especially for relatively low CO 2 concentrations (<5%) which is common in summer in French soils [33]. At a CO 2 concentration greater than 9-10%, two behaviours exist. Some measurements (electric-blue-surrounded triangles in Figure 5(a)) plot close to the CH 4 oxidation line thus suggesting that part of the measured CO 2 is produced by CH 4 oxidation in the soil (2 moles of O 2 are needed to produce one mole of CO 2 + H 2 O). This is coherent with the presence of CH 4 in sufficient amounts to allow for high CO 2 production in soil. Otherwise, with the exception of specific conditions, CH 4 amounts in soils are generally low [40]. Other measurements (blacksurrounded triangles in Figure 5(a)) plot on the replacement line although their CO 2 concentrations may reach very high values (more than 18% for point CLU36). High CO 2 concentrations in soils, produced by biological processes, are sometimes reported (up to 12% [33]; up to 13% [41]; and up to 14% [42]) often in soils having high clay contents [43]. Higher concentrations seem to be exceptional even if locally excessive concentrations of CO 2 (up to 40% above the capillary fringe) are reported in the literature [44]. Point CLU36 being exactly on the replacement line, it must be hypothesized that there are local soil characteristics which allow the high concentrations of CO 2 may be in connection with the landslide phenomenon. Some of the CO 2 could be produced by CH 4 oxidation, but in amounts that are not sufficient for a noticeable deviation from the replacement line towards the CH 4 oxidation line. No influence of dilution of the O 2 +N 2 pool by a deep CO 2 endmember in near surface environments is visible from the CO 2 -O 2 binary plot.
Methane and CO 2 concentrations in soil are not well correlated for any of the sites ( Figure 5(b)). Where biological processes in soil produce CO 2 by using O 2 (replacement line in Figure 5(a)), we would not expect elevated soil CH 4 (dashed rectangle in Figure 5(b) for the JON site and for some CLU points). The situation is different when some CH 4 is present in the soil gas. This is documented only for the CLU site. For the six measurements containing CH 4 , CO 2 appears inversely correlated to CH 4 . The existence of CH 4 oxidation in soil is only established for measurements with the highest CH 4 concentrations (electric-blue triangles in Figure 5(b)). These measurements also show the existence of a positive CH 4 flux at the soil surface (electric-blue triangles in Figure 5(d)). For these data, CH 4 oxidation may superpose to the biological production of CO 2 in soil. For the black-surrounded triangles, the CO 2 -O 2 relationship suggests that this is not the case ( Figure 5(b)). Consequently, it is likely that biological processes are responsible even for very high CO 2 concentrations in soil (up to 18%).
A positive CO 2 flux was always measureable ( Figure 5(c)). The positive correlation between CO 2 flux and soil CO 2 concentration is a common phenomenon in summer when biological activity is at its highest. The CLU site again shows a distinct behaviour. Biological CO 2 flux is less strong than at JON (blue triangles in Figure 5(c)). This can be a consequence of different soil conditions (texture, water content, and vegetation). Higher CO 2 flows (trend 1 compared to trend 2 in Figure 5(c)) demonstrate the additional production of CO 2 by CH 4 oxidation closer to the supposed CH 4 seepage areas. Figure 5(d) corroborates this assumption. The three measurements represented by electric-blue triangles have the highest CH 4 fluxes monitored at the CLU site. Part of the CO 2 release to the atmosphere is thus linked to CH 4 oxidation in soil. The CH 4 fluxes measured at the ROC site are more scattered. Below 10 g·m -2 ·h -1 , it is difficult to define a clear relation between CH 4 and CO 2 fluxes. Two trends appear, one with relatively high CO 2 fluxes that also includes points from CLU (trend a) and one with a lower slope (trend b) reflecting different degrees of CH 4 oxidation. The trend for MG (trend c) has a similar slope to trend a and suggests, here again, a high rate of CO 2 production from CH 4 oxidation, at overall higher CH 4 fluxes, even though this is in apparent contradiction with the low CO 2 concentrations measured for this site (Figures 5(a) and 5(b)). Figure 5(d) shows the exponential decline of fluxes over a short distance (60 cm) from the main vent.
The gas dryness ratio (= C 1 / C 2 + C 3 ) provides some indication on the origin of alkane gases from the newly sampled methane seeps ( Figure 6). The ratio is relatively low, from 34 (MG) up to 72-77 (CLU and JON, respectively). The ratio of ROC is in the same range (45) whereas the ratio of FA is much greater (1300). In the Western Alpine Arc, in Switzerland, ratios close to 50 are found in deep boreholes surrounding the Leman lake (Thônex borehole; [45]), ratios close to 70 have been measured in gas seeps from the Giswil area [5], and ratios close to 20 were reported for boreholes located in the alpine flysch formations (Wilen well; [5]). Such ratios, lesser than 100, are often related to gas production  8 Geofluids through thermogenic processes [46] whereas higher ratios may be related to a wider panel of production pathways or to secondary processes or mixing [47]. For the ROC, MG, and JON sites, situated in similar geological settings (lithology, geological age), similar mechanisms of light alkane production may be supposed. The situation may be different for the CLU site, with a comparable gas dryness ratios but where the seep likely to be associated with host rocks of different age and maturity, parameters known to be of prime importance [2]. For the FA site, if a genetic link of thermogenic methane generation with argillaceous host rocks, rich in organic matter, has certainly to be searched for, the higher dryness points potentially to secondary processes such as chemical fractionation upon transport or to biologically driven processes. This statement applies not only for alkane gas species but also for the origin of CO 2 .

Gas
Emissions to the Atmosphere. On the basis of the data from this study for the JON, MG, ROC, and CLU sites and previous results for ROC [22] and for FA [3], it is obvious that the FA site remains the most important spot source of greenhouse gases (GHG) so far reported in the French Subalpine Chains. Nevertheless, the other seeps cannot be neglected altogether, especially if we consider them as indicators of the diffuse flux potentially related to a large number of undetected macro-or microseeps (see discussion below).
In the absence of flux measurements performed at the JON site, we refer to the other sites only. Nevertheless, the similarities in outcropping lithology, hydrographical, and topographic settings suggest that there may be comparable gas emission patterns for the JON and MG sites. Both are situated in creeks within the Middle Jurassic shales. The case of the MG site is particularly instructive. It is located in a very narrow riverbed (Figure 2), and the CH 4 emission was detected only in a 20 × 60 cm area ( Figure 5(d)). As close as one meter away from this specific location, no CH 4 flux was measured at the soil surface. MG is thus the archetype of a tiny gas escape, with a very restricted seepage area, even though it falls in the macroseep category according to Etiope [2]. Calculation for MG is performed for the observed 0.12 m 2 emission area through which the gas is emitted (not extrapolated to 1 m 2 ). Taking into account the variability of the flux measurement reported above, CH 4 emissions can be estimated between 95 and 100 kg per year (ca. 270 g/day), with minor CO 2 emissions (1 kg). This corresponds to the amount of CH 4 released in two days by the FA seep [3]. This is a negligible absolute contribution to the overall regional methane budget, but not imperceptible when considering the small degassing area. Indeed, Mörner and Etiope [48] reported CH 4 emissions of 20 kg per year at San Vincenzo la Costa (Italy) but for an area of 300 m 2 (gas emission from a mud volcano and not from shales). Etiope et al. [6] reported seeps in the Appalachians where CH 4 emissions may be as low as 1 g/day (Chestnut Creek). Given the size of the seepage area, the MG site is thus remarkable.
Flux measurements performed at the ROC site cover an area of ca. 110 m 2 over which a CO 2 flux was always measureable whereas a CH 4 flux was detected only on approximately half of the area (Figure 7). The 2015 dataset on soil gas measurements is used for comparison [22]. Interpolation uses standard kriging options in Surfer® software (Golden Software). Volume calculation of the amounts of gas released is also done using Surfer® software. The dataset is reduced, the size of the area is limited, so that definition of the data distribution and calculation of a variogram may fail to be successful or representative. The 2015 soil gas data indicated that gas venting was restricted to patchy seeps and two of them were identified as macroseeps. The 2017 flux measurements also revealed patchy gas emissions and poor correlation between CO 2 emissions and CH 4 emissions. It is not possible to identify any clear pattern related to geological or morphological features. When looking at the alkane emissions, little correspondence is found between soil gas and soil flux data. This is likely the consequence of a large time span between the acquisitions and the occurrence of a landslide in between the two surveys ( Figure 2). Nowadays, the ROC site is more emissive on its western part, which corresponds to the upper part of the shales' outcrop. Over an area of ca. 60 m 2 , the global CH 4 emissions (macroseepage + miniseepage [2]) are evaluated at 5.1 tons per year. Gas seeps with similar annual releases are reported in Italy or in Romania [2]. The global CO 2 emission reaches 1.58 tons per year, 0.53 tons being related to gas seepage if a biological contribution to the CO 2 flux of 1.25 g·m -2 ·h -1 is considered [35].
At the CLU site, our investigations followed a more regular grid thanks to the flatness of the grassy terrain. The area of interest is only 50 m 2 large (Figure 8). Concentration measurements were performed over a half of this area. A CO 2 flux was always perceptible whereas a CH 4 flux was measureable over a surface of 12 m 2 only. Some decrease of the O 2 content inside the chamber was nevertheless measured even where no CH 4 was noticeable, suggesting that O 2 can be used as a proxy when the CH 4 efflux is very low. This is the case in the vicinity of points CLU22 and CLU28. As for other seeps, the correspondence between the patterns of CH 4 and CO 2 fluxes is not always obvious: the highest CO 2 fluxes are measured in the northern part of the area, whereas the highest CH 4 fluxes lie in the southeast corner. There is also a poor coherence between patterns of soil CH 4 concentrations and CH 4 fluxes, especially close to point CLU11 (yellow arrow in Figure 8(a)), likely due to varying soil properties. As tortuous the gas emission patterns may be, methane emission in the atmosphere is not negligible on a yearly basis. Over 12 m 2 , global methane emissions are evaluated around 235 kg per year. CO 2 emissions are larger (600 kg per year), a third of this value being related to the gas seep contribution if the biological contribution is assumed at 1.25 g·m -2 ·h -1 [35]. This evaluation needs certainly to be refined as it was not possible to assess the impact of the landslide phenomenon onto the gas emission pattern by a sole 1-day monitoring. Given the historical records, methane emissions can be expected to occur at other locations around the investigated site so that the overall emissions would be currently underestimated.

4.5.
Are there Other Seepage Areas in the French Alps? An interesting conclusion can be drawn from the abovementioned measurements. The sum of ROC, CLU, and MG 9 Geofluids contributions-and potentially the JON contribution if we assume it as similar to that of MG-to the CH 4 budget is close to 5.5 tons per year over a total degassing area of only 72 m 2 . This represents 29% of the yearly CH 4 emission from the FA site [3], but the cumulated seepage area of the other sites is equal to 30% of the surface of the macroseepage and diffuse seepage areas of FA. The similarities of the different sites in terms of outcropping source rocks, topography, gas composition, and emission patterns may lead to the conclusion that we can assume an average annual release of 75 to 80 kg of CH 4 per square meter at a typical subalpine black shale seepage site.  Figure 7: ROC site: interpolated maps for soil gas concentrations (hatched fill; data from [22]) and fluxes (full fill); (a) CO 2 gas specie; (b) CH 4 gas species. The light dot locates the main gas seep as represented in Figure 2 (no soil gas measurement done at this location).

Geofluids
Several locations could not yet been localized and investigated (Figure 1), and many others may still be unknown or unreported. Black shale formations outcrop over hundreds of square kilometers in the French Subalpine Chains, and it is more than likely that other seepages exist in such a large area. Based on the knowledge gained on investigated seeps, the search of new degassing locations should probably focus on thalwegs and indented valleys crosscutting black shales. A rough estimate can be done in the case of the MG site. Over an area of 1 km 2 , centred on the known gas seep, we can identify approximately 3.8 km (planar distance) of small valleys cutting into the outcropping black shales (Figure 9). Extrap-olating this density to the total surface of black shale outcrops at the Subalpine Chains leads to tremendously high distances, which are, in most areas, only accessible by foot, given the rough topography and the dense vegetation on those agriculturally mostly unusable badlands.
The assessment of an environmental baseline with respect to potential uses of methane-containing argillaceous formations as well as any estimation of GHG releases from such formations would require a much more stringent approach to gas seep identification. Gal et al. [3] reported some aspects of gas detection in the context of environmental impact assessment-related shale gas exploration/exploitation. 46

Geofluids
The example of the MG seep is very illustrative in that perspective. Contrarily to the FA site, the MG site represents a small seepage, but the mechanism may be similarly linked to fault structures, given that leakage rates along faults may be highly heterogeneous [49]. The MG site can be considered as point source macroseep. The photographs presented in Figure 9 illustrate how difficult it is to locate such a seepage in the field, using currently available monitoring equipment. Given the natural air movements, a gas detector held 10 cm above the seep will record maximum CH 4 values of only 3%, in spite of the fact that there is enough CH 4 seeping to the soil surface to light a flame directly above the seep. We noted significant variability in the readings even for a very light wind blowing through the narrow channel formed by the creek bed. Moving the gas detector some tens of centimeters away from the source precludes any detection of CH 4 . This makes it highly improbable to detect such gas vents with any method outside a very restricted volume around the point of gas emission. Indeed, the human nose is in the present case one of the best tools owing to its ability to detect trace gases, such as H 2 S, at levels far below the detection limits of common portable equipment [50].
Future detection of unknown seepage areas will thus be a difficult task. At the scale we consider here, neither methods relying on air gas monitoring nor airborne methods will be easily applicable because of the rapid dispersion of the CH 4 in the air. The use of unmanned systems (e.g., drones) may be possible but the rugged topography and often dense natural vegetation will limit their use in regions comparable to the investigated areas of the Subalpine Chains.
At the present technological state of the art, the best way to proceed will be a fastidious ground monitoring with regular stops to check for active leakage. Laser systems seem here to be more appropriate than other detection techniques. The low concentrations imply that the best results can be expected if the sensors integrated in the monitoring equipment point directly, orthogonally to the soil. Direct laser pointing will integrate a CH 4 content (often expressed in ppm) over the distance between the sensor and a reflective source (soil/rock). In the case of the MG site, our flux measurements show that several hundreds of ppm of CH 4 are emitted each second into the atmosphere at the seep (up to 1500 ppm·s -1 ). Even for lower natural fluxes, a laser beam crossing the emission area, even at some distance, will detect such a release. Remote methane leak detectors built for pipeline leakage monitoring may be appropriate candidates. They often have a sensitivity of 5 ppm·m, a working range of 0 to 50,000 ppm·m or more, and working distances from a few decimeters to 30-80 m [51][52][53][54][55]. Some sensors also exist as drone-mounted versions [54]. In the case of the MG site, such sensors should be able to detect the observed seepage at 10 to 20 m of distance (corresponding to max. 150 ppm·m and 75 ppm·m, respectively), longer distances being not compatible with the lengths of the thalweg sections. The counterpart is that such kind of work implies a huge workload and is time and cost intensive. It is not adapted to operational day-to-day site surveillance of hydrocarbon plays.

Conclusions
The three newly investigated gas seeps situated in zones where argillaceous gas-bearing Jurassic host rock outcrop in the French Subalpine Chains complete our previous investigations reported for the most prominent alpine "eternal fire," the Fontaine Ardente du Gua [3]. Again, methane is the dominant gas phase emitted to the atmosphere, over areas of variable size ranging from tens of square centimeters to tens of square meters. The integration of all data so far available for the French seeps leads to mean flux estimations with some statistical significance. Considering information from all macroseepage and miniseepage areas where a positive CH 4 flux is measured, mean CH 4 release is calculated at 76-79 kg per square meter per year. This is far above microseepage rates recently reported by Etiope et al. [9] thus confirming the superposition of diffuse and focused degassing processes at macroseeps, highlighting the role of faults and fractures in gas emissions even from shale formations. Contrarily to most oil-and gas-bearing sedimentary basins with known gas and oil seeps (e.g., the recent inventory for China [56]), the French alpine seeps seem not linked to tectonic trap structures, seeping through younger brittle deformations. All are directly related to outcropping argillaceous host rocks, but alpine thrusting and faults certainly play an important role in creating pathways for gas accumulation and migration, similarly to what was postulated for the Appalachian Mountains [6]. Linking structural analysis of fault patterns to gas emissions is certainly one of the most promising approaches for targeting ground-based investigations aiming at discovering new emitting structures in the restrained timeframe of environmental baseline assessment. The gas seeps described so far were found accidentally, by local inhabitants through spontaneous or induced ignition, bubbling in waterbearing creeks or by detecting the particular smell of higher alkanes or H 2 S. Systematic "blind" monitoring of black shale outcrops will probably fail to succeed if ground investigations are not guided, e.g., by airborne analysis of fault patterns. Future work will be oriented towards the testing of groundbased and airborne laser-based detectors to check their capacities of detection on known seeps. If these tests are conclusive, then trials in uncovered areas will be performed.
The systematic detection of seeps becomes crucial in the context of monitoring the potential impact of shale gas exploration or exploitation as well as of underground storage of natural gas and CO 2 . Environmental baseline assessment is now a well-established prerequisite of Carbon Capture and Storage (CCS). As we attempt to demonstrate here, it is highly hypothetical that baseline investigations, as good as they may be from a technological point of view, will reveal all natural small-scale gas seeps. When then a site will shift to an operational stage, leakage from the exploitation infrastructure may be alleged if so far unreported natural seepage is discovered. This was the case at the CCS site of Weyburn in 2011 [57,58]. Several studies have demonstrated the natural origin of the discovered CO 2 emission, related to specific processes in the soil formations close to the surface [59][60][61]. Nevertheless, eight years after the allegation has been raised, research is still ongoing aiming to discard or confirm CCS-induced leakage [62,63]. In the context of shale gas, the situation is even more complex as the use of chemicals in hydraulic fracturing of the argillaceous host rocks has raised strong public concerns on health effects, both of flowback water spills and leaks to surface water or groundwater resources and of stray gas. [64,65]. Much effort is put into optimizing water and groundwater monitoring [66,67]. Monitoring during the activity of the site may be better focused than initial investigations of gas emission baselines over large areas. However, as shown for the Weyburn case, and other CCS test sites or demonstrators, as well as for areas with active unconventional gas exploitation, intrinsic fingerprinting of gas species as well as investigating chemical and isotopic characteristics of groundwater bears a great potential for system failure detection [68][69][70][71][72][73] and will usefully complete the identification and quantification of gas emanations.

Data Availability
Data are given as supplementary material.

Conflicts of Interest
The authors declare that there is no conflict of interest regarding the publication of this paper. 13 Geofluids