3.1 Data sources

Table 1 summarizes how the four categories of geo-CH₄ sources were elaborated, showing data sources, the parameters used to define the “activity” (spatial distribution), the “emission factors” (fluxes) and the attribution of the isotopic CH₄ values. The list and web links of the sources of databases are reported in the Supplement (Sect. S6).

3.2 Gridding procedure

The gridding procedure is the same for each geo-CH₄ source category. Geo-CH₄ emission and isotope datasets were imported in an ArcGIS environment and saved in either point (OS and GM) or polygon (SS and MS) shapefile format, including coordinates and attributes (i.e. type of emission, area, emission factor, isotopic CH₄ values, plus geographical information, such as country and region). The grid was then joined with single OS, SS, MS and GM shapefiles. The final csv files include data fields that define the coordinates of each cell centroid, the variable name and its unit of measurement (t yr⁻¹ per cell for CH₄ emission and ‰ for δ¹³C-CH₄, according to VPDB, Vienna Pee Dee Belemnite, standard). For the grid cell values, the number zero (0) is used to indicate an actual or best emission estimate of zero (no seepage), whereas −9999 indicates lack of knowledge of the emission. Specifically, in the CH₄ output files, the following are used.

• A zero (0) value is used for

  • all offshore cells of the onshore seepage shape files (OS, MS and GM);

  • all onshore cells of the offshore seepage (SSs);

  • all onshore cells outside the potential MS area;

  • onshore cells without OS or GM sources;

  • offshore cells outside the SS areas.

• The number −9999 is used for

  • cells within SS areas where emissions are unknown.

The categories OS, MS and GM, due to the method of emission derivation (see related sections below), always have an emission value.

In the isotope files, the following points are considered:

• An isotopic value is reported in each cell that has a flux value.

• Where specific values are not available (as occurred for OSs and SSs), the global weighted average δ¹³C for the relative emission category is reported.

• Four decimals are used for global weighted average isotope values; this can help to trace back which cells are based on cell-specific data (with one decimal), and which contain weighted averages (four decimals).

• The value −9999 is only used for cells with no emissions in the corresponding CH₄ output files.

The application of such rules is described in the specific chapters of the four emission categories. Once individual OS, SS, MS and GM maps/files were produced, they were merged into a unified, total geo-CH₄ gridded map: emissions per cell were summed and δ¹³C values were averaged.

Figure 2Global distribution of onshore seeps (OSs).

Download

4.1 Global seep count and distribution

The spatial distribution (activity) of onshore seeps is derived from geographic coordinates of 2827 seeps, from 89 countries, reported in a global onshore seep dataset, which includes 1119 oil seeps, 846 gas seeps, 741 mud volcanoes and 121 gas-bearing springs. This dataset is an updated version of a previous inventory (named GLOGOS, reporting 2100 seeps) available from CGG (2015) and described in Etiope (2015). The global distribution of OSs is reported in Fig. 2.

The seeps listed in the dataset generally refer to individual focused vents (single macro-seep manifestations), but in several cases they refer to groups or clusters, or even wide zones of multiple seep points. A total of 612 seeps (569 gas–oil seeps and 43 mud volcanoes) could not be geographically located with precision and they are listed without coordinates (in addition to the 2827 seeps). The dataset, therefore, actually mentions a total of 3439 seeps or seepage sites, including 3396 gas–oil seeps and 784 mud volcanoes. The total number of 3439 OSs represents about 30 % of total seeps assumed to exist on Earth (≈10 000 was proposed by Clarke and Cleverly, 1991), but the present dataset includes the largest and more active seeps (especially for MVs) because they are more easily documented and have attracted attention for scientific research, petroleum exploration and natural heritage protection. Small or inactive seeps tend to be less observed and reported. In particular, the MV inventory is almost complete, probably missing smaller MVs in Asia. The gas–oil seeps in the dataset likely contribute more than 50 % of the previously estimated total gas–oil seep emission. Africa and South America likely host a larger number of gas–oil seeps and springs not documented in the dataset because of the paucity of specific investigations, especially in remote areas.

4.2 Attribution of CH₄ emissions to individual seeps

The attribution of CH₄ emission magnitudes to individual seep locations follows two different approaches for (a) gas–oil seeps or springs and (b) mud volcanoes (MVs).

4.2.2 Emission of mud volcanoes (MVs)

For MVs, emission values refer to the continuous quiescent degassing; i.e. they do not include emissions during episodic eruptions, as these are practically impossible to estimate for each MV. Eruptions were considered separately for the global emission estimate as discussed below. The quiescent emissions were attributed to each MV following the activity (area) and emission factor approach as explained below.

A precise evaluation of the MV areas was performed using accurate image (Google Earth) analysis. For each MV visible on Google Earth images, the area of the entire MV structure, including central craters and flanks, was estimated by drawing a polygon encompassing the mud cover and mound flanks. For smaller MVs, not visible on low-resolution Google Earth images or covered by vegetation, photos or information from published literature or web sources were considered to define the order of magnitude of the MV size. From two repeated image analyses the global MV area resulted to be about 680±40 km².

The MV emissions were then assessed using an updated dataset of fluxes measured from 16 MVs in Azerbaijan, Romania, Italy, Taiwan, China and Japan (Table S2), distinguishing between the macro-seepage (the focused emission from craters and vents) and miniseepage. Regression analysis between MV area, miniseepage and macro-seep flux of these measured MVs was used to derive miniseepage and macro-seep flux (and thus the total CH₄ emission) for each MV of the inventory, whose area was determined as previously indicated. The procedure is described in detail in the Supplement (Sect. S1.1)

Figure 3Distribution of the order of magnitude of methane emission from onshore seeps.

Download

4.2.3 The “big emitters”

There is a total of 76 OSs with emissions in the order of 10⁴ t CH₄ yr⁻¹ (i.e. that may emit at least 10 000 t CH₄ yr⁻¹ individually), and they can be considered big emitters. They typically refer to large, very active and frequently erupting MVs so their emission is estimated based on the emission factor and area approach described in the previous section. The 76 big emitters likely dominate the spatial distribution of CH₄ emissions (they represent 63 % of the total OS emission) and the weighted global mean isotopic value. As shown in Fig. 3, it is clear that, on a global scale, the Caspian and Middle East regions represent the main OS emission areas.

Figure 4Gridded map of OS methane emission. This map refers to the csv file “OS_output_2018”.

Download

4.3 Attribution of the δ¹³C-CH₄ value

For each seep the δ¹³C-CH₄ value is (a) measured, as indicated in the literature (available in the OS inventory; CGG, 2015), or (b) estimated on the basis of isotopic values, found from one of the following three sources, in priority order:

• from similar seeps occurring in the same basin (when these data are available);

• from reservoir gas in the same petroleum field, from the Sherwood et al. (2017) dataset or literature;

• suggested by local petroleum geology (existence of microbial gas, thermogenic gas, oil), when the previous procedures cannot be applied.

The OS emission-weighted value (Sect. 4.5.2) was used for gridding where the isotopic value could not be assessed. The global distribution of three classes of δ¹³C-CH₄ value is shown in Fig. S4.

4.4 OS gridding

The OS shapefile generated in ArcGIS was spatially joined to the $\mathrm{1}{}^{\circ }\times \mathrm{1}{}^{\circ }$ vector square grid. OSs occur in 616 cells, for a total emission of 3.9 Tg yr⁻¹ (Fig. 4). This is about 0.1 Tg yr⁻¹ higher than the actual sum of the seep emission in the inventory because of multiple counting of 57 seeps that occur exactly on the boundary of a cell.

Table 2Estimates of global CH₄ emission (Tg yr⁻¹) from OSs (onshore MVs and other seeps).

The abbreviation nd refers to values that are not determined. ^a Values include shallow submarine MVs; therefore they can be considered as upper limits for onshore emission.

Download Print Version | Download XLSX

4.5 Evaluation of global OS emission and δ¹³C-CH₄

4.6 OS uncertainties

Spatial distribution uncertainty. In the $\mathrm{1}{}^{\circ }\times \mathrm{1}{}^{\circ }$ grid, the uncertainty of the geographic distribution of the OSs is practically zero, as all identified seeps have geographic coordinates within an error <1^∘.

Emission uncertainty. The uncertainty of the modelled oil–gas seep emission (based on the method of value attribution described in Sect. 4.2.1) is maximum 90 % (1±0.9 Tg yr⁻¹). The uncertainty of global MV emission (48 %) was estimated by summing (a) the uncertainty of the estimated MV areas (6 %; see Sect. 4.2.2) and (b) the uncertainty of the modelled MV emission factor (42 %; see Sect. S1.1). Because oil–gas seeps and MVs account for 26 % and 74 % of total OS emission, respectively, the overall gridded OS emission uncertainty is about 58 % (3.8±2.2 Tg yr⁻¹).

δ¹³C-CH₄ uncertainty. The uncertainty of measured δ¹³C values (from literature) practically corresponds to laboratory analytical uncertainty (typically <0.1 ‰). The maximum uncertainty of the estimated δ¹³C values (based on criteria described in Sect. 4.3) is approximately within 15 ‰, i.e. half of the range of δ¹³C values for typical microbial (−80 ‰ to −60 ‰) and thermogenic (−50 ‰ to −20 ‰) gas. The uncertainty of the emission-weighted mean (−46.6 ‰) is mainly induced by the 76 big emitters, for which the δ¹³C values are available or estimated with good approximation (<±5 ‰), leading to a mean value of −46.7 ‰. The difference between global emission-weighted and 76 big emitters' average δ¹³C values is 0.1 ‰. The average order of magnitude of ±1 ‰ can be considered for the uncertainty of global emission-weighted δ¹³C value.

Figure 5Distribution of submarine seep (SS) areas. SS numbering refers to the list in Table S4 (circle symbols mark small area sites).

Download

5.1 Assessment of global SS area

A specific dataset of offshore seepage areas, in coastal regions and shallow seas (typically <500 m deep, which is generally the maximum depth of seeps that may affect the atmosphere; e.g. Solomon et al., 2009), was developed based exclusively on published literature (Table S4). The dataset includes the following:

• a.

  Submarine seeps (including mud volcanoes), where gas was observed to reach the sea surface via bubble plumes and the emission to the atmosphere was estimated, are included. Flux emission estimates are available from 15 zones (from focused, point-source, manifestations to wide regional areas) in the seas of Australia, Bulgaria, Brunei, Canada, Chile, China, Denmark, Georgia, Greece, Norway, Spain, Romania, Russia, Turkey, Ukraine, UK and USA.

• b.

  Submarine seeps in shallow seas, where gas was actually observed (also through hydro-acoustic images) to reach the surface but the output to the atmosphere was not provided, or where, due to the shallow seabed (<400–500 m), the methane is expected to enter the atmosphere, are included. These areas (16 zones) are offshore of USA, Canada, Mexico, the Netherlands, Denmark, France, Italy, Greece, Russia, Azerbaijan, Turkmenistan and Pakistan.

The dataset does not include deep-sea seeps or areas with gas-charged sediments (e.g. as those inventoried by Fleischer et al., 2001) that may release methane into the water column, but for which the possibility of injection into the atmosphere is scarce or unknown. The area and methane flux estimates reported in several papers were used here without critical evaluation. Geo-referenced polygons were created for each area (Fig. 5).

5.2 Attribution of seepage levels

CH₄ fluxes from the original publications (Table S4) are used in the gridded emission dataset.

5.3 Attribution of the δ¹³C value

The δ¹³C-CH₄ values of SSs are attributed on the basis of available literature or considering the geological setting (type of petroleum system, origin of the gas) of the seepage areas (italic values in Table S4) following the same criteria adopted for OSs. For four areas in Table S4 (China–Brunei offshore, Laurentian Channel and Grand Banks Downing Basin in Canada and eastern Kamchatka Shelf in Russia) it was not possible to attribute any theoretical δ¹³C value because the gas may actually be derived from either microbial or thermogenic sources, covering a wide range of isotopic values. In these cases, the global emission-weighted δ¹³C value of SSs (see Sect. 5.5) was used for these regions in the δ¹³C grids. The global map of δ¹³C for SSs is shown in Fig. S5.

Figure 6Gridded map of SS methane emissions. This map refers to the csv file “SS_output”_2018.

Download

5.4 SS gridding

The SS grid dataset was generated digitizing polygons of the SS areas from literature maps (see references in Table S4). The final shapefile contains 31 polygons characterized by the following variables: country, longitude and latitude of the polygon centroid, CH₄ output flux, area and average δ¹³C value of the emissions in each polygon. The value −9999 is reported for the missing emissions at 16 sites (sites 16 to 31 in Table S4). The SS layer was joined with the $\mathrm{1}{}^{\circ }\times \mathrm{1}{}^{\circ }$ vector grid and the resulting map is shown in Fig. 6.

5.5 Evaluation of global SS emission and δ¹³C-CH₄

The sum of CH₄ emissions from the 15 SS areas in Table S4 (which refer to published estimates) is ∼3.9 Tg yr⁻¹. This represents about 20 % of the theoretical estimate of global SS emission to the atmosphere (∼20 Tg yr⁻¹), derived by process-based models, proposed by Kvenvolden et al. (2001). SS emissions also occur in the other 16 areas reported in Table S4 and likely in many other sites not investigated yet. Among the areas with missing emission values, the Gulf of Mexico and the Caspian Sea are very likely major methane emitters, followed by the northern US Atlantic margin. It is difficult to evaluate whether by adding these missing SS emissions the total sum would approach the Kvenvolden et al. estimate of 20 Tg yr⁻¹ (however, we consider this global value as a theoretical reference for our SS gridded emission, not a target or actual value to reach). Evaluation of the SS emission factor (based on the reported area and total fluxes in Table S4) is also difficult because the areas indicated in the several works (see References in Table S4) often refer to the surveyed area and not to the actual area of seepage; in these cases, using the surveyed area would result in a strongly underestimated emission factor. However, using the relationship observed for the 15 “investigated” sites between the area (actual seepage or surveyed) and emission factor, the other 16 sites would yield total emissions of about 1 Tg yr⁻¹. This would bring the total CH₄ emission from the 15+16 sites of Table S4 to about 5 Tg yr⁻¹. Assuming that (a) SSs generally do not take into account the release of dissolved methane (i.e. only methane bubbles are accounted for) and (b) today unknown SS areas (not listed in Table S4) may have a seepage extent that does not exceed that of the investigated areas, it is plausible that global SS emission exceeds 5 Tg yr⁻¹. If the upper estimate for the East Siberian Arctic Shelf, 4 Tg yr⁻¹ (Berchet et al., 2016; i.e. twice the mean used in Table S4), is considered, then the global SS emission would exceed 9 Tg yr⁻¹. The SS emission-weighted mean value of δ¹³C-CH₄ is −59 ‰. The non-weighted mean value is −51.2 ‰.

5.6 SS uncertainties

Spatial distribution uncertainty. In the $\mathrm{1}{}^{\circ }\times \mathrm{1}{}^{\circ }$ grid, the uncertainty of the geographic distribution of the SSs is practically zero, as all seepage zones have geographic coordinates within an error <1^∘.

Emission uncertainty. The main uncertainty and control on the global gridded 3.9 Tg yr⁻¹ value is associated with the estimate of CH₄ emissions from the East Siberian Arctic Shelf, for which we used the central value (2 Tg yr⁻¹) of the range indicated by Berchet et al. (2016), i.e. 0–4 Tg yr⁻¹ (not very dissimilar from the estimate of 2.9 Tg yr⁻¹ suggested by Thornton et al., 2016). The other 15 SS areas, totalling ∼1 Tg yr⁻¹, have variable uncertainty, often not defined in the individual publications. With a ±2 Tg yr⁻¹ uncertainty for the Siberian Arctic Shelf, and arbitrarily assuming 10 % uncertainty for the other estimates, the overall SS gridded emission uncertainty would result in ±2.1 Tg yr⁻¹ (54 %).

δ¹³C-CH₄ uncertainty. The maximum uncertainty of the estimated δ¹³C values (based on criteria described in Sect. 5.3) is approximately within ±15 ‰, i.e. half of the range of δ¹³C values for typical microbial (−80 ‰ to −60 ‰) and thermogenic (−50 ‰ to −20 ‰) gas. The uncertainty of the emission-weighted mean (−59 ‰) is mainly controlled by emissions from the East Siberian Arctic Shelf, North Sea and Black Sea, whose δ¹³C values are available or estimated, ranging from −50 ‰ to −63 ‰. The overall uncertainty of the global emission-weighted mean is thus reasonably <±7 ‰.

6.1 Assessment of global MS area

The diffuse exhalation of CH₄, called microseepage (MS), is widespread throughout onshore petroleum fields all over the world. It is systematically observed in correspondence with anticlines and marginal (faulted) areas of gas–oil fields (Etiope and Klusman, 2010; Tang et al., 2017). The existence of macro-seeps (OSs) in a given region also implies a high probability that MS exists in that region, even if that region falls outside a known petroleum field. Therefore, as a proxy of the activity (spatial distribution) of MS, we considered the global area of petroleum fields and a global area including OSs defined as described below. This criterion is conservative as MS may also occur in sedimentary basins without known petroleum fields and OSs (Klusman et al., 2000; Etiope and Klusman, 2010). The assessment of the global petroleum field area (PFA) and global OS area (OSA) is discussed in the Supplement (Sect. S3.1 and S3.2). The total potential MS area (PMA) resulted in PFA + OSA = 13 033 000 + 85 900 = 13 118 900 km².

6.2 Attribution of MS levels

The level of MS CH₄ emissions was established on the basis of a statistical analysis of a MS flux dataset (see Sect. 6.2.1) and considering the theory of seepage migration mechanisms, for which the gas flux greatly depends on the permeability of the rocks, especially when induced by faults and fracture networks (Etiope and Klusman, 2010; Etiope, 2015; Tang et al., 2017). Accordingly, the attribution of the flux within the PMA (PFA + OSA) was done considering the presence/absence, in each cell, of three major geological factors, which are proxies of methane seepage and gas permeability, i.e. OSs, faults and seismicity, as explained in Sect. 6.2.2.

6.3 Attribution of the δ¹³C value

Measured and published data of δ¹³C-CH₄ in gas MS are scarce and only available for a few petroleum fields. However, during the seepage process (migration driven by pressure gradients, i.e. advection), the CH₄ isotopic composition does not change significantly, so that surface CH₄ flux basically has the same δ¹³C value of the original gas in the reservoir (e.g. Etiope et al., 2009). Therefore, for each field or basin, the MS δ¹³C value was taken from published data related to subsurface reservoirs. A limitation of this strategy is that in a given basin the MS gas may actually come from shallower reservoirs, not necessarily or not dominantly from the deep productive reservoirs, which are more frequently the literature source of the isotopic value. Therefore, in some cells the real isotopic value could be lighter than that used in the grid maps.

Accordingly we adopted the following procedure:

• When one or more seeps (OSs) occur in a petroleum field (in the “Petrodata” list), the average δ¹³C-CH₄ of those seeps was used for MS.

• In the absence of seeps, reservoir δ¹³C-CH₄ data were used; they were taken from the inventory described by Sherwood et al. (2017) or published literature. For the fields (in the Petrodata list) whose δ¹³C-CH₄ value is not reported either in Sherwood et al. (2017) or literature, a theoretical δ¹³C-CH₄ value was estimated on the basis of the type of gas (microbial or thermogenic) and maturity of the petroleum system.

The file contains 349 δ¹³C-CH₄ data points (from 891 petroleum fields). It was not possible to estimate a specific δ¹³C value for the remaining 542 petroleum fields. In these cases, the global emission-weighted isotopic value was used in the resulting empty cells, as described in Sect. 6.4.

6.4 MS gridding

PFA and OSA (described in the Supplement) were intersected with a high-resolution ($\mathrm{0.05}{}^{\circ }\times \mathrm{0.05}{}^{\circ }$) global grid. The $\mathrm{0.05}{}^{\circ }\times \mathrm{0.05}{}^{\circ }$ cell dimension corresponds to the maximum resolution that can be obtained using ArcGIS software (the software cannot handle shapefiles >2 GB). The high-resolution gridding was used to match, as much as possible, the PFA: gridded PFA is in fact 14 791 897 km², while the original PFA was 13 033 750 km². The high-resolution gridding also served to reduce the boundary effect, and thus the overestimation of the areas with MS enhancing factors, i.e. faults, earthquake and seeps (the larger the cells, the higher the probability that the cells include MS enhancing factors).

Figure 7Gridded map of MS methane emission. This map refers to the csv file “MS_output_2018”.

Download

As discussed in Sect. 6.2.1, only 57 % of PFA cells were considered to host MS. It was then necessary to delete 43 % of PFA cells. The cells were randomly deleted only among those that do not host MS enhancing factors (faults, earthquakes and seeps), i.e. empty cells (which are 93 % of total PFA). The overall PFA reduction of 43 % was obtained by deleting 54 % of the empty cells (resulting in a PFA of 8 408 360 km²). Combining PFA and OSA results in EMA (Table S6). The sequence of MS modelling is summarized in the block diagram of Fig. S10. The MS levels were then assigned to the $\mathrm{0.05}{}^{\circ }\times \mathrm{0.05}{}^{\circ }$ gridded EMA according to the presence of the factors influencing MS: (a) the presence of faults, (b) the presence of seismic activity and (c) the presence of macro-seeps (OSs), as follows.

• Level 1 was applied to cells without any geological factors.

• Level 2 was applied to cells with faults or earthquakes.

• Level 3 was applied to cells with faults plus earthquakes or oil seeps or gas-bearing springs.

• Level 4 is applied to cells with gas-seeps or mud volcanoes.

The resulting global MS CH₄ emissions are about 24 Tg yr⁻¹. The emissions per cell range from 14.7 t yr⁻¹ (cells of about 30 km²) to 29 446 t yr⁻¹ (cells of about 169 km²). The grid was then converted into $\mathrm{1}{}^{\circ }\times \mathrm{1}{}^{\circ }$ resolution for atmospheric modelling applications (Fig. 7). MS emissions occur in 3039 cells, ranging from 15 to 471 000 t yr⁻¹. The cell with the highest emission is located in the Caspian region (Azerbaijan). The sensitivity of the MS modelling is discussed in the Supplement (Sect. S3.3).

6.5 Evaluation of global MS emission and δ¹³C-CH₄

The global MS emission derivable by summing the emission from the cells of the four MS classes, about 24 Tg yr⁻¹ (Table S6), is within the range, 10–25 Tg yr⁻¹, previously suggested by Etiope and Klusman (2010). The emission-weighted δ¹³C-CH₄ resulting from gridded MS is −51.4 ‰ (non-weighted average is −46.4 ‰). This value is mostly influenced by areas with elevated MS of microbial gas, such as the Po Basin (Italy), the Transylvanian Basin (Romania) and the Powder River basin (USA). The global emission-weighted value was applied to cells without isotopic values.

6.6 MS uncertainties

Spatial distribution uncertainty. The uncertainty of the spatial distribution of MS depends on the assumption, supported by field measurements, that MS occurs significantly only within petroleum fields (PFA) and areas with seeps (OSA). The uncertainty of PFA depends on the Petrodata dataset of Päivi et al. (2007), discussed in Sect. 5.1, and it cannot be quantified. The uncertainty of OSA depends on the buffer applied to individual seeps, which was however defined by geospatial analysis (see Sect. S3.2).

Emission uncertainty. The uncertainty of the MS emission depends on the activity (EMA) and on the process-based model of attribution of the seepage levels (emission factors), and their statistical elaboration, discussed in Sect. 6.2 (see also Fig. S10). Changing activity by ±20 % and the emission factor by the 95 % confidence interval of the median, with different combinations, resulted in a total MS output ranging from 15 to 32.7 Tg yr⁻¹, with a mean of 23 Tg yr⁻¹, matching the first estimate (see Supplement). We can therefore set, approximately, a maximum uncertainty in the total MS output of ±9 Tg yr⁻¹ (about ±38 %).

The model was then tested comparing its output values with measured values. This comparison was possible for nine areas where the coordinates of the measurement points were identified. In all cases, measured and modelled values have the same order of magnitude, and in many cases the range of the MS level attributed by the model includes the mean value measured.

δ¹³C uncertainty. The uncertainty of individual MS δ¹³C values depends on the assumptions discussed in Sect. 6.3. The uncertainty of the emission-weighted mean (−51.4 ‰) is mainly controlled by the cells with larger MS emissions where δ¹³C values are estimated. When the cells with emission-weighted mean are excluded, the remaining 536 cells (at $\mathrm{0.05}{}^{\circ }\times \mathrm{0.05}{}^{\circ }$, over a total of 192 166) have emission ranging from 5623 to 8296 t yr⁻¹ and δ¹³C values from −65 ‰ to −35 ‰ (mean −53.4 ‰). The difference of this value with the emission-weighted mean, i.e. 2 ‰, may be considered as approximate expression of the uncertainty of the global emission-weighted mean.

7.1 Global GM distribution

The global distribution of CH₄-emitting geothermal–volcanic sites (GMs) generated here is based on an inventory of volcanoes and geothermal sites developed by the Global Volcanism Program (2013) (see sources of databases in the Supplement). This inventory reports all major volcanic–geothermal systems on Earth (2378 sites; Fig. S12). They include both Holocene systems (1307 sites distributed in 128 countries), and older, Pleistocene volcanic systems (1071 sites distributed in 119 countries), which represent geothermal areas. In order to convert the point data into more realistic areal data (polygons), an arbitrary buffer area of 4 km of radius was created for each GM point (the buffer area does not influence the overall emission estimate, being only a parameter guiding the gridding). It is important to outline that this inventory reports the “zones” of volcanic–geothermal sites, and does not list individual manifestations: for example, the numerous geothermal manifestations in central Italy are cumulatively included in a few lines, e.g. “Vulsini complex”, “Sabatini complex” and “Vulture”. Therefore each emission value, attributed as explained in Sect. 7.2, represents a “regional” GM emission.

7.2 Attribution of CH₄ emission levels

Methane flux measurements and regional total estimates in GMs are available only in a few cases (<100 sites), mostly in Europe (as reviewed by Etiope et al., 2007). The GM inventory refers to geothermal–volcanic areas where GMs are expected to occur, but their actual surface area is unknown. Therefore, even assuming an emission factor (from the limited flux dataset), it cannot be translated into emission for each GM site. In this work, theoretical numbers were adopted considering three classes of regional emissions: 500, 5000 and 10 000 t yr⁻¹, as central values of the ranges 100–1000, 1000–10 000 and 5000–15 000 t yr⁻¹, respectively. These ranges were derived from emission factors ranging from 1 to 150 t km⁻² yr⁻¹ (Etiope et al., 2007) applied on an area of 100 km², as an average order of magnitude of the extension of geothermal–volcanic zones (derived from the Global Volcanism Program, 2013). Although the GM emission grid developed here is expected to improve global CH₄ inverse modelling (as it includes previously neglected GM sources), the total GM emission estimate suggested by the gridding because of the uncertainty of the theoretical emissions, is not meant to update or refine the previous global GM emission estimate (derived by process-based modelling; Etiope, 2015).

The emission level was attributed based on

• a.

  the location of the geothermal site, which may be within or outside a sedimentary basin (Fig. S13);

• b.

  the concentration of CH₄ measured in the geothermal fluids, within and outside a sedimentary basin.

The amount of methane in a geothermal–volcano area depends, in fact, on the presence of sediments rich in organic matter, which may be a source of thermogenic gas in addition to the geothermal abiotic gas. The CO₂ ∕ CH₄ ratio of emissions to the atmosphere is in the order of 1000–10 000 in volcanic sites, with limited sedimentary contribution, and it ranges from 1 to 100 in geothermal systems characterized by important sedimentary covers. Sediment-hosted geothermal systems in sedimentary basins (e.g. Etiope, 2015; Mazzini and Etiope, 2017) show the highest CH₄ concentrations (lowest CO₂ ∕ CH₄ ratio). In addition, sedimentary basins hosting petroleum fields reasonably contain larger amounts of methane. The three classes of methane emissions are reported in Table S9.

Figure 8Gridded map of GM methane emission. This map refers to the csv file “GM_output_2018”.

Download

7.3 Attribution of the δ¹³C value

A specific dataset was compiled listing 98 published δ¹³C-CH₄ values of various, geographically dispersed, geothermal–volcanic systems in the world. The isotopic δ¹³C-CH₄ values range from −43.2 ‰ to −6.4 ‰, with an average of −26.7 ‰. The double-sided Grubbs test (Grubbs, 1969) identified four outliers; the mean δ¹³C-CH₄ value of the 94 values excluding the outliers is −26.5 ‰. It is known that geothermal methane in sedimentary basins, due to the presence of organic matter and related thermogenic gas, has a lower δ¹³C-CH₄ value compared to magmatic, sediment-free systems (e.g. Welhan, 1988). The NPP of the δ¹³C-CH₄ data shows a sharp deviation at about −29 ‰ (Fig. S14). This value is actually consistent with the isotopic boundary of dominantly thermogenic gas; we therefore used this value as the limit between GMs falling outside sedimentary basins and GMs within sedimentary basins. The mean values of the two classes (excluding the outliers) are summarized in Table S8.

7.4 GM gridding

The GM shapefile generated in an ArcGIS environment was spatially joined to the $\mathrm{1}{}^{\circ }\times \mathrm{1}{}^{\circ }$ vector square grid. The result is reported in Table S9 and mapped in Fig. 8.

7.5 Evaluation of global GM emission and δ¹³C-CH₄

The 2378 GM sites yield a total methane emission of about 5.7 Tg yr⁻¹, which is within the range of the latest global GM emission estimate (2.2–7.3 Tg yr⁻¹; Etiope, 2015). The emission-weighted mean value of δ¹³C-CH₄ for the GM emission is −30.6 ‰ (non-weighted mean is −27.5 ‰).

Figure 9Gridded map of total methane emission from OSs + SSs + MS + GMs. This map refers to the csv file “Total geoCH4_output_2018”.

Download

Table 3Global gridded, global extrapolated and global published geo-CH₄ emissions.

^* Including estimates from notes a, b and c. See also text below. ^a Not including MV eruptions. ^b Partial (estimated <50 %) gas–oil seeps emissions. ^c Excluding unidentified or not-investigated offshore seepage sites. ^d Etiope et al. (2008); Etiope et al. (2011) (see also Table 3). ^e Kvenvolden et al. (2001). ^f Etiope and Klusman (2010). ^g Etiope (2015).

Download Print Version | Download XLSX

7.6 GM uncertainties

Spatial distribution uncertainty. The uncertainty of the spatial distribution of GMs has the same uncertainty as the global distribution of geothermal–volcanic areas, derived from the Global Volcanism Program (2013).

Emission uncertainty. The gridded GM emission, equivalent to the sum of individual regional values attributed as described in Sect. 7.2, has an uncertainty of about 75 % (5.7±4.3 Tg yr⁻¹).

δ¹³C uncertainty. The uncertainty of emission-weighted GMs δ¹³C may refer to the average of the two values corresponding to the 95 % confidence interval of the means of the two groups of isotopic data (outside and within sedimentary basins) discussed in Sect. 6.3, i.e. ±2.5 ‰.

8.1 Global geo-CH₄ emission

The global geo-CH₄ emission distribution, obtained merging OS, SS, MS and GM grids, is shown in Fig. 9. The total gridded CH₄ emission is 37.4 Tg yr⁻¹ (Table 3, second column). The extrapolated gridded emission estimate including the factors not considered in the gridding procedure (i.e. mud volcano eruptions, existence of onshore and offshore seeps not included in the OS–SS inventories) is between about 43 and 50 Tg yr⁻¹ (Table 3, third column). These values are within the published global bottom-up estimates (Table 3, fourth column). The global extrapolated geo-CH₄ emission is then compatible with recent top-down estimates (about 50 Tg yr⁻¹ by Schwietzke et al., 2016; see also Sect. 9 for a wider discussion addressing the temporal variability of geological methane emissions). The scope of columns 3 and 4 in Table 3 (extrapolated and published emission estimates) is only to show that gridded emissions do not necessarily represent the actual global geo-CH₄ emission because the datasets developed for the gridding may not be complete or may not contain the information necessary for improving previous estimates. Considering the four geo-CH₄ source categories individually, the gridded MS and GM emission totals are, however, within published ranges. The differences between gridded and published OSs and SSs are largely due to the following:

• an incomplete OS dataset (it represents only 30 % of global number of seeps assumed to exist on Earth);

• a lower estimate of the global MV area (680 km² instead of 2800 km² assumed in previous works);

• an incomplete SS flux dataset (flux data missing from at least 16 areas with known gas emissions).

The gridded emissions may only represent an updated assessment of the global emissions for MS and MVs (part of OSs) because the gridding implied a careful assessment of the spatial distribution and emission factors for these types of geo-sources.

Figure 10Gridded map of integrated δ¹³C values of OSs + SSs + MS + GMs (emission-weighted within each category). “Weighted” (grey) refers to OS and MS sites where the weighted δ¹³C value (Table 4) is used to replace −9999. This map refers to the csv file “Total geoCH4_13C_2018”.

Download

8.2 Global geo-CH₄ δ¹³C

Based on the emission-weighted δ¹³C value for each category of emission (using the respective emissions from Table 3), the global geo-CH₄ emission-weighted average δ¹³C is −49.4 ‰, considering global emission estimates and −48.5 ‰ for gridded emissions (Table 4). The global distribution of the isotopic signature is shown in Fig. 10.

8.3 Uncertainties of gridded geo-CH₄ distribution, emission and isotopic values

The overall uncertainties of the spatial distribution of the geo-CH₄ sources, CH₄ emissions and emission-weighted average values of δ¹³C, depend on individual uncertainties of the four categories of seepage, as discussed in Sects. 4.6, 5.6, 6.6 and 7.6. These are summarized in Table 5.

Table 4Global emission-weighted δ¹³C values (‰).

Download Print Version | Download XLSX

Table 5Summary of uncertainty factors for the four types of gridded geological emissions.

^*Uncertainty of the global emission-weighted δ¹³C values of Table 4.

Download Print Version | Download XLSX