Introduction

Methane (CH4) is a ubiquitous substance that occurs in adsorbed, dissolved and free gas forms in a range of aquifer, surface water, soil and atmospheric environments1,2. In surface waters and the shallow subsurface, CH4 production and consumption are mediated by microbial processes that are stimulated by changes in redox conditions, availability of suitable fermentation substrates and electron acceptors2,3,4. The relative abundance of heavy and light stable isotopes of carbon (12C/13C) and hydrogen (2H/1H) that comprise CH4 is influenced by certain processes including: the type of methanogenic (production) and consumption pathways; transport processes such as diffusion and desorption; competition for substrates with other reducing organisms (e.g. SO4-reducers); and thermodynamic conditions3,5,6,7,8,9,10,11,12. As a result, isotopic data of CH4 can be ambiguous, particularly when interpreting data from larger scales, where multiple sources of CH4 exist, where CH4 has potentially moved or where thermodynamic conditions change. There has been a range of research dedicated to understanding these complexities3,4,5,6,8,13,14,15,16. Yet, while information on the complex behaviour of CH4 isotopes is readily available, it creates some uncertainty about the value of CH4 isotopes as indicators of broader processes, such as fugitive emissions or aquifer connectivity (see Figure S1 for examples of δ13C-CH4 values under a range of different pathways and conditions).

Recent rapid development of unconventional gas resources such as coal seam gas (coal bed methane) or shale gas (CH4 sorbed under pressure in coal measures or shale deposits) has spurred interest in understanding the extent of unconventional gas resources and the potential for gas migration within and between aquifers. Dissolved gas can migrate within and between aquifers either via advection or by a diffusion process12,17. Some previous research in shale-gas-bearing basins has shown that CH4 can migrate with brines from underlying gas-bearing aquifers but, in the absence of hydrocarbon reservoirs, it can also be generated in situ18,19,20.

This paper investigates origins and transport of dissolved CH4 in a hydrogeological setting where a shallow coal seam gas (CSG) reservoir underlies an important alluvial water resource (Condamine River catchment, Surat and Clarence-Moreton basins, Australia). We use a novel approach, employing a combination of hydrochemical, CH4 and isotope data with isometric log ratios (reactants and products) and Gibbs free energy calculations from key biological reaction processes to describe the thermodynamic constraints on CH4 in the alluvium and underlying coal measures. This approach addresses the complexity of CH4 production and consumption in the subsurface and the range of associated isotopic responses. The approaches/techniques used to address this complexity are outlined in Table 1. Previous interpretations of δ13C-CH4 data from a similar area21 are compared to new data presented here and conclusions regarding CH4 migration are reviewed. A minimum suite of parameters required to assess CH4 within and between aquifers is proposed for future monitoring and data collection.

Table 1 The applications of the combined approaches used to understand the origins and controls on CH4 in this study.
Full size table

Hydrogeological Setting

The Condamine River alluvium (the Condamine alluvium) occurs in the Condamine River catchment, which is a large subcatchment (30,451 km2) in the headwaters of the Murray-Darling Basin in southeast Queensland, Australia. This study focusses on the upper, central Condamine alluvium (Fig. 1a). Hydrogeology and hydrochemistry of the alluvium are summarised in previous published work22,23,24. The alluvium overlies the Walloon Coal Measures (the coal measures) and, on the western alluvial flank, parts of the Kumbarilla Beds, which are Jurassic sedimentary features of the Surat and Clarence-Moreton basins (Fig. 1b). At the bedrock-alluvial interface an impervious clay layer (termed: the “transition layer”) is proposed to limit interaction with the underlying coal measures; however, the spatial extent of this transition layer is not well known. In some cases, the alluvium has incised the coal measures by up to 130 m within a paleovalley (QWC25) (Fig. 1c). Weathered bedrock materials, including coal fragments, occur throughout the alluvium22. The alluvium is exploited for water reserves for use in large-scale irrigation, mainly cotton. Higher quality water is generally found in upstream areas of the study area, near Cecil Plains where hydraulic conductivity is higher22.

Figure 1
figure 1

Hydrogeological setting and study area, showing: (a) location of the Condamine River catchment and Surat/Clarence-Moreton basins in eastern Australia; and (b) conceptual cross section of the Condamine alluvium and adjacent sedimentary features. For (a), the basin boundary is defined as the Kumbarilla Ridge28 and references therein30. (a) was prepared using ArcGIS v 10.1 (www.esri.com) and modified in Adobe Illustrator CC 2014. For (b), the land surface and alluvial depth profile is real, as taken from the Condamine Groundwater Visualisation System (GVS)38: the outcrops and depth extent of the olivine basalt and sedimentary bedrock features have not been mapped in detail and are represented as conceptualisations based on interpretations of existing literature22,25,39 by the co-authors.

Full size image

Coal seam gas (CSG) reserves in the underlying coal measures are a significant economic resource and gas production in the study area is focussed on areas in the south west at depths of ~300–500 m, (see Fig. 1). Production of coal seam gas requires water to be extracted from the coal seam which has raised concerns about aquifer connectivity. The CH4 gas in the gas reservoirs is typically biogenic, it tends to be concentrated at geological structures, and coal seams are discontinuous26,27,28,29,30,31. Using δ13C-CH4 of free gas that was collected from degassing alluvial wells during pumping, a recent study concluded that CH4 leakage from the coal measures to the alluvium was occurring in some areas and this was used to infer aquifer connectivity21. However, this study did not collect any CH4 data (free or dissolved) from the underlying coal measures for reference. A previous study that examined δ7Li within and between coal measure and basalt aquifers found very low concentrations of Li in the alluvium when compared to the coal measures32. Assuming conservative behaviour of the Li ion33,34,35,36,37, these results suggest large-scale solute transport between these aquifers is not occurring. While the gas reservoir that underlies the Condamine catchment is relatively shallow compared to some other areas in the Surat Basin, the commercially viable gas reservoir that directly underlies the Condamine catchment is relatively deep (typically 300–500 m) when compared to the maximum alluvium depth (130 m). In this paper we refer to two areas of the coal measures as follows: (1) the CSG or gas reservoir (200–500 m) of the coal measures where commercial gas reserves are found; and (2) the shallow coal measures: shallower zones of the coal measures (<200 m) that are up gradient of the gas reservoir, but which are underlying or adjacent to the alluvium (see Fig. 1b).

Results and Discussion

Redox and salinity conditions

The deep gas reservoir is characterised by highly reduced SO4 (typically less than detection limit (DL) (1 mg/L, or 0.02 meq/L)), and brackish water (Cl = 1000–4500 mg/L or 28–127 meq/L). In the shallower coal measures the SO4 and Cl concentrations are more variable and show a positive relationship, ranging from <0.1 mg/L to 488 mg/L (10 meq/L) for SO4 and 82 mg/L (2.3 meq/L) to 4680 mg/L (131.8 meq/L) for Cl. The majority of shallow coal measure samples have SO4 concentrations below 50 mg/L (1 meq/L). A single coal sample from the shallow coal measures underlying the alluvium at Cecil Plains showed small amounts of pyrite; however, SO4 concentrations at this site ranged from 8–12 mg/L (0.16–0.25 meq/L), indicating that SO4 is not completely reduced at this site. Salinity in the alluvium is also highly variable (Cl ranging from 35–8700 mg/L, or 1–245 meq/L) and also shows a positive relationship with SO4, which ranges from <1 mg/L to 988 mg/L (20 meq/L). Peak Cl and SO4 concentrations are found in shallow (~20 m) wells. This is consistent with the findings of Owen and Cox23, which showed higher salinity is related to evapotranspiration processes.

NO3 is low in all aquifers: typically <0.05 mg/L (0.001 meq/L), with 8 samples having NO3 below DL (0.01 mg/L or 0.00016 meq/L) for the shallow coal measures, and ranging from 0.02 mg/L (3.23e-04 meq/L) to 2.3 mg/L (3.64e-02 meq/L), with three samples below DL (0.01 mg/L or 1.61e-04 meq/L), for the alluvium. With the exception of one shallow coal measure sample underlying a basalt outcrop (P19) and a shallow (~27 m) alluvial well (ID GM1076), all samples that contained CH4 had NO3 concentrations below 0.006 meq/L (0.37 mg/L) which is favourable for methanogenesis38. We found no NO2 above DL (0.01 mg/L) in any aquifer: this shows that significant denitrification is not occurring in these aquifers.

Data on dissolved Fe2+/Fe3+ and Mn species were not available, however, total dissolved concentrations of these ions were low in all aquifers in the study area. In the alluvium, Fe above DL (0.05 mg/L) was found at only 5 sites, (0.11–4.86 mg/L), while Mn concentrations were above DL (0.001 mg/L) at only 12 sites, the majority of which had Mn concentrations <0.01 mg/L. In the shallow coal measures, 8 samples had Fe above DL, with 6 of these being <0.8 mg/L, while only Mn concentrations were <0.09 mg/L at the majority (n = 12) of sites. These low values compare with production water which is highly reduced (SO4 < 1 mg/L) (see Supplementary Information Table S1).

Tritium

Tritium analyses (DL = 0.02 TU) were performed at selected sites: shallow coal measures (n = 5) and alluvium (n = 9). Significant 3H was observed for only one shallow coal measures well (P12, 0.95 TU): this well occurs on the basalt ranges under a thin basalt outcrop in a recharge area and contained no CH4. Only 2 shallow alluvial wells (~27 m (GM1076) and 41 m (GM1338)) were found to have detectable3 (0.05 TU and 0.22 TU, respectively): these were located ~6 and 16 km from the river, respectively. Only one of these wells contained CH4 (GM1076: 0.05 TU). Alluvial wells with no detectable tritium ranged from 18 m–89 m in depth, including 4 wells <40 m. The absence of tritium in the majority of shallow wells indicates limited to no modern recharge. While river recharge is considered important in this alluvial system24, we found no tritium in a 57 m deep alluvial well (GM0057) located approximately 1.4 km from the river.

Stable isotopes of chlorine (δ37Cl)

δ37Cl was measured for all samples containing CH4 within and between aquifers to provide an additional parameter for understanding possible CH4 migration via a diffusion pathway. δ37Cl for these samples ranged from −2.52‰ to −0.1‰ in the CSG reservoir; −1.11‰ to 0.8‰ in the shallow coal measures; and −0.72‰ to 0.89‰ in the alluvium.

Dissolved organic carbon

Dissolved organic carbon (DOC) is typically low in coal measures and alluvium aquifers, ranging from 0.3–1.6 mg/L, and 0.1–3.9 mg/L, respectively. The shallowest well (GM1073: 18 m) had the second highest DOC in the alluvial data set 0.4 mg/L, although DOC of all alluvial and coal measure samples could be considered low compared to other studies19,39,40. We found no relationship between DOC and CH4 concentrations in either aquifer.

The two alluvial wells that contained detectable tritium (GM1076 and GM1338) contained 0.2 and 0.4 mg/L of DOC, respectively. Two wells that occur at the alluvial-coal measure interface had DOC concentrations of 0.4 mg/L (GM1193:110 m) and 7 mg/L (IND2:85 m). The shallower sample contained no CH4 and presented as an anomaly in the dataset. δ7Li for this sample32 and lithological analysis confirm it is in the upper layer of the coal measures (~7 m below the alluvial basement). The other sample at the alluvial-coal measure transition zone (110 m) contained CH4 and occurs in an area where the alluvium appears to have incised the coal measures; drill logs show it contains both sand and coal fragments.

CH4 within and between aquifers

Dissolved CH4 concentrations in the deep CSG reservoir ranged from 2000 μg/L–25000 μg/L (n = 21). In total, 7 of the 14 shallow coal measure wells contain dissolved CH4 above DL (10 μg/L): concentrations ranged from 95–18000 μg/L. Of the 23 wells sampled in the alluvium, only 5 were found to contain CH4, with concentrations ranging from 10–535 μg/L. All alluvial samples with dissolved CH4 were found in monitoring wells which were sampled using low flow techniques.

In this study, CH4 is predominantly the sole hydrocarbon above DL (10 μg/L), with only 2 samples in the shallow coal measures (IND1 and IND3) containing small concentrations of ethane and propane above DL (10 μg/L).

Figure 2a,b conceptualise the spatial distribution of CH4 in the shallow coal measures and the alluvium, respectively. The dissolved CH4 in the alluvium occurred over a large depth range (~20–110 m) and over a large spatial area: CH4 distribution in the alluvium is relatively sparse, and peak alluvial-CH4 concentrations do not show any spatial relationship with peak CH4 concentrations in the underlying coal measures.

Figure 2
figure 2

CH4 μg/L contours for: (a) the coal measures, including the gas reservoir and the shallow coal measures; and (b) the alluvium. CSG production sites are marked by arrows in (a). Points in (a,b) represent sample locations. Maps were prepared using ArcGIS v 10.1 and modified using Adobe Illustrator CC 2014. Contours were determined using the spline tool in ArcGIS v 10.1 (www.esri.com), which interpolates a raster surface from points using a two-dimensional, minimum curvature spline technique and which passes a contour line through each measured point. Contours marked as 0 are based on measured CH4 below DL (<10 μg/L) at relevant points. This representation of CH4 μg/L distribution should be used for conceptual/visualisation purposes for this data set only, as methanogenic and methanotrophic conditions may only be favourable at discrete locations and because contours between points represent a conceptual change in the concentration gradient between measured samples only.

Full size image

δ13C-CH4 and δ2H-CH4 within and between aquifers

While thermogenic methane typically has more enriched δ13C values, biogenic CH4 can also have δ13C values within what is considered a typical thermogenic range. For example, the dominance of acetoclastic methanogenesis2,5,13,26,41,42, shifts in seasonal availability of the substrate43,44, enrichment of the CO2 pool as a result of on-going methanogenesis3, and anaerobic (AOM) or aerobic oxidation of CH48,45,46 can all produce CH4 that is relatively enriched in δ13C (see Figure S1 for a summary). Fractionation factors can offer insights into production and consumption pathways: αDIC-CH4 of ~1.07 and αH2O-CH4 of ~1.2 are typically indicative of CO2 reduction pathways, while of αDIC-CH4 ~1.04 and αH2O-CH4 ~1.4 are typically indicative of acetoclastic methanogenesis or an oxidation pathway3,5,47.

In the gas reservoir, the δ13C-CH4 and δ2H-CH4 values ranged from −58‰ to −49‰, and −210‰ to −198‰, respectively, and correlate with positive δ13C-DIC values (+9‰ to +23‰) (Fig. 3a). The αDIC-CH4 and αH2O-CH4 of CSG production water are consistently around 1.07 and 1.2, respectively, and there is a positive relationship between the δ13C-CH4 and the δ13C-DIC (Fig. 3a,b). This, in combination with no other hydrocarbons above DL, is indicative of a biogenic CO2-reduction pathway in a closed system (limited CO2 pool). This is synonymous with gas trapping on geological structures in closed environments, such as anticlines and synclines. The predominance of biogenic CH4 in the coal measures in this basin is supported by a number of other studies, with the most recent work suggesting microbial CH4 in these reservoirs was generated since the late Pleistocene26,27,28,29.

Figure 3
figure 3

Comparisons between: (a) δ13C-CH4 and δ13C-DIC; (b) δ2H-H2O and δ2H-CH4; and (c) δ18O-H2O and δ2H-H2O for samples that contain CH4 > 10 μg/L between aquifers in the study area. For (c): GMWL = Global Meteoric Water Line; LMWL = Local Meteoric Water Line at Toowoomba93. For (b): long-dashed lines = range of combined hydrogen isotope effects for CO2-reduction as reported in Whiticar3, being δ2H-CH4 = δ2H-H2O–165‰ (±15‰); and short-dashed lines = range of combined hydrogen isotope effects for acetoclastic methanogenesis in sulfate-poor systems as reported in Waldron et al.48, being δ2H-CH4 = 0.675 × δ2H-H2O–284‰(±6‰). For (b), arrows represent the range of isotope effects where a combination of methanogenic pathways has potentially influenced isotopes as reported in Whiticar3. For (a), Alluvium–Free CH4a = free gas samples taken from the well-head space of irrigation bores after extended pumping (up to 3 months) near Cecil Plains, as reported in Iverach et al.21.

Full size image

In the shallow coal measures, the δ13C-CH4 and δ2H-CH4 ranged from 80‰ to −50‰, and −310‰ to −210‰, respectively. In the case of the alluvium, the data showed a similar range of −78‰ to −49‰, and −315‰ to −186‰, for δ13C-CH4 and δ2H-CH4, respectively. The range of δ13C-DIC values was also similar between the shallow coal measures and the alluvium: −15.9‰ to −3.5‰, and −15.3‰ to −6.6‰, respectively. The δ13C-CH4 and δ2H-CH4 and associated fractionation factors indicate CO2 reduction is the dominant pathway in the shallow coal measures, but variability in this data for the shallow coal measures and alluvium suggest there may be multiple production and/or consumption pathways influencing CH4 in these aquifers. The enriched δ13C-CH4 (−50‰) and highly depleted δ2H-CH4 (<310‰), and carbon and hydrogen fractionation factors of ~1.4 for a single shallow coal measure (P7) and alluvial sample (IND4) (Fig. 3a,b), are synonymous with acetoclastic methanogenesis3,48. These occur in isolation: under basalt sheetwash near Bowenville (shallow coal measures sample), and on the opposite side of the alluvium near Stratheden (alluvial sample) (see Figs 2 and S2). Acetoclastic methanogenesis has not been observed before in the Walloon Coal Measures in the Surat and Clarence-Moreton basins, although evidence of this pathway has been observed at basin margins in other areas26,27,29,41,49,50,51. In both cases these samples are found at relatively higher salinity and SO4 concentrations: for the shallow coal measures sample, Cl = ~1775 mg/L or 50 meq/L, and SO4 = 480 mg/L or 10 meq/L; for the alluvial sample, Cl = 5990 mg/L or 168 meq/L, and SO4 = 144 mg/L or 3 meq/L.

The only coal measure samples that contained hydrocarbons in addition to CH4 were found at a nested site at Cecil Plains: these samples (IND1 and IND3) contained small concentrations of ethene (70 and 25 μg/L). ethane (30 and 25 μg/L) and propene (24 and <10 μg/L) which suggests a mixed thermogenic/biogenic gas component. However, the depleted δ13C-CH4 of these samples (−71‰ and −65‰, respectively) as well as the αDIC-CH4 and αH2O-CH4 indicate biogenic CH4 is dominant at these sites.

δ18O and δ2H in water

The CSG production water tends to be more isotopically depleted (ranging from −7.2‰ to −5.2‰, and −44.1‰ to −33.1‰, for δ18O and δ2H, respectively) than the shallow coal measures (ranging from −5.5‰ to −4.3‰, and −36.2‰ to −28.2‰, for δ18O and δ2H, respectively) and the alluvial water (ranging from −5.9‰ to −4.2‰, and −38.2‰ to −26.8‰, for δ18O and δ2H, respectively). This indicates that these deeper areas of the coal measures were recharged during cooler climates than the shallow coal measures and alluvium (Fig. 3c). These values are within the range previously reported for production water in the Surat Basin, which suggests recharge during the last glacial period in south east Queensland27,28. We found no evidence of a spatial relationship between the similarities in the stable isotopes of water from the alluvium and shallow coal measures samples and those from the gas reservoir (CSG production water): for example, the alluvial sample with depleted stable isotopes of water is found in a shallow well (18 m) located on the north eastern flank of the alluvium and is not related to the gas reservoir. Similarly, the most depleted shallow coal measures sample occurs in the ranges near a basalt outcrop. Some caution needs to be applied to interpretations of the stable isotope of water in CSG production water results because high rates of methanogenesis can influence the δ2H-H2O in closed systems41,51,52.

Assessing potential migration from the CSG reservoir to the shallow coal measures

At the depth interface between the gas reservoir and the shallow coal measures there is a distinct shift in the relatively enriched δ13C-CH4 values of the gas reservoir samples, towards more depleted isotope values for samples from the shallow coal measures. Diffusion of CH4 may lead to lighter δ13C-CH412 and a depletion of CH4 along a diffusion pathway12. Similarly diffusion of Cl would also lead to a distinct depletion of δ13Cl in combination with a decrease in TDS. However, for these data, an upward diffusion scenario from the CSG reservoir to shallower areas is not evidenced from the δ13Cl, TDS, CH4 or δ13C-CH4 data (Fig. 4a–d). This distinct change in the δ13C-CH4 values indicates that there is no evidence of leakage from the deeper gas reservoir to overlying shallow zones in the coal measures, either via diffusion or ebullition/advection. Therefore, the variability of the δ13C-CH4 in the shallow coal measures must be the result of changes in methanogenic pathways and/or consuming processes.

Figure 4
figure 4

CSG groundwater and other groundwater samples that contain CH4 > 10 μg/L, showing: (a) TDS versus screen depth; (b) δ37Cl versus screen depth; (c) CH4 versus screen depth; and (d) δ13C-CH4 versus screen depth.

Full size image

The influence of SO4 on CH4 in the shallow coal measures

A decrease in the αDIC-CH4 as the δ13C-CH4 become more depleted in the shallow coal measures suggests influences of different methanogenic pathways (Fig. 5a). This is generally associated with a depletion of SO4 (Fig. 5b). The presence of SO4 can limit methanogenic activity, particularly for CO2 reducers, because SO4-reducing organisms are better at accessing both H2 and acetate1,4,5,14,53,54. In most cases the SO4 reducers maintain H2 levels below a threshold at which CO2 reducers can compete, resulting in complete inhibition of CO2 reduction. In contrast, acetoclastic methanogens, despite having a slower growth rate, can compete for acetate with SO4 reducers to the point where both organisms can co-exist54. Therefore, as SO4 depletes in the coal measures, we can expect changes in the methanogenic community, from one where CO2 reduction is inhibited by SO4 reducers and where some acetoclastic methanogenesis can occur, to one where CO2 reduction dominates over acetogens. This has implications for the δ13C-CH4 and could explain why the αDIC-CH4 changes as the δ13C-CH4 depletes (Fig. 5a). This hypothesis is supported by a decrease in SO4 concentrations as the CH4 increases (Fig. 5e). A positive relationship between CH4 concentrations and δ13C-DIC (Fig. 5c) demonstrates active methanogenesis and, where SO4 becomes depleted and CO2 reduction becomes dominant, higher CH4 concentrations indicate higher rates of methanogenesis via the CO2-reduction pathway. Data do not indicate an influence of DO concentrations on CH4 or associated isotopes (Fig. 5d), although methanogens can tolerate low concentrations of DO55. Spatially variable CH4 in the coal measures is supported by other recent studies which suggested variability in recharge as a possible influence56,57. In this study, variability of recharge may be contributing SO4 (either through discharge or pyrite dissolution) and DO, particularly in the shallower zones.

Figure 5
figure 5

(a) δ13C-CH4 versus αDIC-CH4 values for all dissolved and free gas samples (regression line is for shallow coal measures (grey squares) only), as well as dissolved CH4 samples for shallow coal measures showing: (b) log SO4 meq/L versus αDIC-CH4 values; (c) δ13C-DIC versus log CH4 μg/L; (d) DO mg/L versus log CH4 μg/L; and (e) log SO4 meq/L versus log CH4 μg/L. For (a), Alluvium–Free CH4a = free gas samples taken from the well-head space of irrigation bores after extended pumping (up to 3 months) near Cecil Plains, as reported in Iverach et al.21.

Full size image

Thermodynamic controls on CH4 in the shallow coal measures

In order to further explore the potential dynamism between CH4 production pathways, SO4 reduction, potential anaerobic oxidation of CH4 (AOM) and their influences on carbon and hydrogen isotopes at these large scales, we use a novel combination of thermodynamic information and changes in the activities of reactive species and isotope data expressed as isometric log ratios. A key aspect of this approach is understanding the behaviour of H2, which is a rate-limiting reactant for both CO2 reduction and SO4 reduction, while other reactants, such as HCO3 and SO4 may also provide favourable/unfavourable conditions for certain microbial pathways in coal seams4,5,6,58,59.

A sequential binary partition is used to calculate each isometric log ratio60. The sequential binary partition for each reactant shown in equations (6–8) (CO2 reduction, SO4 reduction and AOM, respectively) is shown in Tables 2, 3 and 4, respectively. The activity of H2O is ignored in relevant reactions, since it is always ~1. All ilr-coordinates are calculated using equation (1). In all isometric log ratio (ilr) calculations, the first ilr represents the compositional changes in the reaction pathway (products versus reactants). As a result, the first ilr (ilr.1) for each reaction pathway is similar to the reaction quotient (Q) used to calculate the change in Gibbs free energy. Using this approach, the principles of compositional data analysis and the law of mass balance holds, such that changes in the composition of species subsequently change the composition of the reactants and products. Where the reaction pathway is limited by the availability of one or more reactants, the ilr.1 is expected to follow a linear relationship with the changes in Gibbs free energy. The remaining ilr-coordinates partition the reactants into subcompositions, thus describing the availability of reactants for the reaction.

Table 2 Sequential binary partition for the CO2 reduction pathway.
Full size table
Table 3 Sequential binary partition for the SO4 reduction pathway.
Full size table
Table 4 Sequential binary partition for the anaerobic oxidation of CH4 (AOM) pathway.
Full size table

For the CO2 reduction pathway scenario (Fig. 6a(i–iii)), a decrease of H2 relative to other reactants (H+ and HCO3−) (CO2-ilr.2) occurs as the reaction pathway proceeds (ΔG/e− become less negative). This can be interpreted as the consumption of H2 as methanogenesis proceeds and as SO4 is depleted. The inverse relationship with the CO2-ilr.1 (reactants vs products) shows that the availability of H2 in higher SO4 environments is limiting CO2 reduction pathways. A depletion in the relative R-δ13C-CH4 and enrichment of R-δ2H-CH4 isotope along this pathway support a shift from acetoclastic methanogenesis in higher SO4 environments where competition from SO4 reducers is higher to one where CO2 reduction becomes dominant in lower SO4 environments. In addition to low SO4 concentrations, low H2 and low HCO3 concentrations can also create more favourable conditions for CO2 reducers59.

Figure 6
figure 6

Microbial reaction pathways in the shallow coal measures (<200 m) for: (a) methanogenesis via CO2 reduction; (b) SO4 reduction; and (c) anaerobic oxidation of CH4 (AOM), showing comparison of isometric log ratios (ilr) derived from the partitioning of reactants and products (ilr.1–solid circles) and the partitioning of reactants (ilr.2–open circles) (see Tables 2, 3, 4 for respective SBPs) and: (i) changes in Gibbs free energy standardised to the number of electrons transferred for each reaction (8) (ΔG/e−); (ii) Rayleigh fractionation of δ13C-CH4; and (iii) Rayleigh fractionation of δ2H-CH4. Dashed arrows in (i) represent the direction in which the thermodynamic reaction proceeds (approaches equilibrium). Under acetoclastic methanogenesis the δ13C-CH4 can be relatively enriched, yet should become more depleted as CO2 reduction proceeds. In the same context, the δ2H-CH4 is highly depleted under acetoclastic methanogenesis, yet more enriched under CO2 reduction. As a result, a reciprocal response between carbon and hydrogen isotopes is expected as the reaction pathway changes: therefore, the R-δ13C-CH4 is defined as R = Ri f (1−α), while R-δ2H-CH4 is defined as R = Ri f (α−1), where f = 0 = min CH4. The acetoclastic sample is marked with a cross in (i). *Samples containing ethene (70 and 25 μg/L), ethane (30 and 25 μg/L) and propene (24 and 0 μg/L).

Full size image

For the SO4 reduction pathway (Fig. 6b(i–iii)), poor relationships between all SO4-ilr.2, ΔG/e− and isotopic responses was observed. This indicates different controls on the SO4 reduction pathway: it does not appear to be limited by the availability of reactants, including H2 and, with the exception of the obvious acetoclastic sample, does not appear to accompany a distinct carbon or hydrogen isotopic response.

Increases (less negative) in the ΔG/e− for the AOM pathway are accompanied by an increase in the relative concentration of CH4 to SO4 (AOM-ilr.3), which shows that as AOM proceeds, the system moves towards one where the CH4/SO4 ratio increases (Fig. 6c)(i–iii). This indicates that, as the AOM reaction approaches thermodynamic equilibrium, the amount of SO4 available for the reaction decreases, yet CH4 must continue to be produced. The availability of SO4 appears to be a limiting reactant for the AOM pathway. These relationships are accompanied by a relative depletion of R-δ13C-CH4 and enrichment of R-δ2H-CH4 values as CH4 concentrations increase. Any CH4 oxidation in higher SO4 environments, as well as the slow growth rate of acetoclastic methanogens, is likely to contribute to the relatively lower CH4 concentrations in higher SO4 environments. In some cases AOM can occur in tandem with methanogenesis61. However, due to generally low S2− and HS− being <DL for all samples, we do not expect the influence of AOM to be significant when compared with the influence of SO4 and shifts from acetoclastic methanogenesis to CO2-reduction. At an isolated site underlying a basalt outcrop (P19), NO3 concentrations were slightly above DL (0.01 mg/L) at 0.02 mg/L (3.23e-04 meq/L), but the highly depleted δ13C-CH4 (−80‰) at this site does not suggest oxidation via denitrification is occurring.

Assessing potential migration of CH4 from the shallow coal measures to the alluvium

Nested sites

Three (n = 3) nested sites that include wells in the underlying coal measure and overlying alluvium wells were sampled: (1) Cecil Plains; (2) Stratheden; and (3) Dalby (see Figure S2). At all sites CH4 was observed in the underlying shallow coal measures, or the Kumbarilla Beds, but no CH4 was found in the alluvial wells.

At the Stratheden nested well site (IND4, IND5, IND6), water levels are similar, indicating there is not a significant pressure gradient to induce groundwater flow, and the absence of CH4 in the intermediate well does not suggest upward CH4 at this site. The δ13C-CH4 of the alluvial CH4 at this site is more enriched (−50‰) when compared with the deeper Kumbarilla CH4 sample (−68‰): the highly depleted δ2H-CH4 (−315‰) of the alluvial sample at this nested site (IND4) indicates acetoclastic methanogenesis3,48.

At the Cecil Plains nested site (P20, IND1, IND2, IND3) the sample with peak DOC (7 mg/L) (IND2) occurred in the alluvial-coal measure transition zone (85 m, ~7 m below the alluvial basement), but the DOC of the overlying alluvial sample was significantly lower (0.3 mg/L). On the same note, the CH4 samples from the two coal measures samples at this site is accompanied by small concentrations of ethene (70 and 25 μg/L). ethane (30 and 25 μg/L) and propene (24 and <10 μg/L); yet we found no other hydrocarbons in the alluvial sample (DL for all hydrocarbons = 10 μg/L). Similarly, at the Dalby nested site (GM1390, GM1074) no CH4 was found in the alluvial well (see Figure S2 for details of nested sites). We conclude that no CH4 migration from the underlying coal measures into the alluvium is occurring at these sites.

Dissolved CH4 in the alluvium

Results show that the CH4 in the shallow coal measures that directly underlie the alluvium are depleted in δ13C-CH4 (−80‰ to −65‰), and have δ2H-CH4 between −222‰ and −209‰. Fractionation factors and thermodynamic results indicate CH4 in the shallow coal measures is generated predominantly via the CO2 reduction pathway, with SO4 concentrations being a major control. Subsequently, the assessment of potential migration of CH4 from the coal measures to the alluvium must consider this CH4 of the shallow (underlying) coal measures as the appropriate end member. The lack of evidence of CH4 leakage from the gas reservoir to the shallow coal measures, and the abrupt shift from enriched δ13C-CH4 (−58‰ to −49‰) of the gas reservoir to the depleted δ13C-CH4 of the shallow coal measures indicate that the migration of CH4 from the gas reservoir to the alluvium at these sites is not a plausible scenario based on these data. Variability in the TDS and CH4 concentrations, as well as the δ37Cl (Fig. 4), and similar ranges of δ13C-CH4 and δ2H-CH4 between the underlying coal measures and deep alluvial samples (Fig. 7d,e) also do not suggest diffusion of CH4 from the underlying coal measures to the alluvium.

Figure 7
figure 7

Comparison of various parameters in the alluvial depth profile for sites with dissolved CH4 > DL (10 μg/L), showing: (a) CH4 (μg/L), SO4 meq/L and DO mg/L; (b) Br/Cl ratio, NO3 mg/L and DOC mg/L; (c) ΔG/e− for CO2 reduction, SO4 reduction and AOM pathways; (d) the carbon isotopes of CH4 and DIC phases and their respective fraction factors; (e) the hydrogen isotopes of CH4 and H2O phases and their respective fractionation factors; and (f) CO2.ilr-2 (the ilr of [H2]/[reactants of CO2 reduction/SO4 reduction pathways] (see Tables 2 and 3)), and the saturation of indices of kaolinite and gypsum. Circles in (a,d) represent the sample point depth: this corresponds to the sample point for all parameters in all plots. Shaded areas in (d,e) represent the ranges of δ13C-CH4, δ2H-CH4, respectively for the coal measures that directly underlie the alluvium.

Full size image

No relationship between depth and CH4 concentration in the alluvium was observed at sites sampled in this study, with the highest concentrations occurring at ~60 m (Fig. 7a). Thermodynamic conditions in the alluvium are suitable for all reaction pathways to occur (Fig. 7c); however, the variability of CH4 concentration in the alluvium is related to the inverse of SO4 concentration (Fig. 7a), demonstrating the influence of SO4 reduction on methanogenic activity. High concentrations of SO4 accompany high TDS (salinity) and large decreases in the Br/Cl ratio (Fig. 7a,b). This shows that different controls on salinity influence these high SO4 concentrations. The relatively consistent δ2H-H2O at maximum salinity shows that the CH4 and related hydrochemical conditions are not necessarily related to different sources of water or simple evaporation processes, and are more likely to reflect the accumulation of salts, including gypsum, or transpiration at a less permeable zone. Interestingly, the high-SO4 zone coincides with depleted δ2H-CH4 values (-315‰) that indicate acetoclastic methanogenesis (Fig. 7e). This explains the more enriched δ13C-CH4 for this sample (IND4) and it is a similar scenario to that which we observed in the coal measures.

δ13C-CH4 and δ2H-CH4 values for two deep alluvial samples (GM1193 and GM0057: 110 m and 57 m, respectively) are within a similar range to that of the underlying coal measures (Fig. 7d,e). The deepest site (GM1193) is at/near the alluvial-coal measure transition zone. As stated previously, there are small coal fragments in the sandy alluvial deposits at this site, which could provide a methanogenic substrate. Furthermore, the δ18O of this sample is the most enriched of the CH4 data set and does not suggest a coal measure source (Fig. 3c). Similarly, the δ13C-CH4 and δ2H-CH4, as well as the αDIC-CH4 and αH2O-CH4, are also consistent with in situ CO2 reduction at the deepest site (Fig. 7d,e), and do not suggest an oxidation pathway or CH4 sourced from another area/zone.

Where peak CH4 concentrations occur (~57 m: GM0057), the αDIC-CH4 values are as low as ~1.04 (Fig. 7d), but the depleted δ13C-CH4 and enriched δ2H-CH4 do not support acetoclastic methanogenesis at this site. While αDIC-CH4 values of ~1.07 are typical of CO2 reduction, a fractionation factor of 1.04 is still within the range observed for CO2 reduction5,62,63,64. These fractionation factors can change between sites and as a function of in situ conditions5,6. Low ΔG/e− values at this site may also suggest some AOM has occurred (Fig. 7c). Alternatively the acetate- and H2-dependent methanogenesis may also occur concurrently during acetate fermentation in some cases65,66. Well GM0057 is located near the river and may also receive some river recharge. This well, and well GM1193, occur in a sandy area of the aquifer where pumping rates and recharge are likely to be relatively higher than areas around Dalby; this could explain slightly higher DO concentrations (Fig. 7a). Methanogenesis may also persist in the presence of low DO concentrations55, and mixing of slightly oxygenated water (river recharge) and the dissolution of carbonates may explain the relatively lower αDIC-CH4 values at GM0057.

In the shallow alluvial zones, fluxes in the type and rate of methanogenesis could be influenced by wetting and drying periods that result in dissolution or precipitation of minerals such as gypsum (Fig. 7f). In addition, clay mineral content has also been shown to influence CH4 concentrations, with high clay content capable of trapping CH467. Furthermore, some clays (e.g. kaolinite) preserve organic matter better than others68. For samples analysed in this study, kaolinite saturation indices are highest in the high-SO4 zone where acetoclastic methanogenesis dominates (Fig. 7f), which also accompanies a peak in DOC concentrations (Fig. 7b). The presence of kaolinite clay lenses in shallow areas may have a dual effect on methanogenic activity by preventing flushing and promoting salinization that result in higher SO4, as well as the preservation of some organic matter that allows fermentation processes to persist. The proportion of [H2] to other reactants ([HCO3] and [H+]) (CO2.ilr-2) in the more saline/high SO4 zone increases (ratio of H2 to HCO3 increases), despite CH4 being low (IND4): this indicates a fermentation process by SO4 reducers and acetate-dependent methanogens.

While AOM is thermodynamically favourable, slightly more depleted δ13C-CH4 and αDIC-CH4 values ~1.07 in the shallower zones do not suggest significant AOM is occurring (Fig. 7d,e). Mixing processes in the shallow alluvium may create scenarios where water with CH4 is mixing with water with higher concentrations of redox species. For one sample (GM1076), a small increase in the NO3 concentration is evident (Fig. 7b) and the oxidation of small amounts of CH4 via denitrifying bacteria cannot be completely ruled out69,70, although we found no NO2 above DL at any sites, and fractionation factors support a CO2 reduction pathway. Substrate depletion may also explain enriched δ13C-CH4 in these shallow zones3. Some caution should be applied when drawing conclusions for the shallowest alluvial samples (GM1073 and GM1076) because the measured δ13C-CH4 was at or near the limit of quantification (0.8 nanomoles) for the analytical method (this is not the case for δ2H-CH4). However, we note that deeper alluvial wells in this area of the alluvium, which is adjacent to the deep gas reservoir, did not contain CH4 above DL (10 μg/L). As a result, CH4 migration from the underlying coal measures in this area does not seem likely at these sites.

A conceptual model of CH4 within and between aquifers

A conceptual model (Fig. 8) summarises the major controls on CH4 within and between aquifers, including:

  1. 1

    Closed system conditions leading to enriched δ13C-CH4 and positive δ13C-DIC in the deep gas reservoir (200–500 m); and

  2. 2

    The presence SO4 concentrations and its influence on methanogenic pathways, including shifts from the acetoclastic pathways in shallow, brackish- and high SO4- zones to a dominance of CO2 reduction in deeper, low SO4 zones, in both the shallow coal measures and the alluvium.

Figure 8: Conceptual model of the behaviour of carbon and hydrogen isotopes in CH4 and respective DIC and water phases in the alluvium and the underlying coal measures.
figure 8

The two graphs at the top are related to the alluvium only. The δ13C-CH4 in the deep gas reservoir (200–500 m) is typically enriched, relative to the δ13C-CH4 in the shallower (<200 m) parts of the coal measures. In both the alluvium and the shallow coal measures CH4 concentrations are controlled by SO4 concentrations, with acetoclastic methanogenesis detected in shallow high-SO4 zones. CO2 reduction is the dominant pathway in both the coal measures and the alluvium. In the shallow coal measures anaerobic oxidation of CH4 and acetoclastic methanogenesis maintain depleted (~-50‰) δ13C-CH4, but, as SO4 depletes with depth, CO2 reduction becomes the dominant pathway and the δ13C-CH4 subsequently depletes (~−80‰). A single CH4 sample was detected in the Kumbarilla Beds, but there are insufficient wells in this formation to accurately assess controls on CH4; however, a shallower well in Kumbarilla beds at the same site contained no CH4.

Full size image

The inverse relationships between CH4 and SO4, and associated isotopic responses and thermodynamic conditions, in the shallow coal measures and alluvium are consistent with in situ CH4 production in other freshwater and brackish environments3,4,5,6,71. This, combined with results at nested sites and an absence of CH4 > DL (10 μg/L) in the alluvium, does not suggest large-scale migration of CH4 from the underlying coal measures is occurring.

Due to the complexity of methanogenesis and methantrophy in the subsurface, different pathways and sources can result in similar δ13C-CH4 values (see Figure S1), and CH4 and δ13C-CH4 are not likely to be spatially consistent. Enriched δ13C-CH4 from CSG production water can pertain to gas trapping scenarios at discrete locations3,26,72 and these values are not necessarily representative of CH4 in the entire aquifer. For future studies that are concerned with understanding CH4 behaviour in the subsurface over large areas and/or associated with CSG, we propose the following parameters as a minimum standard for data collection: δ13C-CH4 and δ2H-CH4, δ13C-DIC, major ions, pH and SO4 and S2− (other redox species, such as Fe and NO3, may also have some value). Researchers are encouraged to prepare comprehensive data sets of a range of parameters that allow conceptual models of the extent, and influences on, CH4 within and between aquifers to be described and built upon over time. The conceptual model outlined here (Fig. 8) provides a basis for doing this in this catchment. More sampling to identify the presence of methanogenic consortia (culturing studies) within and between aquifers, including the extent of acetoclastic methanogens, would build on the information collected in this study.

Comparisons with free gas measurements from alluvial wells

The results presented here are not in agreement with another study in the Cecil Plains area which used δ13C-CH4 of free CH4 taken from multi-screened irrigation wells during pumping to infer CH4 leakage from the coal measures to the alluvium at four sites21. That study proposed the following be met to infer CH4 migration from the underlying coal measures:

  1. 1

    DOC > DL, and 3H < QL (0.04 TU), where QL is quantification limit (this relationship inferred a potential source of coal measure groundwater/CH4); and

  2. 2

    Samples must sit on a mixing line between 1/CH4 and δ13C-CH4, with a y-axis intercept with a δ13C-CH4 value of −55.9‰.

That study assumed that the δ13C-CH4 value of −55.9‰ used in their mixing line is representative of the CH4 in entire coal measure aquifer, and that there are only two sources of DOC: river recharge or discharge from the coal measures. A δ13C-CH4 value of −50.8‰, based on a single atmospheric measurement downwind of a CSG production water storage pond, was also used as a reference (end-member) value for the coal measures aquifer. However, that study did not take any samples from the coal measures, either underlying the alluvium or in other areas, for reference.

Relationships between DOC and CH4

We found DOC in the alluvium (and the coal measures) to be relatively low, yet within a consistent range, regardless of distance from the river or tritium activity. Advanced analytical techniques are required to confidently detect tritium at low TU. We used a highly sensitive tritium analytical technique (DL = 0.02 TU)73, yet only found tritium >DL at two alluvial wells (GM1076 and GM1338). Iverach et al.21 suggested that, where tritium was below QL, yet DOC is present, a source of DOC, in addition to river recharge, must be present. These authors proposed that “upwards migration of CH4 from the coal measures would be the most likely source” of DOC in the alluvium at these sites; however, CH4 is not part of the DOC pool.

Other studies have indicated that a diffuse recharge component over the alluvium is possible in this catchment23,74,75, which may contribute to the DOC pool in the alluvium. In addition, DOC can diffuse through clays and/or be preserved by some clays such as kaolinite, and DOC can also be generated in situ in the subsurface from sedimentary sources39,68,76. Therefore, DOC and CH4 are likely to be associated with different sources and transport mechanisms. Furthermore, CO2 reduction is the dominant methanogenic pathway in the coal measure and alluvial aquifers, and this pathway does not rely on DOC as the energy source, rather it uses H22 (equation 6). We suggest that a more comprehensive research approach is needed to better understand relationships between DOC, age tracers, such as tritium, and methanogenesis within and between aquifers before combinations of these parameters can be used to confidently validate aquifer connectivity, particularly when working at large scales.

Describing the coal measure CH4 end member

The enriched δ13C-CH4 value (55.9‰) estimated for the regression line used by Iverach et al.21 to infer CH4 leakage from the coal measures to the alluvium is within the range of the δ13C-CH4 observed for the deeper gas reservoir (>200 m) sampled in our study (−58‰ to −49‰), and other gas reservoirs in the Surat Basin27,28. However, it contrasts with the depleted δ13C-CH4 (−80‰ to −65‰) that we observed for the shallow (<200 m) coal measures that underlie the alluvium. In this study area the CSG reservoir occurs in deeper zones (>200 m) of the coal measures where gas trapping occurs on the north-western flank of the alluvium (Figure S2). These conditions produce enriched δ13C-CH4 and high, positive δ13C-DIC (Fig. 4a) in the gas reservoir that do not occur in the shallower coal measures directly under the alluvium. High and positive δ13C-DIC can be an excellent indicator of CSG production water migration77; yet highly enriched/positive δ13C-DIC values were not found in the shallow coal measures or the alluvium, either in this study or by Iverach et al.21. The only enriched δ13C-CH4 (~50‰) we observed for the shallow coal measures was an isolated acetoclastic CH4 sample (P7) that underlies basalt sheetwash (see Figure S2).

The majority of δ13C-CH4 of free CH4 measured in Iverach et al.21 are similar to background CH4 concentrations observed in that study and for ambient air in the southern hemisphere observed in other studies78,79,80. It is possible that most of the CH4 analysed in Iverach et al.21 were composed of atmospheric CH4. Additional sampling (preferably using low flow techniques) to measure the degassing rate/potential from alluvial groundwater would also assist in more accurately describing the proportion of atmospheric versus degassed CH4 in the well-head spaces measured by Iverach et al.21. Where mixing with atmospheric and subsurface-derived CH4 is shown to occur, simple mixing lines may be inadequate to understand mixing of the three theoretical end members that should be considered under these potential inter-aquifer CH4 migration scenarios: i.e. (1) atmospheric CH4; (2) alluvial-derived CH4; (3) CH4 that has migrated from other aquifers.

Hydrogen isotope analyses can reduce uncertainties associated with interpretations that are based solely on δ13C-CH4 values, as presented in Iverach et al.21. Atmospheric CH4 tend to be much more enriched in δ2H-CH4 values (−82‰) compared to biogenic CH4 (−160‰ to >−400‰)3,48,80. In addition, CH4 oxidation could partly explained the enriched δ13C-CH4 values (−47.4‰ to −38.8‰) measured in Iverach et al.21. Hydrogen isotope data can also provide information about the influence of oxidation, as well as different production pathways, on the isotopic composition of CH43,48.

Conclusions

Using a comprehensive data set (dissolved CH4, δ13C-CH4, δ2H-CH4, δ13C-DIC, δ37Cl, δ2H-H2O, δyO-H2O, Na, K, Ca, Mg, HCO3, Cl, Br, SO4, NO3 and DO) this study described hydrochemical/thermodynamic controls on CH4 in a deep coal seam gas (CSG) reservoir (200–500 m), shallower areas of the same coal-bearing formation (the Walloon Coal Measures) (<200 m) and the overlying Condamine River alluvium (Surat/Clarence Moreton basins), eastern Australia. Enriched δ13C-CH4 (−58‰ to −49‰) and positive δ13C-DIC (+9‰ to +23‰) in the deep gas reservoir are synonymous with biogenic methanogenesis in closed-system conditions and gas trapping on geological structures. Evidence of leakage from the deep gas reservoir, either via diffusion or ebullution/advection, was not observed, with δ13C-CH4 of the shallow coal measures underlying the alluvium being depleted (−80‰ to −65‰). Importantly, this study demonstrates that, when evaluating potential CH4 migration associated with CSG, the enriched δ13C-CH4 of CSG CH4 is not necessarily the appropriate isotopic end member because the δ13C-CH4 in areas outside of gas reservoirs, yet within the same sedimentary formation, can be distinctly different due to different hydrogeological and microbial conditions. We found the δ13C-CH4 of the alluvium falls within a similar range to that of the shallow coal measures.

Using a novel application of isometric log ratios, this study demonstrated a simple method of providing insight into the microbial controls on δ13C-CH4 and δ2H-CH4 isotopes in the subsurface. The major controls on CH4 in the shallow coal measures and the alluvium were found to be: (a) the presence of SO4 and associated competition between SO4 reducers and CO2 reducers; and (b) shifts from acetoclastic methanogenesis in shallow, high-SO4 zones to the dominance of the CO2 reduction pathway in low-SO4 environments. AOM was found to be thermodynamically favourable but there was no evidence of large-scale, significant AOM in the shallow coal measures or the alluvium. Overall, this study did not find conclusive evidence of CH4 migration to the alluvium from the underlying (shallower <200 m) coal measures, but results do suggest small concentrations of CH4 are likely to be generated in situ in the alluvial aquifer at these sites. This study provides a comprehensive assessment using novel samples. More research and sampling in the area, including culturing studies of methanogenic consortia, will improve our understanding of the nature and extent of CH4 within and between aquifers.

Methods

Sample collection

Samples were collected from 61 wells, including: (a) monitoring wells and stock and domestic wells where there was sufficient space to lower a bladder pump81; (b) irrigation/domestic wells that already contained submerged electric pumps; and (c) production water wells. Where a bladder pump could be lowered into a well, the low-flow sampling technique was applied using a flow-through cell81. For government monitoring wells, where monitoring wells had multiple screens, the bladder pump was placed at the interval of the lowest screen. Infrastructure at irrigation/domestic wells prevented well dipping: in these cases a conservative water level estimate of ~75% of well depth was applied to consider a purging volume. Sampling coincided with landholders pumping schedules and, as a result, the majority of irrigation/domestic wells had been purged by at least 3 x well volume upon arrival on site. In all cases (low-flow sampling and irrigation/domestic-well sampling) sampling was only initiated after hydrochemical parameters (pH, temperature, specific conductance and DO) were stabilised81,82. Due to limited infrastructure, two operating windmills were sampled in recharge areas on the ranges (P9 and P16): in these cases sampling was conducted after a minimum of 3 days of consistent moderate-strong wind (consistent pumping to purge the well) and after stabilisation of hydrochemical parameters (pH, temperature, specific conductance and DO) were confirmed83. Coal seam gas production wells (deep gas reservoir) are constantly pumping and were considered adequately purged upon arrival on-site. Production water from CSG wells was sampled at an outlet of the extraction well prior to the gas-water separator.

Samples were taken from alluvium (n = 23), the Kumbarilla Beds (n = 3), the shallow Walloon Coal Measures (WCM) (<200 m) (n = 14) and from the deeper (200–500 m) gas reservoir in the coal measures (n = 21). Two of the alluvial samples taken were at the alluvial-WCM interface (see section 2). Wells were selected based on drill log information and previous interpretations of hydrogeology in the catchment23.

Samples for dissolved CH4 were collected in glass vials with rubber septums and no headspace (preserved with sulfuric acid). Samples for δ13C-CH4, δ2H-CH4 and δ13C-DIC were filtered through 0.2 micron filters and collected in 12ml entertainer vials with rubber septums and no headspace. Samples for cations, dissolved metals, δ37Cl and Br were filtered through high capacity in-line 0.45 μm polyethersulphone filters. Cation and dissolved metal samples were preserved in the field using HNO3 to pH <2. Samples for 3H and anions, NO3/NO2, S2− and unionized HS were collected in unfiltered 1L Nalgene bottles, and HDPE bottles respectively (APHA Table 1060:1). NO3 and NO2 samples were preserved in the field using H2SO4 to pH <2. S2− and unionized HS samples were preserved in the field with Zn acetate/NaOH. Bottles not containing a preservative were rinsed three times with sample water prior to collecting a sample. All samples, with the exception of 3H, were placed immediately on ice and stored on ice in the field, then in dark cold rooms (<4 °C) until analysis.

Sample analysis

Samples were analysed for pH, DO, specific conductivity (conductivity) and temperature using a YSI physico-chemical meter in the field (YSI Professional Plus). Water samples were taken and analysed in the laboratory for major and minor ions (APHA 2320; APHA 3125B) including SO4 (APHA 4500 SO4-E–laboratory 0.45 μm filtered), and Br (APHA 4110 B) as well as unionised HS (APHA 4500-S2-H) and S2− (APHA 4500-S2–D), NO3 and NO2 (APHA VCl reduction 4500 NO3− + NO2−B), Fe and Mn (APHA 3125B ORP/ICP/MS Octopole Reaction Cell) and dissolved CH4 concentrations (including C1–C4 gases, DL = 10 μg/L: ALS EP033) at the Australian Laboratory Services laboratory, Brisbane, Queensland, and at Queensland Health Scientific and Forensics services laboratory (Br). Bicarbonate values are reported as bicarbonate alkalinity. All major and minor ions, and dissolved C1-C4 hydrocarbons were analysed within ~7 days of collection in the field.

δ2H and δ18O were measured using a Los Gatos Research Water Isotope Analyzer (QUT, Institute for Future Environments), with replicate analyses indicating an analytical error of 0.04‰ to 0.45‰, and 0.001‰ to 0.7‰, respectively.

δ13C-CH4 and δ2H-CH4 were measured using a ThermoScientific PreCon concentration system interfaced to a ThermoScientific Delta V Plus isotope ratio mass spectrometer at the UC Davis Stable Isotope Facility. Standard error of analysed samples was ~0.1‰, for δ13C-CH4 and ranged from 0.9–1.7‰ for δ2H-CH4, and limit of quantification = 0.8 and 2 nanomoles respectively. δ13C-DIC were also measured at the UC Davis Stable Isotope Facility using a GasBench II system interfaced to a Delta V Plus isotope ratio mass spectrometer. Standard error of δ13C-DIC ranged from 0.04–0.09‰: limit of quantification = 150 nanomoles.

δ37Cl were measured using a stable isotope ratio mass spectrometer at Isotope Tracer Technologies in Waterloo, Canada. Standard error ranged from 0.03–0.16‰.

The DOC analyses were performed on a Dohrmann DC-190 Total Carbon Analyzer at the Earth and Environmental Sciences at the University of Waterloo, Canada. DOC storage times (0.45 μm filtered, dark storage <4 °C) prior to analysis ranged from 260–620 days. Data Tables S1 and S4 report the measured and corrected DOC values, as per Peacock et al.84: these values are broadly similar with modelled loss of DOC being minimal due to low DOC concentrations.

Tritium (3H) samples were vacuum distilled and electrolytically enriched prior to liquid scintillation spectrometry analysis by Quantulus ultra-low-level counters at GNS, New Zealand73. The sensitivity is now further increased to a lower DL of 0.02 TU (two sigma criterion) via tritium enrichment by a factor of 95, and reproducibility of tritium enrichment of 1% is achieved via deuterium-calibration for every sample. The precision (1σ) is ~1.8% at 2 TU.

Data preparation

All major ion data was above DL (DL) of 1 mg/L, with the exception of SO4 (n = 24). All S2− measurements with the exception of 1 coal measures sample (well IND3; S2 = 0.5 mg/L, or ~0.008 meq/L) were below DL (DL = 0.1 mg/L, or 1.56e-03 meq/L). Low SO4 and S2− concentrations are expected for reduced environments where methanogenesis occurs. Values below DL were imputed using the R package zCompositions via the log ratio Data Augmentation function (lrDA): this function is based on the log-ratio Markov Chain Monte Carlo MC Data Augmentation (DA) algorithm85.

Deriving isometric log ratios

The isometric log ratio (ilr) uses a sequential binary partition (Table 5) to describe orthonormal bases to which correspond D-1 Cartesian coordinates (ilr-coordinates): these orthonormal coordinates, called balances, are orthogonal86. This technique removes potentially spurious correlation caused by scaling and allows the ratios of parts and subparts to be elucidated, even when the concentrations of different parts are relatively small compared to other parts. Here we use ilrs to investigate subcompositional relationships between reactants and products in a number of thermodynamic reaction pathways (CO2-reducing methanogenesis, SO4 reduction and anaerobic oxidation of CH4 (AOM)). These relationships are compared to isotope fractionation of the δ13C-CH4 and δ2H-CH4 under various conditions. This approach allows subcompositional behaviour of dissolved constituents to be compared to isotopic responses, in order to demonstrate the relationship between methanogenic activity/pathways, thermodynamic conditions and hydrochemistry.

Table 5 Sequential binary partition of a four-part composition (x1, x2…….x4) deriving three orthonormal coordinates (z1, z2 and z3) for ilr calculation.
Full size table

Each partition divides the composition into separate parts (xi and xj). For thermodynamic reaction pathways, we use the first partition to separate the activity of the element (represented by [element]) from reactants and products in each reaction, with the following partitions separating the reactants. Once a sequential binary partition is described, the i-th ilr balance is computed as

where ri and si are the number of parts coded in the sequential binary partition as +1 and −1, respectively86. Isometric log ratios were calculated using CodaPak 2.1087.

Describing isotope partitioning (fractionation factors)

The partition of isotopes between phases, e.g. between the dissolved inorganic carbon (DIC) and CH4 phase, can be described in a number of ways. For simplicity and reproducibility, here we define the isotope partition as the fractionation factor that is simply described as:

where δX = δ13C-DIC or δ2H-H2O and δCH4 = the δ13C-CH4 or δ2H-CH4, respectively: such that αDIC-CH4 = the carbon isotope fractionation factor, and αH2O-CH4 = the hydrogen isotope fractionation factor.

Rayleigh equations

Where Rayleigh equations are presented, we use the Rayleigh equation described as:

where R = change in isotope fractionation relative to the initial value, Ri = the initial isotope delta values, f = the residual reservoir (e.g. CH4). The use of the Rayleigh equation allows simple comparisons of the isotope partitioning as a defined reservoir (e.g. CH4) changes. Biogenic methanogenesis tends to operate at thermodynamic equilibrium, rather than being limited by kinetic controls6 so this approach is considered appropriate here, especially when working at large scales where multiple influences on the carbon and hydrogen isotope may occur. Here we use the Rayleigh equation to simply define the change in carbon and hydrogen isotope partition as CH4 changes relative to a CH4 end member (we do not propose that the δ13C-CH4 or δ2H-CH4 behaviour in the shallow coal measures always follows a simple Rayleigh fractionation process).

Microbial pathways and Gibbs free energy values

Gibbs free energy values (ΔG°) were calculated for a number of microbial pathways (equations (6–8)) using enthalpy and entropy values for each reaction listed in Stumm and Morgan88 and corrected for the temperature of each sample (equation 4).

Where ΔH is the change in enthalpy and ΔS is the change in entropy for each reaction, and T is the temperature in Kelvin for each sample.

Changes in Gibbs Free Energy ΔG were calculated via equation (5).

where R is the universal gas constant, T is the temperature in Kelvin and Q is the reaction quotient for each reaction.

For each reaction the activities of reactants and products [activity] were used to calculate Q. The activities of the reactants and products were calculated using PHREEQC Interactive 3.1.7–9213 using the phreeqc.dat database.

The reaction pathways for CO2-reduction (hydrogenotrophic methanogenesis), SO4 reduction and anaerobic oxidation of CH4 (AOM) are as follows:

CO2 reduction

SO4 reduction

Anaerobic oxidation of CH4 (AOM)

Gibbs free energy values (ΔGoT) for the CO2-reduction pathways and SO4 reduction pathways were ~−229 kJ mol−1 and −262 kJ mol−1, respectively, which is consistent with calculations made for other studies59,89,90.

Additional Information

How to cite this article: Owen, D. D. R. et al. Thermodynamic and hydrochemical controls on CH4 in a coal seam gas and overlying alluvial aquifer: new insights into CH4 origins. Sci. Rep. 6, 32407; doi: 10.1038/srep32407 (2016).