Controls on Gas Domains and Production Behaviour in A High-Rank CSG Reservoir: Insights from Molecular and Isotopic Chemistry of Co-Produced Waters and Gases from the Bowen Basin, Australia

: This paper uses hydrochemical and multi-isotope analysis to investigate geological controls on coal seam gas (CSG) saturation domains and gas well production performance in a high-rank (vitrinite reﬂectance (Rv) > 1.1) CSG ﬁeld in the north-western Bowen Basin, Australia. New hydrochemical and stable isotope data were combined with existing geochemical datasets to reﬁne hypotheses on the distribution and origins of CSG in two highly compartmentalized Permian coal seams. Stable isotopic results suggest that geographic variations in gas content, saturation and production reﬂect the extent of secondary microbial gas generation and retention as a function of hydrodynamics. δ 13 C and δ 2 H data support a gas mixing hypothesis with δ 13 C-CH 4 increasing from secondary biogenic values to thermogenic values at depth ( δ 13 C − 62.2% (cid:24) to − 46.3% (cid:24) ), whereas correlated methane and carbon dioxide carbon isotope compositions, ∆ 13 C(CO 2 –CH 4 ) values and δ 13 C DIC / alkalinity trends are largely consistent with microbial CO 2 reduction. In addition, below 200 m, the majority of δ 13 C-CO 2 values are positive ( δ 13 C: − 1.2% (cid:24) to 7.1% (cid:24) ) and δ 13 C DIC shows an erratic increase with depth for both seams that is characteristic of evolution via microbial activity. The progression of carbon isotope values along the CO 2 reduction fractionation line suggests progressive depletion of the CO 2 reservoir with increasing depth. Faults clearly segment coal seams into areas having signiﬁcantly di ﬀ erent production, with results of geochemical analysis suggesting that pooling of biogenic gas and waters and enhanced methanogenesis occur north of a faulted hinge zone. Comparison of our expanded analytical dataset against updated reservoir, production and structure datasets allowed improved geology-based interpretations and the identiﬁcation of statistically signiﬁcant geochemical di ﬀ erences within and between the upper (P)- and lower (GM)-producing coal seams. New hydrochemical and multi-isotope evidence indicates that coal bed waters from the production zone (100–500 m depth) are likely to be entirely meteoric in origin and evolved via a combination of low temperature ﬂuid-rock interaction processes and microbial activity. Although new δ 13 C and δ 2 H data support a gas mixing trend hypothesis with δ 13 C-CH 4 increasing from more negative secondary biogenic (microbial CO 2 reduction) to less negative thermogenic values at depth ( δ 13 C − 62.2% (cid:24) to − 46.3% (cid:24) ), geographic variations in gas content / saturation and production more likely reﬂect the extent of secondary microbial gas generation and retention as a function of hydrodynamics. Below 200 m, the majority of δ 13 C-CO 2 values are positive ( δ 13 C − 1.2% (cid:24) to 7.1% (cid:24) ) and δ 13 C DIC shows an erratic increase with depth and increasing alkalinity for both seams that is characteristic of evolution via microbial activity. Although there is possible evidence of minor thermogenic input, ∆ 13 C(CO 2 –CH 4 ) values and the overall progression of carbon isotopes along the CO 2 reduction fractionation line suggests progressive depletion of the CO 2 reservoir with increasing depth for both seams on both sides of the hinge zone.


Introduction
Understanding the origins and co-evolution of coal seam gases and waters is essential for coal seam gas (CSG) exploration and development, coal mine planning and related water management. CSG production performance is a function of well design implementation and several hydro-geological factors, such as gas reservoir size and geometry, gas content, gas saturation and coal permeability. It is widely acknowledged that hydro-geological controls on CSG generation, migration and retention can be decoded from the chemical and isotope compositions of the gases and associated fluids [1][2][3][4][5][6][7][8]. Coal bed reservoirs typically behave as fractured aquifers that produce significant volumes of water throughout

Previous Geochemical Investigations
Kinnon et al., [3] presented the first published investigation of geochemistry relative to reservoir geology and production behaviour in the area. That early work suggested that coal bed waters in the study field are predominantly meteoric with a possible connate influence at depth,

Previous Geochemical Investigations
Kinnon et al., [3] presented the first published investigation of geochemistry relative to reservoir geology and production behaviour in the area. That early work suggested that coal bed waters in the study field are predominantly meteoric with a possible connate influence at depth,

Previous Geochemical Investigations
Kinnon et al., [3] presented the first published investigation of geochemistry relative to reservoir geology and production behaviour in the area. That early work suggested that coal bed waters in the study field are predominantly meteoric with a possible connate influence at depth, and that the δ 13 C-CH 4 values of co-produced gases reflect a mixing trend between shallow microbial gas and deeper thermogenic gas. Good gas saturation was interpreted to arise from microbial gas formed from microbial communities resident in the coal formation water and stimulated by some degree of meteoric recharge over geologic time. Correlations were identified between water quality, stable isotope data, production behaviour and structure, but the dominant hydrochemical processes and δ 13 C isotope systematics require further clarification.
In that study, a total of N = 9 upper (P) seam wells and N = 15 lower (GM) seam wells were sampled north and south of the hinge zone, with the majority producing from <300 m. However, additional data are needed to elucidate the geochemical and microbial interactions and the interrelationships with reservoir geology in each of the 4 four domains (P and GM seam, north versus south of the hinge zone).

This Study
The main aim of our study was to increase the spatial coverage of the data, particularly south of the hinge zone and from greater depths (down to 474.5 m), to allow for more detailed spatial analysis of gas and water isotopes and to make statistical comparisons with updated reservoir and production datasets for both seams. Existing [3] and new (N = 5 P wells; N = 8 GM wells) hydrochemical and stable isotope data are analyzed against reservoir properties, structure and production data to test the overarching hypothesis that microbial gas generation is responsible for zones of higher gas saturation and production.

Sampling and Methods
This study draws on a geochemistry database that includes 36 analyses of co-produced water chemistry and 27 gas analyses collected over two sampling trips in 2008 and 2016 (Tables 1 and 2). Well data from 2008 were first presented in Kinnon et al., [3] and sampling details can be found in that paper.
The most recent coupled water and gas samples were collected from 13 previously unstudied wells in March 2016 ( Figure 1). Twelve water samples were collected for water quality and stable isotope analysis, 8 from the GM seam and 4 from the P seam. P seam well P040V was sampled for gas but produced no water after 20 min of pumping. The wells were also sampled for gas stable isotope analysis but shallow well P037V had insufficient gas for analysis. Gas samples were collected using either flow-through gas canisters or FlexFoil© bags. Water samples and onsite readings were taken only after pH and conductivity measurements had stabilised. Once pH, conductivity and temperature were measured, a suite of water samples was collected and treated onsite as appropriate for the analysis required, i.e., some samples were filtered, acidified and chilled as per standard analytical protocols. Gas-and water-stable isotopic analysis was conducted at the University of Queensland (UQ) School of Earth and Environmental Sciences Stable Isotope Geochemistry Laboratory (UQ-SIGL) while standard water analyses were performed at ALS Laboratory Group, Environmental Division Brisbane (Stafford, QLD, Australia). Analytical procedures for the 2016 UQ samples are outlined below.

Standard Water Chemistry
The anions fluoride (F − ), sulfate (SO 4 2− ) and chloride (Cl − ) were determined by discrete analyser with detection limits for F − being 0.1 mg/L, and SO 4 2− and Cl − being 1 mg/L. Alkalinity was determined by PC titrator and has detection limits of 1 mg/L. Dissolved major cations were determined by inductively coupled plasma optical emission spectrometry (ICP-OES) with Na + , Ca 2+ , Mg 2+ and K + having detection limits of 1 mg/L. All water chemistry analyses were carried out by ALS-Chemex Laboratory Group, Environmental Division Brisbane (Stafford, QLD, Australia).

Isotopic Analyses
Hydrogen, oxygen and carbon stable isotopic analyses were conducted using gas source isotope ratio mass spectrometry in the UQ-SIGL.

Water Oxygen and Hydrogen Isotopes
Water samples for oxygen and hydrogen isotope ratio determinations were filtered through 0.45 µm filters into 30 ml glass McCartney bottles and immediately chilled. Samples were analysed using an Isoprime dual inlet isotope ratio mass spectrometer (DI-IRMS) coupled to a multiprep bench for online analysis. δ 2 H values were analysed after online equilibration at 40 • C with Hokko coils. δ 18 O values were analyzed as above, but after equilibration with carbon dioxide. δ 2 H and δ 18 O values (reported in per mil (% )) were normalised to the standard mean ocean water (VSMOW-SLAP) scale, following a three point normalisation [23] based on four replicate analyses of three laboratory standards per analytical cycle. All laboratory standards were calibrated against IAEA (VSMOW, SLAP, GISP) and USGS (USGS45, USGS46) international water standards, such that the δ 2 H and δ 18 O values of SLAP are −427.5% and −55.5% , respectively [24]. Accuracy and precision were better than ± 2% for δ 2 H and ± 0.1% for δ 18 O at 1σ. Both isotopic compositions were measured as deviations relative to VSMOW.

Gas Hydrogen and Carbon Isotopes
All gas samples were analysed using an Isoprime/Agilent gas chromatograph-combustion-isotope ratio mass spectrometer (GC-c-IRMS). For δ 13 C determination of CH 4 , all samples were analysed using a 20:1 split. The gas chromatograph (GC) (with a 50 m × 320 µm × 5 µm CP-PoraBOND Q column) was set to a flow of 1.2 ml/min with an oven temperature of 40 • C. After the sample eluted from the GC, it was passed through a quartz furnace packed with copper oxide set to 990 • C, which oxidised the CH 4 to CO 2 for measurement in the IRMS. Any CO 2 in the sample passed through unaffected. For δ 2 H determination of CH 4 , each sample was analysed utilising the same GC conditions as for δ 13 C determination but after eluting from the GC, samples were then pyrolysed by passing through a ceramic furnace set at 1410 • C.
Isotopic analysis of samples with known CO 2 molecular compositions indicates that δ 13 C-CO 2 data are unreliable below CO 2 concentrations of 0.2 mol% for current instrument settings. In the study area, the CSG is typically >98% methane with minor CO 2 (<2%) and ethane [3], precluding δ 13 C-CO 2 analysis of one of our samples (M189GMV). In this context, laboratory gas standards SIGL1 and SIGL2 were considered a better matrix match for the study samples than T-ISO2. The δ 13 C and δ 2 H values of SIGL1 and SIGL2 had initially been determined by two independent laboratories (Geoscience Australia, Canberra, Australia; Intertek-Geotech Laboratories, Perth, Australia), by calibration against international standards NGS1, 2 and 3 and NBS-19. Thus, all laboratory gas standards were calibrated against international standards defining the VPDB and VSMOW scales [24]. Precision is quoted at ±0.3% for δ 13 C and ±3% for δ 2 H at 1σ. Dissolved inorganic carbon (δ 13 C DIC ) was determined by collecting filtered water samples in serum bottles without headspace. Samples were immediately chilled after filtration and kept chilled until analysis. δ 13 C was determined online using a Thermo Scientific GasBench via the standard method. δ 13 C DIC values were normalised to the VPDB scale using IAEA standards NBS-18 and NBS-19. Analytical uncertainty was determined as ±0.1% at 1σ.

Gas Isotopes
Production gas samples were analysed for compound specific stable isotopes (Table 1). Figure 4 is a gas genetic diagram of the study samples plotted by seam and position north or south of the hinge zone. Across all 2016 samples, the carbon isotope compositions for P and GM seam methane range from δ 13 C −57.4% to −46.3% and average −53.3% (n = 12). The corresponding hydrogen isotope compositions for methane range from δ 2 H −220% to −207% and average −216.1% . These ranges are typical of methane sourced from Queensland CSG fields [2,14] and overlap previously reported values for the study field that were analysed at Geoscience Australia [3]. Methane stable isotope values plot within the mixed thermogenic/microbial CO 2 reduction and solely microbial genetic fields on a plot of δ 13 C versus δ 2 H, and mainly within the secondary microbial field on a plot of δ 13 C-CO 2 versus δ 13 C-CH 4 (Figure 4a,b). The addition of new geochemical data has revealed strong depth trends and deep compartmentalisation along strike. Figure 5a shows that the carbon isotopic compositions of produced methane from this study increase with increasing depth for both seams. This agrees with the findings of Kinnon et al., [3] and supports a mixing trend hypothesis with δ 13 C-CH 4 increasing from more negative secondary biogenic (microbial CO 2 reduction) values to less negative thermogenic values at depth. New data also confirm that the most depleted ("biogenic") δ 13 C-CH 4 values for both seams occur in the SW part of the study area. Contrary to expectation, δ 13 C-CH 4 values in the P seam are generally less negative, rather than more negative for a given depth. Across all datasets, the only significant δ 13 C-CH 4 outlier is shallow well P011 (Figure 5a).
Hydrogenotrophic methanogenesis is associated with 13 C-enriched residual CO 2 and DIC compositions, whereas thermogenic CO 2 derived from organic-rich shale or coal is characteristically depleted in 13 C. δ 13 C-CO 2 values from the 2016 study range from 2.4% to 7.1% (n = 11), and from δ 13 C -13.1% to 7.1% (n = 26) for the greater dataset. Figure 5b suggests that the majority of CO 2 is residual from microbial CO 2 reduction, with minor input from inorganic and/or organic sources. At a given depth, P and GM wells north of the hinge zone have more positive δ 13 C-CO 2 values than wells south of the hinge zone. A similar, more tentative trend was also observed for P seam δ 13 C-CH 4 and δ 2 H-CH 4 values (Figure 5a,c). Overall, δ 13 C-CO 2 tracks δ 13 C-CH 4 , but not δ 2 H-CH 4 , which instead shows a broad inverse trend with depth.
Carbon isotopic difference [∆ 13 C(CO 2 -CH 4 )] values range from 48.6% to 63.5% (n = 26) with considerable overlap between the P and GM seams (Figure 5d). The [∆ 13 C(CO 2 -CH 4 )] range in this study falls on the acetate fermentation-CO 2 reduction boundary as commonly defined (i.e., carbon isotopic difference between CO 2 and CH 4 (α ≥ 1.06 = CO 2 reduction [25])). The ∆ 13 C(CO 2 -CH 4 ) values are variable above 200 m but are restricted to a comparatively narrower range at depth. From 200-300 m, GM north wells have higher ∆ 13 C(CO 2 -CH 4 ) values than GM south wells. This distribution suggests influence from a range of surficial CO 2 sources at shallow levels and potentially, dominance of hydrogenotrophic methanogenesis at mid and deeper levels. δ 13 C gas isotope data are plotted by gas production class in Figure 6. There is considerable overlap between the well classes in a plot of δ 13 C-CH 4 versus depth (Figure 6a), although most good wells plot in the biogenic gas field and lie north of the hinge zone, whereas poor wells mainly plot in the mixed field and lie south of the hinge zone ( Figure 6b).  Blank space: no data. Milkov and Etiope [26]. Data are from Kinnon et al., [3]; this study.  Figure 1 for well locations. Fields on plots from Milkov and Etiope [26]. Data are from Kinnon et al., [3]; this study.  [3]; this study.

Water Isotopes
A total of 12 new co-produced water samples from 12 wells were analysed for δ 2 H-H2O and δ 18 O-H2O (Table 2). Despite being sampled eight years later, the 2016 values span a similar range to the Kinnon et al., [3] dataset that was also analysed at UQ SIGL (Table 2). Across all samples, δ 18 O values range from δ 18 O −8.3‰ to −5.0‰ and δ 2 H −49‰ to −30‰ (n = 36). The majority of wells plot on or to the left of the Global Meteoric Water Line (GMWL), which is typical of shallow meteoricsourced waters or formation waters that have undergone low temperature water-rock-microbial reactions or mixing with connate waters, e.g., [1][2][3]27] (Figure 7). Of the 11 samples plotting to the right of the GMWL (6 P, 5 GM), 8 are shallow wells south of the hinge zone, and 3 are shallow wells north of the hinge zone. All but one of the high water producing wells plots in this less negative field ( Figure 7a). Many processes can shift meteoric water compositions to the right of the GMWL but the shallow depths of the samples are more consistent with evaporation effects during recharge than high-temperature fluid-rock interaction or mixing with basinal brines, e.g., [2,28,29]. A plot of δ 2 H-H2O versus δ 18 O-H2O coded by gas production shows considerable overlap between wells with different production behaviour; however, the majority of higher gas producing wells plot well above the GMWL, suggesting that these samples have been more heavily impacted by low temperature water-rock-microbial reactions (Figure 7b).

Water Isotopes
A total of 12 new co-produced water samples from 12 wells were analysed for δ 2 H-H 2 O and δ 18 O-H 2 O (Table 2). Despite being sampled eight years later, the 2016 values span a similar range to the Kinnon et al., [3] dataset that was also analysed at UQ SIGL (Table 2). Across all samples, δ 18 O values range from δ 18 O −8.3% to −5.0% and δ 2 H −49% to −30% (n = 36). The majority of wells plot on or to the left of the Global Meteoric Water Line (GMWL), which is typical of shallow meteoric-sourced waters or formation waters that have undergone low temperature water-rock-microbial reactions or mixing with connate waters, e.g., [1][2][3]27] (Figure 7). Of the 11 samples plotting to the right of the GMWL (6 P, 5 GM), 8 are shallow wells south of the hinge zone, and 3 are shallow wells north of the hinge zone. All but one of the high water producing wells plots in this less negative field ( Figure 7a). Many processes can shift meteoric water compositions to the right of the GMWL but the shallow depths of the samples are more consistent with evaporation effects during recharge than high-temperature fluid-rock interaction or mixing with basinal brines, e.g., [2,28,29]. A plot of δ 2 H-H 2 O versus δ 18 O-H 2 O coded by gas production shows considerable overlap between wells with different production behaviour; however, the majority of higher gas producing wells plot well above the GMWL, suggesting that these samples have been more heavily impacted by low temperature water-rock-microbial reactions (Figure 7b). Marker sizes denote average gas production as lower, mid or higher. Data from Kinnon et al. [3]; this study. (b) All Δ 13 C(CO2-CH4) versus δ 13 C-CH4 data coded by gas production and position north or south of the hinge zone. Fields on plot based on Smith and Pallasser [18] and Whiticar et al. [25,30].  Marker sizes denote average gas production as lower, mid or higher. Data from Kinnon et al. [3]; this study. (b) All ∆ 13 C(CO 2 -CH 4 ) versus δ 13 C-CH 4 data coded by gas production and position north or south of the hinge zone. Fields on plot based on Smith and Pallasser [18] and Whiticar et al. [25,30]. Marker sizes denote average gas production as lower, mid or higher. Data from Kinnon et al. [3]; this study. (b) All Δ 13 C(CO2-CH4) versus δ 13 C-CH4 data coded by gas production and position north or south of the hinge zone. Fields on plot based on Smith and Pallasser [18] and Whiticar et al. [25,30].  Inset schematic shows likely shifts associated with different geochemical processes. Global Meteoric Water Line (GMWL) from Craig [31]. Labelled wells are referred to in the text. Blank space-no data.
Overall, the water-stable isotope compositions of P and GM seam coal bed waters show systematic depletion trends with increasing depth. This depletion trend is typical of shallow to mid-depth meteoric-sourced groundwaters in general, as older waters will have more have more negative 2 H and 18 O values reflecting the stable isotopic composition of precipitation during past cooler climates. The shift in 2 H and 18 O values between the P and GM seam datasets parallels the GMWL, which is likely more indicative of the climatic conditions during meteoric recharge as opposed to subsurface microbial or fluid-rock interaction processes and suggests that there is no or limited intermixing of the waters. On the other hand, southern GM samples are significantly offset along the GMWL but were taken from comparable depths to the northern GM samples; locally, southern GM samples M188GMV and M189GMV are also offset along the GMWL despite being from similar depths. This is further illustrated in Figure 8, where cross-plots of δ 18 O and δ 2 H versus depth are plotted by seam and location. δ 18 O and δ 2 H show negative correlations with depth for both seams. Two wells are offset from the main trends: P041V (north) and M189GMV (south), which is the deepest well sampled in this study and exhibits poor production behaviour for its depth. At a given depth, GM wells north of the hinge zone appear to have more negative water isotopes values than wells south of the hinge zone. Less negative δ 2 H-H 2 O and δ 18 O-H 2 O values are commonly, but not always, associated with higher water production and lower gas production. For example, δ 2 H-H 2 O and δ 18 O-H 2 O values for 2016 well P041V near the hinge zone are anomalously high for depth, and this well has low water production and medium gas production (Figures 7a and 8a).
Dissolved inorganic carbon (DIC) δ 13 C values in the study area vary between δ 13 C +4.5% and +15.5% (n = 12) ( Table 1). Within seam variation is greater than the difference between the P and GM seams, and both seams show a positive correlation with depth (r = 0.7, 0.85, respectively, excluding one outlier) (Figure 9). In general, strongly negative δ 13 C DIC values (δ 13 C DIC : −30% ) in coal bed waters indicate thermogenic influences, whereas positive δ 13 C DIC values together with high alkalinity are typical of methanogenic waters [32,33]. In high-alkalinity waters, more negative δ 13 C DIC values can also reflect bacterial sulfate reduction [34]. The study samples have enriched, positive δ 13 C DIC values characteristic of hydrogenotrophic methanogenesis, except for GM041V (north) that has a low positive δ 13 C DIC value of +4.5% and was sampled close to the hinge zone. This well has sulfate <1 mg/L, a high alkalinity value (3719 mg/L, as HCO 3 -), and a positive δ 13 C-CO 2 value (+6.9% ), suggesting neither bacterial sulfate reduction nor thermogenic influences are likely in the GM seam at that location.    [3]; this study. Figure 9. δ 13 CDIC versus depth for the 2016 study samples, plotted by seam. Figure 9. δ 13 C DIC versus depth for the 2016 study samples, plotted by seam.

Water Chemistry
Water chemistry data for the produced waters are shown in Table 2 and Durov plots of water quality by seam, geographic position and production behaviour are shown in Figure 10. At a high level, the similarity in water types between samples from both seams suggests that similar geochemical processes are controlling their major ion chemistry. The groundwater chemistry of the study area is dominated by two major ion hydrochemical facies: a sodium-chloride (Na-Cl) facies at shallower depths and a sodium-bicarbonate (Na-HCO 3 ) facies at depths >~350 m. All samples are low-brackish, i.e., <20,000, >1500 mg/L total dissolved solids (TDS). Wells with lower gas production define a curve from more acidic, less sodic, Na-Cl type at shallower depths to more alkaline Na-Cl type waters with depth. Wells with mid and higher gas production (the latter being northern wells) can be Na-Cl or Na-HCO 3 type but tend to have high pH values and TDS concentrations for their depth.
Na concentrations range from 1070 to 3060 mg/L. For both seams, Na concentrations are very high (P median: 1490; GM median: 2200 mg/L), and are significantly higher in the north. Figure 11a is a cross-plot of depth versus sodicity (SAR = ([Na + ]/(([Ca 2+ ] + [Mg 2+ ])/2) 0.5 ) coded by seam and position relative to the hinge zone. Overall, sodicity increases with depth, suggesting a gradual mixing of less sodic meteoric waters proximal to recharge zones with saline connate waters (formational brines). However, this trend may also indicate downdip fluid communication and progressive evolution via low temperature fluid-rock and microbial reactions. Southern wells fall within a narrow range of 60-100, whereas northern wells span a wider range with 75% of all data values and higher gas-producing wells (189-414 m depth) plotting above 80. Four GM south wells plot off the main trend and show a slight decline in sodicity with depth, suggesting a lack of hydraulic continuity between waters from these wells and the rest of the field. Interestingly, the two deep outliers are newer, poor and medium-performing wells, and the deepest outlier has an anomalous δ 2 H-H 2 O value (high for depth) and a more enriched δ 13 C-CH 4 value (high for depth) (see Figures 7a and 8b).
The median alkalinity of southern GM wells (13.43 meq/L) is less than one-half that (34.77 meq/L) of the northern wells. In contrast, the alkalinity of the P seam is statistically similar on both sides of the hinge zone. Alkalinity does not correlate with gas production on an individual well basis (Figure 11b). GM seam calcium concentrations range from 9-92 mg/L, while P seam calcium concentrations span a similar range from 12-109 mg/L. Magnesium ranges from 2-72 mg/L and 3-48 mg/L in the GM and P seams, respectively. In the north, both seams exhibit lower mean Ca/Mg molar ratios typical of methanogenic waters (GM: 1.3, P: 1.4; cf. <1.5) [35]. Fluoride is present in all water samples, ranging from 1 to 4 mg/L. Fluoride concentrations are statistically similar on both sides of the hinge zone. The majority of sulfate concentrations are below detection limit (<1 mg/L). Four northern GM wells have low sulfate concentrations from 1 to 3 mg/L. In these wells, sulfate concentration shows no relationship with depth, gas or water production behaviour.
Chloride (GM: 702-4440 mg/L; P: 1090-4530 mg/L) shows a broad decrease with depth for both seams and is inversely correlated with fluoride and total alkalinity (as HCO 3 − ), suggesting that fluoride is sourced from fluid-rock interactions. The majority of wells have a Na-excess, suggesting an additional Na source from silicate weathering and/or ion exchange reactions. Na/Cl molar ratios are highest (>1.6) at deeper levels, in 7 wells to the north and 3 to the south of the hinge zone, respectively (9 GM, 1 P). Northern GM waters have the lowest mean Cl concentration of any reservoir domain (2174 mg/L). Overall, the study samples reflect the same downdip major ion trends observed in other CSG basins characterised by Ca + Mg depletion and Na and HCO 3 − enrichment, e.g., [2,7,36]. Geosciences 2020, 10, x FOR PEER REVIEW 17 of 26 The same dataset coded by average gas production. Open circle marker sizes on the pH and total dissolved solids (TDS) plots denote average water production. Lower water production circles are the same size as the gas symbols. Data from Kinnon et al., [3]; this study. Data from Kinnon et al., [3]; this study.

Isotopic Evidence for Thermogenic Gas Charge
At a high level, δ 13 C and δ 2 H data support a gas mixing trend hypothesis with δ 13 C-CH4 increasing from more negative, secondary biogenic (microbial CO2 reduction) values at shallow levels to less negative, "thermogenic" values at depth (δ 13 C −62.2‰ to −46.3‰), based on the earlier gas genetic classification fields of Whiticar et al., [25] and Whiticar [37], and the recently revised fields of Milkov and Etiope [26]. Coal rank (Rv >1.1) and gas molecular composition data support the idea that the gas isotope compositions at least partly reflect cracking of coal bitumen and higher hydrocarbons and/or early thermogenic gas generation. Ethane is variably present, albeit at very low levels (<0.01%) in both seams at a range of depths north and south of the hinge zone; over the depth range investigated (100-500 m), production gas samples have dryness (C1/C1+) values approaching 1 ([2,3], Mazumder, unpubl. data, see acknowledgements). Hydrodynamics can facilitate migration of biogenic and thermogenic gas to permeability barriers that locally increases gas contents, e.g. [11]. However, the possibility that higher gas content/gas production domains in the north of the study area reflect pooling of thermogenic gases seems unlikely, based on: (1) uniform δ 13 C-CH4 versus depth trends and positive δ 13 CDIC values for both seams on both sides of the hinge zone; (2) more positive δ 13 C-CO2 values for both seams north of the hinge zone; and (3) the generally high alkalinity values, particularly north of the hinge zone that is somewhat uncharacteristic of thermogenic plays [35]. A more likely explanation is that geographic variations in gas content/saturation and production largely reflect the extent of secondary microbial gas generation, retention and lateral migration, and as such are largely a function of hydrodynamics. This is discussed below.

Isotopic Evidence for Extent of Methanogenesis
Our 2016 samples show a strong positive relationship between δ 13 CDIC and alkalinity that is characteristic of evolution via microbial activity [2,32,38]. Excluding one significant outlier (well

Isotopic Evidence for Thermogenic Gas Charge
At a high level, δ 13 C and δ 2 H data support a gas mixing trend hypothesis with δ 13 C-CH 4 increasing from more negative, secondary biogenic (microbial CO 2 reduction) values at shallow levels to less negative, "thermogenic" values at depth (δ 13 C −62.2% to −46.3% ), based on the earlier gas genetic classification fields of Whiticar et al., [25] and Whiticar [37], and the recently revised fields of Milkov and Etiope [26]. Coal rank (Rv > 1.1) and gas molecular composition data support the idea that the gas isotope compositions at least partly reflect cracking of coal bitumen and higher hydrocarbons and/or early thermogenic gas generation. Ethane is variably present, albeit at very low levels (<0.01%) in both seams at a range of depths north and south of the hinge zone; over the depth range investigated (100-500 m), production gas samples have dryness (C 1 /C 1+ ) values approaching 1 ( [2,3], Mazumder, unpubl. data, see acknowledgements). Hydrodynamics can facilitate migration of biogenic and thermogenic gas to permeability barriers that locally increases gas contents, e.g. [11]. However, the possibility that higher gas content/gas production domains in the north of the study area reflect pooling of thermogenic gases seems unlikely, based on: (1) uniform δ 13 C-CH 4 versus depth trends and positive δ 13 C DIC values for both seams on both sides of the hinge zone; (2) more positive δ 13 C-CO 2 values for both seams north of the hinge zone; and (3) the generally high alkalinity values, particularly north of the hinge zone that is somewhat uncharacteristic of thermogenic plays [35]. A more likely explanation is that geographic variations in gas content/saturation and production largely reflect the extent of secondary microbial gas generation, retention and lateral migration, and as such are largely a function of hydrodynamics. This is discussed below.

Isotopic Evidence for Extent of Methanogenesis
Our 2016 samples show a strong positive relationship between δ 13 C DIC and alkalinity that is characteristic of evolution via microbial activity [2,32,38]. Excluding one significant outlier (well GM041V near the hinge zone), deeper northern waters appear to be more microbially evolved than waters at similar depths in the south, with shallow southern samples representing the other end-member ( Figure 12). Although δ 13 C DIC data are limited, this relationship suggests that domains of differing gas production (i.e., north versus south of the hinge zone) have different δ 13 C DIC /alkalinity signatures. The four GM wells in the "Deeper, S" field plot off the main sodicity trend for the study area (Figures 11a and 12). In this well cluster, sodicity decreases rather than increases with depth, and coal permeability [39] and water production are low. As mentioned previously, one of these wells has an anomalous δ 2 H-H 2 O value (high for depth) and a more enriched δ 13 C-CH 4 value. At this site, the two study seams may be in hydraulic communication via faults or fractures. The other wells in the cluster do not display anomalous isotope signatures. This suggests that relatively limited methanogenesis, as opposed to other possibilities such as more open hydrologic flow conditions and/or meteoric dilution may be responsible for the δ 13 C DIC /alkalinity signature of the GM seam in this deeper part of the field. The lower positive δ 13 C DIC outlier value near the faulted hinge zone (δ 13 C +4.5% ) may reflect a stronger relative influence of acetoclastic methanogenesis at that location [40], but could also reflect carbonate dissolution in a partially open system.
There is consensus that relative 13 C-enrichment of methane can result from: (1) microbial oxidation [41]; (2) a shift toward the acetate fermentation pathway of methanogenesis [25]; (3) CO 2 system limitations [16,30]; and (4) water stagnation [35,42,43], but disagreement on the relative influence of these processes at the basin-scale in major unconventional gas plays. A key divergence is the interpretation that thermogenic-like methane δ 13 C values in deeper parts of the prolific Powder River Basin, USA reflect a change in methanogenic pathway and/or mixing with transitional thermogenic gas [44,45], as opposed to the extent of methanogenesis [35,42]. In a given coal seam, multiple metabolic pathways may operate concurrently [44,46,47] or successively [44]. Traditional isotope approaches [∆ 13 C(CO 2 −CH 4 ) and ∆ 2 H(H 2 O−CH 4 )] are useful to determine the dominant metabolic pathway but controversy surrounds their strict application, particularly when used to calculate contributions from different metabolic pathways operating at the same location [8,[42][43][44].
Geosciences 2020, 10, x FOR PEER REVIEW 19 of 26 GM041V near the hinge zone), deeper northern waters appear to be more microbially evolved than waters at similar depths in the south, with shallow southern samples representing the other endmember ( Figure 12). Although δ 13 CDIC data are limited, this relationship suggests that domains of differing gas production (i.e., north versus south of the hinge zone) have different δ 13 CDIC/alkalinity signatures. The four GM wells in the "Deeper, S" field plot off the main sodicity trend for the study area (Figures 11a and 12). In this well cluster, sodicity decreases rather than increases with depth, and coal permeability [39] and water production are low. As mentioned previously, one of these wells has an anomalous δ 2 H-H2O value (high for depth) and a more enriched δ 13 C-CH4 value. At this site, the two study seams may be in hydraulic communication via faults or fractures. The other wells in the cluster do not display anomalous isotope signatures. This suggests that relatively limited methanogenesis, as opposed to other possibilities such as more open hydrologic flow conditions and/or meteoric dilution may be responsible for the δ 13 CDIC/alkalinity signature of the GM seam in this deeper part of the field. The lower positive δ 13 CDIC outlier value near the faulted hinge zone (δ 13 C +4.5‰) may reflect a stronger relative influence of acetoclastic methanogenesis at that location [40], but could also reflect carbonate dissolution in a partially open system. There is consensus that relative 13 C-enrichment of methane can result from: (1) microbial oxidation [41]; (2) a shift toward the acetate fermentation pathway of methanogenesis [25]; (3) CO2 system limitations [16,30]; and (4) water stagnation [35,42,43], but disagreement on the relative influence of these processes at the basin-scale in major unconventional gas plays. A key divergence is the interpretation that thermogenic-like methane δ 13 C values in deeper parts of the prolific Powder River Basin, USA reflect a change in methanogenic pathway and/or mixing with transitional thermogenic gas [44,45], as opposed to the extent of methanogenesis [35,42]. In a given coal seam, multiple metabolic pathways may operate concurrently [44,46,47] or successively [44]. Traditional isotope approaches [Δ 13 C(CO2−CH4) and Δ 2 H(H2O−CH4)] are useful to determine the dominant metabolic pathway but controversy surrounds their strict application, particularly when used to calculate contributions from different metabolic pathways operating at the same location [8,[42][43][44]. Figure 12. Alkalinity versus δ 13 CDIC for the P and GM seams, north and south of the hinge zone. All higher gas producing wells in the study field are located to the north of the hinge zone.
Methane oxidation is unlikely to be responsible for the 13 C versus depth trends observed in this study as this would enrich the residual methane pool in both 13 C and 2 H and be more likely to deplete than enrich the residual CO2 pool in 13 C. A systematic shift towards acetoclastic methanogenesis with increasing depth also appears unlikely based on available data. The decrease in δ 2 H-CH4 with increasing depth occurs over a narrow range, and southern GM seam δ 2 H-CH4 values exhibit no correlation with depth and overlap the northern GM dataset, despite having similar 13 C versus depth profiles. Furthermore, our lowest δ 2 H-CH4 values are enriched by ~60‰ Figure 12. Alkalinity versus δ 13 C DIC for the P and GM seams, north and south of the hinge zone. All higher gas producing wells in the study field are located to the north of the hinge zone.
Methane oxidation is unlikely to be responsible for the 13 C versus depth trends observed in this study as this would enrich the residual methane pool in both 13 C and 2 H and be more likely to deplete than enrich the residual CO 2 pool in 13 C. A systematic shift towards acetoclastic methanogenesis with increasing depth also appears unlikely based on available data. The decrease in δ 2 H-CH 4 with increasing depth occurs over a narrow range, and southern GM seam δ 2 H-CH 4 values exhibit no correlation with depth and overlap the northern GM dataset, despite having similar 13 C versus depth profiles. Furthermore, our lowest δ 2 H-CH 4 values are enriched by~60% compared to the upper threshold of the empirically-based acetate fermentation field of Whiticar et al., [25] (δ 2 H −280), and co-produced δ 2 H-H 2 O versus δ 2 H-CH 4 values for both seams plot on or slightly below the CO 2 reduction fractionation line (alpha = 1.20) (Figure 13a). The majority of results below the line (with slightly higher alpha values) are from shallow depths or close to major structures, i.e., areas with greater potential for overhead recharge via preferential pathway flow along faults and fractures, and therefore more likely indicate secondary effects such as sulfate reduction or methane oxidation. Figure 13b is a plot of δ 13 C-CH 4 versus δ 13 C-CO 2 plot of all data ( [3]; this study), coded by depth below ground level. Shallower samples display a range of fractionations on the acetate fermentation-CO 2 reduction boundary, whereas mid-level and deeper samples primarily follow fractionation lines for CO 2 reduction [30,43,48]. Although acetoclastic methanogenesis does not appear to be the dominant process, two of our shallower samples have lower alpha values (<1.06) conventionally associated with acetate fermentation. Groundwater sulfate concentrations for these two wells are 3 mg/L for the lower δ 13 C-CH 4 value and <1 mg/L for the higher value (δ 13 C −61.7% and −48.0% , respectively). Rather than acetate fermentation, the former (GM002-shallow, hinge zone) may indicate active sulfate reduction and the latter (PO11-shallow) methane oxidation due to enhanced surface communication [2,35]. Shallower, more negative δ 13 C-CH 4 values on the fractionation line may partially reflect water stripping of 13 CH 4 but the wide range of δ 13 C-CH 4 values suggests this process is not a primary influence on methane δ 13 C values in the study area [3,49].
Geosciences 2020, 10, x FOR PEER REVIEW 20 of 26 compared to the upper threshold of the empirically-based acetate fermentation field of Whiticar et al., [25] (δ 2 H −280), and co-produced δ 2 H-H2O versus δ 2 H-CH4 values for both seams plot on or slightly below the CO2 reduction fractionation line (alpha = 1.20) (Figure 13a). The majority of results below the line (with slightly higher alpha values) are from shallow depths or close to major structures, i.e., areas with greater potential for overhead recharge via preferential pathway flow along faults and fractures, and therefore more likely indicate secondary effects such as sulfate reduction or methane oxidation. Figure 13b is a plot of δ 13 C-CH4 versus δ 13 C-CO2 plot of all data ( [3]; this study), coded by depth below ground level. Shallower samples display a range of fractionations on the acetate fermentation-CO2 reduction boundary, whereas mid-level and deeper samples primarily follow fractionation lines for CO2 reduction [30,43,48]. Although acetoclastic methanogenesis does not appear to be the dominant process, two of our shallower samples have lower alpha values (<1.06) conventionally associated with acetate fermentation. Groundwater sulfate concentrations for these two wells are 3 mg/L for the lower δ 13 C-CH4 value and <1 mg/L for the higher value (δ 13 C −61.7‰ and −48.0‰, respectively). Rather than acetate fermentation, the former (GM002-shallow, hinge zone) may indicate active sulfate reduction and the latter (PO11-shallow) methane oxidation due to enhanced surface communication [2,35]. Shallower, more negative δ 13 C-CH4 values on the fractionation line may partially reflect water stripping of 13 CH4 but the wide range of δ 13 C-CH4 values suggests this process is not a primary influence on methane δ 13 C values in the study area [3,49].  [30,48]. (b) δ 13 C-CH4 versus δ 13 C-CO2 plot of all data ( [3]; this study), coded by depth below ground level. Shallower samples display a range of fractionations on the acetate fermentation-CO2 reduction boundary whereas mid-level and deeper samples follow fractionation lines for CO2 reduction [30,43,48]. The samples with the highest δ 13 C-CO2 values are located in the north.
Although there is evidence of minor thermogenic input, the overall progression of carbon isotopes along the fractionation line suggests progressive depletion of the CO2 reservoir for both seams on both sides of the hinge zone. Parallelism between δ 13 C-CO2 and δ 13 C-CH4 versus depth, coupled with more enriched δ 13 C values north of the hinge zone are consistent with this hypothesis when interpreted in terms of 13 C pool depletion effects [16,17,30]. In the study area, CO2  [30,48]. (b) δ 13 C-CH 4 versus δ 13 C-CO 2 plot of all data ( [3]; this study), coded by depth below ground level. Shallower samples display a range of fractionations on the acetate fermentation-CO 2 reduction boundary whereas mid-level and deeper samples follow fractionation lines for CO 2 reduction [30,43,48]. The samples with the highest δ 13 C-CO 2 values are located in the north.
Although there is evidence of minor thermogenic input, the overall progression of carbon isotopes along the fractionation line suggests progressive depletion of the CO 2 reservoir for both seams on both sides of the hinge zone. Parallelism between δ 13 C-CO 2 and δ 13 C-CH 4 versus depth, coupled with more enriched δ 13 C values north of the hinge zone are consistent with this hypothesis when interpreted in terms of 13 C pool depletion effects [16,17,30]. In the study area, CO 2 concentrations are low (<2%) and show a slight, erratic increase with depth [50]. This hinders the possibility that 13 C pool depletion effects are associated with lower CO 2 concentrations. More likely explanations for a pool depletion isotope shift include increased rates of methanogenesis, decreased flushing of CO 2 , or both. In a study of Pennsylvanian coals and the New Albany Shale, Illinois Basin, Schlegel et al., [35] attributed progressive depletion of the carbon reservoir to longer groundwater residence times and decreased groundwater flushing rates with increasing reservoir depth. In our study field, this mechanism potentially explains both the depth and geographic trends in carbon stable isotopes and supports the previous interpretation that the faulted hinge zone acts as a barrier to water and gas flow along strike [3,22].

Hydrogen Isotope System
The overall covariance of the δ 2 H-H 2 O versus δ 2 H-CH 4 values in Figure 13a provides further evidence that the methane generated in situ with the formation waters. The calculated hydrogen isotopic difference values between study field waters and methane samples [∆ 2 H(H 2 O−CH 4 )] range from 165% to 188% (αH 2 O-CH 4 = 1.21-1.24). Geographically, southern αH 2 O-CH 4 values are higher for the GM seam but lower for the P seam, although the latter pattern may be a product of the number of samples. In the GM seam, these higher southern αH 2 O-CH 4 values are largely the product of more enriched δ 2 H-H 2 O values. Southern GM seam δ 2 H-CH 4 values show no trend with depth and overlap the northern GM dataset (Figure 5c). Systematic depth trends in the stable isotopic and elemental hydrochemistry of southern GM waters suggest that overprinting by preferential pathway flow, or cross-measures groundwater communication are unlikely to be major causes of this scatter. One possibility is that it reflects lateral reservoir heterogeneity within the GM seam itself. By contrast, northern P seam δ 2 H-CH 4 values are significantly more enriched than the south, which may reflect a greater extent of methanogenesis and potentially, the age of groundwaters.

Hydrogeochemical Indicators of Microbial Methanogenesis
Considering all information from stable isotopes, the most likely explanation is that the majority of methane generated in situ via microbial CO 2 reduction. Diagnostic hydrochemical parameters (high alkalinity, low Cl, Ca, Mg and sulfate concentrations) indicate that microbial methanogenesis is supported in both seams north and south of the hinge zone but that geochemical conditions are generally more favourable in the north, particularly in the GM seam that has very high alkalinity, suppressed Cl concentrations and the lowest Ca/Mg ratios in the study field. South of the hinge zone, weaker depth trends and significant differences in water chemistry suggest contrasting groundwater recharge conditions, flow and hydrochemical evolution. A positive correlation between δ 18 O-H 2 O and Cl for both seams and generally higher TDS and cation and anion contents in the deeper, GM seam suggest a possible connate influence at greater depths; however, multi-isotope evidence indicates that coal bed waters from the production zone (100-500 m depth) for both seams are likely to be primarily meteoric in origin and evolved via a combination of low temperature fluid-rock interaction processes and microbial activity. Interestingly, this study found that higher gas production correlates with higher TDS. In the GM seam, higher TDS correlates neither with an increase in the ash content of the coal, which is highest in the central and southern parts of the field [3,39], nor with the presence of a thick roof mudstone in the north (Figure 2b). Less is known on lateral variability in P seam coal character, but the observation that P seam production performance generally follows that of the GM seam [39] suggests that spatial variation in ash content is not a dominant control on coal-bed water chemistry. A more plausible interpretation is that wells with higher gas production have higher TDS due to increased fluid-rock interaction and microbial activity, which supports trends in the stable isotope data that postulate longer groundwater residence times and pooling in the north.
The highest alkalinity values accompany the most negative water isotopes and most positive gas isotopes in the area of highest gas contents, saturation and production, north of the hinge zone. Contrasting gas and water isotopes, sodicity and δ 13 C DIC /alkalinity signatures for deeper wells south of the hinge zone reflect a cutout in permeability and a lower extent of methanogenesis. High gas production occurs at shallow to intermediate levels, which is probably due to low overburden pressure while deeper, the seams display low production (and anomalous geochemical signatures) controlled by a decrease in permeability. Similar trends have also been observed in the Sydney Basin [9,51,52]. Overall, our new results support the hypothesis that the faulted hinge zone acts as a barrier to fluid communication along strike (NW to SE) [3] and suggest that comparatively restricted fluid flow from the north deprived the southern section of laterally migrated gas and meteoric ingress with necessary waters and methanogens for secondary CSG generation, reducing gas contents and saturation ( Figure 14). gas isotopes in the area of highest gas contents, saturation and production, north of the hinge zone. Contrasting gas and water isotopes, sodicity and δ 13 CDIC/alkalinity signatures for deeper wells south of the hinge zone reflect a cutout in permeability and a lower extent of methanogenesis. High gas production occurs at shallow to intermediate levels, which is probably due to low overburden pressure while deeper, the seams display low production (and anomalous geochemical signatures) controlled by a decrease in permeability. Similar trends have also been observed in the Sydney Basin [9,51,52]. Overall, our new results support the hypothesis that the faulted hinge zone acts as a barrier to fluid communication along strike (NW to SE) [3] and suggest that comparatively restricted fluid flow from the north deprived the southern section of laterally migrated gas and meteoric ingress with necessary waters and methanogens for secondary CSG generation, reducing gas contents and saturation ( Figure 14).

Value of Expanded Hydrochemical (Major Ions) and Stable Isotope Datasets for Understanding Gas Domains and Production Behaviour
An intrinsic focus of this study was to address the value of expanded hydrochemical (major ions) and stable isotope datasets for understanding gas domains and production behaviour in the study field. For a given depth, differing production behaviour north and south, and among closely spaced wells, can be correlated with several different geochemical and isotopic signatures (notably, CH4 and CO2 isotopes, sodicity, alkalinity, TDS, Ca/Mg ratios and O and H water isotopes). With the addition of new data, previously proposed correlations between Δ 13 C(CO2-CH4) and gas production, and δ 2 H-H2O and δ 18 O-H2O values and water production [3] were not observed on an individual well basis; however, all three tracers provided insights into variable gas production behaviour.
Combination of new δ 13 CDIC data with alkalinity data revealed a novel relationship with production, requiring further analysis. Further detailed studies on the isotopic chemistry of the coal bed waters (particularly δ 13 CDIC, but also 87 Sr/ 86 Sr, 36 Cl, 14 C, δ 34 S) are needed to assist hypothesis testing for one or more phases of microbial gas generation and gain a better understanding of in situ microbial processes, fluid-rock interaction processes, finite reservoir effects, and groundwater flowlines and residence times.

Value of Expanded Hydrochemical (Major Ions) and Stable Isotope Datasets for Understanding Gas Domains and Production Behaviour
An intrinsic focus of this study was to address the value of expanded hydrochemical (major ions) and stable isotope datasets for understanding gas domains and production behaviour in the study field. For a given depth, differing production behaviour north and south, and among closely spaced wells, can be correlated with several different geochemical and isotopic signatures (notably, CH 4 and CO 2 isotopes, sodicity, alkalinity, TDS, Ca/Mg ratios and O and H water isotopes). With the addition of new data, previously proposed correlations between ∆ 13 C(CO 2 -CH 4 ) and gas production, and δ 2 H-H 2 O and δ 18 O-H 2 O values and water production [3] were not observed on an individual well basis; however, all three tracers provided insights into variable gas production behaviour.
Combination of new δ 13 C DIC data with alkalinity data revealed a novel relationship with production, requiring further analysis. Further detailed studies on the isotopic chemistry of the coal bed waters (particularly δ 13 C DIC , but also 87 Sr/ 86 Sr, 36 Cl, 14 C, δ 34 S) are needed to assist hypothesis testing for one or more phases of microbial gas generation and gain a better understanding of in situ microbial processes, fluid-rock interaction processes, finite reservoir effects, and groundwater flowlines and residence times.

Conclusions
This study used hydrochemical (major ions) and multi-isotope analysis (δ 13 C, δ 2 H, δ 18 O) to investigate geological controls on coal seam gas saturation domains and gas-well production performance in a small, high-rank (Rv > 1.1) CSG field in the northwestern Bowen Basin, Australia. Previous work on the study field suggested that: (1) coal-bed waters are predominantly meteoric with a possible connate influence at depth; (2) the δ 13 C-CH 4 values of co-produced gases reflect a mixing trend between shallow microbial gas and deeper thermogenic gas; and (3) CSG production rates are highest at intermediate depths where the CSG is a mixture of microbial and thermogenic gases. Results of this study strengthen the hypothesis that microbial gas generation is responsible for zones of higher gas saturation and production, but the addition of new analytical data has provided enhanced spatial coverage, leading to refined hypotheses on the distribution and origins of the CSG and the compositional variation of associated waters on opposite sides of a faulted hinge zone.
Comparison of our expanded analytical dataset against updated reservoir, production and structure datasets allowed improved geology-based interpretations and the identification of statistically significant geochemical differences within and between the upper (P)-and lower (GM)-producing coal seams. New hydrochemical and multi-isotope evidence indicates that coal bed waters from the production zone (100-500 m depth) are likely to be entirely meteoric in origin and evolved via a combination of low temperature fluid-rock interaction processes and microbial activity. Although new δ 13 C and δ 2 H data support a gas mixing trend hypothesis with δ 13 C-CH 4 increasing from more negative secondary biogenic (microbial CO 2 reduction) to less negative thermogenic values at depth (δ 13 C −62.2% to −46.3% ), geographic variations in gas content/saturation and production more likely reflect the extent of secondary microbial gas generation and retention as a function of hydrodynamics. Below 200 m, the majority of δ 13 C-CO 2 values are positive (δ 13 C −1.2% to 7.1% ) and δ 13 C DIC shows an erratic increase with depth and increasing alkalinity for both seams that is characteristic of evolution via microbial activity. Although there is possible evidence of minor thermogenic input, ∆ 13 C(CO 2 -CH 4 ) values and the overall progression of carbon isotopes along the CO 2 reduction fractionation line suggests progressive depletion of the CO 2 reservoir with increasing depth for both seams on both sides of the hinge zone.
Domains of differing gas production have different δ 13 C DIC /alkalinity signatures. Contrasting isotope, sodicity and δ 13 C DIC /alkalinity ratios for deep wells south of the hinge zone are consistent with a cutout in permeability. For a given depth, differing production behaviour among closely spaced wells can be correlated with different geochemical and isotopic signatures across mapped faults. Overall, stable isotope and water-quality data suggest that waters south of the hinge zone are not as microbially evolved as waters to the north. Faults clearly segment coal seams into areas having significantly different production, with results of geochemical analysis suggesting that pooling of biogenic gas and waters and enhanced methanogenesis occur north of the hinge zone. More enriched production gas δ 13 C-CH 4 values to the north of the hinge zone are consistent with this hypothesis when interpreted in terms of 13 C pool depletion effects.
The interpretation of a strong microbial influence, despite the high rank of the coal, has implications for interpreting and predicting production behaviour. Overall, our new results support the interpretation that the faulted hinge zone acts as a barrier to fluid communication along strike (NW to SE) and underline the need for further isotopic studies on the study field. Funding: Arrow Energy provided partial funding for the field work and geochemical analyses. This research received no other external funding.