Introduction

Dissolved methane is often present in shallow groundwater (e.g., Barker and Fritz 1981; Darling and Gooddy 2006; Moritz et al. 2015; Humez et al. 2016a; Loomer et al. 2019), and although it is not toxic, its release from groundwater through or around wells or directly to overlying soils presents an explosion hazard to well houses and underground structures (Engelder and Zevenbergen 2018). Within an aquifer, the oxidation of methane can lead to the mobilization of trace metals such as arsenic (Cahill et al. 2017; Glodowska et al. 2020), and taste-odour issues (Gorody 2012). In addition, the release of methane from groundwater to the atmosphere, particularly through wells or where groundwater discharges to surface-water bodies (e.g., Heilweil et al. 2013), contributes a potent greenhouse gas.

Elevated concentrations of methane in shallow groundwater may occur through natural processes, such as: (1) the production of microbial methane via methanogenesis within the aquifer (e.g., Molofsky et al. 2011; Moritz et al. 2015; Humez et al. 2016a; Bordeleau et al. 2018a); (2) the release of thermogenic gas trapped within the aquifer itself (e.g., Molofsky et al. 2011; Bordeleau et al. 2018a); and (3) the migration of gas or groundwater through natural geologic pathways from deeper units containing methane, including unconventional hydrocarbon reservoirs (e.g., Warner et al. 2012; Sherwood et al. 2016). However, there are also widespread concerns over new or enhanced transport of methane to shallow aquifers during the development of oil and gas resources (Jackson et al. 2013; Vidic et al. 2013). Causes may include gas leaks associated with faulty casings of oil and gas wells (e.g., Stein et al. 2003; Bexte et al. 2008; Whyte et al. 2021) or improperly abandoned wells (e.g., Chafin 1994), and new or enhanced fracture pathways created by hydraulic fracturing of unconventional reservoirs (e.g., Vengosh et al. 2014).

To identify methane contamination of shallow groundwater caused by new oil and gas development, many jurisdictions now conduct groundwater sampling of existing water wells (usually domestic or farm water wells; Jackson and Heagle 2016) located within a given radius prior, during, and/or after unconventional hydrocarbon activities. Groundwater samples are typically analyzed for general groundwater chemistry parameters as well as concentrations of methane (C1) and possibly higher alkanes, like ethane (C2), propane (C3), butane (C4), and pentane (C5), with the higher alkanes associated with oil and gas reserves. Further analysis of the stable isotopic composition of these hydrocarbon gases can help distinguish both their origin (i.e., microbial versus thermogenic) and source (i.e., shallow aquifer versus a specific, deeper bedrock unit) (e.g., Hirsche and Mayer 2009; Humez et al. 2016a). Likewise, analysis of noble gases can provide valuable additional information on the source of the methane (e.g., Darrah et al. 2014). Indeed, there are a variety of developing geochemical methods that can be applied for this purpose, as recently outlined by McIntosh et al. (2018); however, many of these additional analyses are more costly and may not be required by regulations. In most cases, sampling conducted prior to development is meant to establish the baseline or the natural variability in these groundwater gas parameters within the area, with subsequent changes following development indicating a potential impact (e.g., Woda et al. 2018; Whyte et al. 2021).

Beyond this specific regulated sampling, researchers may employ groundwater sampling to improve our understanding of the potential natural variations in methane concentrations in aquifers (e.g., Humez et al. 2016b). Such understanding would be helpful for interpreting site-specific monitoring data with newly installed oil and gas wells (as described in the preceding) and for assessing potential impacts to groundwater resulting from legacy oil and gas development.

This reliance on periodic sampling of groundwater has some potential drawbacks, including that short-term changes in groundwater gas conditions may be missed. It is also important that baseline or background monitoring be performed over sufficiently long periods to increase the likelihood of capturing a more complete range of natural variation in groundwater methane concentrations. Humez et al. (2016b) also raised this issue, noting that long-term assessments “have rarely been systematically documented over observation periods spanning several years”.

An additional concern is the potential for degassing losses and atmospheric (air) contamination during groundwater sampling (Ryan et al. 2015). Dissolved gases can build up in groundwater until the total dissolved gas pressure (PTDG; sum of all partial pressures of dissolved gases) reaches the bubbling pressure—PBUB; sum of water or hydrostatic pressure (PW), atmospheric pressure (PATM) and capillary pressure, which is typically considered negligible in aquifers (Roy and Ryan 2013). Given that more gas can dissolve in groundwater at greater depths (termed “gas-charged” for PTDG > PATM), the dissolved gases will tend to exsolve or bubble out from the groundwater with decreasing PBUB, for instance with pumping or by bringing a sample to surface (Roy and Ryan 2010, 2013). Gas-charged groundwater within an open well—i.e., one with no hydraulic packer (herein packer) to seal the well from the atmosphere, to prevent it acting as a “gas chimney”—can also experience degassing, and potentially some air contamination, without pumping (Roy and Ryan 2010).

Sampling at ground surface, where the vast majority of these samples are collected (see Hirsche and Mayer 2009 for various methods), is prone to air contamination (e.g., McLeish et al. 2007; Evans 2017) and to degassing losses (due to the large decrease in PBUB; Barker and Dickhout 1988; Hirsche and Mayer 2009; Evans 2017). An exception is surface sampling with Isoflasks, single-use plastic bags that are directly screwed to a spigot, providing complete capture of liquid and exsolved gas without contacting the atmosphere. More complicated down-hole samplers (e.g., McLeish et al. 2007; Gardner and Solomon 2009) avoid degassing losses and atmospheric contact at surface, but are rarely (if ever) used because the equipment in residential or farm wells impedes their deployment. However, all current sampling methods without exception are prone to issues from exsolution within the well during pumping (Roy and Ryan 2010; Ryan et al. 2015; Humez et al. 2016b; Evans 2017) or gas exchange with an open well (Roy and Ryan 2010).

Many studies have used methane concentrations from groundwater sampling at surface to investigate potential impacts of hydrocarbon activities on shallow aquifers (e.g., Molofsky et al. 2013; McIntosh et al. 2014; Moritz et al. 2015; Siegel et al. 2015; Humez et al. 2016a, b; Sherwood et al. 2016; Currell et al. 2017; Nicot et al. 2017; Bordeleau et al. 2018b; Rivard et al. 2018a; Woda et al. 2018; Loomer et al. 2019) or to assess long-term variation in background or baseline conditions (e.g., Humez et al. 2016b). Some of these have shown considerable changes in concentrations (even >100%) between samples collected in a given well over time, even over very short periods such as a few days or even hours (Gorody et al. 2005; Hirsche and Mayer 2009; Smith et al. 2016; Rivard et al. 2018; Botner et al. 2018). However, it is uncertain whether issues of sample collection accuracy and frequency, as already noted, have affected their results, and if so, to what extent. For instance, the 8-year study by Humez et al. (2016b) using gas samples collected from a single well revealed marked variabilities in methane concentration, but which they suggested was some mix of natural and method-associated variations.

A method for monitoring groundwater gas conditions that may alleviate these issues is currently available. It requires the use of a probe with a total dissolved gas pressure sensor (PTDG-probe), which allows for continual downhole measurement of PTDG. Pumping for groundwater sampling is only required occasionally to determine the dissolved gas composition, which is less affected by degassing than concentrations. The PTDG-probe’s application in a well below a packer can also reduce issues with open well degassing and atmospheric exchange. The use of a PTDG-probe for measuring dissolved gases in wells was first proposed by Manning et al. (2003). Although they suggested PTDG could be used to monitor the relative abundance of a ‘highly abundant gas’, they made no field measurements for methane. Subsequently, the only applications to methane-rich groundwater were for short-term investigations in wells in Alberta (western Canada) by Roy and Ryan (2010) and Evans (2017). Despite a call for its broader use for long-term monitoring of methane in groundwater by Roy and Ryan (2013), there have been no previously published reports of the application of a PTDG-probe for this purpose.

The objectives of the current study are (1) to demonstrate and evaluate the use of a PTDG-probe for long-term monitoring of methane-rich groundwater, and (2) to investigate and document the temporal behaviour of dissolved methane in an aquifer containing gas-charged groundwater. This work involved in situ continuous measurements of PTDG for >2 years and approximately bimonthly groundwater sampling to measure the dissolved gas composition over time, including methane and some higher alkane gases, to determine whether PTDG sensor measurements alone could represent a good proxy for methane concentrations in this region. This study used three monitoring wells from a previous project (Bordeleau et al. 2018a, b; Rivard et al. 2019) that studied potential impacts of shale gas activities on shallow groundwater in the Saint-Édouard area (eastern Canada). The shallow bedrock aquifer of this region is known to contain elevated methane concentrations (Bordeleau et al. 2018b).

Study area

This study was performed in the Saint-Édouard area, which is situated on the south shore of the St. Lawrence River, mostly within the St. Lawrence Platform, in southern Quebec, eastern Canada (Fig. 1). In the upper 2-km depth of the study area, black organic-rich mudstones (Utica Shale and Sainte-Rosalie Group – Lotbinière and Les Fonds formations) are capped by shallowing-upward flysch (Lorraine Group – Nicolet and Pontgrave formations; Lavoie 2008). The Lotbinière, Les Fonds and Nicolet formations generally consist of calcareous mudstone and siltstone, with variable organic matter content, and were shown to be in the oil window (Lavoie et al. 2016). The Upper Ordovician Utica Shale was explored from 2007 to 2010 to assess its potential for shale gas production, until a de facto moratorium came into force. No commercial production ever took place in the St. Lawrence Lowlands and no activities have resumed since 2010.

Fig. 1
figure 1

Location of the study area and of the three monitoring wells used for this study (based on Bordeleau et al. 2018a). The background map illustrates the shallow bedrock units. The Utica Shale, at the location of the shale gas wells (A267 and A275), can be found at a depth of ~2 km

Full size image

Surficial sediments in this region are usually thin (<10 m) and made up of reworked tills and near-shore sediments of the former Champlain Sea, except in a few areas where fine-grained marine sediments have accumulated in local lows of the paleo-topography (Rivard et al. 2018). Most residential wells in this region are completed into bedrock, with depths ranging from 30 to 80 m. Similarly, the monitoring wells that were drilled for the previous Saint-Édouard project (Bordeleau et al. 2018a, b; Rivard et al. 2019; Bordeleau et al. 2019) were open-hole bedrock wells with depths between 30 and 149 m. These wells were only cased across unconsolidated units and to 2 m within the bedrock. Borehole geophysical logging revealed that the upper 60 m of bedrock, and especially the upper 30 m, contain substantial fractures (Crow and Ladevèze 2015), and is believed to be the active zone of groundwater flow (Ladevèze et al. 2018). Hydraulic conductivities obtained using slug tests carried out with compressed air confirmed that these black shale and siltstone formations are moderately to poorly permeable, having values ranging between 2 × 10−9 and 1 × 10−5 m/s (Ladevèze et al. 2016). Groundwater from wells from this region is typically of good quality and is generally of the Na–HCO3 or Na–Cl type, although some shallower wells also had Ca–HCO3 water (Bordeleau et al. 2018b). Further details of the geology and hydrogeology of the study area are presented by Lavoie et al. (2014) and Rivard et al. (2018, 2019).

In the previous study, methane was found in the vast majority of the residential wells and monitoring wells sampled, with widely variable concentrations (Bordeleau et al. 2018b). When sampled at the surface, many of these wells showed bubbles in the pump tubing, a sign of exsolution of gas-charged groundwater. Most of the dissolved methane present in shallow groundwater of this study area was of microbial origin (Bordeleau et al. 2018b, 2019). However, some thermogenic gas was found in a few wells and was shown to come from the shallow rock aquifer itself (Bordeleau et al. 2018a), which was once buried 3–4 km deep (hundreds of millions of years ago), meeting the required conditions for thermogenic gas production.

Methodology

Well instrumentation

Three wells from the previous project (Bordeleau et al. 2018a, b; Rivard et al. 2019) with high methane concentrations (gas-charged groundwater) and accessible year-round, but with varied contexts, were chosen for this work (Table 1; Fig. 1). Well F2 was completed in the Appalachian piedmont (Bourret Formation) to 52 m depth; however, a piece of rock came off the borehole wall in 2016 or 2017 and now partially blocks the well at a depth of 17 m (still below the surface casing). Both wells F4 and F21 were drilled into the Les Fonds Formation. Well F4 was drilled to 60 m depth in a small bedrock valley where rock was found at a depth of 40 m; therefore, the bedrock section is only 20 m. Well F21 is the deepest well that was drilled, at 149 m, and experiences upward flow within the well. It is close to exploration gas wells A267 (vertical) and A275 (horizontal). All three wells are open boreholes drilled during 2013 or 2014, with surface casing placed through the overburden and ~2 m into the bedrock.

Table 1 Well and equipment deployment characteristics (m below surface) and summary of in situ pressures measured ~1.5 m below the hydraulic packer within the three studied wells: average hydrostatic or water pressure (PW) and bubbling pressure (PBUB) outside of pumping periods, maximum total dissolved gas pressure (PTDG), and the ratio of maximum PTDG to average PBUB. Mean barometric pressure was 100.5 kPa (10.25 m H2O)
Full size table

All three wells were equipped with a custom-made PTDG-probe connected to an external datalogger (CR1000; Campbell Scientific) and solar panel. Note that the sensor consists of a gas-permeable water-impermeable silicon membrane that isolates a gas-filled chamber containing a pressure transducer (Roy and Ryan 2010). In addition, a commercial PTDG-probe attached to a Hydrolab MS5 sonde (OTT Hydromet), with its own battery-power and data logging capabilities, was placed in well F2. All PTDG-probes were calibrated prior to installation for values ranging between 101.3 and 507 kPa (1–5.1 atm; 10.3–52.7 m H2O).

When monitoring gas-charged groundwater, Roy and Ryan (2010) recommended sealing the well above the measurement interval using a packer to ensure that groundwater does not continuously degas from the open well. A packer was installed in wells F4 and F21 in June 2016, and then reinstalled for the start of the study period in June 2017, at depths of about 48.5 and 47.5 m, respectively, given that borehole geophysics had shown the presence of flowing fractures slightly below these depths (Crow and Ladevèze 2015). Well F2 was equipped with a packer in June 2017 at a depth of about 14 m (above the rock blockage), above a fractured horizon. Pressure in these packers was verified at each visit; only F2 had to be slightly reinflated on two occasions (in November 2018 and March 2019), but the results showed no effect from this small pressure loss.

Well F4 was equipped with a bladder pump (below the packer), with “quick-connect” connectors used on the piping to avoid contact with the atmosphere, while dedicated tubing to allow pumping via a peristaltic pump at the surface was installed in wells F2 and F21. Below each packer (Table 1) was positioned a custom-made PTDG-probe, as well as a pressure transducer (Solinst) to measure water pressure and temperature. The Hydrolab sonde in well F2 was installed directly above the packer (~13 m) to allow for regular removal from the well to download data and change batteries. All logged measurements were collected every 15 min for F4 and F21, while PTDG was measured every 30 min for the custom-made sensor and every hour for the HydroLab sensor in well F2. A barometer (Solinst) was installed in well F2 to measure barometric pressure. A schematic of the set-up for well F2 is shown, as an example, in Fig. 2. The bottom of the pipe comprising the internal conduit of the packer, through which the cables and tubing pass, was sealed with a Styrofoam plug to prevent gas exsolution to the atmosphere.

Fig. 2
figure 2

Example of monitoring and sampling setup for well F2 (not to scale). The setup was similar for wells F4 and F21, but for differences in the installation depths (Table 1), the absence of a Hydrolab probe above the packer, and the use of a bladder pump below the packer in well F4

Full size image

Groundwater sampling and analyses

Groundwater samples were collected from these wells approximately every 2 months using low flow pumping (Bordeleau et al. 2018b). Once physico-chemical parameters (pH, electrical conductivity, temperature, dissolved oxygen, redox potential) stabilized, the groundwater was collected in a set of three vials at surface using the semiclosed sampling technique, which involves filling 40-ml glass vials in an upright position at the bottom of a larger container being overfilled with the same water (see Bordeleau et al. 2018b, 2019). Teflon-coated silicon septa and open-top caps were used. After filling, the vials were rapidly immersed upside down in a larger plastic bottle filled with the same groundwater to minimize diffusive gas loss during shipping and storage. Samples were kept cool in a cooler with ice or in a refrigerator at all times between collection and analysis. They were shipped using express delivery to the laboratory in Burlington, Ontario (Canada), where analysis typically occurred within 5–7 days after sampling.

All three vials per sample were prepared and sampled using the procedure described in Capasso and Inguggiato (1998) using a 10-ml helium headspace. Any initial gas phase volume in the vials was estimated and added to the total headspace volume for calculations. Major gas species, including N2, O2 (includes Ar), methane, and CO2 were analyzed using gas chromatography (SRI 8610C GC with multigas No. 1 configuration, Torrance, CA, USA) coupled to a thermal conductivity detector (TCD) in series with a helium ionization detector (HID). In addition, ethane, propane, butane, pentane and hexane were analyzed with an Agilent 6890 GC Flame Ionization detector (FID), Santa Clara, CA, USA. Due to a series of parts malfunctions with the GC-FID, starting 5 June 2018, ethane was determined by SRI GC/TCD/HID, and then starting 2 April 2019, propane data was also determined by SRI GC/TCD/HID and butane, pentane and hexane were not determined. Results were averaged across the three vials for a given sample, except if any showed notable differences in concentrations with indications of air contamination (i.e., elevated N2 and O2+Ar, with lower methane); in this case, the result from the suspect vial was not considered.

Calculation of dissolved methane concentrations

The concentrations of the dissolved gases in the groundwater samples were calculated from the % volume of each gas determined by the laboratory analysis, combined with the barometric pressure in the lab (~100.3 kPa), measured volumes of liquid (groundwater) and headspace in the vials, and the individual Henry’s constants of each gas at laboratory temperature (22°C).

The in-situ groundwater methane concentration was also determined from the measured groundwater PTDG and the gas composition determined from the groundwater samples. This calculation considers Dalton’s law, which states the total gas pressure (or total dissolved gas pressure) is equal to the sum of all dissolved gas partial pressures plus the water vapour pressure (1.04 kPa at 7°C), which in this case is approximately:

$$ {P}_{\mathrm{TDG}}={P}_{\mathrm{N}2}+{P}_{\mathrm{C}\mathrm{O}2}+{P}_{\mathrm{O}2}+{P}_{\mathrm{Ar}}+{P}_{\mathrm{C}\mathrm{H}4}+{P}_{\mathrm{C}2\mathrm{H}6}+{P}_{\mathrm{H}2\mathrm{O}} $$
(1)

It also assumes application of the ideal gas law:

$$ PV= nRT $$
(2)

where V is volume, n is moles of a given gas, T is absolute temperature, and R is the ideal gas constant. From these, a ratio of the methane partial pressure to PTDG results in:

$$ {P}_{\mathrm{CH}4}=\left({n}_{\mathrm{CH}4}/\sum {n}_{\mathrm{Gas}}\right)\ {P}_{\mathrm{TDG}}={X}_{\mathrm{CH}4}\kern0.5em {P}_{\mathrm{TDG}} $$
(3)

where ΣnGas is the sum of moles of all dissolved gases, including water vapour, and XCH4 is the mole fraction of methane (equilibrated gas phase) versus all gases dissolved in the groundwater. From this, the methane concentration for the groundwater can be determined with Henry’s law for the temperature measured in the well (~7°C for all three wells). Values for the Henry’s solubility constant for the dissolved gases considered here for fresh groundwater aquifers (total dissolved solids <2,000 mg/L), for both laboratory and field temperatures, were determined from Sander (2015).

Results and discussion

Total dissolved gas pressure measurements

The total dissolved gas pressure (PTDG) from the custom-made probe and the bubbling pressure (PBUB), calculated from the water pressure and atmospheric pressure data, for the three study wells are plotted versus time in Fig. 3. These apply to well conditions just below the packer. The PTDG data from the HydroLab sensor installed above the packer in well F2 are also included. The PTDG was notably above atmospheric pressure (10.3 m H2O) for all three wells, indicating gas-charged groundwater conditions prevailed throughout the study period, reaching approximately 13.9, 40, and 53 m H2O (1.3, 3.9, and 5.1 atm) for wells F2, F4, and F21, respectively.

Fig. 3
figure 3

Continuous PTDG values obtained from the custom-made (custom) PTDG-probes and bubbling pressure (PBUB; hydrostatic pressure plus atmospheric pressure) from below the hydraulic packer in wells a F2, b F4 and c F21 (note some lost data), and PTDG from the HydroLab sensor situated above the packer in well F2 (a). Note the different y-axis ranges; arrows indicate notable PTDG changes outside of pumping periods

Full size image

The PBUB is much larger for wells F4 and F21 than for well F2, as monitoring was performed much deeper in these wells (42–46 m below the static water level, compared to 12 m for well F2; Table 1). The lower PBUB may be contributing to the much smaller PTDG values for well F2 (Fig. 3). For each well, there is a notable sharp decline in PBUB at times of pumping (ca. bimonthly), which results from the decrease in water pressure due to drawdown. This decline is generally <5 m for F2 and F21, but reached ~30 m for well F4. Although a very low pumping rate (typically <200 ml/min) was used, the combination of the presence of a packer restricting the interval to be pumped to the lower part of the well combined with the low transmissivity of the shale aquifer resulted in significant drawdown. Without a packer, drawdown in well F4 typically remained <3 m, with a median of 0.6 m (Rivard et al. 2018). Across the monitoring period, excluding the short periods of pumping, the measured PTDG as a percentage of the associated PBUB was around 60% (reaching 62%) for well F2, ranged from 59–76% for well F4, and reached 94% in 2019 for well F21 (Table 1).

An assessment of the values and patterns of PTDG for each well follows below. It includes consideration of long-term trends and short-term changes in PTDG, including during and following pumping for sample collection. The observed patterns reveal the responsiveness of the readings to induced changes and potential complications affecting data interpretation that are associated with the PTDG monitoring method. They also provide insight into natural variation in dissolved gas conditions that may occur in methane-rich bedrock aquifers.

Well F2

The PTDG of the groundwater below the packer for well F2, as measured by the custom-made PTDG-probe, was the most stable over the 2-year monitoring period, remaining around 13.2 m H2O before showing a slow rise to 13.9 m H2O across the final ~8 months (Fig. 3a). The cause of the small slow rise is not known. On a few occasions, very small increases in PTDG followed pumping (an example is provided in Fig. 4a), revealing some input of groundwater at slightly higher PTDG; this could be from a slightly shallower depth, with likely higher permeability and PTDG (as measured by the Hydrolab above the packer). Otherwise, there were no observed short-term changes in PTDG. Although sensors in this well were installed in the groundwater active zone for this study area (i.e., upper 30 m of bedrock), groundwater below the packer in well F2 is likely relatively stagnant, with the water level (reflected by PBUB) varying at most by ~0.5 m H2O.

Fig. 4
figure 4

Examples of custom-made (c-m) PTDG-probe and PBUB (reflecting water pressure) responses during and after sampling events for a well F2 (March 2018), b well F4 (February 2018), and c well F21 (July 2017). Note the different y-axis ranges

Full size image

The Hydrolab probe gave a similar long-term average PTDG (15 m H2O), but showed more short-term variation than the custom-made probe. This variability is likely associated with mixing in the top portion of the borehole (i.e., above the packer; Fig. 2) from different and more permeable zones in the shallow system, but could also be influenced by recharge or open well degassing (Roy and Ryan 2010).

Well F4

In contrast to well F2, the custom-made PTDG-probe in well F4 showed marked decreases in PTDG during pumping, with a subsequent rebound, typically to values higher than before pumping started (Fig. 3b). This occurred for all sampling events. A more detailed look at one example pumping event (early Feb. 2018; Fig. 4b) reveals an initial small increase in PTDG (but still often >1 m H2O) immediately following the onset of drawdown, prior to the PTDG decline after pumping ceased. This pattern occurred for most of the sampling events. Over the following weeks, PTDG would then typically slowly decline (Fig. 3b). These observations illustrate the responsiveness of the PTDG-probe to short-term changes in downhole PTDG, noting that the relatively slower changes in PTDG relative to PBUB may be partly due to methane sorption onto the silicon membrane (Balilehvand et al. 2012). They also provide some insight into potential variability in PTDG related to the monitoring set-up.

The initial increase in PTDG (Fig. 4b) suggests pumping was drawing groundwater with higher PTDG to the PTDG-probe. This could be from deeper layers, or more likely, the more permeable overburden/bedrock interface located a few meters above the packer. Note that the recorded electrical conductivity (EC) typically declined from its initially stabilized value (after <20 min) before stabilizing again (after another 30 min). Once the pumping stopped, there was greater opportunity for degassing of the groundwater in the well around the probe, caused by PBUB being below the PTDG due to drawdown (Roy and Ryan 2010), which caused the subsequent decline in PTDG. The following slow rise to higher PTDG upon well recovery likely reflects slow flushing of the well and probe by the higher PTDG groundwater pulled into the surrounding formation under the natural flow gradient. Finally, the eventual slow decline would then illustrate the return to the background groundwater conditions of that depth. Such mixing of groundwater of different layers was noted by Loomer et al. (2018) as a likely cause of the larger temporal variation in methane concentrations for wells in their study area (Sussex, New Brunswick, Canada). This indicates that natural changes in flow to the well could cause similar changes in PTDG.

Interestingly, there were a few occasions of sudden changes in PTDG that were not associated with sampling (Fig. 3b). There were two instances of a small decrease in PTDG: 12–16 March 2018 with a decline of ~1.5 m H2O and 27–28 April 2018 with a decline of ~1 m H2O. These changes cannot be caused by nearby pumping because this well is located in a forested area where only a few hunting cabins (without running water) are present. Furthermore, there was no consistent change in water pressure at those dates (not shown). The cause of these PTDG decreases is unknown, but they may reflect a change in the surrounding groundwater flow direction or rate, perhaps related to spring recharge events. There was also a larger sudden PTDG increase (~4 m H2O) during 6–16 January 2019, which coincided with a short-term (~1 week) decline in water pressure (~0.3 m), to about the lowest value observed during the monitoring period. The cause of this change in water level is uncertain, but it may have induced a change in the vertical hydraulic gradient. At the subsequent pumping event, the rebound in PTDG did not rise above its prepumping value as was typically observed (Fig. 3b), further suggesting a change in the local groundwater conditions. Alternatively, the low water pressure may have triggered an ebullition event, with a gas phase moving up from depth and dissolving into the local groundwater.

Well F4 also exhibited evidence of long-term variability, though this is somewhat confounded by the PTDG rebound and subsequent slow decline after the pumping events (Fig. 3b). However, there were periods with constant PTDG, including through Oct–Nov 2017 and the end of 2018, indicating readings could stabilize within a month or two of pumping. Also, there was often a change in slope of the slow decline in PTDG, perhaps indicating a change from a flushing effect to true behavior. Considering the more stable periods, PTDG changed by at least 5 m H2O during the monitoring period (Fig. 3b). The driver(s) of this change is uncertain, but it does not appear to be the water pressure, which was at its highest in late spring and declined to its lowest values (but only by ~0.5 m) in early autumn. The PTDG did not show a related pattern.

Well F21

Unlike the previous two wells, well F21 shows a notable long-term trend in PTDG, with a slow build from ~42 m H2O in June 2017 to what seems a more stable condition for October 2018 onwards of 51–53 m H2O (Fig. 3c). If such a change in PTDG, ~1 atm over 1 year, or a similarly large change in methane concentration from groundwater sampling, occurred in a domestic or farm well being monitored near shale gas production, it would surely raise concerns over the possible slow dissolution of a recently mobilized gas phase into the groundwater supplying this borehole. Well F21 is located 120 m away from an exploratory shale gas well (Fig. 1). It is also possible for gas to have mobilized naturally. However, it seems most likely that this change is an artefact of the monitored system, reflecting the slow flushing of degassed groundwater from this deep borehole (149 m, with packer at 47.5 m).

The hydrogeologic conditions are well suited to this latter scenario. First, it seems likely that this highly gas-charged groundwater would have been degassed from the open well (i.e., the chimney effect; Roy and Ryan 2010) as it recovered from its initial complete purging (so very low PBUB originally) and in the ~2 years prior to packer installation. Second, minimal purging of the well was performed prior to each sampling and low-flow sampling was used, which both limit the flushing of degassed water from the large well volume of the borehole below the packer (~0.78 m3). Third, a low natural groundwater flushing rate seems likely given the very low permeability of the shale formation (overall hydraulic conductivity (K) of 3.35 × 10-9 m/s), and which is likely even lower for the isolated bottom section of the well. This point is further supported by the large drawdowns (2–5 m) that occurred with low-flow pumping. Finally, the noted upward flow in the well (see section ‘Study area’) with a potentially large reservoir of degassed water existing below the PTDG-probe and packer could explain the slow climb to stable conditions.

Under this scenario, the plateau level of PTDG reached during the latter half of the monitoring period likely represents the true dissolved gas conditions of the deep groundwater of well F21, at least at that time. However, this means that long-term trends cannot be assessed with the current data set. The PTDG of >50 m H2O is very high, even surpassing the highest PTDG previously reported of ~40 m H2O for a monitoring well completed in formations containing coal in Alberta, Canada (Roy and Ryan 2010). Indeed, the groundwater bubbled vigorously during the sampling events of well F21. The fact that the final PTDG has nearly reached PBUB (93%, Table 1) means that there could be gas phase (i.e., bubbles) present when PTDG reaches PBUB just above the monitoring depth.

The likely slow flushing of well F21 complicates the assessment of short-term changes in PTDG. During pumping, the PTDG remained below PBUB for nearly all sampling dates and thus showed only a muted (positive or negative) response. However, there was a notable decline in PTDG (up to ~1 m H2O) approximately 2–3 days following pumping for over half of the events, with slow recovery taking a week or more. An example of this pattern is shown for the sampling of 18 July 2017 (Fig. 4c). Considering the upward flow observed in this borehole, this pattern may reflect inputs of degassed or naturally lower-PTDG groundwater from the aquifer to deeper parts of the borehole during pumping, which then slowly ascended to the probe during the recovery period.

Outside of these pumping periods, two small rises in PTDG of ~0.5 m H2O over several days were observed in late February and late April 2018, with no notable change in water levels (Fig. 3c). There also appears to be substantial short-term fluctuations in the PTDG readings (up and down over a few days). The amplitude of these fluctuations grew larger over time and continued through the steady PTDG period, reaching >0.5 m H2O. The PBUB also fluctuated, though only by about 0.1 m, and with no apparent relationship with the PTDG. These PTDG increases and fluctuations might reflect slight changes in flow into and within the well and thus variable mixing of groundwater from different zones.

Investigators could avoid or limit such monitoring artefacts associated with degassed groundwater and long open boreholes by installing a second packer at a lower depth to isolate a shorter section of borehole, though it would be more technically challenging. Additionally, greater purging or longer slow pumping of the well could be employed for those with decent permeability (unlike well F21). Also, simultaneous long-term in-well monitoring of parameters such as EC and pH could help reveal contributions of groundwater from different zones, as suspected here.

Dissolved methane in groundwater

Methane molar ratios

The molar ratio of methane to other gases present in the groundwater, which is required for the calculation of methane concentrations from PTDG measurements (Eq. 3), was determined from groundwater samples collected at surface from all three wells. Figure 5 shows the molar proportions of the four most dominant dissolved gases (methane, CH4; nitrogen, N2; carbon dioxide, CO2; and oxygen combined with argon, O2+Ar), with ethane included for well F21. Methane is the dominant gas for all three wells, typically >80% of the total gases. Dissolved O2 (plus Ar) concentrations were generally low in these samples (further discussed in the following), as was reported for bedrock wells in this region (Bordeleau et al. 2018b). The longer-chain hydrocarbons propane, butane, and pentane, were only detected in samples from well F21, but at very low levels (not shown).

Fig. 5
figure 5

Proportions of dissolved gases from groundwater samples collected from the three monitoring wells a F2, b F4, and c F21, over the course of the study. Note: O2 includes Ar

Full size image

Due to the very low permeability of these bedrock units, distance from the water table (i.e. no direct recharge), and lack of urban or energy development in the area, rapid or major changes in gas composition were not expected. However, some variation across samples was observed for each of the three wells. Some of this variation is likely due to gas exchange while sampling at surface, which would lead to a loss of methane from the sampled groundwater combined with a gain of atmospheric gases (i.e., N2 and O2). The latter can also increase the proportion of CO2 while simultaneously lowering the methane and O2 compositions, if methane oxidation occurs in the vials during storage. There are several samples collected from each well (but particularly for well F21) with notably higher proportions of O2 and/or CO2 along with N2, which also have relatively lower methane, which together are suggestive of air contamination. Other samples may still have been affected (rarely were O2 concentrations zero), but to a lesser extent.

Potential degassing losses from the vials during storage via diffusion across the Teflon-coated silicon septum will have less of an impact on composition (one-way transfer largely), but could possibly lead to slightly lower methane composition, noting that McLeish et al. (2007) calculated that the diffusion coefficient for a silicon membrane was four and two times greater for methane than for N2 and O2, respectively.

Issues with atmospheric contamination and degassing, along with mixing of groundwater from different zones or within long open boreholes (e.g., well F21), make it difficult to determine true temporal changes in the dissolved gas composition. However, overall, the percent contribution of methane to the total gas concentration in these three wells was within a fairly narrow range, approximately 85–95% for well F2, 90–97% for well F4, and 85–95% for well F21 (plateau period; August 2018 onward), ignoring some obvious air-contaminated samples. It is not clear that the actual methane percentage changes over these periods; these ranges may just represent variation of the measurements.

Methane concentrations

For groundwater with methane as the dominant dissolved gas and with a methane molar ratio that is quite constant over time, a single methane molar ratio may be used in the calculation of methane concentration from PTDG data, making PTDG a good proxy for methane concentration behaviour over time. Otherwise, if the methane molar ratio is changing or its stability is unknown, values determined from individual groundwater sampling events can be applied to different parts of the PTDG data time series. However, if the composition changes are substantial, then interpolation between sampling events may be questionable.

In this study, the gas composition was reasonably stable (typically ±15%) for each well over the applicable PTDG data record (July 2017–July 2019 for wells F2 and F4; but only mid-2018 onward, the plateau region, for well F21), with any minor temporal trend obscured by the uncertainty associated with the individual gas composition results, noted above. Thus, it is assumed that the methane molar ratio remained essentially constant in time; however, to account for the uncertainty in the composition data (including potential real and measurement variation), the aforementioned ranges were applied to the respective PTDG data to calculate an upper and lower bound for each well’s methane concentrations (Table 2). The resulting PTDG-derived methane concentrations are all at or above methane solubility for fresh water at 7°C and atmospheric pressure (32 mg/L; determined from Sander (2015)), and are especially high for wells F4 and F21, which reflects the greater capacity to dissolve gases under higher water pressure (i.e., higher PBUB) at their greater monitoring depths.

Table 2 The methane molar ratio range used to derive methane concentrations from PTDG measured in situ (PTDG-derived) and the methane concentrations determined from laboratory analysis of groundwater samples (analytically derived) collected July 2017–July 2019 for wells F2 and F4, and August 2018–July 2019 for well F21, excluding data with suspect atmospheric contamination
Full size table

The temporal patterns of the methane concentrations are illustrated in Fig. 6 by applying the maximum methane molar ratio across the time series. Actual concentrations might be slightly lower at times if and when the true methane molar ratio does decline below its maximum, but the general temporal pattern should apply nonetheless. Obviously, these time series follow that of the PTDG measurements (Fig. 3), being broadly steady (but noting the temporally limited results for well F21) with some evidences of small short-term and potentially seasonal changes. Such muted patterns are not surprising given the low permeability of the aquifer and its lack of response to recharge, the stable groundwater temperature (affecting methanogenesis) throughout the year (<1 °C change; data not shown), and the lack of any substantial oil and gas activities and water extraction nearby, which might alter groundwater flows or induce gas phase mobility from deeper formations. The steady concentrations also fit with observations of consistent methane isotopic composition for a given well and depth over time determined in past work in this study area (Rivard et al. 2018).

Fig. 6
figure 6

Groundwater dissolved methane concentrations for wells a F2, b F4, and c F21, determined from PTDG data of the custom-made probes combined with molar methane gas ratio data (maximum values from Table 2) from groundwater samples, according to Eq. 3 (line), and determined analytically from groundwater samples (circles)

Full size image

Finally, the methane concentrations determined directly from the laboratory analysis of groundwater samples (circles in Fig. 6) exhibited substantial variability (up to >200%) and were up to approximately 3, 4, and 12 times lower than the PTDG-derived values for wells F2, F4, and F21, respectively (Table 2). This demonstrates the extensive and variable degassing losses that can lead to underestimations of dissolved gas concentrations for samples of gas-charged groundwater, with potential losses from the well during pumping, at surface during the sample collection, and during shipping and storage prior to analysis. The losses may be more extensive for well F21 because its groundwater PTDG is higher and near the PBUB, and so substantial ebullition is likely occurring not only in the pump tubing up to the surface, but also in the well below the packer.

Advantages and limitations of P TDG in methane monitoring

The findings for the three wells studied here provide insight into the effectiveness and potential challenges of monitoring methane concentrations in methane-rich aquifers with this PTDG method. Clearly one of the main benefits in comparison to collecting and analyzing groundwater samples is that the PTDG method provides continual in situ measurements rather than infrequent spot measurements. Thus, it is more likely to detect sudden and short-term changes (as observed here for well F4 and F21), while also providing more information on the nature of such a change, which might help identify its cause. In addition, the PTDG method of monitoring avoids or lessens changes brought about by disturbance of the system by pumping (such as mixing from different zones), exacerbated in low permeability settings like these, so long as gas composition sampling is not required (i.e., it is known and stable) or is required less frequently. Even when groundwater sampling is required, degassing losses that commonly occur when sampling gas-charged groundwater have less of an effect on gas composition than on gas concentration. Here, the PTDG-derived methane concentrations were up to 12 times higher than concentrations determined by laboratory analysis of groundwater samples (though with a “semiclosed” method more prone to degassing losses). Other sampling methods (e.g., Isoflasks; Molofsky et al. 2016) would have reduced these losses, but not completely if long-term degassing occurred in the open well or surrounding formation, which was likely the case for at least one well here (F4). Other benefits may include reduced field trips, especially if remote data access systems were employed, and lower analytical costs. Finally, the continual measurements of PTDG during pumping while sampling was also helpful in revealing potential degassing and mixing issues that might have affected the groundwater samples’ gas composition data. To the best of the authors’ knowledge, no current commercial or custom-made sensors are able to record methane concentrations directly at these high concentrations.

The PTDG method also has some limitations. It requires that methane be the dominant dissolved gas and that gas composition be fairly stable; the limits on those two requirements has not yet been explored. As for all dissolved gas monitoring, it has an upper limit of measurement of PTDG = PBUB, at which point a gas phase is likely present in the well and/or surrounding formation. One major issue with this approach is that the PTDG-probe must be placed within a well with a packer installed above (unless continually purged). This means that wells that contain equipment (e.g., a pump) cannot be readily used, including domestic water wells, which are currently relied upon for monitoring by the oil and gas industry (Jackson and Heagle 2016). Installation of packers can also be complicated and they require monitoring for depressurization, on top of the equipment costs. This study also experienced some challenges to monitoring PTDG in long open boreholes in low-permeable formations, despite employing a packer to prevent open-well degassing. Limited flow through the borehole led to delays in freshening the borehole groundwater following initial installation and after pumping events (which appear to have increased groundwater mixing from different zones). Also, probable vertical gradients causing flow in the borehole during nonpumping periods complicated the interpretation of PTDG trends. However, these are likely manageable with moderate changes to the preparation (e.g., purging and pumping patterns), installation (e.g., possibly using two packers), and sampling procedures.

Conclusions

This study provides the first demonstration of the use of PTDG-probes to measure continually in situ (downhole) PTDG for long-term monitoring of high-methane groundwater. The PTDG-probes and associated hydraulic packers and water pressure transducers were installed for >2 years in three bedrock boreholes at depths ranging from 15 to 50 m. The PTDG was high (>1 atm, so gas-charged) for all three wells, including one well with the highest PTDG value yet reported in the literature at 53.0 m H2O (5.1 atm; 520 kPa). The continual PTDG readings revealed long-term trends, as well as short-term changes associated with pumping (degassing or within-well mixing) and nonpumping periods (up to 4 m H2O in well F4) that reflected some uncertain natural or background dynamics in the groundwater gas or flow conditions.

This study also demonstrated and discussed how the PTDG monitoring data can be used to derive continual long-term groundwater methane concentrations for methane dominant groundwater, given a known or measured methane molar ratio of the dissolved gases (in this case, determined by groundwater sampling and dissolved gas composition analysis). For these methane-dominant groundwaters, PTDG provided a good proxy for the methane concentration temporal patterns for these wells. Advantages of monitoring groundwater methane concentrations using PTDG in comparison to collecting and analyzing groundwater samples, as well as limitations of the method and possible solutions to these, were also discussed.

This study highlights a valuable new tool for continual monitoring of groundwater methane conditions that could be applied to areas where oil and gas development activities occur, as is now conducted in many jurisdictions (Jackson and Heagle 2016). For this purpose, the PTDG method could be applied as a specialized tool in dedicated monitoring wells or unused domestic/farm wells with pumps removed to better address problem areas. A move to greater use of monitoring wells by the industry or regulators could allow a broader application of this method, to provide a much clearer picture of potential changes occurring in production areas or around individual production wells. Likewise, the PTDG method could be applied to aquifers with naturally elevated methane concentrations as a management tool to limit adverse impacts. Additionally, it could be applied as a scientific tool to improve the understanding of groundwater methane dynamics under background conditions, which is important to inform the interpretation of the aforementioned monitoring, and possibly also for evaluating this source of greenhouse gas emissions.

Finally, the observations from this study also reveal the nature of background groundwater gas and specifically methane concentrations within this shale aquifer located above the Utica Shale in the St. Lawrence Lowlands (eastern Canada). The PTDG-derived methane concentrations were high, even surpassing 150 mg/L for the deepest well, and were generally stable, which seems reasonable for these wells completed within a low-permeability bedrock aquifer with low recharge, and with no major groundwater extraction or oil and gas development nearby. Given this hydrogeological context being typical of aquifers located above unconventional shale reservoirs, it is assumed that the observed patterns in PTDG and methane concentrations obtained in this study will be informative of groundwater conditions in many similar regions.