Utah Water Science Center
USGS IN YOUR STATE
USGS Water Science Centers are located in each state.
Development of stream-based monitoring methods for assessing effects of natural gas development on water resources
Project Chiefs: Victor Heilweil, Bert Stolp, and David Susong, USGS Salt Lake City, Utah
Figure 1. Aerial photo of a region undergoing extensive oil and gas development in the Upper Colorado River Basin.
Natural-gas production in the United States has increased rapidly because of technological advances that have allowed extraction from unconventional resources (Schnoor, 2012; Dammel and others, 2011). Horizontal drilling and hydraulic fracturing (the process by which specific formations adjacent to a well are fractured to increase permeability) have made existing natural gas reservoirs more productive and have allowed the development of new shale gas and shale oil plays.
The rapid development and widespread application of hydraulic fracturing have resulted in significant public concern about the environmental effects on watersheds, ecosystems, and surface and groundwater resources (Pelley, 2003; Kargbo, 2011). Groundwater contamination from hydraulic fracturing is possible if fluids and (or) gases migrate along faults, fractures, or wells (Dammel and others, 2011; Lustgarten, 2009). Recent studies have established a possible link between increased methane concentrations in overlying aquifers with shale-gas horizontal drilling and hydraulic fracturing (Osburne and others, 2011; Jackson and others, 2013), or traditional vertical oil and gas wells that were improperly completed (Mufson, 2009; Renner, 2009).
Conceptual Model and Objectives
Our conceptual model of methane transport from an underlying natural gas reservoir into an overlying aquifer is shown in figure 2. Methane can move into overlying aquifers either as a dissolved gas in upwardly migrating fluids or as stray gas through fractures and improperly completed well bores. Groundwater from these aquifers can discharge to wells, springs, or along gaining stream reaches. Inflow to surface water along different groundwater flow paths provides an integrated signal of groundwater quality which can include dissolved methane and other potential contaminants from natural gas development activities.
Figure 2. Conceptual model of methane transport from a hydraulically-fractured gas reservoir to an overlying aquifer/stream system.
Adopting stream reach mass-balance techniques developed for other groundwater constituents (Cook and others, 2003), the concentration of methane in the stream can be described as the balance between the rate that methane enters the stream through groundwater inflow and loss of methane from the stream to the atmosphere and possibly also as bacterial consumption (fig. 3). By measuring stream methane (Cstr), gain to the stream from groundwater inflow (Qgw), and using the gas transfer coefficient for methane (λ), the groundwater methane concentration (Cgw) and methane load to the stream can be determined using a 1-D stream transport model with gas transfer.
Figure 3.Stream methane mass balance. Cgw is the methane concentration of groundwater inflow, Qgw is the volumetric flux of groundwater inflow, Cstr is the methane concentration in the stream, λ is the gas exchange coefficient for methane, and Catm is the air-equilibrated methane concentration in the stream.
The objective of this study is to develop simple yet robust methods for using dissolved methane measurements in streams to assess impacts of oil and gas development. Such methods have the distinct advantage of using stream-integrated chemical signatures to indirectly monitor groundwater processes and potential contamination at the watershed scale. The ultimate goal is to implement an integrated methane monitoring approach for a diverse range of streams.
Relevance and Benefits
The development of stream-based methane assessment methods for evaluating groundwater impacts from natural gas development will have a wide range of applications for both (1) the initial regional reconnaissance stage for determining areas of impact, and (2) detailed studies evaluating point-source fluxes of groundwater methane entering gaining streams. The approach utilizes baseflow conditions of a gaining stream as an integrated average of watershed-scale groundwater quality. These methods are cost-effective and relatively easy to implement, in comparison to the installation and sampling of monitoring-well networks. The biggest benefit in implementing this monitoring protocol will be in areas not yet disturbed by natural gas development efforts. Baseline stream water quality can be determined prior to development and followed up by repeat base-flow sampling to determine trends in methane (and other potential contaminant) concentrations.
To assess methods for monitoring methane in a stream, experiments were conducted at two sites using the following approach. After identifying a gaining stream reach (either with flow-meter measurements or stream gauge records): (1) measure stream methane concentrations and stream discharge; (2) conduct a stream tracer injection with a conservative tracer using ion dilution/synoptic sampling methods (Kilpatrick and Cobb, 1985; Kimball and Runkell, 2009) to accurately determine groundwater inflow along multiple segments; (3) conduct a stream dissolved-gas injection to determine the gas transfer coefficient (Genereux and Hemond, 1992; Stolp and others, 2010); and (4) develop a 1-D stream transport model with gas transfer (Cook, 2003) to determine the methane concentration of groundwater inflow.
Methods Development: Nine-Mile Creek, Utah
This approach was tested in Nine-Mile Creek, Utah (fig. 4) during low-flow conditions (Heilweil and others, 2013). Bromide dilution calculations showed a gain of about 35 percent in stream discharge due to groundwater inflow along the upper 1,500 m of the study reach. Methane was injected into the stream along the same reach and achieved a peak concentration of about 16 ppb, a more than 5-fold increase over background concentrations. Although the injected methane eventually diminished due to atmospheric loss, figure 5 shows that methane persisted in the stream for more than 1.5 km. This decline in methane, along with the groundwater inflow quantities determined by bromide dilution, was used in a 1-D stream transport model to determine an apparent methane gas transfer velocity (potentially including bacterial consumption) of about 4.5 m/d. Modeling the increase in background stream methane concentrations along the lower end of the Nine-Mile Creek study reach using the same gas transfer velocity indicates groundwater concentrations of up to about 200 ppb. The Nine-Mile Creek experiment demonstrated that dissolved methane persists at the kilometer scale and can be measured, even in a high-gradient and well-aerated stream. This indicates that the method is applicable to a diverse range of gaining streams as a tool for evaluating methane discharge to streams.
Results of Nine-Mile Creek are published in:
Methods Refinement: West Bear Creek, North Carolina
A follow-up injection experiment was conducted along West Bear Creek, North Carolina in March 2013 in collaboration with an NSF-funded study entitled "Evaluating how the sampling integration scale affects field estimates of groundwater transit time and nitrogen fluxes", being conducted by North Carolina State University and the University of Utah. West Bear Creek is a low gradient stream in the Piedmont Province having very different flow and watershed characteristics compared with Nine Mile Creek in Utah. This experiment included bromide, methane, and krypton injections (krypton has the same theoretical gas transfer velocity as methane) to independently evaluate stream methane loss to bacterial consumption. The injected methane persisted downstream for over 2.5 km (fig. 6) and the krypton is currently being analyzed by the University of Utah Dissolved and Noble Gas Lab to evaluate the bacterial consumption of methane.
Application: Marcellus Formation, Pennsylvania
The USGS Pennsylvania and Utah Water Science Centers have recently completed a pilot-scale stream methane study in the Marcellus Shale Gas development area of northeastern Pennsylvania. Reconnaissance samples collected from 15 streams during May and June of 2013 yielded dissolved methane concentrations ranging from < 0.5 to 68.5 µg/L. Detailed stream reach-mass balance methods were applied at Sugar Run, one of the watersheds with elevated stream methane and included stream discharge and methane sampling profiles for May, June, and November 2013. Peak stream methane concentrations were about 20, 67, and 28 µg/L during these three campaigns (fig. 7). Groundwater collected from shallow piezometers and a seep near the location of the observed peak in dissolved methane yielded groundwater end-member concentrations ranging from 2,300 to 4,600 µg/L. In order to refine estimated amounts and locations of methane-laden groundwater discharge to the stream, the lower part of the study reach was targeted during successive synoptic studies. Spacing between stream sampling for the successive synoptic studies was reduced from 800 m in May to 400 m in June to 200 m in November. The field data, combined with 1-D stream transport modeling, indicate a possible range of gas transfer velocities from 7.4 to 30 m/d and groundwater methane loads of 1.8 ± 0.8, 0.7 ± 0.3, and 0.7 ± 0.2 kg/d, respectively, entering at Sugar Run (fig. 8). This study illustrates the feasibility of the stream methane method for monitoring impacts associated with shale-gas development by showing that a simple reconnaissance sampling campaign can be used to identify streams receiving methane-laden groundwater discharge. Subsequent more-detailed stream and shallow groundwater methane sampling and discharge measurements along one stream (Sugar Run) during baseflow conditions, coupled with 1-D stream transport modeling, yielded estimates of both groundwater methane concentrations and methane loads. Repeat synoptic sampling campaigns along this stream during different seasons gave consistent results for two low-flow periods and larger methane loads during higher flow conditions. This indicates that the method can be used to assess both seasonal variations and longer-term trends in methane-laden groundwater discharge to streams. For high-gradient streams that have large gas-transfer velocities and a relatively small fraction of groundwater inflow, such as Sugar Run, synoptic sampling at closely spaced locations (perhaps every 100-m downstream) may be required to adequately capture peak stream methane concentrations. Alternatively, if a stream’s gas transfer velocity is low and it receives a large fraction of groundwater inflow, sample locations could be spaced farther apart.
Potential future application: Sanford Basin, NC
The USGS – North Carolina Water Science Center is currently developing a proposal for using these stream methane techniques for establishing baseline conditions prior to natural gas development in the Sanford Basin of central North Carolina. This would piggy-back upon an EPA-funded grant that was recently awarded to the North Carolina Division of Water Quality for characterizing baseline surface-water conditions in the Sanford Basin.
Monitoring and design implementation
The gas injections along Nine-Mile (Utah) and West Bear Creek (North Carolina) have shown the feasibility of stream sampling for determining groundwater methane concentrations. Its application along Sugar Run (Pennsylvania) demonstrated how preliminary reconnaissance sampling, followed by more detailed stream measurements, can be used for watershed-scale monitoring. In designing a stream-based methane monitoring program, the first steps would be to identify gaining stream reaches during baseflow conditions (using stream discharge measurements, tracer-dilution, or 222Rn techniques) and analyze stream methane concentrations to establish baseline conditions prior to gas development. Kilometer-scale spacing may be appropriate for initial reconnaissance sampling, but higher resolution could be necessary for evaluating point sources, especially in geologic settings with heterogeneous properties (e.g. bedrock fractures or lithologic variability). An intermediate step would also include higher-resolution stream discharge measurements (we suggest using an acoustic Doppler current profiler) for obtaining stream parameters including depth, width, and velocity that would be used in a preliminary stream-transport/gas-exchange numerical model for an initial estimate of methane fluxes. As gas development proceeds, additional seasonal or annual stream methane sampling during baseflow conditions could be used for trend evaluation. If temporal increases are observed, groundwater methane concentrations and loads to a stream can be more-precisely determined by calculating the gas transfer velocity using both conservative ion and tracer-gas injections coupled with stream-transport numerical modeling. Because of the spatial and temporal variability of groundwater-stream interactions (including hyporheic flow and transient storage), multiple measurement periods and detailed sampling may be required for a complete characterization of groundwater methane loading to streams. To detect impacts from unconventional gas development, the method would require substantially larger methane anomalies than the observed natural variability in background stream concentrations.