Advances in fractured rock characterization and monitoring in the past 20 years have improved understanding of fractured rock, particularly as it relates to the geometry and transport properties of rock fractures and to fluid exchange between rock fractures and matrix. Characterizing hydraulic properties of rock fractures is now standard in practice and provides critical information about flow paths and physical processes that affect chemical migration. Equally important are characterization and monitoring of groundwater geochemistry, including spatial and temporal variabilities, and experimental procedures that identify the significance of biogeochemical processes in contaminant fate and transport.
A number of methods for detecting fractures and their characteristics were outlined by the National Research Council (NRC) in 1996 and are still used, including wireline logs, tracer tests, seismic methods, and flowmeters (NRC, 1996). Recent advances include automated, longer-term, and autonomous data collection that allow physical, chemical, and biological processes that control the transport and fate of contaminants to be better quantified. A revolution in sensor technologies and low-cost techniques has occurred in the past two decades. This chapter focuses on techniques for use beneath the vadose zone that have evolved significantly over that time.
Significant difficulties remain in characterizing and monitoring fractured rock systems: models and decision making are not always informed by data, data are collected over a wide range of scales, biological process characterization remains in its infancy, and the collection and analysis of those data remain difficult for many practitioners. Characterization at depths greater than a few hundred meters compounds the difficulties because of the need to account for increases in pressure, temperature, and changes in salinity. The lack of accessibility to greater depths means more reliance on remote or indirect measurement of properties. Many of the same surface and downhole methods used commonly for nearer-surface characterization and monitoring, however, are also applied or planned for deep rock characterization (e.g., Arnold et al., 2012). Research related to the feasibility of geologic sequestration of carbon dioxide necessarily focuses often on issues related to characterization and monitoring at depth (see Box 5.1 for a brief description).
Improvements in practice will occur when parameter estimation methods for fractured rock systems and models are widely adopted. Such methods include joint inversion techniques that make it possible to synthesize all available information in large, complex, multi-disciplinary data sets. This chapter begins with a general discussion on the importance of geomechanical characterization and then describes geometric, hydraulic, geophysical, geochemical, and biological characterization techniques. The focus of these latter sections is on areas that show great promise in the characterization of the fractured rock environment.
Geomechanical characterization of fractures is important to understand the geometry, flow, and transport properties of existing natural fractures, as wells as the coupling of fracture flow and transport properties with rock stress and deformation. Conventional rock geomechanical characterization uses empirical methods (Marinos and Hoek, 2000; Zafirovski et al., 2012) to estimate in
situ rock mass strength and modulus. Because rock fracture geometry and properties are a major component of rock mass behavior, it is at least theoretically possible to utilize these empirical rock characterization techniques to estimate fracture geometry and properties. Even if the fundamental processes that govern fracture generation are known, the geometric characteristics of the void space within individual fractures and the three-dimensional spatial distribution of fractures cannot be mapped over large volumes deterministically. Geomechanical information can indicate primary fracture orientations and the likelihood of certain fractures being more open and transmissive to groundwater.
The oil and gas industry employs geomechanical characterization techniques such as microseismic methods for information about fracture geometry and in situ stress (e.g., Okada, 2003; Grob and van der Baan, 2011); borehole image and wireline methods used to determine in situ stress and rock mass mechanical properties (Zoback, 2010); and three-dimensional seismic methods for determining rock mass modulus and anisotropy (Dradjat et al., 2012). Borehole imaging
and wireline methods are currently used for environmental site characterization, but the geophysical methods described could also be useful in the characterization of fractured rock sites to inform the design and construction of engineered facilities and contaminant remediation efforts.
Characterization of lithologic geomechanical properties and the evolution of local and regional stress distributions can provide insight into mechanisms of fracture generation and the probable modes of fracturing. Fracture orientation and size, for example, can be defined by the geomechanics of deposition and orogeny, and by the history of tectonic stress and strains. The characterization techniques provide valuable but frequently qualitative and statistical, rather than quantitative and deterministic, information about geomechanical processes and degree of variability.
Fracture geomechanical characterization is an area in need of further development, particularly in the determination of local in situ effect stress fields and fracture geomechanical and coupled geomechanical/hydrogeologic properties. Research is ongoing in the mining (e.g., Bahrani and Tannant, 2011), petroleum (e.g., Warpinski et al., 2013), and nuclear waste management industries (e.g., Liu et al., 2012).
Systematic fracture mapping is an essential tool to delineate the geometries of discrete pathways and the nature of both fracture pathways and rock matrix. Systematic fracture mapping of the broad geologic setting through visual geometric characterization is an early step of any site characterization. Not all fractures at the 1-meter scale need to be mapped if such fractures are part of a single major shear structure that controls a kilometer-scale flow process; determining which fractures to map are a function of the process(es) of interest. Decisions regarding the need to map in more detail individual or sets of fractures can be better informed as more data are collected and site conceptual models are refined. See Chapter 4 for discussion on modeling and Chapter 7 for discussion on decision making. Information can be garnered through visual geometric examination of surface outcrops, boreholes, and bore cores. Informative fracture mapping always includes information about characteristics such as fracture roughness, infilling, porosity, damage zones, and mineralization.
Efficient fracture mapping begins with large-scale lineament mapping techniques such as evaluation of available orthophotography, light detection and ranging (LIDAR),1 conventional satellite imagery (e.g., LANDSAT/SPOT), and high-resolution airborne geophysical methods (e.g., magnetics and electromagnetics [EM] such as very-low frequency EM and frequency-domain EM). These methods, combined with suitable image processing, may make it possible to map fracture traces, shapes, spacings, and orientations (for non-planar surfaces) for entire outcrops and may yield quasi-three-dimensional representations of exposed fracture surfaces. Smaller-scale mapping (5–50 meters) may then provide greater detail of mechanical stratigraphy and fault architecture.
Traditional visual means to characterize fractures include simple direct or inferred measurements of fracture size, shape, orientation, and spacing from fractures exposed in outcrops. The total length of fractures may not be determined directly, but trace lengths (i.e., the length of the intersection of a fracture with the outcrop surface) can sometimes be determined with tools as simple as a measuring tape. Fracture size can then be inferred with suitable bias corrections (e.g., Mauldon, 1998; Zhang and Ding, 2010).
1 LIDAR is a remote sensing technology that combines light (via laser) and radar, illuminating a target with laser and analyzing reflected light with concepts from radar analysis to make high-resolution maps that can differentiate subtle topographic features.
Fracture shape is difficult to map from an outcrop unless the outcrop is three-dimensional (non-planar), but it may be predicted based on visual mapping, as in the case of some layered sedimentary rocks. The orientation of the fracture planes in space typically is defined through measurement of the strike and dip of a fracture (i.e., using a geologic compass, or sampling circle). The spacing between fracture traces belonging to a set of parallel or sub-parallel fractures can be measured along sampling lines that are usually oriented in the direction of the mean pole of the fracture (trace) set. Fracture aperture, roughness, and deviation of fractures from a given plane (e.g., waviness) can sometimes be measured with a ruler on an outcrop. Sometimes the rock bounding a fracture can be removed to expose the fracture surface. In such cases, the deviation from an ideal plane can be measured with tools such as a laser profilometer.
Current characterization and modeling strategies rely heavily on observations from boreholes, especially when characterizing sites for deep sequestration of wastes. Boreholes and bore cores yield important characterization information including fracture orientation (either dip or inclination relative to the bore axis if the bore core was oriented), spacing, and aperture. However, such observations only describe heterogeneities in the fractured rock environment at the borehole wall and may generate biased data due to the impacts of drilling the borehole into the formation. Care is necessary to differentiate between naturally formed fractures and those formed as a result of boring. Because coring boreholes is more expensive than drilling, site-specific approaches should define borehole and sampling strategies. If not precluded by cost, then it might be useful to consider using borings drilled in different directions and angles to help resolve sampling bias issues. Judicious and strategic use of coring to recover rock samples for physical and chemical analyses and fine-scale investigation of lithology and fractures can inform site hydrostructural models. Porosity and mineralogy can be analyzed from core samples, and mineral precipitates or weathering may indicate groundwater flow in specific fractures.
Borehole geophysical methods, especially wireline borehole logging, have been used for many years to characterize the fracture location, lithology, and fluid flow. Imaging methods (e.g., borehole cameras and video; acoustic and optical televiewer logging) can be used to identify fractures along the borehole wall. Borehole diameter logging using mechanical or acoustic caliper methods is also fundamental to quantifying the aperture of fractures and in the choice of hydraulic testing approaches.
Sampling bias needs to be accounted for whether fracture mapping data are gathered from surface or borehole/bore core observations. Visible fractures, for example, may be oversampled and their importance overstated in a hydrostructural model. Discrete fracture networking simulation techniques (i.e., Mauldon and Mauldon, 1997) could be useful to quantify the errors and uncertainties associated with these biases for all fracture mapping and fracture logging applications.
Boreholes provide not only opportunities to visually inspect the subsurface, but also an environment in which to analyze ambient and dynamic hydraulic conditions, conduct tests to estimate hydraulic properties, and collect water samples to evaluate groundwater chemistry and microbiology. Aquifer tests conducted in boreholes can provide explicit information about select groundwater pathways, but they cannot provide complete information about all possible groundwater paths. Test results need to be coupled with a strong knowledge of the geologic conditions and probable fracture distributions to extrapolate hydraulic conditions over larger aquifer volumes.
Uncertainties in conceptualizations need to be acknowledged if no direct evidence from hydraulic testing is obtained.
Transient hydraulic test response, for example, is controlled by the ratio of transmissivity to storativity (i.e., hydraulic diffusivity) and by the connectivity and geometry of the responding fracture network. In porous medium hydrogeology, assumptions are made frequently about storativity to obtain values of transmissivity from the diffusivity. This can be a source of significant error and uncertainty when applied to fractured rock. It is therefore useful to supplement transient hydraulic test information with other data, such as correlations between hydraulic aperture and fracture size, to better constrain both transmissivity and storativity. Hydraulic test interpretations need to move beyond the assumption that the value obtained from a single transient test is representative of all fractures. Interpretations need to include consideration of the spatial and stochastic variability of fracture properties.
In addition to the characterization of fracture and rock matrix hydraulic properties, it is also important to characterize features that behave as flow boundaries, recharge/infiltration boundaries, and discharge boundaries. Conventional hydrogeologic techniques for characterization of fractured aquifers generally can be applied to these boundary features, but their discrete natures still need characterization for accurate modeling. In the petroleum industry, model boundary conditions are based generally on stratigraphic contrasts in permeability—for example, sealing faults or shale layers are defined as no-flow boundaries. This can be problematic in fractured rock, where individual discrete features may connect beyond these faults or low permeability shale layers. The petroleum industry addresses this problem through pressure transient analysis (Streltsova, 1988).
The geometric and hydraulic properties of fractures intersecting a borehole can vary over many orders of magnitude and over short intervals with respect to the length of the borehole. Furthermore, the borehole itself acts as a highly permeable feature in the formation that connects multiple fractures. If the characterization goal is to understand rock transmissivity and groundwater characteristics over a significant thickness of an aquifer (e.g., to determine aquifer capacity to supply groundwater through pumping), then a borehole itself does not necessarily conflict with test objectives because contributions of individual fractures are not being tested.
In contrast, at sites where groundwater is contaminated, or if the goal is to determine the suitability or effectiveness of the geologic environment for waste isolation, characterization of individual fractures and the groundwater flow paths through multiple connected fractures is critical. Because monitoring needs to account for the discrete and potentially tortuous nature of groundwater flow in fractured rock, the variability in the hydraulic head over the length of the borehole, the transmissivity of fractures that intersect the borehole, and variations in groundwater biochemistry in the fractures need to be characterized. Hydraulic communication between multiple fractures in long open intervals in boreholes can yield hydraulic head measurements that are averaged over many hydraulically significant fractures. Additionally, hydraulic communication can also result in cross contamination (see Box 5.2). Methods to isolate individual fractures or sections of boreholes to characterize localized hydraulic and geochemical characteristics are discussed in the next section.
Some regulatory jurisdictions recognize the potential for cross contamination in monitoring boreholes with open intervals and recommend or mandate the open interval lengths to limit such
potential.2,3,4,5,6,7,8 The lower ranges of recommended interval length are within 5 to 10 feet. A fixed monitoring interval, applied most often when monitoring porous media, is intended to eliminate wider spread of contamination through the monitoring boreholes. Monitoring intervals should be selected based on the synthesis of hydrogeologic and biogeochemical information rather than on arbitrarily assigned values for open interval lengths. They need to be chosen to minimize the number of conductive fractures in a single well screen, avoid connecting flowing fractures of differing hydraulic characteristics (heads), and avoid connecting fractures with different geochemical characteristics (e.g., concentrations, geochemistry).
Isolating Discrete Intervals
Sections of a fractured rock system can be isolated by installing multiple boreholes completed at different elevations and open only over short intervals. Alternatively, pneumatic or mechanical packers (see Figure 5.1) or flexible liners (see Box 5.3) can be installed in a single borehole. When packers are used for hydraulic tests, monitoring of the hydraulic response above and below the packers and of the response in the isolated section is needed to identify connectivity between fractures and to detect leaks around the packers. Perturbing hydraulically (e.g., injecting or pumping water) and monitoring the hydraulic response in the packed-off interval allows estimation of the transmissivity within that interval through a simplified interpretation of the groundwater flow regime (quasi-steady, radial flow) near the borehole. More complex flow regimes associated with single-hole tests have also been developed, but the lack of spatially distributed data related to hydraulic responses along heterogeneous fractures result in great uncertainty in these interpretations.
2 Wisconsin Department of Natural Resources. n.d. Groundwater Monitoring Well Requirements, Well Screen. Administrative Code, sec. NR 141.09. Available at https://docs.legis.wisconsin.gov/code/admin_code/nr/100/141/09 (accessed June 19, 2015).
3 Massachusetts Department of Energy and Environmental Affairs. n.d. Standard References for Monitoring Wells. sec. WSC #91-310. Available at http://www.mass.gov/eea/agencies/massdep/cleanup/regulations/wsc91-310-standardrefs-monitoring-well.html (accessed June 19, 2015).
4 U.S. Environmental Protection Agency, Science and Ecosystem Support Division. 2013. Design and Installation of Monitoring Wells. SESDGUID-101-R1. Available at http://www.epa.gov/region4/sesd/fbqstp/Design-andInstallation-of-Monitoring-Wells.pdf (accessed June 19, 2015).
5 New Jersey Department of Environmental Protection, Division of Water Supply, Bureau of Water Systems and Well Permitting. 2007. Well Construction and Maintenance; Sealing of Abandoned Wells. N.J.A.C. 7:9D. Available at http://www.nj.gov/dep/rules/rules/njac7_9d.pdf (accessed June 19, 2015).
6 Florida Department of Environmental Protection, Bureau of Water Facility Regulation. 2008. Monitoring Well Design and Construction Guidance Manual. Available at http://www.dep.state.fl.us/water/groundwater/docs/monitoring-well-manual-formatted-final.pdf (accessed June 19, 2015).
7 Florida Department of Environmental Protection, Bureau of Petroleum Storage Systems, Petroleum Cleanup Program. 2005. Design, Installation, and Placement of Monitoring Wells. Available at http://www.dep.state.fl.us/waste/quick_topics/publications/pss/pcp/MW-SOP-Final-Ap15.pdf (accessed June 19, 2015).
8 California Department of Environmental Protection, Department of Toxic Substances Control. 2014. Well Design and Construction for Monitoring Groundwater at Contaminated Sites. Available at https://www.dtsc.ca.gov/PublicationsForms/upload/Well_Design_Constr_for_Monitoring_GWContam_Sites1.pdf (accessed June 19, 2015).
Although fracture transmissivities are almost certainly spatially heterogeneous, monitoring perturbations over short durations (minutes to tens of minutes) is intended to represent hydraulic properties of fractures over a relatively small rock volume. Perturbations monitored over longer durations are likely to interrogate heterogeneities within individual fractures as well as those associated with the complex connectivity of multiple fractures in the spatially extensive fracture network. Under those conditions, assuming homogeneity is likely unrealistic. The spatial response to hydraulic perturbation associated with heterogeneous hydraulic properties cannot be captured by monitoring at a single location. A relatively new approach to directly measure the magnitudes and directions of cumulative water and contaminant fluxes without perturbations is use of the fractured rock passive flux meter (FRPFM; Hatfield et al., 2004; Annable et al., 2005), described later in this chapter.
More recently, hydraulic characterization of fractures that intersect boreholes is also conducted using commercially available blank flexible liners (see Figure 5.1c and Box 5.3) to eliminate vertical flow in the borehole. Unlike packers, flexible liners can seal the entirety of the borehole semi-permanently but can be removed if needed. Flexible liners are commercially available but, because of cost, not in general use. Box 5.3 describes how the permeability and transmissivity of fractures can be estimated during liner installation.
Characterization of Multiphase Flow Properties
The characteristic curves that describe multiphase flow in fractures and rock matrix are important for contaminant transport and remediation in both shallow and deep environments. Procedures have been developed to characterize fluid saturations, capillary pressure-fluid saturation, and relative permeability-fluid saturation relationship properties for rock matrix (e.g., Jurgawczynski, 2007), but procedures to characterize these in the fractures themselves are still being developed (e.g., Habana, 2002; Zhang and Fredlund, 2003). The petroleum industry is making significant advances in the use of available data to calculate relative characteristic curves for fractured rock multiphase flow properties (e.g., Lian et al., 2012; Han and Zhao, 2015).
Characterizing Flow Paths at Different Scales
The hydraulic properties of complexly connected fracture systems are sometimes inferred through interpretation of single-hole hydraulic tests using methods that assume fractional flow dimensions (i.e., dimensions of groundwater flow between linear, radial, and three-dimensions). Such assumptions are intended to capture qualitatively the heterogeneous character of the flow regime near the borehole. The fractional flow dimension determined at one location, however, does not necessarily represent behavior at other locations.
Longer-duration hydraulic tests in fractured rock are intended to locate permeable fractures and identify their connectivity over dimensions beyond the immediate borehole vicinity. Hydraulic properties have been estimated over tens of meters or greater using classic aquifer testing approaches since the 1950s and are still used widely for petroleum reservoir characterization. These methods involve hydraulic pumping at a single location under quiescent ambient hydraulic conditions and monitoring the hydraulic responses at a number of monitoring wells. This approach was adopted initially because of computational limitations of interpreting tests from closed-form analytical solutions to groundwater flow equations. Homogeneous aquifer properties or highly idealized heterogeneity typically were assumed. Such approaches have been used widely to characterize fractured rock aquifer properties over sufficiently large physical dimensions that effective hydraulic properties are meaningful. Such classic aquifer test interpretations continue to be used because
of their simplicity, but their application should be closely scrutinized when applied to most fractured rock aquifers. When groundwater flow paths need to be defined explicitly to characterize the extent of groundwater contamination, it is inappropriate to rely on a conceptual model with homogeneous aquifer properties or highly idealized interpretations. Additionally, characterizing fractured rock systems at depths of a kilometer or greater offer additional challenges because of their remoteness and because they tend to be subject to larger and more complex stress states that can influence conductive fracture pathways strongly. It is particularly important, then, to characterize the pattern of total and effective stress as spatially varying three-dimensional vectors in these systems (see Box 5.4).
With the advent of numerical algorithms that solve groundwater flow equations in fractured rock systems (e.g., using discrete fracture networks or heterogeneous continua with spatially variable hydraulic properties), hydraulic characterization testing and interpretation can be applied beyond simple and restrictive spatial conceptualizations. For example, at many groundwater sites where contaminated groundwater migration is inhibited through pumping, the spatial connectivity of permeable features needs to be interpreted within the constraints of active pumping because pumping cannot be terminated. However, there may be flexibility and opportunities for gathering information while pumping is under way. Tiedeman et al. (2010) manipulated the pumping rates of a pump-and-treat operation to identify the connectivity of permeable features and estimate their hydraulic properties using available numerical algorithms.
Within a Borehole
Flowmeter logging in bedrock boreholes is a powerful and simple standard practice to quickly interpret the locations of water-producing fractures and to estimate the hydraulic head and transmissivity associated with the most permeable fractures intersecting a borehole. Heat-pulse or electromagnetic flowmeter logging employs a calibrated sensor to measure vertical flow in the borehole under ambient hydraulic conditions and to estimate differences in hydraulic head associated with discrete fractures (usually above and below fractures identified using imaging methods; see Figure 5.2). For example, the heat-pulse flowmeter has been successfully used in fractured rock studies for many years (Paillet, 1998). A packet of hot fluid is introduced into a borehole, and upward or downward movement of the packet is monitored using thermistors9 to estimate the direction and magnitude of flow.
Recently published algorithms and software enables data collected under ambient and pumping conditions to be combined to estimate the hydraulic head and transmissivity of permeable fractures intersecting the borehole (e.g., Day-Lewis et al., 2011). These interpretive methods assume quasi-steady radial flow in fractures intersecting the borehole. Methods to interpret borehole flowmeter logging data collected under hydraulically stressed conditions have also been developed to infer fracture connectivity between adjacent boreholes.
Although flowmeter logging indicates gross vertical groundwater flow in a fractured rock system, better understanding of relative flows within individual fractures, or fractured regions, the degree of interconnection of these regions, and the spacings of these regions could be gained through semi-quantitative methods. However, unless other approaches are used (e.g., active line source temperature logging, described below), flowmeter logging under ambient and stressed conditions should be employed at all sites during the development of the hydrostructural model.
9 A resistor, the resistance of which varies greatly with temperature.
Meters to Hundreds of Meters
Groundwater flow paths over meters to hundreds of meters can be inferred by monitoring responses to hydraulic perturbations in multiple boreholes. The spatial extent over which a perturbation can be monitored will depend on the strength of the perturbation, the magnitude of the hydraulic diffusivity (defined as the ratio of the transmissivity to the storativity10), and the influence of natural groundwater sources or sinks that mask the perturbation. Explicit characterization of groundwater flow paths through hydraulic testing is dependent on the spatial distribution of monitoring wells. However, the cost of borehole installation and completion and, in some cases, the desire to avoid creating additional contaminant migration pathways through the boreholes themselves can discourage additional well installation and, therefore, the ability to identify permeable fracture connections over large volumes of the aquifer.
Controlled hydraulic perturbations may not be viable at the regional scale, and it may be necessary to rely on ambient hydraulic stresses, regional sources and sinks of groundwater, and a sparse number of monitoring locations. Ambient hydraulic stresses, however, may not be suitable to distinguish preferential flow over regional-scale dimensions because fluid pressure responses from ambient hydraulic stresses are dissipated. The most reasonable option may be to infer bulk
10 Storativity is defined as the volume of water released from compressive storage per unit decline in the hydraulic head per unit area of the aquifer.
hydraulic properties of the rock from regional groundwater flow modeling. This may be sufficient for characterizing a regional water balance, but it may not accurately portray the groundwater flow regime or address regional-scale chemical transport.
Seismic and Microseismic Characterization
Seismic methods have been used for more than 40 years to characterize large-scale structures (faults) and stratigraphic contacts. Conventional seismic detection methods rely on detection of seismic waves reflected and refracted from a small number of induced seismic events. Characterization of the geometry of such large-scale features (i.e., hundreds of meters) is essential, particularly at depth where these may provide the most significant flow pathways and flow barriers. Recent developments in three-dimensional seismic processing and microseismic monitoring (e.g., Jones et al., 2014) make it possible to characterize the “sub-seismic” scale fractures both deterministically and statistically. Examples of techniques from the petroleum industry include seismic anisotropic processing to characterize fracture intensity and orientation (Treadgold et al., 2008), ant-tracking, and coherence analysis to detect the location and geometry of sub-seismic fractures (20–200 meter scales) (e.g., Pampanelli et al., 2013). These techniques often allow characterization of sub-seismic fractures to significant depths, even in complex geologic settings, and advances enhance the potential of these methods.
Microseismic techniques utilize changes in fluid pressure to detect fractures and have been used in the petroleum industry to detect fractures at resolutions of meters to depths of thousands of meters. These techniques are based on the concept of “critical stress,” in which seismic energy is released when fluid pressures bring a portion of a fracture to its shear failure criterion (Zoback, 2007). Microseismic techniques are increasingly used in the mining and petroleum industries from both boreholes and shallow surface installations (e.g., Kilpatrick et al., 2010). Seismic emission tomography (SET) integrates energy over time from many millions of microseismic events (Cornette et al., 2012). This makes it possible for SET to resolve discrete features that are orders of magnitude smaller (meters rather than hundreds of meters), even at great depths and in complex geologic settings. SET is used increasingly in the oil industry, both as an active technique with seismic energy provided by conventional seismic thumpers and hydraulic fractures and as a passive seismic method (PSET), relying on integrated microseismic signals in geomaterials due to ongoing stresses and strains such as earth tides and traffic. The underlying technologies of SET and PSET have been shown to detect fractures to resolutions of 2 to 3 meters at depths of more than 2 kilometers (Shemeta et al., 2012).
Joint inversion of hydraulic and seismic data can be used to improve the characterization of both seismic and sub-seismic fractures. With the availability of more powerful computer resources, the environmental and petroleum industries are advancing this field through the development of new inversion algorithms (i.e., four- and five-dimensional inversions) applied to new types of seismic processing, such as for coherence attributes (e.g., Will et al., 2005; Alcolea and Renard, 2010; Landa and Kumar, 2011; Suman, 2013).
Methods to interpret spatially distributed hydraulic properties from multiple hydraulic perturbations—referred to as hydraulic tomography—have been successfully applied in unconsolidated porous media. A variety of algorithms have been used to conceptualize the distribution of hydraulic
properties and infer aquifer heterogeneity between monitoring points. The application of hydraulic tomography to fractured rock has been considered (Illman et al., 2008, 2009; Sharmeen et al., 2012; Berg and Illman, 2013; Illman, 2014); however, the resolution of conductivity and storage tomograms depends on the density of pumping and monitoring locations, the quality of data, and how these data are used to produce (usually smooth) maps of conductivity in the inverse problem. To improve the resolution of tomography and therefore the acceptance of these tools, new devices are required that allow higher-density pressure response monitoring at discrete borehole intervals. Additional field trials are needed, particularly those that integrate data from single-hole tests, borehole flow meter profiling, and tracer tests with tomographic results.
Greater Understanding of Flow Paths
Best practices for characterizing groundwater flow paths are described in Box 5.5. Current practice, however, is limited to describing only a small percentage of fractures with the highest transmissivities. Determining the hydraulic significance of less transmissive fracture networks and the rock matrix requires hydraulic monitoring in fractures over a wider transmissivity range and longer duration than currently possible. Groundwater flow likely occurs over a much wider range of transmissivities than hydraulic testing can identify. Expanding hydraulic characterization capacity would clarify the potential for fluid exchange between the most permeable and less transmissive fractures. Over durations of days and months, such groundwater exchange may not be significant. Chemical exchange over decades or centuries, however, may be important.
Geochemical measurements of tritium, radiocarbon, or chlorofluorocarbons that provide information on water age have been used to estimate natural flow paths (e.g., Szabo et al., 1996). Unlike artificial tracer tests, these do not require perturbation of the system. Such analyses have been useful to measure water age and estimate flow paths in porous aquifers (e.g., Szabo et al., 1996; Price et al., 2003). Tritium has been used with great success to define the depth of active groundwater flow in fractured clayey tills (Ruland et al., 1991). Other isotopes have been used in crystalline rocks (e.g., Tweed et al., 2005; Singhal and Gupta, 2010).
State-of-the-art characterization in fractured environments has moved beyond direct observation and imaging of fracture locations and orientations to time-lapse imaging of geomechanical, hydrologic, and biogeochemical processes, leading to a better understanding of heterogeneity, contaminant fluxes, and the contaminant locations. Geophysical methods may allow remote characterization of fractured rock systems at distances from boreholes. This, in turn, may inform more effective and efficient remediation. Geophysical methods, therefore, are worth additional research investment. The methods described in the following sections, beyond those more commonly used and described in NRC (1996), are of particular interest. Many other methods, including seismic, are outlined in detail in that text.
Distributed Temperature Sensing
Temperature data have long been used in hydrologic studies to estimate flow velocities and identify discharge zones. Temperature data are often generated by discrete measurements, either from boreholes or, for example, from stream systems in which baseflow contributions into a stream are controlled by fracture flow. Fiber-optic distributed temperature sensing methods have been used recently and more commonly to monitor temperature in diverse settings. These instruments show tremendous potential to improve practice given their resolution—they are capable of providing temperature measurements over kilometer-scale reaches, with resolution of 1 meter, 1 minute, and 0.1°C (see Box 5.6). Fiber-optic cable is installed in boreholes, streams, or in trenches and attached to a control unit. Although these instruments are more expensive than standard thermistors or thermocouples, they provide spatially and temporally exhaustive data.
Active line source (ALS) logging is another temperature approach to assist characterization via boreholes. In the ALS approach, a borehole is placed into thermal disequilibrium using a heating cable. Temperatures are logged during both heating and cooling using a chain of thermocouples in the borehole. With two or more logs collected during heating or cooling, an estimate of thermal conductivity is obtained. In the absence of groundwater flow in or around the borehole, variations in the thermal conductivity of the rock are due largely to variable water content, and the ALS log provides a reasonable surrogate for a neutron porosity log (Pehme et al., 2007). When groundwater flow dominates the dissipation of thermal energy, the apparent thermal conductivity is increased. In open boreholes this flow can be both ambient (within the formation itself) and connecting (vertical flow between fractures intersected by the borehole). ALS logs are particularly useful to detect ambient groundwater flow in lined holes with no connecting flow. Alternative methods for flow detection, such as chemical dilution or flow meters, require an open borehole and either have poor vertical resolution or require multiple stationary measurements, often with packers to minimize the effects of connecting flow. The ALS technique is a comparatively simple tool—useful in both open and cased or lined boreholes—and run continuously down the length of the borehole. Fracture resolution on the order of a few centimeters.
Nuclear Magnetic Resonance
Nuclear magnetic resonance (NMR) measurements are commonly used in borehole logging to estimate pore size and porosity distributions in porous media. The past decade has brought improvements in sensitivity and quality of measurements that have allowed more routine use in both down-bore tools and core analyzers. These tools likely are suitable for use in fractured rock settings. NMR has been applied at the lab scale to determine fracture aperture (Renshaw et al., 2000) and how fracture wall morphology affects channeling (Brown et al., 1998; Dijk and Berkowitz, 1999; Dijk et al., 1999). Extension to field applications has yet to be accomplished rigorously.
NMR creates a static magnetic field in the borehole that polarizes hydrogen nuclei in water and some contaminants. The magnitude and decay of the polarization of hydrogen nuclei (precession) reveals information about stored fluid volume and type (e.g., gas, oil, or water) and helps differentiate between free and bound water (i.e., attached to rock surface). With careful calibration, NMR may be able to provide information on porosity and permeability (e.g., Legchenko et al., 2002).
The use of NMR is complicated in fractured rock by signal-to-noise ratio. Because the water content is small in fractured rock, the NMR signal is small. The most important control on resolution is the loop size; with very large loops that might be sensitive to up to 100 meters depth, vertical resolution could be expected on the order of meters. The smallest possible loop is probably 25 meters, which might provide sub-meter resolution (Personal communication, A. Parsekian, University of Wyoming, March 29, 2015). In general, the minimum resolvable water content would be on the order of 3 percent, given favorable noise conditions.
Analysis by NMR also has a strong potential to improve our understanding of the behavior of groundwater contaminants in fractured rock. Key to this is the ability of NMR to image the internal structure of core samples such that important features may be identified and mapped. Core analyzers perform a more advanced suite of experiments than well loggers, including capillary pressure analysis, wettability studies, paramagnetic tracer studies, and full three-dimensional magnetic resonance imaging of the interior of cores. The behavior of fluids at these structures may then be
observed directly using tracers and other NMR techniques, including the characterization of capillary pressure and preferential pathways for fluid migration. Such studies may improve the models available for the characterization of groundwater behavior at fractures and other lithological interfaces. Further development and testing will be required to determine whether these methods have significant value for testing fractured rock systems and to transition from ex situ, core-based NMR to downhole NMR approaches.
Electrical and Electromagnetic Imaging at the Tens of Meters Scale
Electrical and electromagnetic geophysical methods respond to contrasts in electrical properties across interfaces and have been used to image fractured environments in numerous studies. For example, cross-well, ground-penetrating radar (GPR) tomography (e.g., Olsson et al., 1992; Grégoire and Halleux, 2002), single-well borehole reflection-mode GPR (e.g., Lane et al., 1998), combinations of both reflection and tomography surveys (e.g., Seol et al., 2004), and surface reflection-mode GPR (e.g., Grasmueck, 1996; Tsoflias et al., 2004) have been used to detect fractures. Electrical resistivity (ER) methods have also been used successfully in fractured rock environments to, for example, investigate the dominant fracture-strike direction (e.g., Lane et al., 1995). Resolution of these methods depends on the type and setup of the instrumentation and the geophysical characteristics of the subsurface, among other characteristics; consequently, determining resolution in advance can be difficult.
More recently, these tools have been used to image preferential fluid pathways such as fractures by directly monitoring the subsurface migration of electrically conductive solutes such as sodium chloride tracers. In these cases, changes in electrical conductivity are associated with the movement of a tracer and, when differenced from background values, provide a method to map tracer movement relative to geologic variability and systematic noise. For example, time-lapse, surface-based GPR tomography has been used in a number of studies in fractured rock (e.g., Niva et al., 1988; Talley et al., 2005; Tsoflias and Becker, 2008; Becker and Tsoflias, 2010), although the depth of imaging in these studies is usually limited to a few meters. Day-Lewis et al. (2003, 2004) and Lane et al. (2000) imaged tracer migration at the Mirror Lake, New Hampshire site, although the data in these studies were inverted using a continuum representation of the fracture zones. Lane et al. (1996) and Dorn et al. (2011) have imaged tracer transport using single-hole GPR data. Regardless of whether the authors consider single- or cross-hole measurements, the use of saline tracers coupled with geophysical tools can be useful for meter-scale characterization of transport in fractured rock formations.
ER has been used to monitor transport in fractured rock (Slater et al., 1997a; Nimmer et al., 2007), although less frequently, likely because of the poor resolution of ER when compared to methods such as GPR. Imaging fracture paths in unsaturated materials has also been explored (Slater et al., 1996, 1997b; Zaidman et al., 1999). Recently, new methods have led to improved imaging of fracture features based on alternative mathematics for reconstructing geophysical images that allow for sharp resistivity contrasts at fracture locations (Robinson et al., 2013a,b). These methods show considerable promise for better characterization of fractured rock systems, but they require some knowledge of fracture location from boreholes.
Constitutive Relationships Applied to Field Data
Geophysical data provide indirect measures of rock hydrologic properties. ER, for example, measures electrical conductivity, a material’s ability to transmit current, and GPR is largely a measure of dielectric permittivity, a material’s ability to store charge. Relationships between geophysical and hydraulic parameters are commonly developed based on the regression of field data (Kelly, 1977; Klimentos and McCann, 1990), empirical studies performed on lab samples (e.g., Topp et al., 1980), and theoretical considerations (e.g., Bruggeman, 1935; Hashin and Shtrikman, 1962; Moysey and Knight, 2004). The derived relationships are assumed to depend on the local properties of the medium and are independent of spatial location. Recent work has shown that a single constitutive relationship may not capture the complexity of a field site given that aquifers are heterogeneous at multiple scales and that the resolution of geophysical images varies spatially (Day-Lewis and Lane, 2004; Moysey and Knight, 2004; Day-Lewis et al., 2005; Moysey et al., 2005; Singha and Gorelick, 2006). Day-Lewis et al. (2005) indicate that geophysical imaging provides “apparent” constitutive relations specific to the geometry, errors, and physics of the survey. Those authors conclude that spatially variable geophysical resolution must be considered to estimate state variables accurately from tomographic images. This issue is rarely considered in practice and may be especially complicated in fractured rock environments.
Geophysical data are not sufficient to quantify hydrologic properties (e.g., transmissivity, tracer concentration and mass) without complementary data to “calibrate” the images produced by tomography (Binley et al., 2002; Yeh and Simunek, 2002; Singha and Gorelick, 2005) or considering joint inversion methods (Kowalsky et al., 2005; Ramirez et al., 2005; Linde et al., 2006; Hinnell et al., 2010). The difference in resolution between hydrologic and geophysical measurements is one of the foremost challenges facing the application of these tools to fractured rock systems. Additionally, characterization and modeling of fluid saturation, capillary pressure-fluid saturation, and relative permeability-fluid saturation relationships are important for geological repository safety assessments in both crystalline and sedimentary rocks. Despite these challenges geophysical methods can be the most effective means to monitor changes in fractured rock systems, if it is understood that they may not provide exact property estimates. A scientifically rigorous and realistic explanation of uncertainty and limitations of parameters derived from geophysical testing is a necessary part of documentation.
Joint inversion, or consideration of multiple types of data together to build one consistent model, has led to significant advances in imaging (e.g., Vozoff and Jupp, 1975; Lines et al., 1988; Gallardo and Meju, 2003; Kowalsky et al., 2005). Multiple data sets usually are inverted simultaneously using soft mutual constraints between the data types. Any single technique has significant limitations that may be countered by the different sensitivities of another method. Consequently, integrating multiple data sets may offer the most promise for understanding complicated fractured rock systems.
Although still underutilized for fractured rock systems, joint inversion methods have been used recently, for example, to combine seismic and flow meter data to zone hydrologic models in fractured rock (Chen et al., 2006), ground-penetrating radar data with hydrologic and thermal data to understand coupled processes in a fractured geologic repository for nuclear waste (Kowalsky et al., 2008), and seismic and electrical data to map fracture locations and flow paths (Heincke et al.,
2010). Recent work has also explored the use of flux and head data together in synthetic fracture systems to map fracture distributions and velocity fields in a system with known behavior (Zha et al., 2014).
Groundwater samples used in the characterization of fractured rock sites are most often collected from boreholes that intersect fractures. Chemical distribution in fractured rock aquifers is often controlled by fracture orientation, transmissivity, spacing, connectivity, and characteristic dimensions of the void space within fractures as well as the void space in the matrix surrounding the fractures. Mobile dissolved contaminants are most often identified with highly transmissive fractures, and their distribution and migration are controlled primarily by the hydraulic gradient and the transmissivity of intersecting fractures. In cases where density is significantly different from background groundwater (high concentrations of dissolved constituents, non-aqueous phase liquids), fracture orientation and connectivity can directly affect migration direction irrespective of hydraulic gradient. When non-aqueous phase liquids (NAPLs) are present in the vicinity of a borehole used for sampling, high contaminant concentrations may be present in lower transmissivity fractures that have been invaded by NAPLs. The presence of the NAPL will reduce the transmissivity until it dissolves, or until NAPL present in higher transmissivity fractures has been depleted through dissolution or diffusion into the surrounding rock matrix. For a review on flow and transport in fractured media, see Bear et al. (2012).
Spatial Distribution of Contaminants in Fractured Rock
Groundwater sampling of individual or multiple fractures provides only a part of the information needed to understand contamination distribution in a fractured rock setting. For example, investigations at the Naval Air Warfare Center research site in New Jersey indicate that less than 1 percent of the total trichloroethylene mass in the subsurface is mobile in groundwater in fractures (see Box 5.7 for a case study). Where contaminants have been in place for decades, a significant amount of contaminant transfer from mobile fluids in rock fractures to the rock matrix can occur. Contaminant concentrations in the rock matrix cannot be ascertained by sampling only mobile fluids in fractures unless the temporal history of concentrations in the fractures is known, and an accurate hydrostructural model at the single-fracture scale has been developed. Concentration isocontours generated from analyzing mobile fluids often cannot be used to quantify contaminants in the subsurface to evaluate, for example, remediation strategies.
Contaminants can be stored in fractured rock matrix and distributed between the rock matrix pore fluids and pore wall rock surface; both can be potential long-term sources of subsequent groundwater contamination (e.g., Grisak and Pickens, 1980). Large concentration gradients between water in fractures and water in rock matrix promote high rates of contaminant exchange between fractures and the matrix. Although less-permeable fractures may not contribute significantly to the volume of groundwater flow, contaminants will migrate from them to more permeable groundwater flow paths as a result of diffusion and slow advection. Similarly, dissolved contaminants diffuse between water in fractures and the rock matrix.
Direct chemical analyses of fractured rock can be critical when assessing the significance of rock surface chemical processes, for example, the organic carbon content of rock can indicate the rock capacity to sorb organic contaminants (e.g., Allen-King et al., 1996). Rock coring, however,
is expensive. Relatively few cores are drilled and analyzed, and cores may not provide an accurate sampling of heterogeneities in rock matrix properties or contaminant distribution. It may be decided that rock matrix characterization of this sort is necessary for a site only if there is compelling reason to understand the partitioning between pore water and solid phases. If such a reason exists, then the porosity and organic carbon content of core samples need to be analyzed to differentiate the organic compounds in aqueous and sorbed phases and to develop reasonable rate estimates of sorption and desorption (Sterling, 1999; Sterling et al., 2005; Kennel, 2008). If, however, the additional time to remediate sorbed phase contaminants is inconsequential given the slow rates of diffusion and the resulting decades-to-centuries remediation time frames, then rock sampling may be considered unnecessary.
In the absence of contaminant distribution information, the efficiency and cost-effectiveness of contaminant remediation designs for rock matrix cannot be adequately evaluated. New methods to estimate contaminant mass in the rock matrix using existing boreholes and in situ diffusion experiments are needed. Some geophysical applications may be appropriate for quantifying volumes of less-mobile porosities and rates of exchange (Singha et al., 2007; Day-Lewis and Singha, 2008; Swanson et al., 2012), but additional research in such methods is also needed.
Characterizing Groundwater Chemistry in Fractures
When obtaining water samples from boreholes in fractured rock for chemical analyses, the sample may be dominated by water from fractures that have a limited range in transmissivities (biased to highly transmissive fractures) and readily allow sample collection if the sample interval extends to long sections of the borehole (e.g., Shapiro, 2002). However, fractures not readily sampled (lower transmissivity) also may be contaminated.
Pumping from long open intervals of a borehole results in water being drawn preferentially from the most permeable fractures and yields flux average concentrations. Historical sampling practices have required sufficient pumping to purge the water in the borehole creating an averaged sample across the open interval, with the result that the sample obtained following purging may not represent conditions in any given fracture. On an individual fracture scale, fractures with transmissivity greater than 1e-7 m2/s will provide sufficient recharge to the borehole to maintain minimal drawdown, but those with smaller transmissivities generally cause reductions in hydraulic head and exceptionally long water sample retrieval times. In a borehole with a long open interval and a mixture of fracture transmissivities, the dominance of the higher transmissivity fractures essentially masks the contributions from the lower transmissivity fractures, which may be the most contaminated (e.g., Johnson et al., 2002). In addition, because many fractured rock aquifers have small fracture porosities, extended pumping to remove standing water in the borehole may draw water from a large volume of the rock through interconnected fractures and therefore the resultant geochemistry may not be representative of ambient conditions in the fractures intersecting the borehole.
As discussed earlier, packer or flexible liner systems can be used to isolate water samples from specific individual or closely spaced fractures, reducing the averaging effect of borehole-scale pumping. Flexible liners can also be used in conjunction with built-in multilevel samplers (e.g., Cherry et al., 2007) that allow the liners to be used to sample water chemistry in individual fractures or in small volumes of interconnected fractures in close proximity to the borehole.
Fractures containing NAPL can also be identified using a flexible liner system formulated to indicate the presence of NAPLs (Griffin and Watson, 2002; Cho et al., 2008), avoiding the need to identify such fractures through secondary lines of evidence (Kueper and Davies, 2009). The liner fabric reacts with the NAPL to produce a stain where it contacts pure product (described in more detail in next section). The use of such liners is also recommended for the “sealing” phase of single borehole investigations where NAPL is suspected to be present.
Several alternative sampling methods have been proposed that do not rely on individual fracture isolation. Sampling from open boreholes by pumping at low rates (low volume; e.g., Puls and Barcelona, 1996), sampling at a given borehole elevation without pumping (no volume; e.g., Savoie and LeBlanc, 2012), and sampling over longer time frames using diffusion techniques (i.e., diffusion-bag approaches—semipermeable bags used to passively collect groundwater samples; Vroblesky and Hyde, 1997) are examples. In some instances, these methods yield similar results to those obtained through discrete fracture isolation; however, discrete fracture isolation is a more robust approach that does not require secondary assessment to confirm that averaging is not occurring to a significant degree. In any of these methods, confirming that geochemical results represent conditions in individual fractures requires that hydraulic conditions in the borehole are understood. If hydraulic conditions give rise to vertical flow through the borehole, for example, then low-volume, no-volume, and diffusion-bag sampling will not be representative of water in an individual fracture, and correlating results to discrete depths is not possible.
An alternate approach without the potential limitations of pumping or diffusion-bag approaches is the FRPFM. The FRPFM is a relatively new technology that measures directly the magnitudes and directions of cumulative water and contaminant fluxes in fractured rock aquifers (e.g., Hatfield et al., 2004; Annable et al., 2005) as well as contaminant fluxes. The design functions under closed-hole conditions across a defined vertical interval in fractured rock wells and is also easily installed in deep rock wells or deep screened wells. The FRPFM directly measures (1) the location of active or flowing fractures; (2) active fracture orientation; (3) direction of groundwater flow in each fracture plane using tracers; (4) cumulative magnitude of groundwater flux in each fracture plane; and (5) cumulative magnitude of contaminant flux in each fracture plane. See Box 5.8 for a description of how the FRPFM functions.
The FRPFM for field applications is currently constructed to interrogate a 1-meter interval of the rock borehole. Exposing the FRPFM to flowing groundwater for a selected duration gradually leaches the tracer from the sorbent layer, and the dye from the cloth producing a residual distribution of tracer. Visual inspection of the FRPFM sorbent leads to estimates of location, number, strike and dip, and groundwater flow directions of active or flowing fractures. Further analysis of the sorbent for tracer loss and contaminant accumulation at active fractures will yield the cumulative magnitudes of groundwater and contaminant flux in fractures. These meters have been developed recently for fractured rock flux measurements and applied to a synthetic fractured rock system (Acar et al., 2013).
Characterizing Chemical Transport Processes in Fractured Rock
The spatial distribution of contaminants within a fracture system or at a single borehole can be an indicator of groundwater velocity and dilution only in the rare cases that the original contaminant concentrations and time of subsurface introduction are known with reasonable certainty. Despite this potential limitation, monitoring changes in contaminant spatial distribution may yield qualitative information about groundwater velocity if the processes that attenuate contaminant migration can be accounted for. Groundwater contaminant residence or travel times, dilution, or mixing that occurs during contaminant migration cannot be inferred from single and cross-hole hydraulic tests. Those processes are affected by physical properties of the void space that control groundwater flow, as well as chemical processes such as diffusion and sorption that affect the concentration, spatial distribution, and magnitude of chemical fluxes associated with contaminants in the groundwater flow regime. Characterizing the processes is important in evaluating the fate and transport of groundwater contaminants.
Controlled chemical tracer methods not only are proving successful for characterizing some hydraulic properties and for flow path identification, but also have been successful in the characterization and estimation of chemical transport processes and properties. An extensive outline of the use of tracer tests in fractured rock systems was provided in NRC (1996). In these tests, a tracer is introduced into the groundwater, and arrivals at groundwater withdrawal locations are monitored using multi-level samplers or wells. Given the sparse number of boreholes and the complexity of fracture architecture, tracer experiments should be conducted under hydraulically stressed conditions to ensure that tracers do not simply bypass monitoring location.11 Under hydraulically stressed conditions, tracer tests can be conducted in single boreholes (e.g., push-pull tests; Haggerty et al., 1998) or between multiple boreholes in close proximity. The results from tests conducted under stress, however, may not be representative of what would be observed in ambient flow conditions. Applying both approaches is preferable but rarely done because of time and cost constraints.
In general, tracer tests conducted under controlled conditions can provide data sets that allow advection and dispersion in fractured rock systems to be estimated. However, large variability in fracture transmissivity along even localized flow paths will lead to tails in the response that are often improperly characterized as controlled by diffusion. Tracer tests are best conducted using multiple tracers in concert to provide resolution of the slow advective and true diffusive effects (e.g., Becker and Shapiro, 2000). At sites where NAPL contamination is thought to exist, partitioning tracers can be employed (e.g., Annable et al., 1998), whereby the tracer partitions into and
out of the non-aqueous phase, resulting in a response at the monitoring point that can be resolved into evidence of the existence of NAPL along the tracer flow path.
In Situ Characterization and Monitoring
Direct, in situ characterization and monitoring of the subsurface is an ideal shared by many sectors interested in subsurface engineering including the oil and gas industries, the geothermal energy production industry, and the environmental management industry. Tracers that provide information about the environment through which they travel may be a promising source of such in situ information for fractured rock. To date, the use of such “smart” tracers for environmental applications has been largely associated with the use of resazurin in surface and ground waters (e.g., Haggerty et al., 2008). Resazurin is a redox indicator to estimate biological activity and has seen groundwater applications such as for characterization for and monitoring of bioremediation (e.g., Guerin et al., 2001).
Various smart tracers and nanosensors are also being used or developed to help in the physical and chemical characterization of petroleum reservoirs (e.g., Anifowose et al., 2013), to monitor geologic storage of carbon dioxide in saline aquifers (e.g., Würdermann et al., 2010), and to characterize in situ stress thermal properties of engineered geothermal systems (e.g., Alaskar et al., 2010, 2011). Use of such technologies in situ, however, will be dependent on factors such as the ability to emplace, protect, locate the sensors, and, in the case of nanotechnology, miniaturize and power the sensors, store data, and retrieve the sensors or transmit data (Matteo et al., 2012). More interdisciplinary research is necessary before these technologies see practical use in fractured rock.
Microorganisms inhabit all niches of the subsurface environment, including fractured rocks. The activities of subsurface microbial communities can exert significant effects on both physical and geochemical characteristics and may be responsible for a variety of dynamic processes including mineral formation and dissolution, as well as changes in redox chemistry, fluid surface tension, and pH. Pedersen (1997), Geller et al. (2000), Edwards et al. (2005), and Stoner et al. (2005) used lithographic physical models to demonstrate the significant impacts that subsurface microorganisms can exert on fluid flow in fractures by means of biofilm formation and mineral precipitation.
Methods to characterize microbial communities and activities in fractured rock are similar to those used for porous media, so the discussion here will focus on the most commonly applied techniques with specific examples focused on fractured rock matrices. It is important, however, to note specific challenges associated with proper sampling of microbial communities in fractured rock environments, especially in deep formations (Pedersen, 1997). Drilling into these formations to acquire physical samples can be extremely expensive, and special care is needed to avoid potential contamination of the sampled rock and groundwater from drilling fluids and cuttings. Potential contamination can be overcome by strict adherence to careful aseptic sampling and processing protocols as well as the use of microspheres and tracers to detect drilling fluid intrusion (Griffin et al., 1997; Lehmen et al., 2001). Furthermore, to appropriately analyze microbial activities in environmental samples, they must be maintained under conditions similar to those of their environments, which may include elevated pressures, temperatures, and lack of exposure to oxygen (Geller et al., 2000; Lehmen et al., 2001; Edwards et al., 2005; Purkamo et al., 2013).
Microbial characterization of subsurface matrices is most commonly performed using water pumped from the formation; however, it is important to recognize that many subsurface microorganisms occur as biofilms attached to the solid matrix and therefore may be underrepresented in groundwater samples. Direct characterization of collected solids or down-borehole microcosm methods incorporating beads or other solids (e.g., Bio-Traps) are useful for characterizing the attached microorganisms (e.g., Geller et al., 2000; Lehman et al., 2001; Chang et al., 2005).
Measurements of microbial activity in laboratory experiments generally rely on preserving as many of the environmental conditions as possible, and while they often involve static microcosms, efforts are sometimes made to mimic natural fluid flow, given the importance of this characteristic to microbial activities (Stoner et al., 2005). For example, Geller et al. (2000) applied flow-through “geocosms” created from rock sample cores to study microbial volatile organic degradation in fractured basalt and showed that active bacteria significantly impacted fluid surface tensions and flows. Masciopinto (2007) employed a flow-through system constructed of limestone slabs to evaluate microbial transformation of wastewater nitrogen in fractured formations. Silver and colleagues (2010) constructed flow-through traps emplaced directly within deep boreholes in a mafic sill gold mine to evaluate microbial communities there.
Direct measurements in the field are often used to assess subsurface microbial activities in fractured rock. For example, measurements of oxygen, nitrogen forms, ferric/ferrous iron, sulfur forms, organics, and contaminants can all be useful in assessing microbial activity in fractured rock. In addition, stable isotope ratios of carbon, nitrogen, or sulfur species can provide useful information on microbial activities given that biological processes result in dynamic isotopic shifts in reactants and products. For example, tracking of δ-13C and δ-34S isotopes was used to indicate sulfate-reducing activities within deep granite formations below Aspo Island in Sweden as part of a site characterization for the Swedish nuclear waste disposal program (Pedersen, 1997), and compound-specific carbon isotope signatures were tracked to demonstrate active in situ bioremediation in a variety of fractured rock settings (e.g., Song et al., 2002; Chartrand et al., 2005; Lojkasek-Lima et al., 2012).
Estimates of microbial numbers can be determined for solids or groundwater samples by traditional culturing or molecular techniques; however, it is important to recognize the substantial undercounts associated with culturing techniques of environmental communities (Amann et al., 1995), especially in oligotrophic (nutrient-poor) environments such as those expected within most fractured-rock matrices (Pedersen, 1997). The number of microorganisms detected in groundwater within deep granitic fractured rock environments ranged from 103 to 107 cells per mL in several studies (Pedersen, 1996; Lehmen et al., 2001), and only a small fraction of these cells were culturable in the lab.
Non-culturing techniques include both non-specific staining methods that target all cells (such as fluorescent DNA stains) and specific staining methods that target individual microbial groups or functions (such as fluorescence in situ hybridization [FISH]). Molecular methods can be designed to be either specific or non-specific such as quantitative polymerase chain reactions (qPCR) that target either universal bacterial or archaeal DNA, or that focus on one specific species, strain, or even functional gene.
The rapid expansion in the past two decades of molecular techniques available to study microbial communities without the need for laboratory culturing has greatly expanded understanding of subsurface microbial communities and the importance they have influencing flow, fate, and transport within fractured rock. For example, Pedersen (1997) highlighted the important role that
hydrogen-consuming chemoautotrophs (microorganisms that use hydrogen for energy and inorganic carbon for biosynthesis) likely played in the early evolution of life on Earth and the current role shaping biogeochemical conditions in deep fractured rock.
Molecular techniques involve the analysis of cell-associated DNA/RNA/proteins and metabolites. The most common molecular techniques associated with the study of subsurface environmental microbial communities begin with the extraction of DNA followed by polymerase chain reaction (PCR) to generate sufficient copies of the DNA to be identified and sequenced. Commonly, sequences of genes associated with all forms of life, such as those encoding the ribosomes responsible for protein generation in cells, are used to identify the wide variety of bacteria and archaea present in fractured rock environments. Widely diverse heterotrophic (cells requiring organic carbon/energy sources) and chemolithotrophic (cells using inorganic energy sources) bacteria, for example, were unexpectedly detected in groundwater samples from boreholes in deep granitic aquifers in Sweden (Pedersen, 1997). Molecular techniques were also used to demonstrate the difference in matrix-attached and groundwater-associated microbial populations in an acidic crystalline rock aquifer at Mineral Park Mine in Arizona (Lehman et al., 2001). Purkamo et al. (2013) employed molecular techniques combined with a packer sampling method as a mechanism to evaluate indigenous microbial communities in water drawn from geochemically distinct bedrock fractures in a deep aquifer in Eastern Finland, demonstrating that community structure and geochemistry are strongly linked. Additionally, Lima et al. (2012) used molecular tools to demonstrate that microorganisms present in the rock matrices of a fractured sandstone-dolostone were actively participating in significant dechlorination of halogenated organics.
Recently, a variety of high-throughput molecular techniques have become available to characterize the microbial ecology of complex microbial communities associated with fractured rock. They include nucleic acid sequencing techniques, mass spectrometry-based proteomic and metabolomic approaches, and phospholipid fatty acid analysis (Hughes et al., 2000; Venter et al., 2004; Vieites et al., 2009; Bartram et al., 2011; Loman et al., 2012).
The most common high-throughput sequencing techniques include target gene sequencing based on phylogenetic or functional targets and shotgun metagenome sequencing. Both of these techniques investigate the gene content and genetic diversity of microbial communities, but not the functional activity of the cells. For that, metatranscriptomic sequencing of the expressed genes can be performed (Sorek and Cossart, 2010; Moran et al., 2013).
Although recent advances in high-throughput molecular techniques have greatly accelerated the rate of new knowledge development in subsurface microbiology, few comprehensive applications to understand microbial community structure and, more importantly, function in fractured rock environments have been performed. Given the importance of subsurface microbial communities in the geochemical characteristics of fractured rocks, and in the potential for bioremediation of contaminants moving through them, more research in this area is essential and would represent an important investment.