OR/18/012 Intrinsic vulnerability
|Loveless, S, Lewis, M A, Bloomfield, J P, Terrington, R, Stuart, M E, and Ward, R S. 2018. 3D groundwater vulnerability. British Geological Survey Internal Report, OR/18/012.|
This section describes potential pathways for contamination from the source (hydrocarbon source unit) to receptors (groundwater body). The general intrinsic vulnerability factors and methodology relating to each of the pathways are presented after each pathway has been described. Pathways relating to specific sub-surface hydrocarbon activities are detailed in Specific vulnerability.
The intrinsic vulnerability of a receptor is a function of the geological setting (geometrical relationships and hydrogeological properties) within which the receptor and the proposed hydrocarbon source occur.
Intrinsic vulnerability is assessed for each potential receptor in the geological sequence identified in the conceptual model, below the surface within the AOI, according to key factors, sub-factors and parameters that will influence the vulnerability.
For each factor, the full range of possible measurements or values is divided into between three and eight increments depending on the parameter. Each is given a rating value, i.e. 1, 2, 3 etc, a weighting (numeric, pre-ascribed reflecting its contribution to intrinsic vulnerability) and a confidence level (high, medium, low).
The rating (r) and weighting (w) for each factor are multiplied, and the scores for all the factors are summed to produce the receptor’s intrinsic vulnerability score. A higher rating indicates higher intrinsic vulnerability. A confidence level is ascribed to each of the intrinsic vulnerability scores based on the lowest confidence of all the assessed subfactors. The rating is determined from the conceptual model (Development of a geological conceptual model).
Separation of the hydrocarbon source rock and potential receptor
Rocks in the intervening zone between the source and receptor can facilitate or hinder the transport of contaminants. The further apart the source and receptor are, and the increased time taken for contamination to travel from hydrocarbon source to receptor, the lower the likelihood of contamination reaching the receptor, (e.g. US EPA, 2016). Longer pathways may also allow longer exposure to microbial degradation and attenuation, and prevent contamination reaching the receptor. Certain properties of the rock mass (porosity, permeability, attenuation capacity) will make the transport of contaminants more or less likely. This pathway is identified as ‘R’ in Figure 5.4, Figure 5.6 and Figure 5.8.
The groundwater flow in sedimentary formations at depth is up to two orders of magnitude greater in the horizontal than in the vertical direction, and simulations have shown that the majority of flow following hydraulic fracturing is in the horizontal direction (Brownlow et al., 2016). This is due to the permeability anisotropy resulting from sedimentary layering. The movement of contaminants is controlled by the lowest permeability layer.
In the 3DGWV methodology, proximity is divided into two subfactors; vertical and lateral separation distances. It reflects the greater likelihood of contamination through the rock mass and preferential pathways when the hydrocarbon source unit and the potential receptor are closer. The spatial extent of permeability changes resulting from extraction processes, in particular, hydraulic fracturing are also considered (Effect of hydraulic fractures). Different separation distances have been used in other industries, such as mining (Mines) and could be used to modify distances in different locations and for specific industries if this methodology is extended to other sub-surface activities.
There are more categories for the intrinsic vulnerability parameter range for the vertical than for the lateral separation because better estimates of vertical separation can be made using the 3DGWV LFV model, or borehole logs. Since groundwater flow at depth is generally greater in the horizontal direction than in the vertical direction the lateral separation has a greater weighting so that a given distance scores higher (or the same) for the lateral than vertical direction (Table 4.1). There are also fewer categories for the lateral than vertical scores because of the lower resolution.
|Separation distance (m)||Vertical score||Lateral score|
Effect of hydraulic fractures
Induced hydraulic fractures are fractures thought to be micrometres (µm) in width (Younger, 2016), which are created to release gas from shale or other tight rock formations. Due to the orientation of stresses in the sub-surface, hydraulic fractures at depths >1200 m are predominantly vertical and at depths <600 m are predominantly horizontal, with a mixture of vertical and horizontal fractures in the interval between (Fisher and Warpinski, 2012).
Hydraulic fractures could potentially provide preferential pathways for contaminants from source to receptors depending on the height and aperture of the fractures and the vertical separation distance between the hydrocarbon source unit and the receptor. Even if the fractures do not directly link the source and receptor, they can shorten the pathway that a contaminant would have to travel without a preferential flow path (modified separation).
Local rock failure, which occurs as the hydraulic fractures form, creates microseismic events which can provide information on in-situ rock deformation. While geophysical data can be used to image fracture height in the subsurface (Fisher and Warpinski, 2012), data remains relatively limited since only 3% of hydraulic fracturing operations in North America are currently monitored with seismic arrays (Gassiat et al., 2013). Nevertheless, studies assessing induced fracture height from micro-seismic and micro-deformation data for high volume hydraulic fracturing indicate that most hydraulic fractures are less than 100 m in height (Davies et al., 2012; Fisher and Warpinski, 2012). Statistically, less than 1% of hydraulic fracturing stages have fractures that are greater than 350 m in height (Davies et al., 2012). On average, there are seven hydraulic fracturing stages per borehole, thus about one in fourteen boreholes could have a maximum fracture height exceeding 350 m. Monaghan (2014) used a similar cut-off of 305 m (1000 ft) for vertical separation between shale gas activities and coal mines in the Midland Valley based on communications with an experienced US shale gas company. The maximum upward propagation of recorded fractures in the data from five shale gas plays in the US, analysed by Davies et al. (2012), is 588 m in height. This work also concluded that fracture height probabilities are likely to be over-estimated due to difficulties identifying smaller fractures. In addition, Fisher and Warpinski (2012) show that fracture height distributions differ between regions and shale formations and there is currently no information on possible hydraulic fracture heights for England. Hydraulic fractures from lower pressure/volume fluid injection are expected to be smaller in extent (e.g. Flewelling et al., 2013).
There is limited information on the lateral extent of hydraulic fractures. The US EPA (2016) report fractures extending to horizontal lengths of 300 m from borehole data in the Fisher and Warpinski (2012) dataset. Modelling of hydraulic fracturing in sandstone (at a depth of 640 m) indicates a potential fracture length of 244 m (Adachi et al., 2007). Evidence from well communications between closely spaced boreholes might also help to elucidate the fracture half lengths; Jackson et al. (2013a) report a borehole blow-out adjacent to a hydraulic fracturing borehole separated by a distance of 200 m. Interwellbore Communication (IWB) was also found to occur in the Barnett Shale Play in Texas at distances of 340 m and 760 m (US EPA, 2016). Lefebvre (2017) found that the average horizontal distance for well communication at depth was 400 m, with a range from 30 to 2000 m. From 179 wells in Oklahoma, Ajani and Kelkar (2012) (in US EPA, 2016) found that the maximum distance between wells in which an impact was identified was 2590 m (individual fracture length of ~1295 m). The likelihood of communication was <10% for wells 1000 m apart (fracture length of 500 m) and up to 50% for wells <300 m apart (fracture length 150 m).
Hydraulic fractures can also interact with other pathways such as faults or boreholes and seismicity resulting from hydraulic fracturing can impact borehole integrity as seen at Preece Hall, Lancashire (Ward et al., 2015).
VERTICAL SEPARATION OF HYDROCARBON SOURCE UNIT AND POTENTIAL RECEPTORS
This is the shortest (perpendicular) distance between the top of the hydrocarbon source unit and the base of the overlying potential receptor or, where the potential receptor is below the hydrocarbon source unit, the perpendicular distance between the base of the hydrocarbon source rock unit and the top of the potential groundwater body receptor.
There are eight possible vertical separation distance ratings (Table 4.2). The lowest rating is for >1200 m, accounting for the maximum distance of natural hydraulic fracture height (Davies et al., 2012) and would be the minimum depth of hydraulic fracturing below groundwater in SPZ1. Other incorporated boundaries include 100 m (most likely height), 400 m (<1% of hydraulic fracturing stages have fractures >350 m in height) and 600 m (maximum recorded height of induced hydraulic fracture). The weighting for this sub-factor is 1.5. If a receptor does not directly overlie the hydrocarbon activity footprint (but is within the area of interest) this should be given a rating of 1. The weighting for this subfactor is 1.5.
Sources of information
LATERAL SEPARATION OF HYDROCARBON SOURCE UNIT AND POTENTIAL RECEPTORS
Lateral separation is calculated between the hydrocarbon source unit and potential receptor units when they occur at the same horizontal plane in the AOI. This may be due to geological structures such as faults and steeply dipping beds. In the case where an additional unit is introduced to the succession, for example in the hanging wall of a fault or on a deepening succession, this factor can be used to provide a distance between the hydrocarbon source unit and the additional potential receptors even though it is not included in the vertical succession.
Sources of information
Mudstones and clays in intervening zone
Clays, mudstones and shales limit transport of contaminants (Flewelling and Sharma, 2014; Birdsell et al., 2015) due to their low permeability (2.4 ×10-7 to 9.6 x 10-4 mD at depth) as evidenced by their ability to behave as cap rocks for conventional hydrocarbons (Younger, 2016). They also have the ability to adsorb charged particles. Their adsorption properties are largely governed by the nature and quantity of clay minerals present and the available surface area (between clay particles in unconsolidated material and on fracture surfaces in consolidated rocks). In the UK, Early Palaeozoic (Cambrian, Ordovician, Silurian, Devonian) shales and slates would typically contain illite and chlorite with rather low adsorption capacity, whereas Late Palaeozoic (Carboniferous and Permian) shales and mudstones often contain interlayered clays with a higher adsorption capacities. Mesozoic and younger mudstones and clays are typified by increasing amounts of smectite and therefore have the highest adsorption capacity. Clay particles in sandstones and siltstones may also have adsorption properties and will lower bulk permeability (S. Kemp pers. comm).
MUDSTONES AND CLAYS IN INTERVENING ZONE BETWEEN HYDROCARBON SOURCE UNIT AND POTENTIAL RECEPTOR
This factor accounts for potential barriers (mudstones and clays) to contaminant migration between the top/base of the hydrocarbon source unit and the base/top of each potential receptor.
The rating is based on the cumulative thickness of mudstone/clays (Table 4.4). The potential receptor adjacent to the hydrocarbon source unit will always have a rating of 5 as there are no intervening units.
Sources of information
Conceptual model (Development of a geological conceptual model). Thickness of units is calculated from unit depths entered into the 3DGWV methodology spreadsheet. The thickness of mudstone/clay associated with a particular unit can be estimated from the unit thickness (above) and the proportion of mudstone/clay associated in with the unit in the 3DGWV spreadsheet. The proportion of mudstone/clay can be obtained from:
If a unit comprises only mudstone/clay the total unit thickness can be entered. If only a proportion of the unit is mudstone then the total unit thickness should be multiplied by the fraction of the unit that is mudstone/clay. For each potential receptor, the thickness between it and the hydrocarbon source unit is summed. For example, in (Figure 2.2), the cumulative mudstone/clay unit thickness between the upper principal aquifer and the source unit would be:
Unit overlying hydrocarbon source unit: 100% mudstone and 300 m thick
Groundwater flow mechanism
The groundwater flow mechanism (fracture flow or intergranular flow) of the intervening rock can affect the ease with which groundwater flows. Rocks, such as limestone, with predominantly fracture flow are likely to allow faster travel times than rocks with predominantly intergranular flow or multi-layered aquifers such as the Millstone Grit or the Coal Measures. Solution enlarged fissures and conduits (known as karst in carbonate rocks) can potentially create rapid contaminant pathways though the subsurface (e.g. Ruggieri et al., 2017). Solution features may also affect the permeability of the immediately overlying geological unit due to subsidence.
GROUNDWATER FLOW PROCESSES
This factor accounts for key flow processes in the intervening zone between the hydrocarbon source and the top of each potential receptor. Potential receptors designated ‘unproductive strata’ by the EA are not considered.
The rating is based on the cumulative groundwater flow mechanism in the intervening units between the hydrocarbon source unit and the potential receptor, including the potential receptor. There are four possible ratings (Table 4.5).
Sources of information
Thicknesses of units are calculated from unit depths entered into the 3DGWV methodology spreadsheet from the conceptual model (Development of a geological conceptual model).
The groundwater flow mechanism of a particular unit can be estimated from:
Using the example in Figure 2.2, the cumulative groundwater flow mechanism categories would be:
Dissolution features occur in both carbonate and evaporite rocks. The depth of karst development is highly variable and is related to differences in geology and landscape evolution. In England, karst is quite common at shallow depths in parts of the Chalk and in the Carboniferous Limestone. Cave systems are known to nearly 300 m bgl (e.g. Brants Gill, Yorkshire Dales) (Waltham et al., 1997). Karst drainage may have developed at times of lower base levels (sea levels). Some deep caves are formed by water rising up from depth or by geochemical mixing, sulphuric acid dissolution or rising artesian flow through soluble rocks, and are unrelated to modern drainage systems (Farrant, 2008). Palaeokarst systems can be inferred at depth in Carboniferous Limestones, for example at the Buxton Springs, where groundwater circulation is inferred to 1500 m bgl (Aitkenhead et al., 2002) and at the Bath Hot Springs, where it is inferred up to 4000 m bgl (Edmunds et al., 2014). Palaeokarst is more likely to have developed below an unconformity. The maximum depth of karstification in England is limited by the base of the limestone. Gypsum karst has also formed phreatic cave systems, but the rapid solubility rate of the gypsum means that the karst can evolve on a human time scale (Farrant, 2008).
This factor accounts for solution features in the intervening units between the hydrocarbon source rock unit and the potential receptor, including the potential receptor within the AOI. It accounts for evidence of solution features in the AOI and the potential for the development of solution features according to the lithology. There are four possible ratings (Table 4.6).
Source of data
Propensity of geological units to have solution features. This includes places where there are unconformities and disconformities above a unit. A list of areas with important solution features have been identified by Farrant (2008) and included in Appendix 4.
Large volumes of fluids, for example deep brines (Warner et al., 2012; Llewellyn, 2014) and gases (Molofsky et al., 2013; Moritz et al., 2015), have been shown to migrate vertically through rock masses for large distances (up to 2.4 km (Llewellyn, 2014)) over long timescales. However, contaminant migration over the large vertical separation distances between deep hydrocarbon source units and shallow receptors (for example shales with an average depth of 2 km in the US, (US EPA, 2016) is considered unlikely, or would take a very long time, without preferential flow pathways (Lefebvre, 2017). Numerical models by Reagan et al. (2015) have shown that characteristics such as the presence of preferential flow pathways (e.g. faults) and production characteristics might have a greater impact on transport than vertical separation distances.
Faults are planes of movement along which adjacent blocks of rock strata have moved relative to each other. Faults commonly comprise zones, of up to several tens of metres (or greater) in width, of fractures and fault rock. Faults can enhance or hinder fluid flow, or a combination of both (preventing fluids crossing the fault while at the same time allowing fluids to flow parallel to the fault) (Bense et al., 2013). Faults that enhance (are conduits for) fluid flow can allow contaminants to travel along the fault (as a pathway) to a groundwater receptor and can provide vertical pathways through otherwise low permeability bodies of rock. Faults have been found to be conduits for methane (and thermal fluids) even through large thicknesses of shale in British Columbia, Canada (Grasby et al., 2016). Faulting may also bring receptor formations into contact with hydrocarbon source unit formations across the fault zone (known as juxtaposition) and result in laterally variable hydrogeological and rheological properties.
The largest faults can cut all of the brittle rocks in a geological sequence hence pathways provided by faults could be kilometres in length. For example, the Bath thermal springs are believed to flow along a deep fault from between 2.6 and 4 km depth to the surface (Andrews, 1982; McCann et al., 2013). Although faults are often segmented along strike and dip, they tend not to occur in isolation and where large faults occur smaller faults are likely to be present nearby (e.g. Torabi and Berg, 2011). The interaction and connection of these faults can also lead to long pathways.
Faults can also interact with hydraulic fractures, and the longest induced fractures are thought to result from interactions with existing faults (Davies et al., 2012). Monitoring of shale exploitation in Greene County, Pennsylvania, found the maximum height of hydraulically induced fractures corresponds with the maximum height of faults in the region (Hammack et al., 2014).
Natural fractures (where there has been no offset either side of a fracture) can also interact with hydraulic fractures. These were found by Davies et al. (2012) to be predominantly between 200- 300 m in height, with 33% of >350 m in height and a maximum height of 1106 m. The greater vertical extents (than induced hydraulic fractures) possibly result from the greater fluid volumes involved in a natural system and occurrence in more extensively homogeneous lithology (Davies et al., 2012; Lacazette and Geiser, 2013). However, the fracture height probabilities are likely to be over-estimated due to difficulties identifying smaller fractures (Davies et al., 2012).
Faults and fractures have been thought to act as preferential pathways for methane in areas of shale gas exploitation (Warner et al., 2012; Molofsky et al., 2013; Llewellyn, 2014 and Moritz et al., 2015). Numerical modelling has suggested that permeability and overall volume of the connecting fault or fracture have a greater impact on methane transport than separation distance (Reagan et al. 2015). However, Younger (2016) states that in the UK there are no known minewater discharges from natural faults although minor seepages are known to occur along natural faults in close proximity to major mine seepages, even if the fault does not deliver the bulk of the flow.
It is thought that a large number of factors might interact to determine whether or not a fault will enhance or hinder fluid flow, including orientation with respect to the regional stress field, lithology, fault throw and deformation history (history of movement and subsequent diagenesis) (e.g. Bense et al., 2013). Pressure changes surrounding faults, perhaps due to stimulation techniques such as Enhanced Oil Recovery (EOR) or hydraulic fracturing, might also alter the hydraulic behaviour of the fault, for example by fault reactivation, and can also lead to leakage along a fault (e.g. Rinaldi et al., 2014). Westwood (2017) found that the horizontal ‘respect distance’ (minimum lateral distance that hydraulic fracturing should occur from a pre-existing fault in order not to reactivate it) ranged from 63 to 433 m depending on fracture intensity and failure threshold, based on numerical models of hydraulic fracturing at Preese Hall, Lancashire.
This factor accounts for the proximity of faults to the hydrocarbon activity and their hydraulic behaviour. Distances relate to the minimum lateral istance between the hydrocarbon activity and the fault in the AOI since contaminants are more likely to reach a fault if the separation distance is smaller. It is assumed that a fault could cut the entire geological sequence. The distances correspond to lateral separation distances, based on the horizontal extents of high volume hydraulic fractures. This is larger than the maximum horizontal respect distance (minimum lateral distance that hydraulic fracturing should occur from a pre-existing fault in order not to reactivate it) suggested by Westwood (2017) of 433 m.
Since not all faults are permeable, faults that are known to be transmissive are given a higher rating. Evidence for transmissive faults includes discharge of thermal waters and other fluids from depth. There are four possible ratings (Table 4.7):
Sources of information
Mines, for coal and other minerals, can create voids in the subsurface which can provide multiple pathways for contaminants over relatively large volumes (Ward et al., 2015; Monaghan, 2017). The footprint of voids from coal mines can be 50 000 to 200 000 m2 in area (Younger, 2016). Younger (2016) states that minewater discharges overwhelmingly occur via anthropogenic mined features such as shafts, adits or boreholes.
Mining also impacts the characteristics of the surrounding rock, forming an anthropogenic aquifer (e.g. O’ Dochartaigh et al., 2015). Longwall mining, in which a long wall of coal (3 to 4 km in length, and 400 m in width) is mined in a single slice, allows the mine to collapse within two to three years of coal extraction, forming voids filled with goaf (broken rock) (Younger, 2016). As a result of the collapse, bed-parallel fractures can form up to 20 m above the roof of the mined seam (or 1/3 of the distance between underground mine roadways which are typically 100 to 200 m). This fractured zone is overlain by a zone of net compression (and reduced permeability) of up to 1/9 of the distance between the roadways which isolates an upper extensional zone of the same thickness (Younger, 2016). Jones et al. (2004) estimate that the permeability of seams and surrounding strata is increased up to 160–200 m above and 40–70 m below worked seams as a result of previous longwall mining. Nevertheless, Younger (2016) presents the case at Selby Coalfield, Yorkshire, where mines were developed at depth with no connections to shallower workings and ‘complete’ hydraulic isolation from the near-surface hydrogeological environment. Stoop and room mining, in which pillars are left in place and coal mined from around these, can be stable for many years before collapsing (Younger, 2016).
The statutory stand-off interval between longwall workings and the seabed or aquifer is 105 m, reducing to 45 m for supported methods of mining and have been extensively tested including in flooded old workings with head gradients of up to 200 m (Younger, 2016). The current UK criterion for safe longwall mining induced net tensile strain at the base of any overlying aquifer is 100 mm per m, thus a minimum of 60 m of interburden is required regardless of the distance between roads (Younger, 2016).
Hydraulic fracturing of the Marcellus shale is undertaken beneath active coal mines in Kentucky, Pennsylvania and West Virginia in the USA with a vertical separation distance of ~2200 m (Monaghan, 2017). Regulations ensure the special casing and plugging of boreholes through coal-bearing intervals and well plans must be made available to coal operators when the mine is within 90 m of a well, but there are no regulations regarding separation distances (Monaghan, 2017).
This factor accounts for the vertical and lateral proximity of the hydrocarbon activity to mines. The distances correspond to lateral separation distances based on the horizontal extents of high volume hydraulic fractures. Mine shafts can be deeper than the worked coal seams (Monaghan, 2014). There are three possible ratings (Table 4.8).
Sources of information
Boreholes drilled into the subsurface create potential pathways for contaminants to a receptor. While deep hydrocarbon and mineral boreholes are generally completed to prevent leakage, with both steel casing and cement bonding, borehole integrity failures (defects in steel casing, holes in casing joints, mechanical seals and cement e.g. Jackson et al., 2014) can occur. 3% of all hydraulic fracturing operations in the USA involved a downhole mechanical integrity failure (US EPA, 2016). Davies et al. (2014) present data from around the world for the percentage of boreholes (including production, injection, idle and abandoned boreholes) that have some form of borehole barrier or integrity failure. Percentages range from 1.9% (onshore, nationwide CCS/natural gas storage facilities, dates unknown, including well integrity failure only, described as significant gas loss, for 470 boreholes) to 75% (onshore, operational wells in the Santa Fe Springs Oilfield, discovered 1921, including well integrity failures, leakage based on the observation of gas bubbles seeping to the surface along well casing for more than 50 wells). The probability of borehole integrity failure depends on the quality of completion (which will vary over time), the age of the well (degradation) and the exploitation processes but the high variability in recorded barrier or integrity in Davies et al. (2014) also reflects differences in classification of failure (e.g. well, single barrier, significant or bubbles), geological setting and importantly regulation (e.g. Thorogood and Younger, 2015; Davies et al. 2015). Of 143 active wells producing in the UK at the end of 2000, one has evidence of borehole integrity failure (Davies et al., 2014).
In many areas of hydrocarbon interest, there may be existing boreholes which can provide pathways for contamination if they are not properly sealed (for example the casing or cement) or have had a loss of integrity over time (Jackson et al., 2013a; Ward et al., 2015). Borehole leakage rates range from 2% to 50% in the UK (Davies et al., 2014). If abandoned, boreholes might not be monitored and the integrity of their casing will be unknown. In the UK there were 2152 hydrocarbon wells drilled onshore between 1902 and 2013. The ownership of up to 53% of these wells is unclear today and between 50 and 100 are orphaned (Davies et al., 2014).
Hydraulic fracturing has been shown to impact on adjacent wells (US EPA, 2016). In Alberta and British Columbia, 5349 horizontal wells were drilled between 2009 and 2012 and there were 39 reported cases of wellbore connection with existing oil and gas wells, 95% of which were producing in the same geologic unit. Alberta requires that locations of existing oil and gas wells be identified and their capability to sustain increased pressures be verified prior to hydraulic fracturing (Lefebvre, 2017).
This factor accounts for the vertical and lateral proximity of the hydrocarbon activity to boreholes. The distances correspond to lateral separation distances based on the horizontal extents of high volume hydraulic fractures. Deep boreholes can be deviated and hence should be corrected to true vertical depth and also the geographic location of the base. There are three possible ratings (Table 4.9).
Sources of information
ArcGIS layers have been provided in the digital dataset showing the position of mines and all boreholes (including those held confidentially) over 400 m in depth. The depth and logs of all open access boreholes can be viewed via the BGS Geology of Britain viewer http://mapapps.bgs.ac.uk/geologyofbritain/home.html. Some logs are held as ‘management protect’ and although not freely available, the data can be obtained on request. Information for hydrocarbon (management protect) boreholes are available via the Oil and Gas Authority’s website https://www.ogauthority.co.uk/data-centre/access-to-information-and-samples/. Information for water (management protect) and other (management protect) boreholes are available on request from BGS or may be available to the EA from other sources. Other confidential boreholes, included in the layer may also be available to the EA on request to BGS.
- US EPA. 2016. Hydraulic Fracturing for Oil and Gas: Impacts from the Hydraulic Fracturing Water Cycle on Drinking Water Resources in the United States (Final Report). U.S. Environmental Protection Agency, Washington, DC, EPA/600/R-16/236F.
- BROWNLOW, J W, JAMES, S, and YELDERMAN, J R, J C. 2016. Influence of Hydraulic Fracturing on Overlying Aquifers in the Presence of Leaky Abandoned Wells. Groundwater, Vol. 54(6), 781–792.
- YOUNGER, P L. 2016. How can we be sure fracking will not pollute aquifers? Lessons from a major longwall coal mining analogue (Selby, Yorkshire). Earth and Environmental Science Transactions of the Royal Society of Edinburgh, Vol. 106, 89–113.
- FISHER, K, and WARPINKSKI, N. 2012. Hydraulic-Fracture-Height Growth: Real Data, Spe Production & Operations Water, Vol. 27(1), 8–19.
- GASSIAT, C, GLEESON, T, LEFEBVRE, R, and MCKENZIE, J. 2013. Hydraulic fracturing in faulted sedimentary basins: Numerical simulation of potential contamination of shallow aquifers over long time scales. Water Resources Research. Vol. 49(12), 8310–8327.
- DAVIES, R J, MATHIAS, S A, MOSS, J, HUSTOFT, S, and NEWPORT L. 2012. Hydraulic fractures: How far can they go? Marine and Petroleum Geology. Vol. 37, 1–6.
- MONAGHAN, A A. 2014. The Carboniferous shales of the Midland Valley of Scotland: geology and resource estimation. (London: British Geological Survey for Department of Energy and Climate Change). 96 pp.
- FLEWELLING, S A, TYMCHAK, M P, and WARPINSKI, N. 2013. Hydraulic fracture height limits and fault interactions in tight oil and gas formations, Geophys. Res. Lett., 40, 3602–3606, doi:10.1002/grl.50707.
- ADACHI, J, SIEBRITS, E, PEIRCES, A, and DESROCHES, J. 2007. Computer simulation of hydraulic fractures. International Journal of Rock Mechanics and Mining Sciences, Vol. 44(5), 739–757.
- JACKSON, R E, GORODY, A W, MAYER, B, ROY, J W, RYAN, M C, and VAN STEMPVOORT, D R. 2013a. Groundwater protection and unconventional gas extraction: the critical need for field-based hydrogeological research, Groundwater, Vol. 51(4), 488–510.
- LEFEBVRE, R. 2017. Potential impacts of shale gas development on groundwater quality. Wiley Interdisciplinary Reviews: Water, Vol. 4 (1). doi: 10.1002/wat2.1188.
- AJANI, A, and KELKAR, M. 2012. Interference study in shale plays, SPE Hydraulic Fracturing Technologu Conference, The Woodlands, Texas.
- WARD, R, STUART, M E, and BLOOMFIELD, J, P. 2015. The hydrogological aspects of shale gas extraction in the UK. In: Issues in Environmental Science and Technology: Fracking. (London: Royal Society of Chemistry). pp.122–150.
- FLEWELLING, S A, and SHARMA, M. 2014. Constraints on upward migration of hydraulic fracturing fluid and brine. Groundwater, 52(1), 9–19.
- BIRDSELL, D T, RAJARAM, H, DEMPSEY, D, and VISWANATHAN, H S. 2015. Hydraulic fracturing fluid migration in the subsurface: A review and expanded modelling results, Water Resources Research 5, Vol. 1, 7159–7188.
- RUGGIERI, R, FORTI, P, ANTOCI, M L, and DE WAELE, J. 2017. Accidental contamination during hydrocarbon exploitation and the rapid transfer of heavy-mineral fines through an overlying highly karstified aquifer (Paradiso Spring, SE Sicily), Journal of Hydrology, Vol. 546, 123–132.
- WALTHAM, A C, SIMMS, M J, FARRANT, A R, and GOLDIE, H S. 1997. Karst and Caves of Great Britain. (London: Chapman and Hall).
- FARRANT, A, and COOPER, A. 2008. Karst geohazards in the UK: the use of digital data for hazard management. Quarterly Journal of Engineering Geology and Hydrogeology, Vol. 41, 339–356.
- AITKENHEAD, N, BARCLAY, W J, BRANDON, A, CHADWICK, R A, CHRISHOLM, M J I, COOPER, A H, and JOHNSON, E W. 2002. The Pennines and adjacent areas (4th edition), Regional Geology Guide. (Nottingham: British Geological Survey).
- EDMUNDS, W M, DARLING, W G, PURTSCHERT, R, and CORCHO ALVARADO, J A. 2014. Noble gas, CFC and other geochemical evidence for the age and origin of the Bath thermal waters, UK. Applied Geochemistry, Vol. 40, 155–163
- WARNER, N R, JACKSON, R B, DARRAH, T H, OSBORN, S G, DOWN, A, ZHAO, K, WHITE, A, and VENGOSH, A. 2012. Geochemical evidence for possible natural migration of Marcellus Formation brine to shallow aquifers in Pennsylvania, PNAS, Vol. 109(30), 11961–11966.
- Llewellyn, G T. 2014. Evidence and mechanisms for Appalachian Basin brine migration into shallow aquifers in NE Pennsylvania, USA. Hydrogeology Journal, 22(5), 1055–1066.
- MOLOFSKY, L J, CONNOR, J A, WYLIE, A S, WAGNER, T, and FARHAT, S K. 2013. Evaluation of Methane Sources in Groundwater in Northeastern Pennsylvania, Groundwater, Vol. 51(3), 333–349.
- MORITZ, A, HÉLIE, J-F, PINTI, D L, LAROCQUE, M, BARNETCHE, D, RETAILLEAU, S, LEFEBVRE, R, and GÉLINAS, Y. 2015. Methane baseline concentrations and sources in shallow aquifers from the shale gas-prone region of the St. Lawrence lowlands (Quebec, Canada), Environmental Science and Technology, 49, 765–4771.
- BENSE, V F, GLEESON, T, LOVELESS, S E, BOUR, O, and SCIBEK, J. 2013. Fault zone hydrogeology. Earth-Science Reviews, Vol. 127, 171–192.
- GRASBY, S E, FERGUSON, G, BRADY, A, SHARP, C, DUNFIELD, P, and MCMECHAN, M. 2016b. Deep groundwater circulation and associated methane leakage in the northern Canadian Rocky Mountains. Applied Geochemistry, Vol. 68, 10–18.
- ANDREWS, J N. 1982. The Thermal Springs of Bath. Nature, Vol. 298, 339–343.
- MCCANN, C, MANN, A C, MCCANN, D M, and KELLAWAY, G A. 2013. Insights into the origin of the thermal springs of Bath and Bristol, England, from geophysical data. Quarterly Journal of Engineering Geology and Hydrogeology, 46(3), 267–279.
- TORABI, A, and BERG, S S. 2011. Scaling of fault attributes: A review. Marine and Petroleum Geology, Vol. 28(8), 1444–1460.
- HAMMACK, R, HARBERT, W, SHARMA, S, STEWART, B, CAPO, R, WALL, A, WELLS, A, DIEHL, R, BLAUSHILD, D, SAMS, J, and VELOSKI, G. 2014. An Evaluation of Fracture Growth and Gas/Fluid Migration as Horizontal Marcellus Shale Gas Wells are Hydraulically Fractured in Greene County, Pennsylvania, NETL-TRS-3-2014, EPAct Technical Report Series. (Pittsburgh PA.: U.S. Department of Energy, National Energy Technology Laboratory.)
- LACAZETTE, A, and GEISER, P. 2013. Comment on Davies et al., 2012 — Hydraulic fractures: How far can they go?. Marine and Petroleum Geology. 43(0), 516–518.
- REAGAN, M T, MORIDIS, G J, KEEN, N D, and JOHNSON, J N. 2015. Numerical simulation of the environmental 603 impact of hydraulic fracturing of tight/shale gas reservoirs on near surface groundwater: Background, 604 base cases, shallow reservoirs, short-term gas and water transport, Water Resources Research, 605 51:2543–2573.
- RINALDI, A P, RUTQVIST, J, and CAPPA, F. 2014. Geomechanical effects on CO2 leakage through fault zones during large-scale underground injection. International Journal of Greenhouse Gas Control, Vol. 20, 117–131.
- WESTWOOD, R F, TOON, S M, and CASSIDY, N J. 2017. A sensitivity analysis of the effect of pumping parameters on hydraulic fracture networks and local stresses during shale gas operations, Fuel, Vol. 203, 843–852.
- MONAGHAN, A A. 2017. Unconventional energy resources in a crowded subsurface: Reducing uncertainty and developing a separation zone concept for resource estimation and deep 3D subsurface planning using legacy mining data. Science of the Total Environment, Vol. 601–602, 45–56.
- O’ DOCHARTAIGH, B E, MACDONALD, A M, FITZSIMONS, V, and WARD, R. 2015. Scotland's aquifers and groundwater bodies. British Geological Survey Report, 63pp. (OR/15/028) (Unpublished)
- JONES, N S, HOLLOWAY, S, CREEDY, D P, GARNER, K, SMITH, N J P, BROWNE, M A E, and DURUCAN, S. 2004. UK Coal Resource for New Exploitation Technologies Final Report. British Geological Survey Report CR/04/015N.
- JACKSON, R B, VENGOSH, A, CAREY, J W, DAVIES, R, DARRAH, T H, O’SULLIVAN F, PÉTRON G. 2014. The 555 Environmental Costs and Benefits of Fracking, Anny. Rev. Environ. Resoure. 39:327–362, doi 556 10.1146/annrev-environ-031113-144051
- DAVIES, R J, ALMOND, S, WARD, R S, JACKSON, R B, ADAMS, C, WORRALL, F, HERRINGSHAW, L G, GLUYAS, J G, and WHITEHEAD, M A. 2014. Oil and gas wells and their integrity: Implications for shale and unconventional resource exploitation. Marine and Petroleum Geology, 30(1), 1–16.
- THOROGOOD, J L, and YOUNGER, P L. 2015. Discussion of ‘Oil and gas wells and their integrity: Implications for shale and unconventional resource exploitation’ by R J Davies, S, Almond, R S, Ward, R B, Jackson, C, Adams, F, Worrall, L G, Herringshaw, J G, Gluyas, and M A, Whitehead. (Marine and Petroleum Geology 2014). Marine and Petroleum Geology, Vol. 59, 671–673.
- DAVIES, R J, ALMOND, S, WARD, R S, JACKSON, R B, ADAMS, C, WORRALL, F, HERRINGSHAW, L G, GLUYAS, J G, and WHITEHEAD, M A. 2015. Reply: ‘Oil and gas wells and their integrity: Implications for shale and unconventional resource exploitation’. Marine and Petroleum Geology, 59, 674–675.