Shale gas is considered by many to have the potential to provide the UK with greater energy security, economic growth and jobs. However, development of a shale gas industry is highly contentious due to environmental concerns including the risk of groundwater pollution. Evidence suggests that the vertical separation between exploited shale units and aquifers is an important factor in the risk to groundwater from shale gas exploitation. A methodology is presented to assess the vertical separation between different pairs of aquifers and shales that are present across England and Wales.
Sian E. Loveless1, John P. Bloomfield1, Robert S. Ward1, Alwyn J. Hart2, Ian R. Davey2, Melinda A. Lewis1
1British Geological Survey, Maclean Building, Crowmarsh Gifford, Oxon OX10 8BB, UK. 2Environment Agency, Horizon House, Deanery Road, Bristol BS1 5AH, UK
Received: 20 April 2017 / Accepted: 23 January 2018
©The Author(s) 2018
The application of the method is then demonstrated for two of these pairs—the Cretaceous Chalk Group aquifer and the Upper Jurassic Kimmeridge Clay Formation, and the Triassic sandstone aquifer and the Carboniferous Bowland Shale Formation. Challenges in defining what might be considered criteria for ‘safe separation’ between a shale gas formation and an overlying aquifer are discussed, in particular with respect to uncertainties in geological properties, aquifer extents and determination of socially acceptable risk levels.
Modelled vertical separations suggest that the risk of aquifer contamination from shale exploration will vary greatly between shale–aquifer pairs and between regions and this will need to be considered carefully as part of the risk assessment and management for any shale gas development.
Increasing demands for use of the deep subsurface, such as for storage of nuclear waste materials, sequestration of CO2, and the development of conventional and unconventional on-shore hydrocarbon resources can place additional pressures on groundwater resources. One such use of the subsurface, the production of natural gas from shales, or shale gas, has received growing attention in the last few years, with recognition of the range of potential threats to groundwater (e.g. Jackson et al. 2014).
The industry has expanded dramatically since the year 2000 in North America due to cost effective extraction technologies including directional drilling and slick water fracking fluids (US EPA 2016; Gallegos and Varela 2015). Shale gas exploration and production is now taking place in a number of other countries around the world, including Europe (e.g. Scotchman 2016). Low-level resource-assessment activity has been ongoing in the UK since 2011, with two hydraulic fracturing licences now granted.
Groundwater is an important resource in England and Wales, providing an average of 31% of water resources, and up to 100% in some areas of southeast England. Pressures on groundwater resources from the development of shale gas may include water supply issues associated with drilling and hydraulic fracturing of wells (Flavin and Kitasei 2010; Gregory et al. 2011; Wood et al. 2011; Stuart 2012; Vengosh et al. 2014; Kondash and Vengosh 2015).
Potential groundwater contamination has been postulated from a range of possible contamination pathways illustrated in Fig. 1a, including: (1) migration of produced gases and/or fluids used in the hydraulic fracturing process though rock units separating shale-gas source rocks and overlying aquifers (Vengosh et al. 2014; Myers 2012a); (2) contamination via defective production wells; (3) leakage of abandoned wells (Davies et al. 2014); and (4) spills of chemicals, flow back and produced waters at the land-surface that could percolate to shallow aquifer systems (Vengosh et al. 2014; Rozell and Reaven 2011).
Fig. 1a Schematic diagram showing potential source-pathway-receptors resulting from shale gas exploration and production (after Vengosh et al. 2014). Numbers correspond to pathways described in the text. Diagram not to scale. b Schematic diagram illustrating the effect of aquifer–shale separation distances and potential direct pathways through the intervening interval, including migration through the intervening rock mass, and aquifer and hydraulic fractures linking with a permeable fault zone. Green dotted lines illustrate 100 and 600-m hydraulic fracture heights.
Potential scenarios for migration of contaminants from shales to aquifers (pathway 1) are shown in Fig. 1b. The greater the vertical separation between the shale and overlying aquifer, the more likely the intervening rock mass will limit the upward migration of fluids and gases and thereby reduce the risk of contamination of groundwater associated with this pathway.
There will also be a further reduction of risk when rock units with low permeability are present within the overlying rock mass, contributing to a relatively low bulk vertical hydraulic conductivity (Freeze and Cherry 1977), and also where there is an absence of through-going, conductive fracture networks which might connect them (Myers 2012a; Cai and Ofterdinger 2014).
A methodology is presented here for evaluating the vertical separation between shales and aquifers across England and Wales. Its application is demonstrated for two important principal aquifer–shale combinations—the Cretaceous Chalk Group aquifer and the Kimmeridge Clay Formation, and the Triassic sandstone aquifer and the Bowland Shale Formation (the upper shale within the Craven Group).
The Chalk Group and the Triassic sandstone aquifers are the most important aquifers for water supply in southern and northern England respectively (Allen et al. 1997), while the Bowland Shale Formation is currently the principal target for shale gas exploration in England, with planning permission having been recently granted for hydraulic fracturing operations (DCLG 2016).
Context: the concept of ‘safe separation’
A number of studies have sought to quantify a ‘safe separation’ distance between the zone of hydraulic fracturing and overlying aquifers (Davies et al. 2012; Kissinger et al. 2013); however, despite an increasing body of research there is a lack of consensus regarding what might be considered a ‘safe separation’ on the basis of being confident that the risks will be acceptable, i.e. be extremely low. The term itself currently has no commonly agreed meaning and is a relative rather than absolute concept that encompasses considerations of the value of groundwater resources within a particular society and the consequent nature of risk assessment undertaken within widely varying regulatory frameworks (Alberta Energy Regulator 2013; Environment Agency 2013a).
A useful way, therefore, of using the concept of ‘safe separation’ is to aim for a separation distance over which no contaminant breakthrough would be expected to occur but in the unlikely event that it did, the concentrations would be so low that they would not be harmful or of concern to humans and the environment (Myers 2012a). Therefore information about contaminant concentrations at source, timescales of interest in relation to contaminant attenuation, the physical proximity of sources and receptors, and the physical properties of the intervening interval are all relevant to the concept of a ‘safe separation’ (Birdsell et al. 2015).
The present study addresses methods to constrain uncertainty about the separation of sources and receptors and looks at the impacts on the available area of potential shales if suggested safe separation distances are applied. It does not attempt to define safe separation due to the large number of other factors that would need to be considered.
Vertical separation distances between aquifers and shales vary greatly depending on the depositional and tectonic setting and resultant relative position of aquifers and shales in the local stratigraphy. Unlike existing plays in North America, where shale gas plays and overlying aquifer units are typically regionally extensive and often in relatively simple structural settings, in many parts of Europe, including the UK, the geological and hydrogeological settings may be much more complex (Ward et al. 2015).
For example, in the UK there can be multiple potential shale gas targets within relatively complex stratigraphic sequences and structural settings that may alter the spatial relationships between shales and aquifers across a region (Andrews 2013). In these complex geological settings, knowledge of aquifer–shale separation and its variability will be a critical consideration for well-regulated, future development of shale gas resources. Consequently, there is a need for high-level regional-scale screening tools such as the one described here, to characterise and investigate the vertical separation between major shales and aquifers.
Data and methods
Consistent with the source-pathway-receptor conceptual framework used by the Environment Agency (England; Defra 2011), hydraulic fracturing of shales is considered as a potential source of contamination (Jackson et al. 2013), aquifers as potential receptors, and the intervening volume of rock and associated discontinuities the potential pathway. The aim of the current study is to develop an approach to estimate and map spatial variations in the distance (vertical separation) between identified receptors and sources at a regional to national scale. To do this there is a need to identify the aquifers (receptors) and shales (sources) of interest across the area of interest (England and Wales) and then to define their respective boundaries, in this case, the base of the aquifer and top of the shale.
In England and Wales, principal aquifers are defined as ‘geological strata that exhibit high permeability and usually provide a high level of water storage’. Also ‘they are capable of supporting water supply at a strategic scale and are often of major importance to river baseflow’ (EA 2013a). Principal aquifers provide most of the potable groundwater supply across England and Wales and are therefore the focus of the present study. The Environment Agency for England and Natural Resources Wales recognise 11 main bedrock principal aquifers. The outcrop pattern of these is shown in Fig. 2a and their relative stratigraphic position shown in a schematic column in Fig. 2b.
Fig. 2a Principal aquifers and major shales at outcrop in England and Wales (part of the United Kingdom). Rock units are generally younger towards the southeast. Note the principal aquifers are extensive at outcrop, while the shale units are more extensive in the subsurface than at outcrop. b Schematic stratigraphy of these units (scaled to time not unit thickness). For more detail about the aquifer and shale units see Tables S1 and S2 of the electronic supplementary material (ESM)
Principal aquifers are defined in terms of geological strata (EA 2013b). As the base of these aquifers is considered to be co-incident with the base of the geological unit this means that the base of aquifers can be up to at least 7 km bOD (below Ordnance Datum, or sea level)—much deeper than the maximum depth of exploitation, given that generally water quality and groundwater yields reduce significantly with depth (Allen et al. 1997).
In the UK, environmental objectives for groundwater (quality and quantity) are established as part of the EU Water Framework Directive and applied to groundwater bodies. The UK Technical Advisory Group (UKTAG) have issued guidance on the definition and delineation of groundwater bodies (UKTAG 2011), recommending a default maximum thickness for bedrock groundwater bodies of 400 m, unless local knowledge indicates that a different depth should be applied.
The UKTAG recommendation is consistent with the zone of active exploitation of groundwater in England based on the distribution of borehole depths within a given aquifer. Most exploitation of groundwater takes place at depths of a few tens of metres below groundwater level—Fig. S1 of the electronic supplementary material (ESM). However, different aquifers are exploited to different maximum depths (dependent on their respective depth-yield relationships)—for example, Triassic sandstones are exploited to much greater depths (a number of boreholes exceed 400-m depth) than the Chalk Group (maximum depth ~200 m). Consequently, for the purposes of the separation analysis, and consistent with the UKTAG recommendations, it was assumed that the base of a given principal aquifer is either the base of the geological unit forming the aquifer, or if the unit is present at greater than 400 m below ground level (bgl), the base of the aquifer is assumed to be 400 m bgl.
The first extensive review of shale gas prospectivity in the UK (Smith et al. 2010) identified potential shale gas targets as the main organic-rich black shales from Cambrian to late Jurassic age, which could have reached the thermogenic gas window. Using this and related work, reports for the Dept. of Energy and Climate Change (Andrews 2013; DECC 2012; Andrews 2014) highlighted the following six units as potential shale-gas source rocks: the Kimmeridge Clay Formation; Oxford Clay Formation; Lias Group; Marros Group; Bowland Shale Formation; and the Upper Cambrian shales. These units are used in this study and the outcrop of the shales is shown in Fig. 2a, while Fig. 2b shows their position in a schematic stratigraphic column.
In this study, the top of shale units has been identified for the purposes of calculating separations between shales and aquifers. In reality, the potential target or ‘sweet zone’ for hydraulic fracturing and shale gas production will be below this level, within the body of the shale (DECC 2012; Andrews 2014); however, this information is not yet consistently available at the regional- to national-scale, and hence a precautionary approach has been adopted. Since the maximum depth of principal aquifers considered here is 400 m bgl and the shale units are unlikely to present a commercial prospect shallower than this level (and in reality legislation prevents high volume hydraulic fracturing above 1,000 m bgl) the only scenarios considered are those where principal aquifers overlie shale units.
Analysis of aquifer–shale separation based on a three-dimensional geological model
The British Geological Survey (BGS) National Geological Model (NGM) of Great Britain (Mathers et al. 2012, 2014a, b) was used as the basis for modelling the aquifer base and the top of the shale in England and Wales. The NGM is a digital model, developed using the geological modelling software GSI3D (Mathers et al. 2014b). It consists of a series of geological sections (‘fences’) across the UK, typically with spacing of about 30 km and to a depth of up to 5 km. It is built on a common stratigraphic succession for the UK. Depending on the underlying geological data for each section, the location of geological boundaries in each section may have a vertical accuracy of between about 10 and 100 m.
A subset of 84 geological cross sections across England and Wales, totalling ~12,000-km line length, was used to construct the aquifer and shale surfaces of interest (Fig. 3a; Mathers et al. 2014a). Top or base surfaces were generated by applying a simple linear interpolation algorithm between polylines along geological sections and the intersection of the land surface and outcrop.
Fig. 3a Map illustrating the location of the geological sections from the BGS National Geological Model (NGM) used in this work. The section shown in (b) is highlighted in red. b Representative cross section from the NGM, through central southern England and the western end of the Weald. The Chalk Group aquifer is outlined in green, and upper surfaces of the Kimmeridge Clay Formation, Oxford Clay Formation, and Lias Group are highlighted by red, blue and yellow lines respectively
A 3 km × 3 km-grid resolution was used for the interpolation of the surfaces reflecting the degree of uncertainty in the position of surfaces between cross-sections, while still honouring the overall depth distribution data. Where principal aquifers were present below 400 m, the base of these units was modified to show a maximum depth of 400 m bgl. Where principal aquifers were underlain by shale units, spatial queries in ArcGIS were used to calculate vertical separations between the pairs of top shale and base aquifer surfaces.
Figure 4 shows that there are 25 pairs of surfaces where major shales underlie principal aquifers in England and Wales. The complexity of some of these relationships in the geological sequence can be seen along a representative section from central England through the western end of the Weald (Fig. 3b), highlighting the spatial relationships between the Chalk Group aquifer and three different shales, the Kimmeridge Clay Formation, the Oxford Clay Formation and the Lias Group. Figure 5 shows the modelled full crops and vertical separation for the Chalk Group principal aquifer and the Kimmeridge Clay Formation and the Triassic sandstone principal aquifer and the Craven Group (Bowland Shale Formation).
Fig. 4 Matrix illustrating the large number of spatial relationships between the principal aquifer and shale units in England and Wales due to the complex geology and structural history
Table 1 Summary statistics of vertical separations for individual aquifers and shales. OD relative to Ordnance Datum; SD standard deviation
Vertical separations are typically smaller between the Chalk Group and Kimmeridge Clay Formation than the Triassic sandstone and Bowland Shale Formation pair (Fig. 5; Tables 1 and 2), reflecting the proximity of the Chalk Group and Kimmeridge Clay Formation in the stratigraphic sequence over much of England. In addition, in southern and eastern England regional structures are typically relatively simple with more gentle, open fold structures compared with deeper, fault controlled basins and tighter, localised folding in central and northern England.
Fig. 5 Maps showing examples of modelled outputs. Depth is meters above or below Ordnance Datum. a Base of the Chalk Group aquifer, b top of the Kimmeridge Clay Formation, c Chalk Group and Kimmeridge Clay Formation vertical separation, d base of the Triassic sandstone aquifer, e top of the Craven Group (Bowland Shale Formation), f Triassic sandstone and Craven Group (Bowland Shale Formation) vertical separation. Contours and shading (c and f) show the areas where the shale formation is >1,000 m bgl
Consequently, the Triassic sandstone and Bowland Shale Formation separation map (Fig. 5f) shows more spatially complex distributions of vertical separation than between the Chalk Group and Kimmeridge Clay Formation pair (Fig. 5c). These differences are also reflected in the respective histograms of relative frequency of vertical separations,(Fig. 6); however, a common feature of both separation maps is that these aquifers only cover part of the respective shales, 51% in the case of the Chalk Group and Kimmeridge Clay Formation and 26% in the case of the Triassic sandstone and Bowland Shale Formation pairs.
This area is reduced further when considering the minimum permitted depth of high volume hydraulic fracturing in England and Wales, shown by the 1,000-m contour of the shales (Figs. 5c,f), to 2 and 16% respectively. Full crop maps of all the principal aquifers, shales and the respective separation maps are available in the supporting information (Fig. S2 of the ESM), and from the BGS website (BGS 2016).
Table 2 Summary statistics of vertical separations for both pairs of aquifers and shales. OD relative to Ordnance Datum; SD standard deviation
Fig. 6 Distribution of vertical separation for the two aquifer–shale separation map examples shown in Fig. 5; 600 m and 1,000 m represent the maximum height of induced hydraulic fractures and natural hydraulic fractures (e.g. Davies et al. 2012)