Abstract
Humankind, in its technological development, is increasingly utilizing both mineral resources from Earth's interior and developing the rock mass as a resource in itself. In this paper we review the types of anthropogenic intrusion, at different depth ranges, that can modify the physical structure and chemistry of the subsurface. Using examples from across the world, but with emphasis on the UK, and physical models of the induced modifications, we predict what kind of subsurface signatures a geologist of the future might recognize as anthropogenic, including boreholes, tunnels and caverns, waste and resource storage facilities, mineral workings and military test traces. The potential of these anthropogenic signatures to be discriminated from natural analogues is discussed against known or modelled processes of deterioration and transformation over geological timescales of millennia or longer.
A key stage in the development and ultimate success of Homo sapiens has been the capability to utilize the subsurface as a resource. Initially this was dominated by small-scale shallow mining of metals. Major underground development came with the Industrial Revolution, which saw markedly increased and extended mineral working. New technologies, including steam engines used to pump water from mine workings, allowed mining to extend to greater depths (Fig. 1). There was great expansion of coal working, to fuel steam engines, this becoming the volumetrically dominant global mining activity.
Approximate depth ranges of typical anthropogenic structures discussed in this study (modified from Evans et al. 2009).
London, which by the mid-nineteenth century was the largest city in the world, saw the extensive development of the shallow subsurface to support building construction, transport infrastructure and utilities (Fig. 2a; Price et al. 2016). The opening of the first London Underground station in 1863 is seen by Williams et al. (2014, 2019) as a key indicator of the technological evolution of cities, allowing outward expansion of the conurbation. Arguably of greater importance was development between 1859 and 1865 of 160 km of intercepting sewers, fed by 720 km of main sewers and 21 000 km of smaller local sewers. This provided the model by which subsequent megacities have developed.
Schematic representation of historical development of the subsurface in a city: (a) during the Victorian era; and (b) its evolution to multiple historical and current uses. From Price et al. (2016), reproduced with permission.
Technological development during the twentieth century saw new use of underground rock masses (Fig. 2b; Price et al. 2016). These include disposal and storage facilities for wastes or resources such as radioactive waste (radwaste), CO2 Carbon Capture and Storage (CCS), underground gas storage, compressed air energy storage, geothermal energy, subsurface development of habitations and laboratories (Fig. 1; Evans et al. 2009) and modification through warfare and weapons testing. Table 1 compares the ranges of anthropogenic intrusions, from the near surface to thousands of metres depth. Shallow excavations remain the most extensive category of subsurface use by both area and volume.
Typical types of anthropogenic subsurface modifications, their depth of influence and associated material types
The scale of intrusion of the underground rock mass through discrete developments is now significant as a planetary phenomenon, and this study explores aspects of the scale and permanence of these developments.
To what extent will infrastructure associated with these developments provide a lasting (on geological timescales of millennia or longer) signature of anthropogenic activities?
How far does anthropogenic influence of the rock mass extend beyond the rock directly excavated?
Can these far-field physical and chemical alterations of the rock mass be unequivocally distinguished from natural processes?
Subsurface modification includes the construction of voids, the propagation of fractures associated with these voids and the migration of materials to form vein-fill or cements within pores. The types and nature of the voids, discussed by Zalasiewicz et al. (2014) in the context of characterization of the Anthropocene, are summarized in Table 1. Earthworks associated with surface construction and landfill sites, both associated with distinctive artificial deposits readily identifiable as anthropogenic, and diffuse infiltration of contaminants (nitrates, heavy metals or organic pollutants) into groundwater from the ground surface, are not considered here.
Structural impacts of subsurface modification as distinctive anthropogenic features
Anthropogenic (induced) seismicity has many causes: mining of minerals, particularly coal; solution mining, especially of salt; oil and gas field depletion; hydraulic fracturing (‘fracking’) of low-permeability rocks to extract gas and oil; reservoir impoundment; enhanced geothermal operations; and academic research boreholes (Davies et al. 2013; Mitchell & Green 2017). The seismicity commonly results from reactivation of pre-existing ‘natural’ faults. Accepted criteria for recognizing that fault reactivation is anthropogenic depends on knowledge of the timing and spatial relationships of the seismicity and of the injection procedures (Davis & Frohlich 1993). Hence, far-future recognition that a component of displacement is anthropogenic would be difficult, if not impossible. Here, we investigate fracture patterns that may be uniquely anthropogenic in origin, summarized in Table 2.
Mechanical effects of anthropogenic subsurface modification
Structural or mechanical loading by earthworks
Mechanical compaction applies mechanical energy to densify a soil, changing its bulk density. On a larger scale, extensive urban development can increase consolidation of subsurface deposits and enhance subsidence rates, particularly in coastal deltas. In extreme cases, dam construction, with reservoir impoundment and suddenly increased water-loading of the ground surface, can induce earthquakes (Gupta 1985).
In the future it would be exceptionally difficult to differentiate between the effect of mechanical compaction or structural loading by human constructions and that of natural burial consolidation (e.g. uneven glacial loading). Similarly, recognition of an anthropogenic origin for components of displacement on a fault would be difficult, if not impossible, unless observed at the time of deformation.
Structural disturbance by boreholes and hydrocarbon wells
Drilling-induced disturbance associated with boreholes is typically minor, on the scale of a few millimetres (Kelsall et al. 1983), though fractures may extend up to 0.2 m from boreholes (Fig. 3). Weak materials such as chalk and sand can flow into a wellbore, and sandstones can be affected by fracturing and borehole breakouts (Doornhof et al. 2006).
Excavation near Stoke-on-Trent, UK, revealing a former void associated with old coal workings, subsequently infilled with injection grout to stabilize the ground above the workings. The feeder injection borehole, also grout-filled, is associated with a 0.2 m wide zone of closely spaced fractures that appear to be associated with the intrusion by the borehole. Widely spaced fractures evident in the sandstone present in the roof of the former (labelled on the image) coal seam are probably formed by collapse of open voids following coal extraction and before grout injection. Photo reproduced with the permission of Jacobs (from Wilson et al. 1992).
On a larger scale, hydrocarbon and groundwater extraction can cause consolidation, e.g. the subsidence of Venice in the 1940s to 1970s resulted from consolidation of underlying deposits following extraction of groundwater and natural gas, resulting in pore space collapse, grain fracture, the closing of open fractures or initiation of new fractures (Doornhof et al. 2006); groundwater abstraction has resulted in a similar situation in London (Bonì et al. 2016). The flexure of the overburden due to hydrocarbon extraction generates shear stresses that can cause slip along weak shale strata (Davies et al. 2013).
Injection of CO2 as part of a CCS facility can result in increased seismicity and subsidence of the Earth's surface (Cox et al. 1996). Fluid injection to enhance secondary oil recovery or dispose of extracted saline waters reduces consolidation effects, but can cause fault reactivation (Davies et al. 2013; Elsworth et al. 2016). Enhanced geothermal and hydrocarbon extraction involves creation of fracture networks through fracking. The fracturing operation for hydrocarbons commonly involves detonation of explosives perforating the well in controlled orientations to produce finger-like fractures up to 2.5 cm in diameter and extending up to 0.6 m into the shale rock (Cuss et al. 2015). Hydraulic fractures are then induced by pumping saline water under high pressure, which can reactivate faults that cross or are within hundreds of metres of the wellbore (Davies et al. 2013). There remains some uncertainty regarding the morphology and development of fracture networks generated during shale gas production (Cuss et al. 2015). However, stimulated fractures are expected to propagate along planes in the lowest energy orientation, generally orthogonal to the maximum horizontal stress (Gale et al. 2007; Davies et al. 2012), as indirectly observed from microseismic hypocentre location and seismic anisotropy analysis (Simon 2005).
Natural hydraulic fractures are common, ranging from cm-scale sand-filled injectites up to hundreds of metres in height (Hurst et al. 2011). Natural hydraulic fracture pipes cluster into pipe-like features that may propagate upwards by >1 km. Stimulated hydraulic fracture pipes have a maximum upwards propagation recorded to date of c. 600 m, although most are <100 m (Davies et al. 2012).
Methane extraction from unworked coal (Davidson et al. 1995) involves drilling boreholes down to the seam, with permeability enhancement through: (a) fracking to develop new cracks or open existing joints (known as ‘cleat’) within the coal; (b) blasting; (c) injection of CO2 or N2; or (d) chemical reactions, such as the introduction of weak hydrochloric acid. Alternatively, in situ Underground Coal Gasification (UCG) involves injection of a water/oxygen mixture via a borehole, igniting and partially combusting the coal and extracting the gasification products (syngas). This process forms cavities, which as with coal mining, can cause subsidence that could propagate to the surface.
Enhancement of groundwater borehole yields can cause permanent changes to the subsurface. Groundwater abstraction in carbonate aquifers (e.g. chalk) is highly dependent on fractures, which often have limited connectivity and extent. The productivities of boreholes drilled in chalk are improved by introduction of hydrochloric acid under pressure, which enhances weaknesses in the rock and promotes fracture development (Banks et al. 1993).
Structural disturbance of tunnels and caverns
When a shaft, tunnel or storage facility is excavated into a rock mass it produces an Excavation Damaged Zone (EDZ), reflecting stress redistribution around the opening, damage by the excavation process and weathering associated with rock-groundwater interaction (Kelsall et al. 1983). The EDZ includes fracturing of originally intact rock, opening or closing of pre-existing fractures or loosening of crystal structures, particularly in salt. Blasting disturbs the walls of an excavation more than mechanical processes, but typically only to a depth of 1 m from the face (Fig. 4). In fractured rock, Kelsall et al. (1983) show the EDZ resulting from stress relief typically ranges from 0.3 to 0.7 times the excavation radius. The fracture distribution around the tunnel is similar in geometry to that generated by much smaller-diameter boreholes (Blümling et al. 2007). Damage in the sidewalls is generally limited to tensional circumferential cracks; compressive zones may be prone to slabbing or breakout notches if stresses are sufficiently high. More extensive subsidence results from collapse of the generated voids. Such collapse structures in natural systems, such as dissolution of salt or gypsum, commonly generate laterally extensive breccias.
Example of radial fracture pattern (c. 0.5 m in diameter) with central pulverized core (c. 50 mm diameter) associated with blasting during construction of a cutting in basalt, near Arteara, Gran Canaria (Spain). Photo taken by C.N. Waters.
Structural impacts from Geological Disposal Facility (GDF) construction
The EDZ extent is influenced by natural fractures or heterogeneities, the initial stress field, bulk material properties and the geometry of the excavation (Blümling et al. 2007). The resulting distribution and longevity of fractures will vary depending on context, but excavation activity in more plastic materials may not be preserved, while in brittle shales or crystalline rocks (Martino & Chandler 2004; Cai & Kaiser 2005) an EDZ may remain detectable on longer timescales.
Experiments in Underground Research Laboratories give insight into the physical damage generated during excavation of a GDF. Indurated argillaceous rocks at the Tournemire research site, southern France, described by De Windt et al. (1999), show natural sub-vertical fracturing including millimetre-scale microfissures and centimetre-scale fractures sealed with calcite sourced from adjacent shales, and decimetre- to metre-scale faults, including fault breccia. Tunnel construction generated additional millimetre- to centimetre-scale fractures within a 2 m-thick EDZ. These induced fractures lack calcite sealing or pyrite alteration, though currently show macroscopic gypsum crystal growth on fracture planes.
Observations at the HADES test facility, Belgium, indicate the potential transient nature of anthropogenic fractures in more plastic materials. Excavation of horizontal galleries through clays generates large shear planes, with a geometry reflecting stress redistribution (Mertens et al. 2004). Here, the clay tends to self-seal, with a reduction in fracture transmissivity through time (Blümling et al. 2007).
Deformation associated with subsurface solid mineral workings
Collapse of mine workings can deform adjacent bedrock, the first report of induced seismicity due to coal mining dating to 1738 at South Stanford Colliery, England (Li et al. 2007). Deformation of the overlying rock mass is dependent upon the mineral extraction technique used (Waters et al. 1996; Table 1). At very shallow depths (<10 m), extraction was historically undertaken using bell pits. This involved sinking many narrow (c. 1 m) shafts and working the coal for up to 20 m diameter, after which the roof would become unstable, producing a crown-hole or subsidence hollow (Edmonds 1988, Fig. 5). The pillar-and-stall method allowed minerals to be extracted (notably coal) to greater depths of c. 30 m by driving a grid of narrow roadways through the mineral seam, leaving pillars to support the roof (Figs 5 and 6). In West Yorkshire, UK, such workings extended up to 200 m from the shaft, with room sizes up to 5 m and extraction ratios of <70% (Bell 1978). The pillars can support open voids for many decades after working stops. However, pillar failure through ‘robbing’ on retreat from the workings, or subsequent loading, results in roof collapse (Figs 5 and 6), which can reach the ground surface, as a crown-hole (Edmonds 1988). The remaining grid-like pattern of coal and intervening stowed waste is uniquely an anthropogenic signature (Fig. 6). To mitigate the risk of surface subsidence, voids can be grout-filled (Fig. 3). Longwall mining, which involves the void roof being temporarily supported by props that are moved forwards as the face is advanced, permits the total extraction of coal from a panel (Fig. 5). The Commission on Energy and the Environment (1981) detailed the method, which originated in Shropshire in the seventeenth century and, in the UK, typically ranged to 100–1000 m deep in panels 50–250 m long. The process produces a zone of sagging, fracturing and general subsidence within days after coal extraction, with residual subsidence completed after about two years. The zone of disturbance has an ‘angle of draw’ extending c. 35° outwards from the workings to the surface (Bowell et al. 1999, Fig. 5). The worked-out coal seam can often be recognized as chaotic beds of broken shale and coal fragments where the coal was formerly present.
Schematic cross-section showing the types of subsidence associated with different coal mining methods (modified from Waters et al. 1996).
View of a former pillar-and-stall workings for coal, subsequently re-excavated. The coal seam, seen at the base of the backwall, forms pillars present as the dark areas of the excavation floor. The infilled waste materials, mainly shale, form the paler areas between the pillars. A rubble-filled void in the backwall shows the development of a choked chimney. Photo taken by C.N. Waters (BGS © NERC).
Analysis of mining-induced seismicity in China by Li et al. (2007) shows that fracture initiation, propagation and rock mass movement typically occurs along pre-existing fracture planes. Here, rockbursts occur at >200 m depth, becoming increasingly common with greater mining depths. Fractures develop in the roof rock above the mined area in response to flexure under gravity load and these fractures may propagate to the surface. The floor deforms and fractures during mining, resulting in local floor heave (Fig. 5), which can also induce seismicity typically within 1000 m depth below the active mining. Underground coal fires (accidental or to extract coal gas) can also cause roof collapse and surface subsidence, the adjacent shales typically being reddened, with most organic carbon content lost. Whilst induced seismicity may result from mining activity, it seems unlikely that future geologists would be able to differentiate between slip resulting from natural and human-induced events. Although such modern structures are associated with brittle and open fractures and loose collapse debris, with time the fractures will become mineralized and the debris will form breccias.
The development of chimneys of collapsed material or more regional sagging due to subsurface extraction of coal, either through mining or UCG, can have natural analogues in lithologies prone to dissolution, e.g. gypsum or limestone, in which dissolution creates the voids that result in comparable collapse structures (e.g. Edmonds 1988). Although natural subsidence hollows may look superficially similar to these crown-holes, development of such structures in dominantly siliciclastic successions associated with coal workings indicates an anthropogenic origin.
Where excavated evaporite workings are abandoned, within months they reseal as the surrounding salt flows into the cavity, e.g. Boulby Potash mine (Yorkshire, UK). In such circumstances, it may be difficult to detect prior mining activities unless there is a corresponding surface expression, though entombed artefacts will commonly be well preserved.
Structural effects of subsurface weapon detonations
Subsurface detonation of nuclear devices produces very high temperatures and pressures generated over a fraction of a second. This causes the nuclear device and associated hardware and the rock mass adjacent to the charge to be vaporized or melted; a strong shockwave travels outwards, crushing and fracturing nearby rock (US Congress 1989; Fig. 7a).
Schematic illustration of the development of cavities, fractures and collapse structures as a consequence of a confined underground nuclear weapon detonation. (a, b) In profile showing (a) initial maintenance of the explosion cavity until the main collapse occurs after several seconds, culminating in (b) surface subsidence (both modified from Houser 1969). (c) In plan showing radial, concentric and linear fracture patterns (from Barosh 1968).
The effects of underground detonations reflect explosion depth, the yield of the device and local lithology (McEwan 1988). Uncontained subsurface explosions at shallow depths will penetrate to the ground surface (Hawkins & Wohletz 1996), producing a conical crater, tens to hundreds of metres in diameter and depth, surrounded by ejecta. After the explosion, the central cavity may collapse to form a debris chimney, which if it reaches the ground surface forms a bowl-shaped subsidence crater that may be tens of metres to nearly a kilometre wide and several tens of metres deep (Fig. 7b). Typically, the chimney diameter is 10–20% greater than the cavity diameter (McEwan 1988). A broader area of subsidence can extend some hundreds of metres from the test site, with local reactivation of faults (Hawkins & Wohletz 1996).
A detonation at greater depths may be evident at the surface as a shallow crater, a mound or by consolidation of surface strata (McEwan 1988). In deep constrained detonations there are four commonly recognized zones (Fig. 7a) (Adushkin & Spivak 1994; Hawkins & Wohletz 1996), with the radius (r) of detonation effects in rock, scaled to the explosive yield of the detonation in kilotons: (a) the cavity, potentially floored by molten rock with a void radius of r = 4–12 m/kton1/3; (b) the crushed zone surrounding the cavity in which the rock mass has lost all of its former integrity and with r = 30–40 m/kton1/3; (c) the cracked zone in which the rock mass contains radial and concentric fractures (Fig. 7c), with r = 80–120 m/kton1/3; and (d) the zone of irreversible strain with deformation modifying porosity/permeability and material strength, with r = 800–1100 m/kton1/3.
In the USA, Shoemaker (1959) directly compared the Teapot Ess Crater (Nevada), generated by an uncontained subsurface detonation of a 1.2 kton nuclear device 20 m below ground surface, and the Meteor Crater (Arizona), formed by a bolide impact c. 50 000 years ago (Fig. 8). The impact origin of the bolide structure is demonstrated by the ‘high-shock’ mineral coesite within the crater (Chao et al. 1960). However, the generation of similar minerals is expected in the case of nuclear explosion and Shoemaker (1959) noted that nearly all the major structural features at Meteor Crater (Fig. 8a), are reproduced at Teapot Ess (Fig. 8b). Comparable structures include inverted bedrock stratigraphy within debris ranging from micron to >30 m diameter, in turn resting on older disturbed strata dipping at moderate to steep angles in the wall of the crater and locally overturned near the debris contact. A further crater nearby (Jangle U, Fig. 8c), created by a nuclear device of the same yield but at shallower depths of c. 5 m, is structurally distinct, having an anticline rim, as opposed to a syncline seen at Teapot Ess, though an anticlinal rim was also recognized at a shallow meteorite impact site near Odessa (Texas) (Shoemaker 1959).
Comparison of crater structures associated with (a) meteorite impact at Meteor Crater, (b) comparatively deep unconfined nuclear detonation at Teapot Ess and (c) comparatively shallow unconfined nuclear detonation at Jangle U (modified from Shoemaker 1959).
Chemical alteration of the subsurface as a distinctive anthropogenic feature
Anthropogenic geochemical or mineralogical modification of the subsurface rock mass can result from: (a) the migration of leachates from wastes within a void; (b) the interaction of wastes with any containing barrier; (c) the alteration of construction materials used to support and line voids; or (d) injection of novel minerals or mineral assemblages directly into the rock mass. The leachates include dissolved minerals that may precipitate within fractures and pore spaces within the rock mass.
In Table 3 the effects of geochemical or mineralogical modification are summarized for a number of important construction materials and settings, including: bricks; concrete; drilling fluids and infrastructure associated with boreholes and wells; rockmass adjacent to a GDF; rockmass part of a CCS site; mining leachate; and rocks and soils adjacent to zones of atomic weapon detonation.
Chemical and mineralogical changes brought about by anthropogenic modification of the subsurface
Deterioration of brick-built structures
Bricks are commonly used in deep basements and older tunnels, their durability reflecting their composition and the processing of raw materials. Brick-clays typically comprise quartz, clay minerals (illite in combination with kaolinite, chlorite or smectite), and variably subordinate feldspars, carbonates, Fe-bearing minerals and organic materials (Dunham 1992). Fired bricks are heated to crystallize high-temperature mineral phases. The length and temperature of firing determines brick durability, strength, subsequent water absorption and hence expansion and deterioration, with bricks fired at low temperatures being least durable (Hughes & Bargh 1982). However, fired bricks rich in sodium and potassium or illite-rich clay tend to expand when wet and the presence of calcite, dolomite, pyrite and siderite is associated with development of cracks, spalls and pops (Hughes & Bargh 1982). Bricks can suffer salt deterioration resulting in efflorescence, with resultant volume increases causing degradation (Sena da Fonseca et al. 2013). Lime present in calcareous fired bricks can modify to portlandite in the presence of water, causing volume increase and ‘lime blowing’ (Elert et al. 2003).
The oldest fired bricks from Babylonia date to around 4000 BCE (Rapp 2009). This suggests, despite their progressive deterioration, recognition of bricks as anthropogenic structures is likely to persist for many millennia into the future. However, in contrast to slow thermal processes attained naturally, brick clays are heated rapidly and for short durations and may not contain geologically stable mineral assemblages (Dunham 1992) that could ensure preservation in their current state over millions of years.
Mineralization and chemical alteration of concrete
Concrete, commonly used in modern shallow (e.g. spread footing) and deep (e.g. piled) foundations, to line and protect excavations such as tunnels, and as a grout to infill voids, may decompose to affect the host rocks. Boreholes differ in this respect and consequently are described later.
The main phases present in hydrated Portland cement are roughly micron-scale particles of portlandite (Ca(OH)2), Ca-silicate hydrates (e.g. jennite and tobermorite), Ca-aluminate hydrates and Ca-sulpho-aluminate hydrates, e.g. ettringite (Gherardi et al. 2012; Rochelle & Milodowski 2013; Waters & Zalasiewicz 2018).
Construction of subsurface structures can modify and divert groundwater flow (Attard et al. 2016) with geochemical impacts. For example, carbonate-rich water derived directly from leaky water mains or the leaching of lime mortars and cement-grouting (‘concrete carbonation’) results in hyperalkaline groundwater formation (pH >10). During carbonation, phases react (Table 3) to form CaCO3 (calcite, aragonite or vaterite depending on the degree of supersaturation), SiO2 and gibbsite (Gherardi et al. 2012; Rochelle & Milodowski 2013), initially decreasing porosity. After more than about 100 years porosity increases due to loss of primary cement phases and re-dissolution of secondary minerals, e.g. zeolites (Gherardi et al. 2012). Ultimately, on a millennial timescale, the altered cement is a carbonated, low-porosity layer comprising a mineralogy (e.g. calcite, aragonite, quartz, illite, dawsonite) that is naturally common (Waters & Zalasiewicz 2018). This suggests that far-future recognition of an anthropogenic source would be difficult mineralogically, although macroscopic structural fabrics may persist, and novel geochemical or isotopic fingerprints may still be observable long into the future.
Rapidly growing calcium carbonate speleothems have been observed in domestic cellars (Sundqvist et al. 2005), and both abandoned and active railway tunnels (Field et al. 2016), underground aqueducts (Pons-Branchu et al. 2015) and dam site tunnels (Liu & He 1998), but do not extend beyond the void.
Natural mineral systems similar to Portland cement are very rare, limited to low-temperature serpentinization of ultrabasic rocks or retrograde alteration of marble, impure limestones, calcareous shales and carbonaceous sediments such as coal within high temperature–low pressure metamorphic aureoles around igneous intrusions (Alexander 1992). Natural Ca silicate hydrate phases show similar reactions as cement/concrete, but over approximately 10 000-year timescales, involving conversion to CaCO3 and SiO2 and formation of a well-defined reaction front associated with localized Ca migration, increase in matrix porosity and cracking (Milodowski et al. 2009; Rochelle & Milodowski 2013). A close analogue at Maqarin, Jordan, involves bituminous biomicrites and calcareous mudstones heated by combusting bitumen in a hyperalkaline groundwater system (pH 12.5). It forms a plume c. 500 m from the ‘cement zone’ (Savage 2011), buffered by naturally occurring cement minerals, mainly portlandite but also ettringite and tobermorite (Khoury et al. 1992). CO2, and H2O release causes brecciation of rocks overlying the area of combustion, possibly distinguishing this natural system from most anthropogenic cement sources.
Chemical and mineralogical changes associated specifically with boreholes
Boreholes can include complex associations of clay or cement grouts, steel or plastic casing, with injection of drilling muds and fracturing fluids into adjacent rocks and soils extending the geochemical fingerprint of drilling activity well beyond the borehole itself.
Clays, mainly bentonite, or cement grouts are widely used in the construction, geotechnical and water well industries to plug boreholes (Daemen 1996). When pumped down a wellbore to seal it, cement filtrates containing dissolved CaCO3 and sulphate can be forced at high differential pressure into, and re-precipitate within, the surrounding rocks (Hodgkinson & Hughes 1999). Portland-type cements (described above) have been widely used for <150 years, so there is uncertainty as to the long-term alteration of boreholes cement plugs (Rochelle & Milodowski 2013).
Drilling boreholes involves pumping bentonite-rich ‘drilling muds’ down the borehole. In chalk rocks, Hilbrecht & Meyer (1989) recorded injectites of drilling mud and rock cuttings that superficially appear similar to natural marl and solution seams, but may be distinguished by the presence of flow structures, abrasion of cuttings, wedge-shape geometries with sharp tops and bases locally cross-cutting sedimentary bedding and the absence of trace fossils, stylolites and concretions. Drilling muds may also contain water-emulsifying, suspending and filtration-control agents, clay-barite suspensions, salts (sodium chloride and calcium chloride), various detergents, flocculants, organic polymers and bentonite (Caenn et al. 2011) that may leave a diagnostic geochemical signal. Extraction of conventional oil can also be enhanced through the injection of NaOH/KOH-type alkaline fluids (pH 12–13) (Savage 2011). Such drilling muds and solutions may temporarily leak into formation waters, but because of the potential damage to aquifers and the cost of lost drilling materials, such penetration into subsurface strata is typically kept to a minimum.
A typical hydrocarbon well on completion comprises a steel casing, separated from surrounding rock by a cement grout, and a cement plug sealing. The steel can vary in the type and degree of corrosion (Enning & Garrelfs 2014): (a) oxygen and moisture near the ground surface will oxidize the steel casing with time; (b) absence of oxygen and pH >6 – the corrosion of iron is insignificant and is expected to last for centuries; (c) CO2-rich, O2-poor (anticipated during concrete alteration) – siderite (FeCO3) coating initially forms on the casing; (d) anoxic conditions (expected in oil and gas pipelines) – sulphate-reducing bacteria can significantly accelerate corrosion (anaerobic biocorrosion), resulting in iron sulphide precipitation.
Plastic casing/liners are common in monitoring wells. Chemical degradation of plastic polymers at depth in geological strata is poorly known, but may result from alteration of molecular bonds driven by heat or hydrolysis at very high or very low pH (Zalasiewicz et al. 2016). Ultimately over geological timescales plastics will likely alter to hydrocarbons, expelled as oil and gas and leaving a carbonized film on a mould of the plastic artefact.
High-viscosity fracturing fluids associated with fracking (Ferrer & Thurman 2015) typically comprise polymer-based gels and chemical additives, including gelling agents (Guar/xantham gum or hydroxyethyl cellulose), crosslinkers (borate salts), breakers (ammonium persulphate, magnesium peroxide), friction reducers (polyacrylamide, petroleum distillate), surfactants (ethanol, isopropyl alcohol, 2-butoxyethanol), biocides (glutaraldehyde, 2,2-dibromo-3-nitrilopropionamide (DBNPA)), scale inhibitors (ethylene glycol), corrosion inhibitors (isopropanol, acetaldehyde) and oxygen scavengers (ammonium bisulphite). The organic polymers and chemicals may have only short to intermediate residence on grain surfaces or as a filtrate in shales, whereas proppants used to keep the fractures open (Fig. 9) will leave permanent injectite fabrics (Legarth et al. 2005). Superficially, the injectite structures are comparable to natural neptunean dykes, but the presence of the proppants, typically synthetic ceramic spheres, or natural sand of provenance that cannot be traced back to a source within the depositional basin, indicates an anthropogenic origin. Short-term, artificial hydraulic fractures lack the calcite cement seen commonly in natural fractures, though subsequent mobility of Portland cement from the well will ultimately produce a likely source of calcite fill of the hydraulic fractures.
Schematic view of secondary effects associated with hydraulic fracture that may be observed as being of anthropogenic origin (from Legarth et al. 2005).
Over-pressurization during UCG can drive a plume of benzene, volatile organic carbons and inorganic contaminants, leached from the residual ash, into regional aquifers (Burton et al. 2013). It is unclear if this would produce a geochemical signal that would differ from natural combustion of coal.
Deep (>500 m) ‘closed’ groundwater systems, with very slow groundwater flow and limited or no connection to the surface and potable water aquifers (Hickey 1989), have been used since the 1950s for hazardous waste disposal through boreholes, e.g. radioactive, toxic or other heavily contaminated substances. Pressure changes mean that contaminated formation fluids can penetrate other parts of the groundwater system (Lesage et al. 1991).
Temporal chemical and mineralogical changes to a Geological Disposal Facility
Globally, many GDF schemes are under construction or proposed, but completed underground storage of radioactive wastes is limited to Asse II and Morsleben former salt mines (Germany) and near Carlsbad (New Mexico, USA) and active sites to Onkalo (Finland) and Forsmark (Sweden). The intention is for radwaste to be ‘contained and isolated’ until it has similar radioactivity to naturally occurring materials, generally within hundreds of thousands of years. Repository evolution is therefore generally considered over c. 1 million-year timescales. Existing and proposed repository designs show considerable variability, so long-term evolution is likely to differ substantially between sites and be highly dependent on the choice of barriers, host rock, repository geometry, thermal footprint, groundwater behaviour and type of waste. However, there has been much study of the long-term progression of specific processes, including evidence derived from natural analogues.
In most cases, the radwaste would be stored in cement or vitrified form and associated containment structures would be recognizably anthropogenic in origin. Alexander & McKinley (1999) describe various alteration scenarios. To slow devitrification in the Swedish disposal concept, the waste is contained in copper canisters filled with lead and a surrounding bentonite buffer, with in situ corrosive penetration timescales of 10−100 million years. In Switzerland, more corrosive higher sulphate groundwaters and greater pressures necessitated a stainless steel flask (a redox buffer) surrounded by bentonite (a pH buffer) to ensure low radionuclide solubility. Alteration of the bentonite barrier, mainly montmorillonite with minor calcite, pyrite and siderite, progresses very slowly, with significant alteration to illite not expected within the first million years and complete alteration to take 10−100 million years. The steel canisters, by contrast, are likely to have a lifetime of 10 000 years. Devitrification/dissolution of the glass is expected to take c. 100 000−10 million years after canister failure and exposure to groundwater, yielding radionuclide-bearing colloids.
Fluids produced by cement dissolution (section ‘Mineralization and chemical alteration of concrete’) could migrate into the host geology as a ‘hyperalkaline plume’ or ‘alkaline-disturbed zone’ (Berry et al. 1999). This would result in significant dissolution, mainly of aluminosilicates, followed by precipitation of amorphous calcium aluminosilicate hydrate phases crystallizing as zeolites and feldspars as the final product of mineralization (Eikenberg & Lichtner 1992; Savage & Rochelle 1993; Hodgkinson & Hughes 1999; Moyce et al. 2014; Milodowski et al. 2015). Modelled over 50 ka, the zone of alteration can extend up to 3 km from the site (Eikenberg & Lichtner 1992).
A GDF will be associated with a thermal footprint generated by the radwaste; repository design is optimized to achieve a ‘desired’ temperature. Clay mineral stability at elevated temperatures is of particular importance. Above 100°C, smectite fully alters to illite over timescales of <1 million years, while high pH conditions associated with a repository containing cementitious materials could potentially accelerate the transformation, even at lower temperatures (Horseman & McEwen 1996). Weaver (1979) considered that illitization could begin within a few years at temperatures as low as 40°C in saturated conditions and moderate (50–100°C) temperature increases, particularly in the presence of organic material, and will cause kaolinite to form in sandy beds. Chlorite can form by the alteration of other clay minerals at temperatures as low as 70°C and at c. 200°C chlorite may grow in fractures and voids. However, if shales have been subjected to temperatures of 200–300°C for millions of years, as with many Lower Paleozoic rocks, the heat generated by radwaste is unlikely to cause any major chemical or mineralogical change.
To understand the potential for migration, should radionuclides become mobile, there has been research on natural nuclear fission reactors, e.g. the uranium-rich 2.0 Ga old site at Oklo, Gabon described by Gauthier-Lafaye et al. (1996). Here, 90% of the uranium remained in situ after criticality (the point at which a nuclear reaction is self-sustaining), with limited dissolution and chemical mobility of some elements (most actinides, some rare earth elements (REE), Y and Zr). Uranium mineralization, associated with calcite, organic matter and sulphides, occurs in dense networks of bedding-parallel microfractures. During criticality the intense hydrothermal alteration of the host sandstone extended c. 50 m beyond the centre of the natural reactor, with replacement of the sandstone by newly crystallized Mg-chlorite, Al-chlorite and illite. The resultant volume decrease caused a collapse of the host rock both above and below the natural reactor and the associated fractures acted as pathways for the hydrothermal fluids. The study of anthropogenic nuclear accident sites such as Hanford (USA), Chernobyl (Ukraine) and Fukishima (Japan) indicates that some elements have mobilized, e.g. at Hanford, caesium is estimated to have migrated up to 40 m below the storage tanks (McKinley et al. 2001; Serne et al. 2001).
Chemical alteration associated with CCS gas sites
CCS technology involves injection of CO2 into porous and permeable strata at depths likely to be greater than c. 800 m, resulting in a phase change to a supercritical fluid with <0.3% of the volume of the gaseous form (Rochelle et al. 1999). Once injected, trapping of CO2 is expected via four primary mechanisms (IPCC 2005; Fig. 10 and Table 3): (a) immediately after injection, the CO2 will be stored as a free phase within the host rock through structural traps, e.g. within an anticline; (b) CO2 dissolution into the local formation water initiates a variety of geochemical reactions (Rochelle et al. 2004) – simulations suggest that a supercritical CO2 bubble develops in a 5 m radius around a well after 10 years of injection, surrounded by a two-phase zone of c. 650 m radius (André et al. 2007); (c) residual trapping along the migration path of CO2; and (d) mineral trapping, caused by interaction of supercritical CO2 with the reservoir rock, may provide a permanent CO2 sink and can lead to partial dissolution of feldspars, dolomite and anhydrite and precipitation of clays and calcite (Rochelle et al. 1999, 2004).
(a) Components of the CCS storage system showing interaction of the CO2 plume with formation waters and the CO2 bubble with the host and cap rock (schematic cross-section modified from Rochelle et al. 2004). (b) Trapping mechanisms of CO2 with time (IPCC 2005).
The Bravo Dome CO2 field, New Mexico, provides a natural analogue in which CO2 is considered to have degassed from magma entering the reservoir <50 ka ago (Rochelle et al. 2004), with resultant dissolution of anhydrite, gypsum, dolomite, K-feldspar and plagioclase and late-stage precipitation of kaolinite, zeolites and gibbsite, particularly along fault planes (Pearce et al. 1996). For rocks with saline pore fluids, dawsonite can be an important secondary phase, along with quartz and occasionally Ca−Mg−Fe carbonates, e.g. the Bowden–Gunnedah–Sydney, eastern Australia (Baker et al. 1995).
CCS sites will exhibit many of the features present in naturally occurring CO2 accumulations. CO2-rich gas discoveries at the Fizzy Horst, Southern North Sea, demonstrate the potential for CO2 to be contained in the subsurface over long timescales (c. 50 Ma) (Underhill et al. 2015). Within the first few hundred years, it may be possible to differentiate human-made storage sites from natural accumulations, simply by detection of the ongoing pressure-pulse generated by CO2 injection (Noy et al. 2012). On longer timescales, isotopic signatures may indicate an anthropogenic origin. For example, for Fizzy Horst, Heinemann et al. (2013) used carbon and oxygen isotopic compositions to differentiate between dolomite generated before and after the CO2-rich gas was introduced to the reservoir, the novel dolomite being generated by the reaction of CO2-bearing brines and silicate minerals. In Fizzy Horst the carbonate mineral dawsonite, distinctively a product of high CO2 concentrations, has formed in only trace amounts (0.4 ± 0.3% solid volume) even after >50 Ma (Wilkinson et al. 2009). Hence, the majority of injected CO2 is unlikely to be stored by chemical mineralization (Wilkinson et al. 2009; Fig. 10b). On intermediate timescales (c. 1 ka), anthropogenic CO2 will be most apparent by its physical presence, either as a distinct phase, in solution or as a result of potential leakage to the seafloor.
Cold-water geysers at Green River (Utah), provide a natural analogue of the potential long-term fate of anthropogenic CO2 stored in geological reservoirs migrating to the surface (Assayag et al. 2009), potentially evident in the form of chimneys. The geysers show helium and carbon isotopic ratios suggesting that the CO2 is derived from crustal sources. Trace impurities within captured CO2, such as SO2 and NOx, could indicate an anthropogenic source.
Leachates from underground mining activity
Acid mine drainage (AMD) is highly aggressive to native rocks (Simate & Ndlovu 2014). A study by Bowell et al. (1999) showed that underground metal mine workings typically contain water with high Total Dissolved Solids >1000 mg l−1, and in some coal mines sulphate is commonly >2000 mg l−1. Water contaminated by natural interaction with minerals can have widely variable pH. The Levant mine, Cornwall, has acidic groundwater, pH 2–4, and high levels of Fe, Al, Cu, Zn and SO42–, whereas in Yerrington pit, Nevada, mine water is neutral and has low metal content despite reacting with sulphide-bearing host rocks.
AMD in metal or coal mines, reviewed by Akcil & Koldas (2006), is caused by mining introducing an oxidizing environment, with oxidation of FeS2 generating both the acidity of the mine waters and the supply of large quantities of Fe and sulphate-forming ochre precipitates, but relatively low concentrations of toxic heavy metals. Where neutralization occurs, metals and sulphates are precipitated forming a range of minerals dominated by iron oxyhydroxides, oxysulphates such as ferrihydrite (Fe(OH)3), goethite (FeOOH) and iron sulphate polymorphs. AMD can generate very soon after mining commences and persist long after the operations cease (potentially thousands of years if unremediated). AMD can migrate extremely quickly through flooded mine workings and, on intermediate timescales, migrate through natural, or mining-induced fractures and disconformities. In an underground mine, oxidation can be reduced by keeping the former workings flooded, with isolation of pollution sources through the use of plugs (bulkheads) to reduce the flow of water through parts of the mine where contamination is high.
Mineralogical expression of nuclear detonations
Nuclear testing has produced a durable, but localized lithological expression through the conversion of sand into a glass-like substance known in the USA as ‘trinitite’ (Eby et al. 2010) and in Kazakhstan as ‘kharitonchik’. Trinitite is the complex product of surface detonations, and the glass may be vesicular, similar to natural volcanic scoria, but showing compositional heterogeneity at a scale of tens to hundreds of micrometres, with unmelted quartz grains, parts of the nuclear device present as explosion-generated metallic chondrules and radionuclides that are distinctively anthropogenic (Eby et al. 2010). Trinitite from surface explosions generates glass beads similar to tektites produced during meteorite impact explosions (Eby et al. 2010). Nickel–iron crystalline debris and minute spherules, typical of a meteorite impact (Shoemaker 1959), contrast with the products of nuclear detonations described above. Radioactive gases may also diffuse through pore spaces and unsealed fractures of the overlying rock in underground explosions, producing a marked radiogenic signal. However, a more immediately apparent marker resulting from nuclear activities is the isotopic ‘bomb-spike’ signature deposited from both local and global fallout, reviewed in the context as a potential marker for the Anthropocene by Waters et al. (2015).
Conclusions
Many of the physical and chemical products of human subsurface intrusion either do not extend far from the source of intrusion, lack long-term persistence as a signal (Fig. 11) or are not sufficiently distinctive from the products of natural processes to make them uniquely recognizable as of anthropogenic origin. But the scope and complexity of the signals have increased greatly over recent decades, both in areal extent and with increasing depths, and seem set to be a fundamental component of our technological expansion. There will be some clues to the geologist of the far-future, when historical knowledge records may not be preserved, that will help resolve the origin.
Infrastructure associated with subsurface developments over the short term will leave a clear expression of anthropogenic intrusion. Over geological timescales of millennia and longer, the anthropogenic materials associated with such infrastructure will alter diagenetically to minerals that may be common in nature, but the physical form of the intrusion will likely be fossilized as a permanent signal.
Chemical alteration of bricks is not expected to extend beyond the immediate vicinity of the structures and their likely durability for millions of years into the future is both untested and unmodelled.
Cement and concrete, when initially introduced into the subsurface in tunnel or cavern linings or as a grout in boreholes, comprise phases which are relatively rare in nature; where found naturally the presence of brecciation above deposits indicates a natural origin. In the presence of acidic groundwaters, cement and concrete will produce contaminant plumes that may migrate in groundwaters some distances from the source, eventually within hundreds or thousands of years reacting to form naturally common minerals, in which the anthropogenic origin may be determined through complex geochemical or isotopic signatures.
Steel may, in favourable conditions, persist for centuries, but over geological timescales will alter to iron oxides, carbonates or sulphides depending upon the redox environment, potentially fossilizing the infrastructure morphology.
Rates of chemical degradation of plastic polymers in the subsurface is poorly known, but on geological timescales will likely leave a carbonized film on a mould of the plastic artefact.
The generation of crush zones and radial and concentric fractures associated with subsurface detonations (Fig. 4), at sub-metre-scale as part of the early phase of fracking or tunnel excavation, and at the hundreds of metres-scale through deep subsurface nuclear explosions (Fig. 7), has no direct natural analogue. However, shallow subsurface nuclear detonations produce craters similar in geometry to bolide impacts (Fig. 8) and both can contain the mineral coesite. Glass associated with subsurface nuclear detonations will be recognizable through extremely small-scale compositional heterogeneity and incorporation of vaporized metals from the site infrastructure.
Drilling disturbance adjacent to boreholes is typically less than the diameter of the borehole itself (Fig. 3), whereas excavation of shafts, tunnels or underground storage facilities can generate Excavation Damaged Zones with fracturing induced up to 1.4 times the void diameter. Such mechanical processes do not greatly extend the lateral influence of the bores or larger excavations (Fig. 11), but chemically and mineralogically distinctive drilling muds may penetrate a short distance into the surrounding bedrock.
Hydraulic fracturing induced as part of enhanced geothermal and hydrocarbon extraction operations can produce fracture pipes or networks on a scale of hundreds of metres. Though, superficially similar to natural structures, discrimination will be possible on short to intermediate timescales through organic chemicals used to extract hydrocarbons, or permanently through the presence of proppants injected to keep fractures open.
The origin of collapse structures would be mainly resolvable by the nature of the host rocks: subsidence in mainly siliciclastic rocks with coal seams or mineral veins is likely to be associated with anthropogenic void generation and collapse; collapse structures in carbonate- and evaporate-rich successions could be equivocal in origin, but a natural source would be assumed unless infrastructure associated with mining or brine dissolution was present.
The chemically disturbed zone around a Geological Disposal Facility could extend several hundreds of metres, and the hyperalkaline plume well beyond that. Thermal alteration of the host rock around a repository, in which temperatures associated with high level waste are unlikely to exceed 100°C and low and intermediate wastes a few tens of degrees above ambient temperatures, may be very difficult to distinguish from mineral and chemical alterations resulting from natural diagenesis and low grade metamorphism of mudrocks. The most distinctive feature will be the vitrified radwaste and containment structures.
Long-term containment of CO2 in rocks can result in the interaction with the reservoir rock, leading to partial dissolution of some minerals and precipitation of others. These features can be seen in naturally occurring CO2 accumulations, though carbon and oxygen isotopic composition and SO2 and NOx impurities may indicate an anthropogenic origin. However, differentiation from natural accumulations will be most clearly made by the discovery of infrastructure related to injection activities, such as boreholes.
Subsurface mineral extraction can be associated with groundwaters rich in dissolved metals and sulphates, and acid mine drainage – with precipitates of iron hydroxides and sulphates – may not clearly be recognizable as being of anthropogenic origin when found in rocks, even if AMD precipitates can be a distinctive indicator of oxidizing environment in mineral workings when seen in surface springs.
Maximum persistence of key subsurface physical and chemical signals and their likely maximum distance of migration beyond the source anthropogenic feature. CCS, Carbon Capture and Storage; EDZ, Excavation Damaged Zone; GDF, Geological Disposal Facility.
Acknowledgements
All authors, with the exception of JZ, publish with the permission of the Executive Director, British Geological Survey, Natural Environment Research Council.
Funding
This work was funded by the Natural Environment Research Council (NEE4586S),with the support of the British Geological Survey's Engineering Geology science programme.
Scientific editing by Jamie Standing; Richard Ghail
- © 2018 UKRI. British Geological Survey. Published by The Geological Society of London. All rights reserved
References
Recognizing anthropogenic modification of the subsurface in the geological record
