Introduction

Large hydraulic head gradients observed across fault zones in unconsolidated sedimentary aquifers suggest that they often form effective barriers to lateral groundwater flow (Rawling et al. 2001; Bense and Van Balen 2004; Mayer et al. 2007; Bense et al. 2013). These barriers are created by the juxtaposition of aquifer-confining units such as clay beds with aquifer media at the location of the fault (Mailloux et al. 1999) and/or fault zone deformation processes such as cataclasis (Fulljames et al. 1997), diagenesis (Chan et al. 2000; Dewhurst and Jones 2003) and clay smearing (Lehner and Pilaar 1997; Bense et al. 2003; Egholm et al. 2008), which reduce the permeability of the fault zone itself. The role of faults as barriers to groundwater movement has been widely described in the literature but much less documented is the ability of faults to simultaneously act as conduits for sub-vertical flow along the fault (Roberts et al. 1996; Wiprut and Zoback 2000; Bense and Person 2006). Hydraulic head observations are typically not sufficient to infer the magnitude and direction of fluid flow along faults, usually due to the sparse distribution of observation boreholes. However, the application of noble gas tracers has been particularly useful in characterizing fluid flow associated with faults as conduit-barrier structures in both deep (Wiersberg and Erzinger 2011) and shallow (Kulongoski et al. 2003, 2005) settings. In general, the potential for fluid migration along faults is an important consideration for water resources management as well as proposed carbon capture and storage (CCS) programmes (Shipton et al. 2004; Bickle et al. 2007; Gilfillan et al. 2009), shale gas recovery from deep hydrological environments and its implications for shallow groundwater quality (Cassiat et al. 2013), and nuclear waste storage sites (Flint et al. 2001; Dublyansky and Spötl 2010) where faults could act as potential leakage pathways for hazardous fluids.

Geochemical tracers are useful tools for investigating the origin and residence time of deep circulating fluids in aquifer systems worldwide (Kipfer et al. 2002; Lippmann et al. 2005; Ma et al. 2005). Dissolved noble gases and stable isotopes are particularly useful in studies of aquifer systems that contain pre-Holocene ‘palaeo’ groundwater (e.g. Vaikmäe et al. 2001) or fluids that have a deep crustal origin such as those observed in the Great Artesian Basin in Australia (Torgersen and Clarke 1985, 1987; Bethke et al. 1999), the Paris Basin in France (Marty et al. 1993, 2003; Castro et al. 1998; Lavastre et al. 2010) and the Witwatersrand Basin in South Africa (Lippmann et al. 2003; Lippmann-Pipke et al. 2011). Dissolved helium has been used extensively for estimating groundwater residence times and tracing the origin of crustal fluids (Torgersen et al. 1992; Ballentine et al. 2002; Gardner et al. 2012). Deep circulating groundwater typically has a radiogenic source of 4He from in situ α-decay of U-Th series elements. However, modern groundwater (>10 years) can also exhibit high concentrations of helium which originate from previously trapped reservoirs in the subsurface (Solomon et al. 1996). Helium enrichment in palaeo groundwater is reported in the literature at up to six orders of magnitude higher than expected due to solubility equilibrium with the atmosphere (Andrews and Lee 1979; Castro et al. 1998; Bethke et al. 1999). Terrigenic 3,4He is also significant in deep circulating fluids (Griesshaber et al. 1992; Stute et al. 1992; Ballentine and Burnard 2002; Castro 2004). The 3He/4He isotope ratio is often used to distinguish between helium of crustal or mantle origin (Oxburgh et al. 1986; Sano et al. 1986; Kulongoski et al. 2003, 2005). However, due to groundwater typically containing helium from several different sources (Solomon et al. 1996; Zhou and Ballentine 2006), helium age dating is usually only used as a technique for estimating groundwater age rather than a precise quantitative tool (Stute et al. 1992; Castro et al. 2000). More recently, however, Torgersen (2010) suggested that it is now possible to determine uncertainty limits associated with crustal fluxes of helium using the increased number of published data sets, thus improving the accuracy of 4He age estimates.

In the unconsolidated sedimentary aquifers of the Lower Rhine Embayment (LRE), Germany, where large hydraulic gradients exist across many faults, suggesting that they act as barriers to groundwater flow, thermal anomalies in aquifer units flanking the Rurrand fault have provided evidence for significant fault-parallel flow (Bense et al. 2008). Here, hydrogeochemical evidence is presented for the upward migration of deep crustal fluids along the Bornheim fault in the LRE.

Study area

The Lower Rhine Embayment forms part of the Roer Valley Rift System (RVRS), which is the southward extension of the North Sea Basin and part of a Cenozoic mega-rift system that crosses western and central Europe (Ziegler 1994). The German region of the LRE (Fig. 1) covers an area of 3,800 km2 and contains approximately 1,300 m of Oligocene to Pleistocene unconsolidated siliciclastic sediments that form a highly complex multi-layered aquifer sequence within six tectonic blocks. The central Erft block is adjacent to the Köln block to the east and the Rur block to the west, whilst the Krefeld and Venlo blocks are situated northeast and northwest, respectively. Each tectonic block consists of 5–15 different aquifers to a depth of 200 m and the Erft block has as many as 21 separate aquifer units to a depth of approximately 400 m. The LRE is intersected by numerous NW–SE-striking fault zones which have a significant impact on regional groundwater flow patterns (Wallbraun 1992). The formation of the LRE aquifer system began during the Oligocene and is described in great detail by Schäfer et al. (1996). The LRE primarily consists of Oligocene, Miocene and Pliocene marine sediments and the more recent Pleistocene sediments are mainly fluvial deposits derived from the Rhenish Massif to the south (Schäfer et al. 2005). The sediments derived purely from the Rhenish Massif are comprised of gravels that are characterised by having >90 % quartz material in the coarse gravel fraction and the remaining sediment is made up of quartzite (Boenigk 2002). Highly porous loess soils have formed due to the accumulation of silts, sands and clays, and these overlay the Pleistocene deposits (Kemna 2008). Boenigk (2002) and Kemna (2008) provide detailed mineralogical analyses of the shallow Pleistocene deposits, and Pliocene and lower Pleistocene deposits, respectively. The LRE contains two commercially important lignite seams. The main lignite seam has a thickness of up to 100 m and the second ‘upper’ seam has a thickness of up to 40 m (Hager 1993). The Miocene lignite deposits of the LRE form one of the largest reserves in Europe with an estimated 55,000 Mt (Hager 1993) with an overburden thickness up to 300 m (RWE Power AG). The geological structures that underlie the Oligocene aquifers below a depth of 400–800 m are less well known. The oldest sequences identified from borehole records are Palaeozoic. It is known that Carboniferous strata with a total thickness of up to 2,500 m is underlain by approximately 2,000 m of clastic sedimentary rocks, limestone and evaporites, and a Devonian sequence of unknown thickness and composition (Geluk et al. 1994). Lower Jurassic sandstone and claystone sequences with a total thickness of up to 1,500 m are found beneath the Cenozoic sand deposits, and they overlie Permian-Triassic sandstone and conglomerate material (Geluk et al. 1994).

Fig. 1
figure 1

a Location of the study area in central Europe. b Hydraulic head distribution within the deeper aquifer units of the Lower Rhine Embayment (LRE), Germany, with spatial reference to the Hambach, Garzweiler and Inden open cast lignite mines. Contour increments vary due to the lower hydraulic head gradients that exist towards the River Rhine. A cone of depression is evident below the Hambach and Garzweiler mines, and the major graben faults frequently act as barriers to lateral groundwater flow. The shallow aquifer system is less impacted by lignite mine dewatering due to confining layers of clay that create relatively shallow perched aquifer conditions in many areas including the Brühl region. The observation boreholes not illustrated on the map are all situated to the north in an area less impacted by lignite mining

Full size image

Methods

Observation boreholes in the LRE were selected at locations in close proximity to faults, and where some historical monitoring records of well head parameters and hydrochemistry were available. Physical limitations of the pumping equipment restricted borehole selection and generally only shallow (<100 m) aquifer units were sampled. In total 73 piezometers were sampled from 64 observation boreholes during three separate sampling campaigns totalling 23 days in the field. However, only 35 piezometers were successfully sampled for the full range of noble gas, stable isotope and hydrochemical analyses. Boreholes were pumped using a Grundfos MP1 submersible pump and purged to the equivalent of three borehole volumes before groundwater samples were collected. All the observation boreholes were fully cased with filters positioned at specific inflow horizons. The depth of each borehole filter is given in Table 1. Groundwater samples were collected in copper tubes, using clear Portex™ connective tubing for noble gas (He, Ne, Ar, Kr and Xe) analyses following the method described by Beyerle et al. (2000) and back pressure was applied to maintain hydrostatic pressure during sampling using a simple ball valve attached to the outflow pipe of the copper tube. Generally, very few gas bubbles were encountered during sampling with the exception of observation borehole 958063 in the Brühl region, which exsolved large quantities of gas during sampling primarily due to a high dissolved CO2 content.

Table 1 Hydrogeochemical data for the Lower Rhine Embayment, Germany
Full size table

Groundwater samples were also collected and analysed for major ions (Ca2+, Na+, Mg2+, K+, HCO3 −, SO4 2−, Cl− and NO3 −) by the Erftverband at their laboratories in Bergheim, Germany using a range of analytical instruments, as well as for water isotopes (δ2H and δ18O) at the University of East Anglia (UEA), UK. All Noble gas analyses were conducted at the UEA Stable Isotope Laboratory using a quadrupole mass analyser operated in static mode and isotope dilution mass spectrometry. The analytical procedure is described in detail by Poole et al. (1997). Additionally, a small number of groundwater samples from the Brühl region were analysed for 14C (National Science Foundation grant NSF-EAR 0609809), but pMC was not detected, suggesting that the groundwater age was beyond the dating range of 14C.

Results and discussion

Stable isotope data (Fig. 2a) show that the majority of the observation boreholes sampled in the LRE contain groundwater recharge of meteoric origin. The main cluster of data points shown in Fig. 2a displays a limited range of δ2H and δ18O values but exhibits a wide range of hydrochemical compositions illustrated by the Durov plot (Fig. 2b) that is typical of modern groundwater from silicate aquifers in which no single mineral dominates weathering reactions (Appelo and Postma 2005). Generally, the isotopic composition of groundwater from the shallow aquifers of the LRE is characteristic of present-day precipitation for this region (IAEA 2006) confirming a ‘modern’ Holocene recharge. However, seven groundwater samples have relatively low δ18O values, whereas two samples have relatively high δ18O values and display a typical evaporation trend. The small cluster of samples with δ18O values between −9.0 and −9.2 ‰ originates in a single observation borehole but at slightly different depths. Hydrochemical analysis of these samples show that the ionic composition is Na–Cl–(HCO3) dominated with Cl− concentration up to 2,460 mg L−1. This groundwater appears to be of meteoric origin as the samples plot on the global meteoric water line (GMWL), implying that recharge must have occurred under cooler climatic conditions than those that predominate today, and therefore strongly suggests the presence of pre-Holocene ‘palaeo’ groundwater at shallow depth. This groundwater was primarily observed in the Brühl region of the LRE. Previous studies by Schenk (1981, 1982) identified the same body of anomalous groundwater near the Bornheim fault in the Brühl region using hydrochemical data but, without the application of geochemical tracers, Schenk (1981, 1982) was unable to establish the age or origin of the fluid. New hydrochemical data suggest that the palaeo groundwater identified in the Brühl region evolved in a hydrogeological setting that was significantly different from the present-day shallow aquifer system. The groundwater appears to have evolved under conditions that led to a low Ca2+ concentration but relatively high HCO3 − concentration (Fig. 2c). For groundwater to follow this evolutionary path, it would be necessary to assume a high production rate of H+ ions in the aquifer. A mechanism such as the oxidation of pyrite (FeS2) could provide the source of the H+ ions, but it is more likely that the source of H+ ions originates in the production of CO2 in the subsurface because the Brühl groundwater samples are known to have high dissolved CO2 concentrations up to 1,125 mg L−1. Coalification of lignite is a possible source of CO2 which would in turn generate the necessary concentration of H+ ions to account for the observed trend of low Ca2+ and relatively high HCO3 − displayed by the palaeo groundwater. The chemical half-reaction involving the diagenesis of lignite that produces CO2 is given by the following equation:

Fig. 2
figure 2

a Isotopic composition of groundwater in the Lower Rhine Embayment (LRE) with reference to the global meteoric water line (GMWL). Palaeo groundwater has relatively low δ2H and δ18O values in comparison to modern recharge of meteoric origin. b Durov plot displaying the major ion composition of groundwater from the LRE. The ionic composition of LRE palaeo groundwater is Na–Cl–(HCO3) dominated with Cl− concentration up to 2,460 mg L−1. The Durov plot identifies processes such as cation exchange as well as graphically illustrating subtle differences in the hydrochemical composition of modern groundwater recharge. c Cross plot of HCO3 −+SO4 2− against Ca2+. Palaeo groundwater exhibits an anomalous hydrochemical evolution with low Ca2+ and relatively high HCO3 − concentrations that require a high production rate of H+ ions in the aquifer. Processes such as sulphate reduction can also cause deviations from the 1:1 ratio mixing line

Full size image
$$ {\mathrm{C}}_n{\mathrm{H}}_2\mathrm{O}\to \frac{n}{2}{\mathrm{C}\mathrm{O}}_{2\left(\mathrm{g}\right)}+\frac{1}{2}{\mathrm{C}}_n{\mathrm{H}}_{2n+2} $$

Under anaerobic groundwater conditions, the oxidation of organic matter is possibly achieved via SO4 2− reduction.

Measured concentrations of dissolved noble gases (Ne, Ar, Kr and Xe) were interpreted using NOBLE90 (Aeschbach-Hertig et al. 2000; Peeters et al. 2003) in terms of noble gas temperatures (NGTs) and excess air (see Table 1). Comparing NGTs and excess air concentrations to hydrochemical and stable isotope data, it is possible to identify two groundwater types in terms of the long-term recharge history. Modern groundwater recharge is characterised as having lower excess air concentrations in the range of 5–36 % ΔNe and an average NGT of 12.5 ± 1.3 °C. Palaeo groundwater generally exhibits higher excess air concentrations from 32–75 % ΔNe and a lower NGT of 5.6 ± 1.7 °C. Erftverband weather station data show that the current weighted-mean-annual surface air temperature for the region is 10.8 °C, which is 1.7 °C lower than the calculated mean NGT for modern groundwater. Although the uncertainty limits for the calculated NGT could explain most of this difference, it is possible that NOBLE90 does not adequately describe all the mechanisms that control noble gas concentrations at the water table in the LRE. The modelled NGT of the palaeo groundwater is not inconsistent with the findings of Andrews and Lee (1979) and Stute et al. (1995) that show palaeo groundwater displaying a temperature decrease of 5–7 °C, although some studies report a smaller temperature decrease in the range of 2–3 °C between Holocene and Pleistocene recharge (e.g. Osenbrück et al. 1993). However, although much of the data indicate late Pleistocene recharge of meteoric origin in the Brühl region, helium data (Fig. 3) suggest a much longer residence time. A groundwater residence time can be estimated for the Brühl groundwater samples using 4He concentration data. In this study, a terrigenic-diffusive-flux-rate estimate of 4 × 109 atoms 4He m−2 s−1 was taken from the literature presented by Torgersen (2010) for a similar hydrogeological setting to estimate helium input to the deep groundwater system (prior to recent upward movement along faults), but generally many values used in our calculation were under-constrained. Recent terrigenic helium inputs to the shallow aquifer system are assumed to be negligible; however, a 4He age for ‘old’ groundwater at shallow depth in the Brühl region of approximately 107 years was estimated, which is, thus, more ancient than Quaternary palaeo groundwater. In situ production of helium cannot account for the observed 4He concentrations, so a terrigenic flux must be invoked. Further analytical work to determine the 3He/4He ratio of Brühl groundwater would be useful to identify the relative magmatic contribution.

Fig. 3
figure 3

Cross plot of helium concentration against δ18O. The solid lines represent a binary mixing envelope between modern groundwater recharge with a range of δ18O values and palaeo groundwater. The horizontal dashed line indicates the expected helium concentration due to solubility equilibrium with the atmosphere at 10 °C. Fluids at a depth of ∼1,200 m in the Trias aquifer of the Paris Basin display δ18O and helium values (data fall in the shaded area, from Marty et al. 2003) that are similar to the palaeo groundwater observed at very shallow (<100 m) depth in the Brühl region of the LRE

Full size image

A conceptual hydrogeochemical model (Fig. 4) elucidates the observed occurrence of palaeo groundwater with a deep crustal component at very shallow depth in the Brühl region. In situ 4He production derived from α-decay of U-Th series elements in the subsurface as well as a significant external flux of helium are necessary to account for the observed helium concentrations. Hydrochemical analysis indicates that Na+ and Cl− are dominant ions in the palaeo groundwater and that the hydrochemical evolution involved very high P CO2 conditions and a significant subsurface source of H+ ions. The occurrence of deep crustal fluids in shallow aquifers is associated with preferential sub-vertical flow along faults in the LRE. Geochemical and temporal hydraulic head data suggest that the groundwater originates from depth and flows upwards within the fault core material. Conduit flow associated with the Bornheim fault is possibly linked to historic groundwater abstraction in the region. At the location of the Bornheim fault, no thermal anomaly is observed, suggesting that either the flux of upwelling fluid along the fault zone is not vigorous enough to create such a thermal anomaly, or that the plume of fluid observed was emplaced during an episode of enhanced fluid flow caused by groundwater abstraction. A shallow source of helium from U-Th rich lignite deposits could explain the decoupling of helium and heat, but excess helium is absent from groundwater sampled at all other observation boreholes situated near lignite seams, and additional data such as hydrochemistry, water isotopes and 14C-dating all support the notion of deep fluid migration from depth to shallow aquifer units.

Fig. 4
figure 4

Conceptual hydrochemical model illustrated by a cross section of the Bornheim fault indicating sub-vertical flow of helium enriched fluid from depth along the fault during periods of lignite mine dewatering (a) and subsequent dispersal following mine restoration (b) represented by the blue-coloured plume. The cross section illustrates the groundwater flow dynamics in the shallow aquifer system of the Brühl region. The deeper (>250 m) aquifer system is currently influenced by dewatering of the Hambach mine further to the northwest but is effectively isolated from the shallow aquifer system by a very thick confining layer of clay

Full size image

It is hypothesised that the emplacement of deep fluids at shallow depth occurred during the active dewatering of nearby lignite mines and that conduit flow from depth is either reduced or not occurring at present along the Bornheim fault. The emplaced anomalous fluids also appear to be slow to disperse, perhaps due to the small hydraulic head gradients that currently exist in the shallow aquifers of the Brühl region.

The locations of three observation boreholes situated in the Brühl region are illustrated in Fig. 4 with observation boreholes 352721/2 and 957982 in very close proximity to the Bornheim fault and 958062/3 located slightly further west of the Bornheim fault. The direction of groundwater flow is currently from southwest to northeast in the shallow aquifer system. Observation borehole 958063 yields the highest concentration of dissolved helium and the most δ18O-depleted water, and is therefore assumed to represent the end-member composition. It is hypothesised that during historic periods of groundwater abstraction and lignite mine dewatering, the direction of groundwater flow in the Brühl region would have been complex and variable from what is observed today as a cone of depression developed below the nearby lignite mine, which would explain why the end-member composition is presently observed up-gradient from the Bornheim fault. As lignite mining began to decline in the area and finally ceased, the natural direction of groundwater flow slowly restored itself and the plume of anomalous groundwater dispersed to the east of the Bornheim fault in the shallow aquifer system. In addition, the geochemical data indicate that the Bornheim fault is also acting as a barrier to groundwater flow at shallow depth near observation borehole 352722, which could explain the slow rate of dispersion and dilution. The deeper (>250 m) aquifer units in the Brühl region are now influenced by the Hambach mine to the northwest. This deep hydraulic system is effectively isolated from the shallow aquifer system by a thick confining layer of clay shown in the borehole records for 958062/3 and 957981/2. The extent to which groundwater abstraction drives sub-vertical fluid flow along faults in the LRE is unknown.

Conclusions

Palaeo groundwater with a deep crustal signature is potentially widely distributed at depth within the LRE aquifer system, but only observed at shallow depth in a few areas close to fault zones. It is possible that point source emanations similar to those described here occur frequently in close proximity to faults, but remain undetected due to the often sparse distribution of observation boreholes near faults. The influence of groundwater abstraction on sub-vertical conduit flow along faults remains uncertain. Similar crustal fluids have been reported at depth in numerous aquifer systems worldwide (e.g. Marty et al. 2003) but rarely at near-surface depths (<100 m).

It is concluded that this study provides support for a conceptual model of fault zone hydrogeology in unconsolidated sedimentary aquifers in which faults can operate as both barriers for lateral fluid flow, while along the fault, fluid flow can potentially be enhanced. How exactly fluid flow patterns around fault zones develop depends on the way hydraulic stress is imposed onto the groundwater system, for example by groundwater abstraction, CO2 injection or shale gas development. In this particular study, it is demonstrated that time-variant boundary conditions as a result of historical trends in the pumping regime in the study area, have led to geochemical patterns that demonstrate leakage of deep fluids into shallow aquifers via fault zones. The observations thus provide a perspective on how fault zones can behave as hydrogeological pathways between shallow and deep systems as triggered by human activity at shallow depth (e.g. < 100 m), but similar effects could occur via other hydrogeological mechanisms that create hydraulic gradients near fault zones (e.g. fluid injection at depth).