Introduction

About 65% of the world population lives close to a coastline, within less than 150 km. This number could increase up to 75% by 2025 (Hinrichsen 1998) and water supply could become a serious problem. Coastal aquifers are complex salt–freshwater systems, featuring hydraulic interaction with local aquifers. For the future use of coastal aquifers for local water supply, it is essential to understand such systems under present and past hydraulic conditions. In most cases, an anthropogenic impact on natural hydraulic conditions, like for example intensive water withdrawal in coastal areas, will not have an immediate influence on the salinity of the extracted drinking water. However, in the long-term, these aquifers could be lost for water supply for decades if saline water reaches the production wells. Especially for old groundwater resources, which recharged under different climatic conditions than today, a change of hydraulic conditions could lead to serious quality problems.

A number of Dutch scientists have published work about saltwater-intrusion problems, especially on the Dutch coast (Stuyfzand 1996), which is similar to the situation in northern Germany. The River Ems estuary aquifer system in northwest Germany is threatened by the intrusion of brackish water from the tidal-influenced River Ems. Close to the river, production wells of the water supplier Stadtwerke Emden GmbH extract high quality groundwater mainly from a Pliocene aquifer. The mosaic of fresh- and saltwater-dominated aquifer sections within the two-aquifer system could not be explained only by the heterogeneous geological structures and the present-day hydraulic system. It is more likely that time-variable groundwater flow regimes and mixing are the reasons for the present-day salt-freshwater distribution (Führböter 2004). This dating study should contribute to knowledge of the regional-scale groundwater dynamics.

With the set of environmental tracers consisting of 3H, 39Ar, 14C, He isotopes and Ne the groundwater age was studied on a larger scale in proximity to the production wells. The applied tracer set covers a dating range of years to millennia and allows for investigation of the hydraulic conditions with respect to human activity during the last hundreds of years.

Hydrogeology of the Ems estuary aquifer system

The study area is located at the Ems estuary near the city of Emden (Fig. 1). Following the cross-section in Fig. 1 from A to B (shown in the final Fig. 12), the area is divided into the tidal influenced River Ems, the marshlands, relict Geest cores within the marshland, peat bogs at the edges of the Upper Geest, and the Upper Geest in the east. The term Geest is used for the typical landscape in northern Germany, The Netherlands and Denmark, and is formed by glacial sediments of the Pleistocene. The Oldenburg-Ostfriesischer Geest ridge is morphologically characterized by the NW–SE striking structure with a maximum elevation of about 8 m above sea level. In particular, the glacial-fluviatile sediments of the Elster ice age formed aquifers for local water supply. With the beginning of the Holocene North Sea transgression, episodical flooding formed peat bogs at the edges of the Geest and the marshlands between the Geest and today’s coastline. Within the marshland, the Pleistocene sediments are covered by a layer consisting of clay and silt with embedded sandy layers of varying thickness. The bottom of the Pliocene aquifer layer is ~140 m below sea level (mbsl). The lithology is characterized by a varying mixture of fine sand, coarse sand and gravel. The upper Pleistocene aquifer is locally separated from the lower Pliocene aquifer (sand and gravel) by a clay layer called the ‘Tergast Ton’ at a depth of about 50 mbsl near the village of Tergast and 30 mbsl near the village of Neermoor. The Pleistocene glacial-fluviatile sands are interbedded with layers of fine- to medium-sized sands, usually getting coarser, sometimes pebbly, towards the base of the aquifer. Local silt and clay layers with different contents of fine sands are common. The aquifers are silica-dominated sands with a variable content of feldspar and mica and low carbonate content. The average hydraulic conductivities are about 10–5 m/s for fine-sized sands and up to 10–3 m/s for gravel layers in both aquifers.

Fig. 1
figure 1

a Study area with the major structure elements of the landscape of Ostfriesland (Eastern Frisia), northern Germany. The cross-section A–B marks the central part of the study area (Cartesian German National Grid: Gauß-Krüger coordinates). b Location of sampled observation and production wells and major surface water features

Full size image

Hydraulic conditions

The natural hydraulic gradient follows the morphology of the land surface. Groundwater flows from the Upper Geest to the North Sea and River Ems. The natural groundwater surface in the marshland is at sea level. About AD 1000 the natural hydraulic conditions changed. With the construction of dykes the flooding of the marshlands stopped. The artificial drainage of the marshland decreased the piezometric head of groundwater to about 1 m below sea level. Between the River Ems and a band of about a few kilometres from the river to the mainland the natural flow direction changed direction and caused intrusion of brackish river water (Führböter 2004). Since the end of the nineteenth century, the groundwater withdrawal by production wells at the relict Geest core of Tergast (see Fig. 1) increased the hydraulic gradient from the river to the mainland (Führböter 2004).

The major groundwater recharge takes place within the Upper Geest. Relict Geest cores are sources for local groundwater recharge. The groundwater recharge varies from 22–33 mm/yr in the marshland up to 200 mm/yr at the Upper Geest. In the transition zone between Upper Geest and the marshland with the peat bog areas the groundwater recharge varies within a range from 50 up to 100 mm/yr. The mean annual precipitation of 767 mm/yr is distributed equally through the year (Stadtwerke Emden 2006). Evaporation peaks from May to September and amounts to about 500 mm/yr (Hanauer B, Lenz W, Führböter JF, HG Consult/TU Braunschweig, unpublished report, 2008)

The production wells of Tergast and Simonswolde extracted about 3.5 million m3 in 2005 mainly from the Pliocene aquifer (Stadtwerke Emden 2006). There are only estimates of the amount of interaction of shallow groundwater and the drainage canals within the marshland. During intense pumping of the drainage canals, usually during the winter, varying chloride concentrations are measured within the canal system (Schöniger M, Wolff J, Führböter JF, Jagelke J, TU Braunschweig, unpublished report, 2003). This suggests that a small amount of brackish groundwater is removed from the aquifer and mixes with the low mineralized surface water. A general correlation between the chloride concentration in shallow groundwater and varying chloride concentration in the drainage canals was not apparent. The exchange of water between the canals and shallow groundwater depends on the thickness of the cover layer and the depth and construction of the canal bed as well as on the varying hydraulic gradient between groundwater and canal.

Materials and methods

In the area of investigation, 31 observation wells are located in the marshland between the River Ems and the Geest ridge (Fig. 1). Screens are within the upper or the lower aquifer with maximum screen depths listed in Table 1. The screen length is typically 2 m, but for some observation wells length is 4 m. The observation wells are grouped in well galleries. Samples were collected during four sampling campaigns, in 2002, March 2006, July 2006 and May 2007. To cover the complex salt-freshwater distribution between River Ems and the water catchment Tergast, 29 observation wells within the rectangle Oldersum-Tergast-Neermoor-Terborg, two representative production wells (one for each catchment, labelled as PW in Fig. 1b) and two observation wells at the edge of the Upper Geest were chosen. As a reference sample, not influenced by drainage during the last 1,000 years, an observation well (Gk372) on a Geest ridge 50 km to the southeast was also sampled.

Table 1 Data / results of noble gas analysis and age dating
Full size table

Tritium-helium dating

Radioactive tritium (3H) was released in large quantities into the atmosphere between 1955 and 1963 as a result of atmospheric thermonuclear testing (e.g. Solomon and Cook 2000). The 3H concentration in precipitation in the northern hemisphere displayed a maximum in 1963, with concentrations three orders of magnitude above the natural concentrations. A groundwater dating method based on the analysis of 3H combined with its decay product, the lighter and rare helium-3 isotope (3He), was first suggested by Tolstikhin and Kamenskiy (1969). In the saturated zone, 3He produced by 3H decay accumulates in the groundwater and does not undergo any chemical reactions. The derived apparent piston flow 3H–3He ages are therefore independent of the 3H input concentration. The 3H–3He age of a sample refers mainly to the water components with the highest concentration. In mixtures with broad age distributions, the 3H–3He age is therefore biased towards the age of water which recharged during the bomb peak. The 3H–3He dating method has been used in many groundwater studies (e.g. Torgersen et al. 1979; Weise and Moser 1987; Schlosser et al. 1988; Solomon et al. 1993 and Szabo et al. 1996). An overview is given in Solomon and Cook (2000). Stute et al. (1997), Beyerle et al. (1999), Massmann et al. (2008) and Massmann et al. (2009) applied the method to date bank filtrate with an apparent age of several months to decades.

Tritium concentrations in water are preferentially given as hydrogen isotope ratios: 1 tritium unit (TU) equals 10–18 3H/1H. 3He produced by decay of 3H, called tritiogenic 3Hetri, can be converted to TU units: 1 TU equals 2.49·10–12 3Hetri cm3 STP/kg water (STP = standard temperature and pressure). The apparent 3H–3He age (τ) provides a measure for the time difference between the last contact of the water with the atmosphere and the measurement and is given by:

$$ \tau = \frac{1}{\lambda } \cdot \ln \left( {1 + \frac{{{}^3H{e_{tri}}}}{{{}^3H}}} \right) $$
(1)

Where λ = ln(2)/t 1/2 is the decay constant and t 1/2 is the half-life of 3H (12.32 yrs; Lucas and Unterweger 2000). 3Hetri must be separated from other 3He sources in the water which are:

  1. 1.

    3Heequi: Concentration at equilibrium conditions with the atmosphere, which is a function of temperature, salinity and atmospheric pressure during infiltration. 3Heequi is calculated with the solubility function of Weiss (1971). Isotopic fractionation was adapted according to Benson and Krause (1980).

  2. 2.

    3Heexcess: Additional air dissolved in the water from the (partial) dissolution of air bubbles that are trapped in the quasi saturated soil zone (Aeschbach-Hertig et al. 1999). Because neon (Ne) has no sources in the aquifer the excess air component can be derived by the deviation of the Ne concentration in the sample from the calculated air-saturated equilibrium Ne concentration Neequi and is called ΔNe: \( \Delta Ne = \left( {\frac{{N{e_{sample}}}}{{N{e_{equi}}}} - 1} \right) \cdot 100 \) in %. Negative ΔNe values indicate degassing of the water.

  3. 3.

    3Herad: Produced in rocks and subsequently released to the water. 3Herad is produced via 3H by a neutron capture reaction of lithium (6Li(n,α)3H→3He). Neutrons are generated by (α,n) reactions with α-particles from nuclides of the uranium (U) and thorium (Th) decay series and major matrix elements (Si, O, Al, Mg). Concentrations of radiogenic 3He in water are small in general and derived from concentrations of radiogenic 4He. The radiogenic 3He/4He-ratio is about 2·10–8 (Mamyrin and Tolstikhin 1984), a factor of 70 smaller than the atmospheric ratio (Ra = 1.384·10–6).

The tritiogenic 3He is therefore given by:

\( {}^3H{e_{tri}} ={}^3H{e_{sample}} - {}^3H{e_{equi}} - {}^3H{e_{excess}} - {}^3H{e_{rad}} \). Long filter screens, dispersion and, in the case of production wells, convergence of flow paths will form water with a spectrum of ages rather than one age which could be specified by a single value for the 3H–3He age.

Radiogenic 4He (4Herad) is produced by α-decay of nuclides from the U and Th decay series in the aquifer matrix. 4Herad can be used as a qualitative age indicator (Solomon 2000) and is simply calculated as remainder: \( {}^4H{e_{rad}} = {}^4H{e_{sample}} - {}^4H{e_{equi}} - {}^4H{e_{excess}} \). If the resulting 4Herad accumulation rate in the water is constant in time and space within the aquifer, the 4Herad concentration increases linearly with time and represents relative age differences. The method has been applied in old systems with apparent groundwater ages between 103 and 108 years (Andrews and Lee 1979; Torgersen and Clarke 1985). However, Solomon et al. (1996) showed that it is also possible to detect 4Herad in young groundwater with an age of 10–103 years if the groundwater originates from a shallow aquifer of recently eroded sediments. In these sediments 4Herad was accumulated in the source rock and released after erosion subsequently to the water with high rates.

Analysis

Tritium samples were collected in glass bottles. Water for noble gas analysis was pumped with a submersible pump (Grundfos MP1) through a transparent hose into a copper tube (volume 40 ml, outer diameter 1 cm). A regulator clamp was attached to the outlet of the copper tube and used to increase pressure in order to suppress potential degassing. The transparent hose was always checked for bubbles. Pumping proceeded until steady-state conditions with regard to temperature and electrical conductivity were reached. The copper tubes were subsequently pinched off with stainless steel clamps to maintain high vacuum tight sealing.

Analysis of 3H, He isotopes and Ne for 3H–3He dating was conducted at the noble gas laboratory of the Institute of Environmental Physics, University of Bremen (Germany). For the He isotope and Ne analysis all gases were extracted from water. He and Ne were separated from other gases with a cryogenic system kept at 25 and 14 K, respectively. 4He and 20Ne were measured with a quadrupole mass spectrometer (Balzers QMG112A). Helium isotopes were analyzed with a high-resolution sector-field mass spectrometer (MAP 215-50). The system was calibrated with atmospheric air and controlled for stable conditions for the He and Ne concentrations and the 3He/4He ratio. For groundwater samples, the precision of the He and Ne concentrations is better than 1% and for the 3He/4He ratio better than 0.5%. 3H was analysed with the 3He–ingrowth method (Clarke et al. 1976). Samples of 500 g of water were degassed and stored for the accumulation of the 3H decay product (3He) in dedicated He-free glass bulbs. After a storage period of 2–3 months 3He was analyzed with the mass spectrometric system.

For this study a detection limit of 0.03 TU was achieved which allows the detection of the residue of natural, pre-bomb 3H. The uncertainty is less than 3% for samples of >1 TU and 0.03 TU for very low concentrations. For ideal samples, these analytical conditions allow a time resolution of about 6 months for apparent ages less than 10 years. Unknown hydrogeological conditions affecting gas exchange, as for example inaccurate temperatures assumed for gas exchange with the atmosphere lead to a time resolution of about 1.0 year (Kipfer et al. 2002). More details of the sampling, analysis and data evaluation of the method are given in Sültenfuss et al. (2009).

39Ar dating method

39Ar is produced by the interaction of cosmic rays with nuclides of potassium and argon in the atmosphere. The resulting atmospheric equilibrium activity expressed as %modern in relation to recent air (100 %modern) is unaffected by anthropogenic sources and constant at 1.67 × 10–2 Bq/m3 of air (Loosli 1983), corresponding to an atmospheric 39Ar/Ar ratio of ~10–15. With a half-life of 269 years, 39Ar fills the precision dating gap between the young residence time indicators (e.g. 3H–3He, 85Kr) and the radiocarbon method.

The subsurface residence time of groundwater can be simply estimated using the radioactive decay equation and the measured decrease in 39Ar compared to the initial activity in the atmosphere. A possible complication is 39Ar produced in the subsurface due to neutron activation of potassium (39K(n,p)39Ar). If not considered in the age calculation, this would lead to an underestimation of the calculated groundwater residence time. However, elevated 39Ar concentrations due to underground production rates have only been observed in rare cases within fractured rock aquifers with high U and Th concentration in the rock matrix (Andrews et al. 1989; Purtschert et al. 2007).

Sampling and analysis

Because of the very low isotopic abundance of 39Ar, a relatively large amount of water (> 1,500 L) has to be degassed in the field to obtain a sufficient number of 39Ar atoms. In contrast to other gaseous tracers which rely on absolute concentrations of dissolved constituents (e.g. CFC, SF6 or 4He), the 39Ar method is insensitive to the degree of degassing (both in nature and during sampling) and recharge conditions because the method is based on an isotope ratio of the same element (39Ar/Ar). However, for waters with depleted gas concentration, a corresponding larger sample volume is required. The 39Ar activity of the argon separated from groundwater is measured in high pressure proportional counters in an ultra-low background environment (Loosli and Purtschert 2005). Although first attempts have been made to use the accelerated mass spectrometry (AMS) technique (Collon et al. 2004), which would reduce this volume, the detection of 39Ar is currently only practical by the radioactive decay-counting techniques (Loosli 1983). Counting times range typically between 1 and 4 weeks, depending on the extracted amount of argon and the age of the water. The minimum detection level (MDL) is 3–10 %modern for water volumes of 4–1.5 tons of air-saturated water respectively.

14C dating method

Radiocarbon dating of total dissolved inorganic carbon (TDIC) is one of the most widely used applications of radiocarbon dating (Kalin 2000). However, the interpretation of 14C results from groundwater is complex because carbon in groundwater may be derived from several sources with different 14C concentrations. Natural atmospheric 14C is produced in the upper atmosphere by the reaction of thermal neutrons, which are produced by cosmic rays, with nitrogen: \( ^{14}N + n \to {^{14}}C + p \). The resulting equilibrium activity in the atmosphere is 100 % modern carbon (pmC) corresponding to 0.23 Bq/gC (Kalin 2000).

14C is oxidised to CO2 and mixes into the lower atmosphere where it is assimilated in the biosphere and hydrosphere. 14CO2 of the soil zone dissolves in the infiltrating water and 14C decays in the sub-surface with a half life of 5,730 years. The radiocarbon age of the water is calculated based on the radioactive decay law:

$$ {A_{14C}} = {A_0} \cdot \exp \left( { - {\lambda_{14C}} \cdot t} \right) $$
(2)

where λ14C is the decay constant of 14C.

The main complication of the application of the 14C method for groundwater dating arises from the fact that the initial activity A 0, which defines the starting point for the “radioactive decay clock”, is possibly affected by geochemical reactions. This is commonly considered using correction models that are either based on the chemical composition of the groundwater, the 13C content of TDIC or both (Eichinger 1983; Fontes and Garnier 1979; Ingerson and Pearson 1964; Mook 1980; Parkhurst et al. 1990). The simplest correction models are basically mixing calculations of different carbon sources in groundwater. The stable 13C is the most suitable mixing parameter because its behaviour is chemically similar to 14C but it undergoes no radioactive decay. Therefore, the concentration of the stable 13C is normalised to the standard carbonate PDB (carbonate from a Pee Dee belemnite) where:

(3)

Plants that follow the C3 photosynthesis pathway have δ13C values of ca. –25‰ (Fritz et al. 1985; Mook 1980). CO2 from root respiration, which has a similar δ13C value, is dissolved in the water and the resulting carbonic acid drives the dissolution of carbonate minerals in the aquifer. The δ13C value of marine carbonates is, by definition, equal to zero (Eq. 3). Neglecting isotope exchange reaction (Fontes and Garnier 1979), a δ13C value of –12.5‰ in TDIC indicates in this pure mixing approach an initial A 0 of 50 pmC. Dating is even more complicated when CO2 gas from subsurface sources is dissolved in the groundwater. The type of the fermentation reaction determines the isotopic composition of this additional carbon source to the TDIC of the groundwater. For example or CO2-reduction produces CO2 enriched in δ13C (+20‰) and methane (CH4) which is accordingly depleted in δ13C (–70‰; Aravena et al. 1995). The 14C activity of CH4 or CO2 depends on the age of the decomposed organic matter and the point in time along the groundwater flow when those gases were produced. In the case study reported here, CO2 from methanogenesis was considered as a third end member in the A 0 mixing calculation.

Sampling and analysis

14C samples were obtained by direct precipitation of TDIC dissolved in several tens of litres of water. For 14C analysis, the precipitated BaCO3 was converted to CO2 by acidification and CH4 was synthesised for low level gas counting (Oeschger et al. 1976). The achieved detection limit (DL) is 0.5 pmC. Separate samples (1L) were collected for 13C analysis by mass spectrometry.

Results

3H, 3He, 4He and Ne

The results of the age dating are given in Table 1. Tritium concentrations are below 0.1 TU for samples from depths below 20 mbsl, except for samples from the region Neermoor and the wells in Terborg (Tb3) located closest to the River Ems (Fig. 1). The two shallow Tb3 wells contained tritium concentrations significantly above the tritium concentration in precipitation at the time of recharge (1987). At that time the operation of a nuclear power plant upstream the River Ems started and tritium concentrations in the Ems increased more than one order of magnitude above concentrations in precipitation. For these two samples, the high tritium concentration clearly indicates infiltration from the River Ems.

The tritium concentration in precipitation for the Ems region (IAEA 2006), corrected for decay to year 2006, exceeds 7 TU for the period since 1960. The samples with tritium below 7 TU must contain a tritium-free component, i.e. water infiltrated before 1955. As an example, for well Ro2/14, this tritium-free fraction must be larger than 85%. A natural tritium concentration in precipitation of 4 TU was assumed for this study site, which is 20% lower than for central Germany due to the proximity to the ocean (Roether 1967). The deeper tritium-free samples are therefore older than 80 years, because the residual of 4 TU after 80 years is 0.045 TU. The results of all 3H and 3He analyses and the corresponding 3H–3He-ages are displayed in Fig. 2a.

Fig. 2
figure 2

a Concentration of 3H, tritiogenic 3He and 3H–3He age, b radiogenic 4He and ΔNe vs. depth. The vertical bars at symbols for radiogenic 4He display screen lengths. The approximate position of the clay layer is indicated. The broken line indicates ΔNe = 0

Full size image

Ne concentrations are significantly below the solubility equilibrium concentration (ΔNe < 0%) except for the shallow samples of Tergast and the deepest samples in the study area (Fig. 2b). The Ne data prove loss of gases from the water body. Maximum degassing ranges between ΔNe –65 to –70% for samples of mid-depths (21–39 m). These samples are free of tritium; therefore, they infiltrated before 1955. Younger waters from shallower wells are affected by degassing also, but not so much. Hence, the gas loss has to be considered in the calculation of tritiogenic 3He and the corresponding 3H–3He ages, and for the accumulation of radiogenic 4He. To do so, there is a need to investigate (1) whether degassing leads to fractionation of gases and (2) the point in time of degassing. It was assumed, that all degassed samples never contained additional atmospheric air, i.e ΔNe was never positive. Consequences of this assumption are specified in Aeschbach-Hertig et al. 2008. To study the first effect, this work focuses on the noble gas isotopes which would be unaffected during aging or affected only marginally, that is (a) 4He for young samples (tritium ≥ 7 TU), (b) 3He for tritium-free samples and (c) Ne.

  1. a and c.

    Samples with tritium from young water (age < 50 yrs) display a small concentration of radiogenic 4He (Fig. 3). Figure 4a shows the 4He and Ne concentration of three sampling periods. Replicate analyses display the same concentration of He and Ne. Degassed samples have Ne concentrations below 2·10–4 cm3 STP/kg. For degassed samples with low radiogenic 4He, the 4He–Ne relation shows gas loss of equal proportions for He and Ne, indicating no significant fractionation.

  1. b and c.

    Tritium-free samples could not have produced 3Hetri above 10–11 cm3 STP/kg. For tritium-free waters with negligible radiogenic 4He concentration the 3He–Ne relation was used to investigate fractionation (Fig. 4b). Also here, a number of samples showed gas loss of equal proportions and therefore displayed no fractionation during degassing. For comparison a diffusion controlled fractionation (Rayleigh fractionation) for 3He and Ne for degassed water is indicated by the thick blue broken line. All samples fall to the right of this line. For old tritium-free waters a maximum for tritiogenic 3He of 10–11 cm3 STP/kg equivalent to 4 TU (natural tritium in precipitation) is expected. All 3He concentrations are consistent with that upper limit if a small radiogenic component is subtracted (indicated by the black bars at the symbols). According to the long term data records most of old tritium-free non-degassed groundwater samples show tritiogenic 3He significantly below the estimated natural tritium concentration in precipitation. From this one can state: if degassing would have been diffusion controlled, all samples should fall left of the thick blue broken line. As this is not the case it is concluded that degassing reduces the concentrations of the investigated gases (and their isotopes) in equal proportions.

Fig. 3
figure 3

Radiogenic 4He vs. tritium separated for different areas

Full size image
Fig. 4
figure 4

a4He vs. Ne concentrations for all samples. Red dots display solubility equilibrium concentration for fresh water at atmospheric pressure of 1013 hPa for different temperatures. The excess air line indicates the extra air with the atmospheric Ne/He ratio dissolved in a sample which is in solubility equilibrium at 10°C. In the enlargement, some replicate measurements are marked. b3He vs. Ne concentrations for samples with tritium <0.1 TU. Black bars indicate the amount of radiogenic 3He derived from radiogenic 4He with a radiogenic 3He/4He ratio of 2·10–8. An equivalent of natural tritium in the order of 4 TU = 1·10–11 3He cm3 STP/kg should be found in the samples. If degassing would be controlled by diffusion, data points should fall on or left of the thick blue broken line, which was calculated assuming diffusion-controlled degasssing; diffusion constants for He and Ne in water: DHe: 5.68·10–5 cm2/s, DNe: 2.95·10–5 cm2/s at 10°C (Jähne et al. 1987). The diffusion for 3He is derived from 4He by scaling DHe with sqrt(mass4He/mass3He)

Full size image

Data of the same well from different sampling periods are generally in very good agreement as are the replicate samples showing the high reliability and quality of sampling and analysis. This was judged as an indicator that degassing did not occur during sampling.

Another substantial issue for 3H–3He dating of degassed samples is the point in time of degassing (Visser et al. 2007). For reasons deduced in the previous and in the following, it was assumed that degassing took place at the time of infiltration. With this assumption, the calculated 3H–3He ages in the depth range of 12–24 mbsl are between 21 and 41 yrs, as already shown in Fig. 2a.

The maximal variation in 3H–3He ages depends on the difference in tritiogenic 3He between the two extreme scenarios (1) degassing during infiltration and (2) degassing during sampling and is represented in Fig. 5. This difference is derived from Eq. 1 where 3He = 3Hetri:

$$ \Delta \tau = \frac{{\partial \tau }}{{\partial {}^3He}}\; \cdot \Delta {}^3He + \frac{{\partial \tau }}{{\partial {}^3H}}\; \cdot \Delta {}^3H\quad while \;tritium\;is\;not\;affected:\, \Delta {}^3H = 0\quad \Rightarrow \Delta \tau = \frac{1}{\lambda } \cdot \frac{1}{{{}^3He + {}^3H}}\; \cdot \Delta {}^3He $$
Fig. 5
figure 5

Two degassing scenarios affecting tritiogenic 3He. 1: degassing at the time of infiltration, 2: degassing at sampling

Full size image

For tritiogenic 3He for both scenarios from Fig. 5 follows: \( \frac{{^3He(2)}}{{^3He(1)}} = \frac{1}{{1 + \frac{{\Delta Ne}}{{100}}}} \) and for the difference of tritiogenic 3He:

\( {\Delta^3}He{ =^3}He(2){ -^3}He(1) = \frac{{ - \frac{{\Delta Ne}}{{100}}}}{{1 + \frac{{\Delta Ne}}{{100}}}} {\cdot^3}He(1) \) (note ΔNe < 0)

This leads to an estimate for the age difference for the two scenarios depending on the degree of degassing: \( \Delta \tau = \frac{1}{\lambda }\; \cdot \;\frac{{ - \tfrac{{\Delta Ne}}{{100}}}}{{1 + \tfrac{{\Delta Ne}}{{100}}}}\; \cdot \;\frac{{{}^3He(1)}}{{{}^3H + {}^3He(1)}} \)

For ΔNe = –10% the calculated age would increase by about 2 yrs if 3H < < 3He; for ΔNe = –50% the age would increase by 18 yrs.

For samples with tritium below 7 TU, the derived 3H–3He age is not the mean residence time because the age of an old tritium-free component could not be determined. For samples with high concentrations of radiogenic 4He, the separation of tritiogenic 3He is affected by a higher uncertainty, because the 3He/4He ratio of the radiogenic He must be estimated for the aquifer matrix.

The radiogenic 4He is calculated by simply subtracting the atmospheric He portion in the sample from the measured concentration, assuming that the atmospheric He is represented by the atmospheric Ne:

$$ {}^4H{e_{rad}} = {}^4H{e_{sample}} - \frac{{{}^4H{e_{equi}}}}{{N{e_{equi}}}} \cdot N{e_{sample}} = {}^4H{e_{sample}} - \left( {1 + \Delta Ne} \right) \cdot {}^4H{e_{equi}} $$

Concentrations of 4Herad increased continuously with depth below 60 mbsl, indicating increasing ages with depth. Highest concentrations in the order of 6.4·10–4 cm3 STP/kg were detected in well Te2/113, which is the deepest (Fig. 2b). Samples with tritium concentration below 7 TU show a significant 4Herad concentration and therefore could be affected by mixing.

39Ar dating

In the depths below 25 mbsl, no 39Ar above the DL was detected (< 10 %modern; Fig. 6). The groundwater residence times in this depth range must, therefore, exceed 900 years. Furthermore the low values indicate that underground production of 39Ar is low in this system. The detectable 39Ar activities observed in the more shallow samples (Table 1), therefore, were directly converted into ages of 130 ± 50 and 250 ± 60 years respectively.

Fig. 6
figure 6

14C and 39Ar vs. depth. Below the clay layer the samples are 39Ar free (<10 %modern). For the two shallow wells 39Ar ages of 130 and 250 years were determined

Full size image

14C dating

Ten samples from the deepest wells have been analysed for the carbon isotopes 13C and 14C of TDIC. The 14C activities of TDIC range between 24 and 68 pmC and the δ13C values between –21.6 and +1.6 ‰ (Table 1). This large variation in δ13C values is only partly induced by the dissolution of rock carbonates which would cause a much stronger 14C-TDIC reduction than observed (Table 1); Pearson et al. 1991). The investigated waters contained dissolved methane with a δ13C signature of –70 ± 5 ‰ (Table 2), which is typical for bacterially induced methane production (Aravena et al. 1995). Due to isotope fractionation effects that occur during methanogenesis, the simultaneously produced CO2,meth of the reaction becomes isotopically enriched compared with CH4 (Balabane et al. 1987, Klass 1984). It is assumed that a δ13C value of –25 ‰ of decomposed organic matter (Mook 1980) results in a δ13C of CO2,meth value of +20 ‰ which dissolves in the water and contributes to the TDIC pool (Figs. 7 and 8). Elevated δ13C-TDIC values, therefore, imply a high CH4 and CO2,meth production. The quasi-linear correlation between δ13C and ΔNe (Fig. 9) suggest that CO2,meth accumulation is linked to the degassing process. Degassing during sampling is considered to be unlikely as deduced in the previous from the noble gas concentrations. It is, therefore, assumed that degassing occurs by gas stripping in shallow depths during the infiltration process. At a given hydrostatic pressure, CO2,meth is kept much longer in solution than the poorly soluble CH4. Continuous CH4 (+CO2,meth) production will, therefore, induce the formation of CH4 bubbles. In the study samples, the highest measured CH4 concentrations range up to 50 cm3 STP/Lwater (Table 2), corresponding to a dissolved gas pressure of about 1.4 atm (solubility of methane: ~35 cm3 STP/Lwater/atm). Together with the atmospheric gases remaining in these degassed samples, the total dissolved gas pressure is in the order of 1.5–2 atm, indicating that the depth of methane production is about 5–10 m below groundwater surface. The initially pure methane gas bubbles trigger the ex-solution of atmospheric gases (such as Ne) by gas stripping and the subsequent depletion of those gases in the water phase (Fig. 8 and Brennwald et al. 2005). With increasing methane bubble production the Ne concentration in the water decreases (more negative ∆Ne values). The higher CO2,meth input leads to a more enriched δ13C signature in TDIC (Fig. 8).

Table 2 Concentration, isotope composition and age of methane and TDIC
Full size table
Fig. 7
figure 7

Carbon isotope (δ13C,14C) composition of groundwater samples and carbon sources for TDIC. The initial 14C activity for radiocarbon dating is estimated based on the mixing between three carbon sources (1–3) with distinct isotopic signatures. The reference sample represents water that is not affected by methanogenesis (no CO2,meth; source 2). Labels indicate ΔNe (%) that decrease with increasing CO2,meth (see text and Fig. 9). The dotted line corresponds to the initial activity A o as a function of the δ13C value of the sample. The lengths of the vertical arrows represent aging due to radioactive decay

Full size image
Fig. 8
figure 8

Scheme of gas stripping in peat bogs. Formation of pure methane bubbles and subsequent gas partitioning of water dissolved gases and gas bubbles. Solubility equilibrium between the water and the gas phase is achieved in steady state. The carbon isotope signature of fermented organic matter, CO2 and CH4 are also given. CO2 stays preferentially in solution due to the 25 times higher solubility compared to methane

Full size image
Fig. 9
figure 9

δ13C in TDIC vs. ΔNe

Full size image

Soil organic matter is a complex mixture of substances with an extended range of turnover times generally increasing with depth (Charman et al. 1994; Gillon et al. 2009). As a simplifying hypothesis, three main TDIC sources were identified in the study area’s groundwater (Fig. 7).

  1. Source 1.

    At shallow depths and under aerobic conditions, CO2 from root respiration (CO2,soil) and carbon from relatively fresh organic matter is dissolved in the recharging water. The carbon isotope composition is that of the source plant material, i.e. –25 ‰ and 100 pmC for δ13C and 14C, respectively.

  2. Source 2.

    With increasing depth and under anaerobic conditions, δ13C-enriched CO2,meth is increasingly produced as deduced in the previous. The 14C activity of this CO2,meth is related to the age of the fermented bio mass at the location of main methane production. This value is difficult to estimate. It was also observed that 14C ages of dissolved CH4 and TDIC in peat porewater are much younger than the 14C age of the peat at the same depth (Charman et al. 1994). This reflects the fact that carbon dynamic in peatlands is characterised by a complex mixing pattern. 14C-TDIC of 85 pmC was measured in water of the discharge region of a recent peat bog south of the research area. Similar 14C–CH4 and 14C-TDIC values were also found in peat bogs elsewhere at a few meters depth (Aravena et al. 1995). A mean 14C activity of 85 ± 10 pmC was consequently assumed for organic matter at depths of maximal methane production. The resultant CO2,meth is 14C-enriched up to 94 pmC due to isotope fractionation. (Because the 14C–12C mass difference is twice that of 13C–12C, the 14C fractionation is approximately twice that of the δ13C: ∆14C = 2·∆13C = 2·45 ‰ = 9 pmC, 85 + 9 = 94 pmC)

  3. Source 3.

    The carbonic acid accumulated in the recharge zone drives the dissolution of carbonate in the aquifer. In the absence of significant subsurface CO2 sources, this is the reaction which is commonly considered in 14 C correction models (Fontes and Garnier 1979; Mook 1980; Eichinger 1983, Pearson et al. 1991). The 14 C dead carbon from the rock matrix (Crock) would dilute the initial carbon isotopic composition towards 0 pmC and δ13C ~0‰ (Kalin 2000).

The carbon isotope compositions of the three mixing end members (sources 1–3) are depicted in Fig. 7 in a δ13C–14C graph. A 14C-dating-correction procedure requires knowledge of the relative portions (p 1, p 2, p 3) of the three end members for initial 14C activity (A 0) calculation. The remaining activity difference can be attributed to radioactive decay and interpreted in terms of groundwater age. However, because there are three unknowns (p 1–3) and only two measured parameters (δ13C, 14C), the system of equations is underdetermined. The reference sample Gk372 was used in order to constrain the reaction pathway without methanogenesis. This CH4-free well is located in the same aquifer but further inland. The absence of 3H and the 39Ar activity of 80%modern indicate a residence time of 90 years. The low δ13C value of –21.6‰ indicates a limited availability of carbonate in this aquifer in accordance with the low alkalinities and pH observed. The δ13C–14C mixing line between the reference sample and the assumed CO2,meth-endmember (dotted line, Fig. 7) together with the measured δ13C value defines the initial 14C activity of each sample which is used in the age calculation (vertical arrow in Fig. 7).

The 14C activity of the dissolved methane was measured as a further age constraint. The results from five measured samples range between 50 and 69 pmC. Using again an average initial 14C activity of the decomposed organic matter in the peat bog of 85 pmC, a 14C–CH4 fractionation factor of 2 · –45 ‰ = –9 pmC (see the previous) and, thus, an initial CH4 activity of 76 pmC results in CH4 ages which are comparable to the 14C-TDIC ages (Table 2). It is thereby assumed that methane behaves conservatively along the flow path (no isotope exchange with the rock matrix or oxidation). The rough correspondence between 14C–CH4 and 14C-TDIC ages provides additional evidence for the correctness of the basic dating assumptions and the obtained 14C age scale which ranges from 90 yrs for the reference sample up to 9,000 yrs (Table 1).

Discussion

Degassing and dating

Gas stripping or degassing affected the applied dating tracers differently. 39Ar and 3H were not affected at all. 14C, 3He and 4He dating required a careful evaluation of the details of the degassing process. The correlation between δ13C and ∆Ne (Fig. 9) suggests that gas stripping is related to methanogenesis, possibly taking place in peat bogs at shallow depths. The continuous production of methane in a low hydrostatic pressure environment induced the formation of initially pure methane bubbles (Fig. 8). The concentration difference of atmospheric gases (Ne, N2) in air saturated water (ASW) and gas phases (initially pure CH4) caused the ex-solution of atmospheric gases into the gas bubbles. At the beginning, the degassing process is controlled by the relative diffusion coefficients of the ex-solved gases. The lighter and more mobile gases are enriched in the gas phase during this first non-steady-state phase (Brennwald et al. 2005). In the steady-state equilibrium, however, the partitioning between the water and the gas bubbles is controlled by the relative solubilities of the gases. A detailed discussion of diffusion controlled and solubility controlled degassing processes can be found in Aeschbach-Hertig et al (2008).

Figure 4b shows that 3He and Ne concentrations of the degassed samples were not fractionated. Whereas diffusion-controlled degassing implies a significant fractionation between He and Ne due to the strong difference of diffusion constants, solubility controlled degassing only slightly changes the ratio of He to Ne, as both gases have similar solubilities. The degassing process was probably slow enough to allow solubility equilibrium. Such a slow formation and escape of bubbles was observed in lacustrine sediments (Brennwald et al. 2005) and could also be expected for peat bogs within the research area (Strack et al. 2005). In the beginning of flooding during Holocene transgression phases, the peat bogs grow fast because of banked-up water. After flooding with seawater, the peat bogs died and were covered with fine-grained marine sediments (Streiff 1990).

For this study, the application of radiogenic 4He as a dating tool was not affected by degassing, because (1) the water degassed before 4Herad accumulated from the sediments and (2) the atmospheric He component in the sample could be quantified well enough by using the measured Ne concentration of the sample and presuming gas loss of equal portions.

On the much shorter timescales of 3H–3He dating (< 50 yrs), the residence time of the water in the ebullition zone of peat bogs could be significant (years to decades). In principle tritiogenic 3He can escape into CH4 bubbles during that time interval. Therefore, 3H–3He ages refer to the flow time since the water left the peat bogs. However, at this study site, the highly degassed waters were tritium free and, hence, 3H–3He dating was not applicable anyway.

The initial 14C activity of TDIC and methane for dating of the degassed samples is constrained by the carbon isotopic composition of the fermented organic matter in the peat bog. A three component mixing model between CO2,soil (in the root zone under oxidising conditions), CO2,meth (in a reducing environment) and to a smaller portion carbon from the aquifers’ rocks reveal 14C ages which depend on δ13C in TDIC and, thus, on the degree of degassing or CO2 accumulation respectively. The initial 14C activity in methane (76 pmC) is reduced compared to the 14C activity of the peat bog (85 pmC) due to isotope fractionation during methanogenesis. The resulting 14CH4 ages are roughly in agreement with the calculated 14C-TDIC ages (Table 2).

The correlation of 14C ages and 4Herad concentration yield a 4He accumulation rate of 5.3·10–8 cm3/kgwater/yr (Fig. 10). Under the assumption of a similar 4He accumulation rate for all samples 4Herad was used as a regional age proxy. The deduced 4Herad accumulation rate compares well with values found by other researchers: Lehmann et al. (2003): 2–20·10–8 cm3 STP/kg/yr, Schlosser et al. (1989): 3·10–7 cm3 STP/kg/yr, Castro et al. (2000): 4·10–8 cm3 STP/kg/yr and Solomon et al. (1996): 3·10–7 cm3 STP/kg/yr. A study site with glacial-fluviatile sediments, mainly of a similar structure, 400 km to the east provides nearly the same accumulation rate of 5·10–8 cm3 STP/kg/yr derived by comparison with 14C ages (Sültenfuß et al. 2006).

Fig. 10
figure 10

Concentration of radiogenic 4He concentrations vs. 14C ages calculated according to the mixing scheme in Fig. 7. Error bars indicate differences for 14C ages calculated with an initial 14C concentration of 70 pmC

Full size image

Recharge history

Using the constructed 4He ages, the relation of ∆Ne vs. age is shown in Fig. 11. Only the oldest (and deepest) waters were unaffected by degassing. In the age range from hundreds to a few thousand years (or the middle depth range 70–30 m), ∆Ne values are lowest (most negative) and increase again for the shallower and, thus, younger waters. Water was assigned with ∆Ne < 0 to be recharged through marshland or peat bogs. The spatial distribution of peat bogs, however, changed over time. The deepest and oldest waters display ∆Ne > 0 and are supposedly recharged on the Geest ridges. Hence, they were not affected by CH4 production and likewise gas loss. As these samples are from depths below the degassed samples, they originate from furthest away.

Fig. 11
figure 11

ΔNe vs. derived 4He ages. Thick grey line indicates supposed evolutionary pattern

Full size image

Since about AD 1000, men started building dykes along the River Ems and the North Sea and drained the hinterland. Before that time, the complete area could have acted as a recharge area. After the marshland and peat bogs were drained, the recharge here was interrupted. Since that time groundwater recharged mainly over the Geest ridge and at the Geest humps. Shallow groundwater displays young 3H–3He ages which confirm this concept. The comparison of tritium in wells near the River Ems with water from the River Ems with high tritium produced by a nuclear power plant shows that young brackish water entered the aquifer. At present, young river water intrusion, as indicated by elevated tritium concentrations from the tritium contaminated river Ems, is restricted to few locations close to the river where the river bed is permeable.

From the overall tracer examination, the structure of the aquifer is derived as displayed in Fig. 12. This figure is a composition of recent flowpaths of young waters and ancient flowpaths for the older, deeper water. Due to the sparse and scattered sample spacing, it may be doubtful whether this cross section is applicable for the complete study area. However, the main features of the groundwater dynamics should be well covered.

Fig. 12
figure 12

Schematic of the present-day and past flowpaths within the aquifer along the cross-section A–B (Fig. 1). The salt– and freshwater distribution is indicated according to Führböter (2004)

Full size image

Conclusions and summary

The groundwater in the investigated section of a coastal aquifer system covers a large range of residence times. In shallow depths below 27 m, most samples contain measurable 3H concentrations and are, therefore, younger than 50 years. 3H–3He ages of the young water components have been determined to 20–41 yrs. Mixing of young water with tritium and tritium-free older water was detected for some samples in shallower depths. Two tritium-free samples in the shallow layer (depths 22–25 m) show groundwater residence times determined by 39Ar measurements of 130 and 250 yrs. This tracer is the best suited tool for groundwater dating on the century time scale. At greater depths, below a low permeability clay layer the 39Ar concentrations drop below 10 %modern, indicating groundwater residence times above 900 years. Additionally, the low 39Ar concentrations in this part of the aquifer demonstrate that underground production is low and does not affect the 39Ar dating. Groundwater ages on the millennium scale were estimated by means of 14C activities in TDIC and methane. A whole set of absolute dating tracers (3H–3He, 39Ar and 14C) provided consistent groundwater ages and allowed an estimate of the 4He accumulation rate. With a nearly constant, 4He release from the aquifer matrix to the water for a finite study site 4He is a sensitive dating tool for a large range of residence times.

Degassing has been observed in most of the samples by means of depleted atmospheric neon concentrations. It was presumed that degassing occurred during groundwater infiltration through peat bogs which covered the area to a large extent before AD 1000. The effect of this gas loss on the applied age dating methods has been discussed. Because degassing occurred at the beginning of the groundwater flow lines and did not fractionate the gases, the separation of 3Hetri and 4Herad was possible and relatively well constrained. 39Ar and 3H concentrations are insensitive to degassing processes.

The temporal and spatial degassing pattern is used as a proxy for groundwater recharge through peat bogs. Groundwater flow and discharge was towards the sea. With the construction of dykes and the drainage of the marshland the water table fell below the level of the River Ems and caused bank filtration of brackish water. This was identified by the high tritium signal from the river.

In this study, the environmental tracers proved their capability as key tools for studying large-scale hydrodynamics. In particular, their combination gave valuable clues for infiltration conditions. For a more detailed study on the degassing processes, the application of the heavier noble gases would be desirable.