Tritium (3H) and its daughter product 3He have been widely used as tracers in hydrological studies, but quantitative analyses of their behaviour in freshwater lenses and the transition zone in coastal aquifers are presently lacking. In this paper, the fate of 3H and 3He in the freshwater lens and the transition zone as well as the saltwater wedge is studied using numerical variable-density flow and transport models of different degrees of complexity. The models are based on the conditions on the German island of Langeoog, which is uniquely suited for this purpose because of the high 3H concentration of the North Sea. It is found that most bomb-related tritiogenic 3He still resides in the freshwater lens, making it a useful tracer for young (<60 years) groundwater. Differences in dispersive transport between 3H and 3He can cause an apparent age bias on the order of 10 years. Under favourable conditions, 3H from seawater can penetrate deep into the offshore part of the aquifer and has potential to be used as a tracer to study saltwater circulation patterns. Our modelling suggests that the field-observed 3H in the transition zone does not originate from seawater but from freshwater affected by the bomb peak. Yet in models with a low (
Tritium (3H) has been used extensively as a tracer in hydrogeological studies [
When the tritium daughter product helium (3He) is simultaneously analysed and after the so-called tritiogenic contribution (3Hetrit) is separated from other 3He contributions (exchange with the atmosphere, excess air, and underground production), the apparent age
While tritium has been used extensively in terrestrial, fresh groundwater settings, it has seen less application in brackish and saline coastal environments. It has been mostly applied in freshwater lenses [
Tritium has been detected in brackish and saltwater in coastal aquifers in Tahiti [
While the tritium concentrations of most seas and oceans have returned to almost their prebomb era, natural background [
(a) Location of Langeoog Island, (b) map of the water supply area in the Heerenhus dune area showing the positions of the abstraction and observation wells, and (c) apparent bulk resistivity of the subsurface at 10 m below msl based on airborne electromagnetic measurements [
The relationships between tritium, helium, and groundwater age in coastal aquifers have not been explored using quantitative models except for the highly simplified case of the Henry Problem [
First, a set of models of increasing complexity was used to determine the influence for each of the various controlling factors separately. Attention is paid to the implications for submarine groundwater studies. The results are expected to be applicable to other study areas, especially along the North Sea shores, and also to other coastal localities where tritium concentrations persist above the natural background (e.g., [
The island of Langeoog is part of the chain of barrier islands that separate the shallow intertidal Wadden Sea from the North Sea (Figure
Below the superficial dune and beach sands, the subsurface is made up of Holocene deposits of the Wadden Sea, underlain by mostly Pleistocene (glacio-) fluvial sediments [
Three individual freshwater lenses exist on the island. Their separation is the result of occasional catastrophic storm flood inundations, storm floods, e.g., the Christmas flood of 1717, which caused large breaches of the dune belt in two locations [
Extraction is done intermittently at pumping rates of 10 m3/hr per well and is distributed over 20 small wells, with screens installed at varying depths between 10 and 18 m below sea level. As an effect of increasing tourism, water consumption continuously rose until the 1980s. It has significantly decreased since the 1990s due to the implementation of water-saving measures. Peak demand occurred in 1983 with 452,000 m3/yr (Figure
Time series data used in the model: (a) total annual abstraction (
Basic information, such as well locations, groundwater levels, and pumping rates, were obtained from the local water supply company, the Oldenburgisch-Ostfriesischer Wasserverband (OOWV). Details about sampling and analytical techniques used to determine the chemical and isotopic composition of the groundwater samples were presented by Houben et al. [
Transition zone water samples were obtained from observation well 17 (Figure
(a) Chloride concentration, (b, c) 3H concentration, and (d) 3He concentration as a function of depth for observation well 17. The small red square marker symbols in (a) indicate the chloride concentration calculated from the downhole
A two-dimensional section across the freshwater lens was considered (Figure
Cross section showing the model dimensions, boundary conditions, and hydraulic parameters for the layers used in the layered simulations. Parameter values in white font were constant in all model simulations. Along the left, bottom, and right boundaries, no-flow and nondispersive flux conditions were applied. The abstraction (
The term
Nine different scenarios (A1-A9) were considered that each differ from each other by a single model variable (Table
Overview of the model simulations.
A1 | A2 (A31) | A3 | A4 | A5 | A6 | A7 | A8 | A9 | |
---|---|---|---|---|---|---|---|---|---|
Transient recharge | X | X | X | X | X | X | X | X | |
Anisotropic | X | X | X | X | X | X | X | ||
Layered | X | X | X | X | X | X | |||
Clay layer | X | X | |||||||
Discontinuous clay layer | X | X | |||||||
Abstraction | X | X | X |
The effect of the vertical anisotropy on the 3H concentration distribution in the offshore domain due to the recirculation of seawater was investigated in an additional set of simulations labelled A3. The horizontal to vertical hydraulic conductivity ratio was varied between values
The transient simulation spanned the 61-year period from 1 January 1953, the earliest month for which the rainfall 3H concentration could be estimated, to 31 December 2013. Groundwater recharge was calculated based on daily rainfall and evaporation measurements as described in Post and Houben [
The 3H concentration of seawater (Figure
The initial conditions for the model (i.e., the heads and concentrations on 1 January 1953) were determined by simulating the 61-year period twice (i.e., 122 years spanning the period 1 January 1831–31 December 1952). With the exception of simulation A1, the initial conditions therefore do not represent a steady-state situation, the reason being that the temporal variability of the recharge could not be neglected as it is responsible for a significant widening of the transition zone. Using a constant recharge to generate a steady-state starting point would thus result in an unrealistically narrow transition zone, as discussed below. During the 122-year period, all model parameters remained the same except that no abstraction was considered and that the 3H concentration of the meteoric recharge was assigned a constant value of 5 TU, the estimated natural background (Craig and Lal, 1966; Roether, 1967). It was further assumed that seawater had
To include abstraction in the two-dimensional cross-sectional simulations, the reported abstraction volumes were divided by the estimated ground surface area of the well field zone of influence of
Post and Houben [
The results of simulations A1 and A2 are shown in Figure
This result highlights the importance of explicitly modelling rainfall recharge as a transient process for understanding the development of the transition zone in coastal aquifers. The importance of the temporal rainfall recharge variability has also been recognized in other modelling studies of island aquifers. For example, Griggs and Peterson [
The effect of the 3H bomb peak can clearly be recognized as a quasihorizontal, elongated zone of high 3H concentrations, which starts at a depth of 20 to 40 m below sea level near the left model boundary and gradually rises closer to the surface with decreasing distance to the shoreline (Figure
Cross sections showing contour plots of the simulated 3H concentrations (shaded colours) and Cl concentrations (white lines) at the end of the simulation period (31 December 2013). The five white lines represent Cl concentrations equal to 2.5%, 16%, 50%, 84%, and 97.5% of the seawater chloride concentration (17 g/L). Results are shown for simulations (a) A1, (b) A2 (=A31), and (c) A320. The dashed white line in (a) and (b) indicates the position of the vertical profiles shown in Figure
Temporal evolution of the total 3H in Bq (solid lines) and 3He (dashed lines) in the freshwater lens (groundwater with less than 2.5% seawater) for simulations A1 (red) and A2 (blue).
The graph in Figure
Solomon and Sudicky [
(a) Apparent age
Comparing simulations A1c and A2c shows that the transience of the recharge similarly results in a small error of the calculated 3He/3H ages even when the 3H input signal is constant in time. In the case of simulation A2c, the maximum difference of the 3He/3H age compared to simulation A1c is 1.8 years. The difference is due to the greater dispersion in simulation A2c than in simulation A1c, but the magnitude of the effect is only of secondary order compared to that of the bomb peak 3H input.
The apparent age shift due to the nonconstant 3H input is also apparent when the sum of 3H and 3He is plotted as a function of
Comparison of the 3H concentration in precipitation with time (same data as shown in Figure
The recirculation of seawater results in an asymmetric funnel-shaped zone of elevated 3H concentrations in the saline groundwater just seaward of the coastline (Figure
An important factor that determines the circulatory flow pattern of seawater is the vertical hydraulic conductivity [
In contrast to the relative constancy of the freshwater lens geometry, Figure
Dependency of (a) seawater inflow
The practical implication of this finding is that for studies of submarine groundwater discharge it may be possible to use 3H to constrain the recirculated seawater component. For conditions similar to those simulated here (i.e., a sufficiently homogeneous aquifer) and assuming the seawater is well mixed and the vertical anisotropy is constrained, 3H measured in water samples from a multilevel observation well at the shoreline might be used to infer the magnitude of
In simulation A4, the horizontal hydraulic conductivity of the upper 40 m was lowered (Figure
Contour plots of the simulated 3H concentration activities (shaded colours) and Cl concentrations (white lines) at the end of the simulation period (31 December 2013). The five white lines represent Cl concentrations equal to 2.5%, 16%, 50%, 84%, and 97.5% of the seawater chloride concentration (17 g/L). Results are shown for simulations (a) A4, (b) A5, (c) A6, (e) A7, (f) A8, and (g) A9. The location of the clay layer is indicated by the hatched white rectangle. The open red rectangle in (d)–(f) encloses the cells with groundwater abstraction, and the dashed white line indicates the projected location of observation well 17.
A complex distribution of chloride and 3H developed above the clay layer. The reason for this is that the clay has a dispersing effect on the groundwater discharge near the coast. Rather than being focussed in a relatively narrow zone, the discharge extends over circa 200 m in simulations A5 and A6 (as well as A8 and A9). Hence, the upward flow rates decrease, and the propensity for density-driven downward flow becomes greater. Consequently, a complex pattern of upwelling and downwelling occurred, and water fluxes across the aquifer-seawater interface were spatially and temporally variable. Similar effects have been noted in numerical simulations of submarine groundwater discharge by Kooi and Groen [
With respect to the role of the clay layer, it is also interesting to note that simulations A5 and A6 had almost the same chloride and 3H concentration patterns below the dune area (Figure
Simulations A7–A9 included the effect of groundwater abstraction. Comparing Figures
The simulated and measured variations of the chloride concentration as well as the 3H and 3He concentrations with depth are shown in Figures
This outcome is somewhat surprising as simulation A7 did not include the clay layer, which is known to exist in the area albeit that there is uncertainty about its hydrogeological significance. Possibly, this result indicates that simulations A8 and A9 overestimated the effect of the clay layer on solute transport in the lens. This might be because its permeability is higher than the value used in the model (Figure
The depth of the transition zone at the projected location of well 17 was overestimated in simulations A8 and A9. The midpoint of the transition zone was well matched by simulation A7 (Figure
In none of the simulations did 3H originating from intruded seawater reach the location of the observation well, which most likely rules it out as a source of the observed 3H in the transition zone at appreciable inland distances from the shore. This contrasts the findings by Stuyfzand et al. [
But even with a longitudinal dispersivity
The model results presented in this paper demonstrate how atmospherically derived tritium behaves in freshwater lens groundwater systems under different combinations of recharge variability, aquifer anisotropy, lithological heterogeneity, and groundwater abstraction. The results reinforce earlier findings by Solomon and Sudicky [
The model results can be synthesised in terms of groundwater age classes as shown in Table
Time since recharge for freshwater and seawater depending on groundwater 3H and 3He concentration (all numbers in TU) for simulation A7 with
Freshwater | ≥60 yr (3He ≈ 5) | <10 yr | ≥20 yr | <10 yr ( |
Saltwater | ≥40 yr (3He ≈ 0.2) | Mixture of old seawater and |
Mixture of seawater and ≥20 yr freshwater (brackish) | <10 yr ( |
Using a transient recharge input as opposed to a constant recharge had only a secondary effect on apparent ages but did result in a significant widening of the transition zone between fresh- and saltwater. While the width of the transition zone (as measured at well 17) could be approximated by increasing the dispersivity by an order of magnitude (simulation A7), the simulated 3H concentrations for the transition zone underestimated the measured values. This suggests that the model still underrepresented mixing between fresh- and saltwater, despite the adopted dispersivity value of
Despite uncertainty about the magnitude of the dispersion coefficient and processes that control the width of the transition zone, our models suggest that the 3H observed in the transition zone on Langeoog more likely derived from the freshwater lens than from the North Sea. In none of the simulations did 3H from the North Sea reach far enough inland to explain the observed values.
In the offshore part of the aquifer, there can be a pronounced funnel-shaped zone of elevated 3H concentrations that stems from the circulation of seawater driven by dispersion in the transition zone. The vertical hydraulic conductivity and the presence of a clay layer were found to exert a strong control on the maximum depth to which measurable 3H can be found in the subsea portion of the aquifer, as well as on submarine groundwater discharge patterns. Therefore, groundwater 3H measurements at the coastline or below the seafloor could see application as an indicator of the strength of the seawater circulation and SGD in the southern North Sea area and other coastal sites where 3H or 3He variations can be resolved with sufficient analytical precision.
Previously reported field data were used to support this study and are available at [doi
The authors declare that they have no conflicts of interest.
The authors would like to thank the Landesamt für Bergbau, Energie und Geologie (LBEG), of Lower Saxony for the technical support during the sampling of the wells on Langeoog and the Oldenburgisch-Ostfriesischer Wasserverband (OOWV) for providing data. Dr. Matthew Currell provided thoughtful reviewer comments that led to significant improvements of the manuscript.
This file contains a table with the data previously published by Houben et al. [