Groundwater mean residence time of a sub-tropical barrier sand island

island Harald Hofmann1,5, Dean Newborn1, Ian Cartwright2, Dioni I. Cendón3, and Matthias Raiber4 1School of Earth and Environmental Sciences, The University of Queensland, St Lucia QLD, Australia 2School of Earth, Atmosphere and Environment Monash University, Clayton VIC, Australia 3Australian Nuclear Science and Technology Organization, Lucas Heights NSW, Australia 4Commonwealth Science and Industrial Research Organisation (CSIRO), Dutton Park QLD, Australia 5Geotechnical Engineering Centre Civil Engineering The University of Queensland, St Lucia QLD, Australia Correspondence: Harald Hofmann (h.hofmann@uq.edu.au)

bodies. A further consequence is the potential to damage groundwater-dependent ecosystems and destroy a vital groundwater source for coastal regions (Rao and Charette, 2012;Post et al., 2013b;Parizi et al., 2019).
The purpose of this study is to determine groundwater flow paths and MRTs in a Quaternary dune sand aquifer on the world's second largest sand island, North Stradbroke Island, Queensland, Australia (Laycock, 1975;Ulm et al., 2009). The island is an important part of the coastal environment in South East Queensland and provides a vital groundwater resource, recreational space for the large urban areas and most importantly unique coastal ecosystems with many freshwater wetlands, lakes and marine coastal environments . The island also plays a major role in the paleoclimate reconstructions for the east coast of Australia as some of the longest sediment records have been extracted from North Stradbroke Island wetlands (Barr et al., 2013;Tibby et al., 2016Tibby et al., , 2017Barr et al., 2019). Apart from a small number of government reports (Leach, 2011), there are few studies of the groundwater resource of the island and there is no prior work on residence times. This study closes this knowledge gap using major ion, stable isotope and radioactive tracer data from 21 sites along an east-west transect across the main centre of North Stradbroke Island (Fig. 1). 3 H and 14 C are used in combination with major ion chemistry and stables isotopes to estimate mean transit times, flow paths of groundwater and potential inter-aquifer mixing on the island.
A more thorough understanding of the islands groundwater flow paths and transit times significantly improves the current 70 understandings of the hydrogeology of North Stradbroke Island and barrier islands in general. This study is relevant to most barrier islands in terms of water resource management and predicting fresh water resources with a changing climate, sea levels changes and increasing urbanisation in coastal environments.   impellor pump (ThemoFisher Inc. Super Twister) was used for the shallower bores. The samples were collected after purging the bores for approximately 5 bore volumes. Bore RN14400075 was pumped dry and sampled using a bailer a day later. In nearly all cases the high hydraulic conductivities of the sands resulted in no or only minor drawdown during purging. At each bore, 6 x1000 mL samples were collected in high density polyethylene (HDPE) bottles, 4 of which were bottom filled and sealed for 3 H and 14 C analysis. In-situ measurements of temperature, pH, ORP, DO and EC were taken using a GPS AquaRead 135 with an AP-800 probe (ThermoFisher Inc.). Samples from the Wonky Holes were taken in the middle of the holes by pushing an aluminium tube approximately 1 m into the sand to avoid sampling a mix of fresh water and ocean water. Water was then pumped from the tube using a GeopumpTM peristaltic pump (Geotech Environmental Equipment, Inc.). At the end of each sampling day, water samples were titrated for CO 2 and HCO 3 concentrations using a Hach titration kit; precision of these concentrations is 5%. 125 ml of filtered (0.45 µm) samples were taken for stable isotope and major ion anion analysis. 125 ml of sample was separated, filtered and acidified with 70% HNO 3 for later cation analysis. All samples were kept cool until analysis.
The filtered and acidified cation samples were analysed using a ThermoFinnigan quadropole ICP-MS at Monash University.
Anion samples were filtered using 0.45 µm cellulose nitrate filters and concentrations were determined using a Metrohm ion chromatograph also at Monash University. The precision of major ion concentrations based on replicate analysis is 2-5%. and counted using 3 Quantulus ultra-low background liquid scintillation counters (LSC). This analysis had a combined standard uncertainty of 0.04 TU and a quantification limit of 0.05 TU, further analytical details are described in Neklapilova (2008).
Following acid extraction and graphitisation of inorganic carbon, 14 C concentrations were analysed using accelerator mass spectrometry (2MV STAR Tandem Accelerator). The concentrations for 14 C are expressed as percent modern carbon (pMC) and the precision of 14 C/ 12 C ratios is +/-0.05%.

Estimating MRT
Tritium is a suitable tracer for determining residence times of young groundwater. It is part of the water molecule and once isolated from the atmosphere its activities are only affected by radioactive decay. 3 H has a half-life of 12.32 years and may be used to estimate residence times of water that are up to 100 years with a precision of a few years (Morgenstern and Taylor, 2009;Morgenstern et al., 2010). The activities of 3 H in rainfall is known with sufficient precision over time in many areas of 165 the world to derive a local 3 H input function (Tadros et al., 2014). Brisbane is the closest 3 H monitoring station to Stradbroke Island (approximately 35 km from the island) and has a continuous record of 3 H data from the 1960s until 2012 (Tadros et al., 2014; IAEA -Global Network of Isotopes in Pricipitation, 2017).
The 3 H activities peak in the 1950's to 1960's due to the production of 3 H by atmospheric nuclear tests (the 'bomb pulse' 3 H).
Traditionally, the propagation of the bomb pulse has been utilised to trace the flow of water recharged during this period (Fritz 170 et al., 1991;Clark and Fritz, 1997). Because the bomb pulse 3 H peak was several orders of magnitude lower in the southern hemisphere than in the northern hemisphere, 3 H concentrations of remnant bomb pulse water in the southern hemisphere have now decayed well below that of modern rainfall. This situation allows estimates of MRT to be obtained from single 3 H activities (Morgenstern et al., 2010;Morgenstern and Daughney, 2012). Fritz, 1997). While it is not particularly well suited for estimating residence times of young groundwater it may be used to detect the input of groundwater from old water stores, such as low permeability layers, within the aquifer system (Hofmann and Cartwright, 2013). The combined use of 3 H and 14 C may also be used to assess mixing of older groundwater with recent recharge (Cartwright et al., 2007). Groundwater MRT were estimated using lumped parameter models (LPM) (Maloszewski and Zuber, 1982) (Maloszewski, 2000). These models predict the distribution of ages and tracer concentrations in homogenous 180 aquifers with simplified geometries under steady state conditions. The mean residence time represents the average age of the individual water molecules in the sample. The concentration of a radioactive tracer in the sample C out (t) is related to the input over time C inp (t) via the convolution integral: where λ is the decay constant of a radioactive tracer (Farlin and Maloszewski, 2013) and g(τ ) is the transfer function and simulates the distribution of a wide variety of aquifer geometries (Cartwright et al., 2017). It allows for variable degrees of dispersion by adjusting the dispersion parameter Dp, which describes the ratio of dispersion to advection. It approaches zero when advection becomes the dominant process controlling the tracer transport. The EMM is most suitable for homogeneous, unconfined aquifers of constant thickness with uniform recharge. The EPM applies to aquifers that have regions of confined and unconfined flow (Maloszewski and Zuber, 1982;Zuber et al., 2005;Jurgens et al., 2012;Atkinson et al., 2014;Cartwright 195 et al., 2017). The PEM is applicable to the same type of aquifer as the EMM but is used when only the lower part of the aquifer is sampled by a well (Jurgens et al., 2012). The PEM ratio is defined as the ratio of unsampled thickness of the aquifer to the sampled thickness. For bores screened across the total saturated thickness of the aquifer the PEM ratio equals 0 and the PEM equals the EMM.
For this study lumped parameter models contained within the programmable Excel spreadsheet TracerLPM was used (Jur- were estimated by matching the measured radioactive concentrations to those predicted from the lumped parameter models.
While groundwater transit times through aquifer systems are expected to be orders of magnitude larger than short-term rainfall variability, tracer concentrations from accumulated yearly rainfall are the best representation of the tracer input concentration

Groundwater hydraulic heads and flow
Groundwater bore 14400088 is probably screened in a perched aquifer system that surrounds Brown Lake ( Fig. 1) while all 220 other bores are screened within the regional groundwater system. The distinction between perched and regional aquifer system (in this case regional refers to the connected aquifer system across the island) is made on the bases of nested bores 14400088 and 14400087 in this study. On the day of sampling, the hydraulic head in bore 14400088 was 58 m while that in bore 14400087 was 35 m. This 20 m hydraulic head difference indicates the presence of a perching layer. Bores 14400151 and 14400152 are located in the coastal beach/shore dunes between the Eighteen Mile Swamp wetland and the ocean on the eastern side of the 225 island.
Hydraulic heads in the regional aquifer are highest (∼35 m) in the centre of the island and the unsaturated zone is approximately 30 to 60 m thick. The hydraulic head values decline towards the coasts and reach sea level in Eighteen Mile Swamp in the east and close to sea level on the western side nearby Dunwich. Long-term decadal fluctuations of regional groundwater heads are <0.5 m to approximately 5 m. These are not correlated to yearly rainfall and therefore represent longer timescale 230 climate fluctuations (Fig. 2).

Major ion chemistry and stable isotopes
Electrical conductivity (EC) of the groundwater is generally low across the island, ranging from 57 to 257 µS cm  wedge developing underneath the island and only shallow areas in beach dunes and intertidal areas have marine dominated groundwater. Most of the groundwater is acidic due to the limited buffering capacity of the relatively clean quartz sands with pH values ranging from 3.6 to 7.5 with an average of 4.9. Most of the groundwater is oxidised with an average oxidation-reduction 240 potential of +111 mV.
In general, the geochemistry of the groundwater shows only minor variations. Most of the samples (74%) are Na-Cl type groundwater, 21% are Ca-Na-HCO 3 groundwater and 5% are Ca-HCO 3 groundwater (   generally decrease with bore depth and increase with distance from centre of the island towards the coastlines (Fig. 5 B). The decrease is more pronounced towards the east coast then towards the west coast. Groundwater from bores 14400134, which is the deepest bore at 131 m below natural surface in the central part of the island, and bore Testhole C, which is located close to the west coast at a depth of 43 m (Fig. 1), have 3 H activities that are below detection. The water discharging from the two Wonky Holes also has low 3 H activities of 0.12 and 0.15 TU. The highest 3 H activity of 1 TU is from groundwater in bore 265 14400088, which is a shallow bore (4.6 m) in a perched aquifer system close to Brown Lake.
Most 14 C concentrations range from 59 to 111 pMC with the majority of samples having concentrations above 90 pMC.
The lowest 14 C concentrations are in groundwater from bores 14400134 (77 pMC) and Test hole C (63 pMC) ( Fig. 6 A & B), which also had the lowest 3 H activities. Groundwater from the Wonky Holes has 14 C concentrations of 81 and 59 pMC.
Generally, the high 3 H activities and high 14 C concentrations of most of the groundwater imply that it was recently recharged. values of -12‰ (Fig. 6 A).

Discussion
The combined hydraulic head and geochemistry data allow the conceptualisation of groundwater flow across North Stradbroke Island. Groundwater flows from the centre of the island towards the east and the west coasts driven by the large hydraulic gradient. There is groundwater discharge into freshwater wetlands (e.g. Eighteen Mile Swamp), some submarine groundwater 280 discharge via the Wonky Holes into Moreton Bay and probably offshore on the eastern side of the island (Fig. 1). The sand dunes are thickest in the central dune field and the thickness gradually declines towards both coastlines. The thinning of the aquifer and the unsaturated zone allows mixing of groundwater from the centre of the island with more recent recharge closer to the coast towards the discharge zones.
Most of the groundwater has low TDS and low pH. This together with the low Na/Cl ratios (< 1.2) suggests only minor sili-285 cate weathering is occurring, limited by the availability of weatherable silicates. The linear relationship between Ca and HCO 3 ( Fig. 3 G) suggests that some carbonate weathering has occurred. However, the observation that δ 13 C values are generally similar to those expected for DIC derived from the soil zone implies that this is limited. SO 4 and Cl concentrations increase near the coastline suggesting some marine influence (sea spray) in the coastal lowlands of the island. However, the salinity remains low, supporting the argument that the salt water wedge underneath the island is relatively deep close to the bedrock. the exponential-piston flow model assumes that the entire aquifer thickness is sampled, which is not the case. Here we use the partial exponential model and calculate the PEM ratio for each bore using the screen width and depth below the water table. This model assumes that flow system is exponential and so ignores the likelihood of piston flow in the unsaturated zone.
However, as the thickest unsaturated zone is approximately 60 m and the total flow path length is ∼5000 m, the proportion of the piston flow component is very small.
MRT calculated using the partial exponential flow model range from 37 years to >150 years. Below background 3 H ac-300 tivities in groundwater samples from Test Hole C and 14400134 suggest MRT past the age range of tritium (Tab. A2). Deep groundwater in the centre of the island where the sand dunes overlay the bedrock has MRT > 150 years (Fig. 8 B). determining MRT is subject to uncertainties, the relative distribution of older and younger water does not change according to which lumped parameter model is used (Fig. 7 A).
While groundwater was expected to have young MRT some of the groundwater has relatively low 14 C activities. Estimating MRT using 14 C activities requires that the addition of 14 C-free carbon from the groundwater flow system be accounted for.
Significant addition of 14 C-free carbon may dilute the 14 C activities, potentially resulting in MRT being overestimated (Coet-  not possible for waters to have higher 14 C as that would require the initial 14 C activity to be greater than that recorded in the atmosphere. In the case of Stradbroke Island groundwater, this implies that q values cannot be substantially lower than 0.8. MRT were calculated by adjusting the 14 C input function using q values of 0.85-0.95. Some of the adjusted groundwater 345 14 C MRTs are younger than 200 years and are therefore considered to be modern (Fig. 7 B).

Disparities in groundwater MRT
The calculated groundwater MRT from the radioisotope tracers are generally higher than what was initially hypothesised.
Existing MRT were based on flow rates estimated using the regional groundwater model Leach and Gallagher (2013). This assumed lateral hydraulic conductivities of 1 to 40 m day −1 based on Leach and Gallagher (2013). The large variance in 360 hydraulic conductivities comes from the inclusions of isolated peat and clay layers (Leach and Gallagher, 2013). By contrast hydraulic conductivities calculated via Darcy's Law with velocities from the 14 C MRT, porosities from Leach (2011) and the measured hydraulic gradients are generally between 0.25 and 1 m day −1 . There are two likely explanations for these lower hydraulic conductivities; a) some groundwater discharge to the unconfined sand aquifer from the basement units underneath the sand dunes, and b) a larger volume of geological units with lower permeability such as peat, coffee rock and clay with more 365 control on groundwater flow than prior studies suggest.

Potential influence of geological basement
The geological basement underneath the island is comprised of the Woogaroo Subgroup, Rocksberg Greenstone and Rhyolitic intrusions. A regional evaluation of aquifer storage and retention for South-East Queensland identified aquifers within the units summarising some of the general hydrogeological characteristics, such as permeabilities, transmissivities and general 370 groundwater flow (Helm et al., 2009). In the case of the Rocksberg Greenstone water is stored in fractures infilled with clay and the typical yield is less than 0.2 Ls −1 . The Woogaroo Subgroup sandstone has a typical yield ranging from < 1.5 Ls −1 to 6 Ls −1 and was identified to have moderate groundwater storage potential (Neuman, 2005). There is a possibility that the underlying basement units on the island are connected to the lower sand dunes and groundwater enters from these formations the young dune water. The basement isopach, which was extrapolated from the regional geological model for South-East Queensland, indicates the proximity of each sampling location to the bedrock (Fig. 8 B). The closer the groundwater samples are to the basement geology the older their respective MRT. This in itself is not a compelling argument, as it is also in line with the progression of MRT from shallow to deeper sections of the aquifer but having larger MRT in the west of the island where the bedrock is shallow might indicate a small degree of discharge from the basement to the dune aquifer. To the best of our knowledge there are no groundwater bores in the basement on the island but general groundwater heads on the mainland 380 suggest that this could be a possibility, however, the volumes would be minor, if at all, compared to the volume of the sand aquifer.

Potential influence of lower permeability units on MRT
Over Cl, Br and HCO 3 . Almendinger and Leete (1998) found elevations in the same major ions and trace elements when comparing wells above peat layers to wells below peat layers. The large enrichment of elements seen in this sample and its disparity to 405 other MRT suggests that this bore receives groundwater from the peat unit forming Eighteen Mile Swamp where the hydraulic conductivities are relatively low and concentration of the aforementioned ions and elements are high.

Conceptualisation of groundwater flow
Variations in groundwater MRT estimated from 3 H and 14 C throughout North Stradbroke Island support the current idea of groundwater flow from the centre of the island towards both coasts. There is also robust evidence that groundwater has a large 410 vertical flow component (Fig. 9). Nested bore sites show a distinct increase in MRT as bores protrude deeper, while single bores that extend deeply also display increased MRT (Fig. 8 A & B). While the MRT variation is not as strongly pronounced in lateral flow direction, the combination of known hydraulic gradients and known areas of discharge give prove to the idea of lateral flow to be consistent with MRT distributions. Modelling by Chen et al. (2003) and Leach and Gallagher (2013) suggests that groundwater on North Stradbroke island is discharged via: a) movement into coastal wetlands (55%), b) submarine discharge 415 around the coastline (27%), c) groundwater extraction (16%) and d) discharge into blue lake (2%) (Chen et al., 2003). The conceptual model of groundwater flow along the sampled transect is divided into two areas. An area where groundwater flows to the west and an area where it flows to the to the east (Fig. 9). There are minor flow diversions to the north and the south resulting from the centre of the transect also being the highest elevation on the island and topography drops from there into all four directions. It is assumed that recharge occurs across the whole island and as such recent recharge is added to the lateral 420 groundwater flow at all points along the cross section.
The east coast of the island is comprised of a large freshwater wetland system at the foothills of the major dune system. This wetland, Eighteen Mile Swamp, separates the major Quaternary dunes from the fore dunes. Groundwater from the main dunes discharges into the wetland and re-enters partially into the fore dunes along the coastline (Fig. 9). However, it appears that there are at least two aquifer systems that are separated by a lower permeability unit. This lower permeability unit was identified as 425 a marine clay underneath the peat sequences of Eighteen Mile Swamp by Mettam et al. (2011). Nested bores 14400151 (67 m) and 14400152 (17 m) indicate younger MRT and higher salinity (173 mg L −1 TDS) in the shallow part of the fore dune system and older, lower salinity in the deeper parts which is also reflected in bore 14400051 in the main dune system (Fig. 1).
There is also an upward gradient indicating a hydraulic connection of the deeper fore dune to the regional groundwater system and potentially to Eighteen Mile Swamp (Fig. 9).

430
The west coast of the island has generally lower topographic gradients and extended salt marshes that continue into the mangrove tidal areas of Moreton Bay. Sequences of fine, organic matter rich muds overlay the dune sands which lead to semiconfined conditions in the sand aquifer in the direct proximity of the coast underneath the muds (Fig. 9). The Wonky Holes are circular disturbances in the mud sequences where fresh groundwater from the island's sand dunes discharges into the salt water environment of the bay. The controls surrounding the formation of these discharge points come to existence is unknown, but 435 an upward head gradient in the underlying confined system, bioturbation and heterogeneities in the mud sediments possibly created an opening to the surface. The long MRT of water discharging through the Wonky Holes suggest that most of the water derives from the deeper confinded sand aquifer units that are linked to the centre of the island (Fig. 9).
The results of this study indicate that the perched groundwater systems have a significant effect on groundwater flow, recharge inhibition and intermediate water storage. Many lake and wetland systems on the island exist around perched aquifer ? Figure 9. Conceptual sketch of groundwater flow across the island from the centre of the island to the ocean in the east and to Moreton Bay in the west. Parts of the centrally recharged groundwater also flows to the north and thesSouth (Leach and Gallagher, 2013). Lower permeability units in the unsaturated zone can have longer residence times than direct recharge. There also may be groundwater contributions from the bedrock. On the western side some of the groundwater is discharged directly into Moreton Bay, some flows underneath the tidal mud flats in semi-confined conditions and is discharged partly through Wonky Holes into the Bay.

Conclusions
In summary, the combination of cosmogenic isotopes, major ion chemistry and stable isotope geochemistry were used to conceptualise groundwater flow on North Stradbroke Island with varying groundwater MRT in different parts of the island and determine groundwater flow paths through the aquifer systems. The groundwater MRT in this study differ to those found in similar studies of barrier island groundwater systems (Röper et al., 2012). The reason for the differences is that: the dune 450 stratigraphy of North Stradbroke Island is more heterogeneous with inclusion of peat, clay and indurated sands; volcanic and sedimentary basement units are underneath the island; and the island is much larger with a more variable topography, geomorphology and vegetation. MRT estimated using 3 H indicate a strong vertical stratification from 37 years to >150 years.
14 C MRT display similar temporal relationships with much greater ranges. This MRT discrepancy is attributed to different groundwater reservoirs. This study did not produce evidence for contributions from the fractured Woogaroo Subgroup sand-455 stone aquifer but the possibility remains. Water diversion and retention by low permeability units in the dune systems are so far the most likely course for relatively long MRT. The geochemical composition of groundwater remains relatively consistent throughout the island, with the only irregularities attributed to old groundwater stored within coastal peat. The stable isotope composition of North Stradbroke Island groundwater is similar to Brisbane precipitation without any indication of evaporative enrichment. The outcomes of this study can be incorporated in regional groundwater flow models to refine the potential inhi-460 bition and retardation of recharge to test model validities. The position of the islands large fresh water reservoir in a dry and populous South East Queensland means its potential to be used as a water resource is always high and background information on aquifer distribution and groundwater MRT are crucial to better validate impact assessment for water abstraction.    Table A2. 3 H residence times calculated with input function with modern tritium rainfall activities of 1.6 TU and 2.0 TU as well as the average in between the two results. Furthermore, 14 C ages are reported as conventional ages from the laboratory and further ages are calculated using a statistical q factor of 0.85 (Clark and Fritz, 1997) and a correction factor of q =0.95.