Groundwater sources for the Mataranka Springs (Northern Territory, Australia)

The Mataranka Springs Complex is the headwater of the iconic Roper River of northern Australia. Using environmental tracers measured in springs and nearby boreholes, the origin of groundwater contributing to the springs was evaluated to help assess the impact of proposed groundwater extraction in the Cambrian Limestone Aquifer (CLA) for irrigation agriculture and for hydraulic fracturing in the Beetaloo Sub-basin (an anticipated world-class unconventional gas reserve). Major ions, Sr, 87Sr/86Sr, δ18O-H2O, δ2H-H2O, 3H, 14C-DIC were consistent with regional groundwater from the Daly and Georgina basins of the CLA as the sources of water sustaining the major springs (Rainbow and Bitter) and one of the minor springs (Warloch Pond). However, 3H = 0.34 TU in another minor spring (Fig Tree) indicated an additional contribution from a young (probably local) source. High concentrations of radiogenic 4He (> 10–7 cm3 STP g–1) at Rainbow Spring, Bitter Spring and in nearby groundwater also indicated an input of deeper, older groundwater. The presence of older groundwater within the CLA demonstrates the need for an appropriate baseline characterisation of the vertical exchange of groundwater in Beetaloo Sub-basin ahead of unconventional gas resource development.

Stable isotopes of water. Two groups of samples can be distinguished based on δ 2 H and δ 18  H = -50‰) and two groundwater samples (RN034030 and RN034032) from bores in the eastern section of the study area had more evaporative enrichment. However, they were still within expectations relative to regional  www.nature.com/scientificreports/ samples from elsewhere along the Georgina flow path (Fig. 4b). The presence of evaporative enrichment in groundwater is a common feature of arid regions and is consistent with the Georgina flow path being in a more arid section of the CLA relative to the Daly flow path [10][11][12] . The inferred lower isotopic signature for rainfall in the Georgina Basin is probably due to a continental effect 10 (rainfall tends to become more isotopically depleted inland) and an amount effect 13,14 (groundwater recharge in arid climates mostly occurs following very large rain events, which tend to be more isotopically-depleted). Whilst data are limited, the isotopic composition of groundwater samples from the Antrim Volcanics and the Proterozoic Sandstone within or south of the complex were within expectations for the Georgina flow path of the CLA.
Tritium, 14 C and 4 He. The examination of age-dating tracers shows a wide mixture in springs and groundwater and a contribution from a deeper, non-CLA groundwater source (or sources). Most springs had 3 H < 0.05 TU (that is, were at the analytical detection limit), demonstrating a large component of pre-1950 water (Fig. 5a) (Fig. 5a). Most spring and groundwater samples had radiogenic 4 He (that is, had 4 He concentrations in excess of atmospheric equilibrium, 4-5 × 10 -8 cm 3 Standard Temperature and Pressure (STP) g -1 ) with concentrations ranging from 4.6 × 10 -8 to 2.6 × 10 -6 cm 3 STP g -1 (Figs. 5b,c). The highest 4 He concentration was found in the Antrim Plateau Volcanics groundwater sample. Whilst some accumulation of radiogenic 4 He during groundwater transport in the CLA is possible, the very high 4 He concentrations observed in many samples combined with the large spatial variability indicated a localised input of older groundwater from an underlying confined aquifer (or aquifers). However, as the 3 He/ 4 He ratio indicated admixture of pure radiogenic helium in all samples this deeper source was not derived from primordial mantle fluids 17 .
Other noble gases. The springs in the complex are notable for being 'warm' (25-33 °C). The trends in noble gases (here shown through Ne and Xe) indicate the warm temperatures are due to a relatively high ambient soil    4 He and 3 H; (c) 14 C and 4 He. The vertical grey boxes represent the range in 4 He concentration expected for a CLA groundwater sample that would be at equilibrium with the atmosphere (i.e. that has no radiogenic 4 He).
The horizontal stripped boxes represent the analytical detection limit for 3 H. www.nature.com/scientificreports/ temperature at the time of recharge rather than from geothermal heating (Fig. 6). All samples had a relatively large amount of excess air (that is, Ne and Xe concentrations were greater than expectations from solubility equilibrium). Using a combination of the unfractionated excess air model (UA) and the partial re-equilibration model (PR) for a moderate (10 cm 3 STP kg -1 ) and a large (

Discussion
Sources of water for the Mataranka Springs complex. The original conceptual model from Karp 20,21 was for the Mataranka Springs Complex to be fed via the Daly and the Georgina regional flow paths of the CLA. However, a component of local recharge feeding the springs was also anticipated owing the less permeable Cretaceous sediment cover overlying much of the CLA elsewhere being absent in the area surrounding Elsey National Park, which should facilitate local recharge processes. The findings of this study are largely consistent with Karp. Based on salinity, ionic composition, δ 2 H and δ 18 (Fig. 7). Thus, most of the groundwater sustaining the major springs (Rainbow and Bitter) is regional in nature, that is originates from recharge a significant distance away from the springs. A contribution from a deeper source (or sources) inferred from radiogenic 4 He had not been demonstrated previously for the Mataranka Springs Complex. As radiogenic 4 He was more elevated in Rainbow and Bitter springs, the deeper source(s) may also contribute to the baseflow of the Roper River. The timescale of regional groundwater transport within the CLA is too short (< 10,000 years) and the U and Th content of the limestone too low 22 to account for the elevated radiogenic 4 He content. However, the timescale of groundwater transport in underlying formations is unknown but anticipated to be much longer than in the CLA, which would foster the accumulation of radiogenic 4 He. This is consistent with the 4 He > 10 -6 cm 3 STP g -1 measured in Antrim Plateau Volcanics groundwater. There may also be a contribution from other geological formations underlying the CLA owing the presence of faults in the region 9 , including faults that have been reactivated in the recent geological history 23 . The concentrations of radiogenic 4 He found at the Mataranka Springs Complex are within The evidence for partial re-equilibration of the gas content with the atmosphere for some spring and groundwater samples indicates the radiogenic 4 He content provided here is conservative (that is, some of the initial radiogenic 4 He may have been lost during groundwater transit through the complex). The noble gas-inferred recharge temperatures are broadly consistent with the measured spring temperatures, which indicates the deeper groundwater source is not geothermally-heated or that its contribution is small relative to the one from the CLA. A more quantitative interpretation of the source of radiogenic 4 He will require the deeper formations in the complex and in Beetaloo Sub-basin to be instrumented and for near-surface measurements to be carefully corrected for the presence of excess air and for partial re-equilibration effects with the atmosphere.

Salt balance. Deeper groundwater discharge and extensive groundwater transpiration may both contribute
to the increase in groundwater salinity observed in parts of the Mataranka Springs Complex. Whilst springs are the more obvious groundwater discharge feature in the landscape, there is significant groundwater evapotranspiration throughout the complex, consistent with a net annual evaporative flux estimated by remote sensing (~ 250 mm year -1 ) 27 . This net evaporative flux would tend to increase the salinity of shallow groundwater, as is commonly observed in other regional groundwater discharge zones in arid environments 28,29 . Alternatively, the increase in groundwater salinity in the complex could be caused by saline groundwater discharge to the CLA from a deeper groundwater source (or sources). The two processes are not mutually exclusive and deeper groundwater discharge may also be partially transpired when occurring in areas with shallow water tables within the Complex. Thus, better constraining the contribution of the various groundwater sources to the springs will require both defining the geochemical signature of the deeper groundwater source(s) and an independent assessment of groundwater transpiration by phreatophytes within the Complex. Implications for water resources development. The potential impact of groundwater extraction in the CLA on the Mataranka Springs Complex will be commensurate with the extraction rate, the distance of the extraction from the complex, and the nature of the interaction between groundwater and the various groundwater-dependent assets within the complex 30 . Recent evaluations of proposed groundwater extraction scenarios for the Georgina Basin of the CLA suggest a potential 20% decline of the Roper River baseflow over 300 years 31 . However, as groundwater-surface water interactions in the Mataranka Springs Complex are not represented in current hydrogeological models of the CLA, the potential impacts of planned extraction on springs and other groundwater-dependent assets are unclear.
The potential for hydraulic fracturing to increase the vertical connectivity between water resources and deeper geological formations is a key concern of the public relative to unconventional gas resources development in Beetaloo Sub-basin 32 and elsewhere 33,34 . Demonstrating the presence of a deeper groundwater source at the springs (and for the CLA in general 24 ) is a key requirement for pre-development baseline assessments for www.nature.com/scientificreports/ unconventional gas resources [35][36][37] . Whilst it was established here that at least one deeper groundwater source is present pre-development at the springs, the exact origin of the deeper source (or sources), the mode of connectivity with the CLA (faults, contiguous formations, etc.) and the proportion of the spring flow attributable to deeper groundwater could not be evaluated because of the lack of infrastructure (that is, monitoring boreholes) in the geological formations underlying the CLA 24 . Due to a significant interburden thickness, hydraulic fracturing in Beetaloo Sub-basin is unlikely to materially increase the vertical hydraulic connectivity between rocks hosting unconventional gas plays and the CLA 38 . Nevertheless, additional baseline geochemical mapping and a field evaluation of the existing pathways for vertical connectivity are required to guide the development of unconventional gas resources in the Beetaloo Sub-basin 37,39,40 .

Methods
Study area. Climate in the study area ranges from tropical savanna near Mataranka and Katherine to hot semi-arid further south, with a distinct north-south monsoon-influenced rainfall gradient. At Katherine (Bureau of Meteorology Station 014903), annual precipitation varies from 678 to 1575 mm, annual potential evaporation is 2227 mm, and the average daily maximum temperature is 34 °C. The hydrogeology of the CLA is complex owing to the north-south rainfall gradient, distinct geological basins (Wiso, Daly and Georgina) with their own groundwater flow systems, extensive karst features, partial confinement, and the potential for vertical hydraulic connectivity with underlying geological units. At the Mataranka Springs Complex, the CLA is unconfined, relatively thin, but overlain by a layer of tufa (redeposited limestone) with extensive karst features (Fig. 2). Two local confined aquifers, the Antrim Plateau Volcanics basalt and the Proterozoic Sandstone and Siltstone, underlie the CLA near Mataranka and outcrop to the north of the study area. The extent of tufa deposition (> 10 m) at the Mataranka Springs Complex indicates the area has been a regional groundwater discharge zone for millennia. The Beetaloo Sub-basin underlies part of the Georgina Basin and is approximately 150 km south from Mataranka (Fig. 1). For groundwater samples, a submersible pump was lowered to the screen interval and the borehole casing was dewatered for at least three bore volumes before sample collection was initiated. A YSI™ multi-parameter probe (https:// www. ysi. com) was used to measure pH, specific electrical conductance (EC), temperature and dissolved oxygen concentrations. For major and minor elements, samples were filtered through a 0.45 µm nitrocellulose membrane filter and (for cations) acidified with (1% v/v HNO 3 ). Alkalinity was measured in the field using a HACH™ titration kit (https:// www. hach. com). Delta-18 O and δ 2 H samples were stored in 28 mL gas-tight glass bottles (McCartney). Tritium, δ 13 C-Dissolved Inorganic Carbon (DIC) and 14 C-DIC samples were stored in 1 L HDPE bottles, unfiltered with no headspace and with no preservative. Sr (and strontium isotopes) samples were collected unfiltered in 125 mL plastic bottles. Spring and groundwater samples for noble gases were collected using the copper tube method 41 .

Environmental tracer sampling.
Major and minor cations were analysed by a SPECTRO CIROS Radial Inductively Coupled Plasma Optical Emission Spectrometer and anions using a Dionex ICS-2500 Ion Chromatrograph at CSIRO Land & Water Analytical Services, Adelaide, South Australia. The charge balance error on the major ion measurements was ± 5%. Strontium and 87 Sr/ 86 Sr were measured by Inductive-Coupled Mass Spectrometry at the CSIRO Land & Water Analytical Services. Stable isotopes of water were measured by Isotope Ratio Mass Spectrometry at GNS, New Zealand, with a precision of 0.1‰ and 1.0‰, for δ 18 O and δ 2 H, respectively. Noble gases were measured at the CSIRO Land & Water Noble Gas Laboratory using offline separation of gases from water by cryogenic techniques, separation of reactive gases from noble gases using catalyst and getter systems and separation of the noble gas fractions using cryogenic techniques down to 13 K. Total gas content was evaluated using a spinning rotor gauge, the isotopic composition using quadrupole mass spectrometers and a high-resolution Helix MC noble gas mass-spectrometer 42 . Tritium was measured by electrolytic enrichment and liquid scintillation counting with a detection limit of 0.025 TU 43 . Radiocarbon and δ 13 C-DIC were measured by accelerator mass spectrometry and Isotope Ratio Mass Spectrometry at the Rafter Radiocarbon Laboratory with an accuracy of 0.2‰ for δ 13 C and a detection limit of 0.5% Modern Carbon (pmC) for 14 C.
Lumped-parameter models. The age-distribution of spring and groundwater was evaluated using lumped-parameter models 16,44 . For simplicity, only the results of the Exponential Model (EM) are reported here for two different initial 14