Interactive comment on “ Assessment of Halon-1301 as a groundwater age tracer ”

Water age dating is of fundamental importance in hydrology and hydrogeology since it provides quantitative information on the time scales of water movement through catchments. This has implications for both water resources, fluxes and processing times of biogeochemically relevant ions and molecules. Despite the importance of water age dating, there are very few tools that can be used in a quantitative manner to measure water residence times. The paper presented by Beyer et al further investigates the use of Halon-1301 as a new tracer for dating of young water in groundwater systems. Thus the paper makes a potentially important contribution to the available methods for dating groundwater. Technically the paper is sound and in my opinion needs only minor revisions. My main concern is that this new tracer actually gives us no new information to the methods already available. SF6 for example is actually more sensitive and


Introduction
Groundwater dating is a widely applied technique to determine groundwater flow parameters, e.g.recharge source and rate, flow direction and rate, residence time and volume.Age in itself is also increasingly used as a indication for quality and contamination risks (e.g. the New Zealand drinking water standard (Ministry of Health, 2008) and the European Water Framework Directive; EU Legislature, 2000).
Tracers, such as tritium, SF 6 (Sulfur hexafluoride) and various CFCs (Chlorofluorocarbons), are commonly used to infer groundwater age of relatively young groundwater (recharged < 100 years ago) by comparing their atmospheric history to their concentration found in groundwater.How-ever, all tracers have a restricted application range and face individual limitations, which can lead to ambiguous age interpretations (e.g.Allison and Hughes, 1978;Edmunds and Walton, 1980;Visser, 2009;Beyer et al., 2014a and references therein).As examples of these limitations, SF 6 has natural sources (e.g.Busenberg andPlummer, 2000, 2008;Stewart and Morgenstern, 2001;Koh et al., 2007); CFCs have a stagnant input function (Bullister, 2011), have anthropogenic point sources (e.g. in industrial and horticultural areas) (e.g.Oster et al., 1996;Stewart and Morgenstern, 2001;Busenberg andPlummer, 2008, 2010;Cook et al., 2006) and are known to be degradable in anoxic environments (e.g.Lesage et al., 1990;Bullister and Lee, 1995;Oster et al., 1996;Shapiro et al., 1997).Ambiguous age interpretations can be inferred from tritium measurements due to similar rates of radioactive decay and decrease in atmospheric concentration, which leads to similar concentrations of tritium in groundwater recharged at different times.This is particularly true for the Northern Hemisphere, where concentrations in young groundwater are still elevated due to H bomb testing in the 1970s (Taylor et al., 1992;Morgenstern and Taylor, 2009;Morgenstern et al., 2010).Additional interpretation issues follow from both the seasonal variability of groundwater recharge and tritium in rain.In these situations the tritium recharge is often estimated using recharge weighting techniques (Allison and Hughes, 1978;Stewart and Taylor, 1981;Grabczak et al., 1984;Engesgaard et al., 1996;Knott andOlimpio, 2001, Morgenstern et al., 2010).These limitations with ambiguity and input uncertainty can be overcome by time series or multiple tracer observations.To allow for more robust age interpretation of (relatively young) groundwater in future, there is a need for additional, complementary groundwater age tracers.
We have previously and unexpectedly identified the presence of Halon-1301 (CBrF 3 ) in modern water samples.Our paper immediately following this discovery (Beyer et al., 2014b) has detailed this identification, has discussed known Halon-1301 properties and has suggested this compound might have potential as a new, complementary groundwater age tracer (for water recharged < 100 years ago) to join the limited set of established compounds commonly used for this purpose.We have not inferred ages from Halon-1301 concentrations in that paper.However, we have provided a first insight into its performance by approximating Halon-1301 ages derived from corrected CFC-13 data presented in Busenberg and Plummer (2008).In this work, we analysed Halon-1301 in a range of groundwater locations, inferred Halon-1301 ages from its concentration, and compared these to groundwater ages previously inferred from other tracers.We additionally commented on (and analysed where possible) the various properties of Halon-1301 that had not previously been assessed in detail but may affect its wide-scale applicability as an age tracer.
since it is soluble in water (saturation: 30 mg L −1 at 20 • C; in contact with modern air (3.2 pptv): 7.5 fmol L −1 at 20 • C, 10 m elevation) (Deeds, 2008) and its increasing atmospheric concentration has been determined in the atmosphere since the 1970s by NOAA (National Oceanic and Atmospheric Administration) and AGAGE (Advanced Global Atmospheric Experiment) and data from 1969 to 1977 were reconstructed by Butler et al. (1999) (Fig. 1).Open questions remained regarding its conservativeness and contamination potential in groundwater environments.These are the following: -Is Halon-1301 degrading like the structurally similar CFCs in anoxic groundwater (e.g.Plummer and Busenberg 1999) or due to hydrolysis (e.g.Butler et al., 1991;Sturges et al., 1991;Kanta Rao et al., 2003)?
-Does Halon-1301 sorb to organic material in soil or elute from sampling material as suggested for CFCs (Reynolds et al., 1990;Cook and Solomon, 1995)?
-Does its use as a fire suppression agent and occurrence as a by-product during pesticide (Fipronil) production lead to "local" contamination of groundwater?To answer these questions, we analysed Halon-1301 in (river) water samples.The analysis allowed for simultaneous determination of Halon-1301 and SF 6 (Beyer et al., 2014b), which are both gaseous tracers with a similar behaviour in water.In this way, problems such as contamination due to contact with air during sampling or local (anthropogenic) sources could be identified.All groundwater samples have been previously dated with tritium, CFC-12, CFC-11 and SF 6 .We determined piston and exponential piston flow ages for Halon-1301 and SF 6 , as inferred by matching the historic input to the determined concentrations in the groundwater samples.Comparison of inferred Halon-1301 piston flow and exponential piston flow mean residence times (MRTs) to relatively robustly inferred tritium and SF 6 MRTs enabled for direct assessment of the performance of Halon-1301 as a groundwater age tracer.Because Halon-1301 and SF 6 are both gaseous tracers, they are expected to show similar behaviour in the unsaturated zone.Gaseous tracers equilibrate with the atmosphere during transport through the unsaturated zone and therefore do not account for this unsaturated zone travel time.This contrasts with inferred tritium ages, which do account for travel time through the unsaturated zone.Comparison of age information inferred from tritium and four different gaseous tracers (SF 6 , Halon-1301, CFC-12 and CFC-11) allowed for assessment of unsaturated zones processes or potential contamination/degradation of Halon-1301.Since some of the anoxic samples clearly have shown evidence of CFC degradation, comparison of Halon-1301 from these samples enabled a first understanding of the potential for degradation of Halon-1301 in anoxic groundwater systems.

Water samples
This study took advantage of the relatively well-defined age information of New Zealand groundwater inferred from time series tritium and SF 6 (and CFC) observations, particularly for confined aquifers (Morgenstern andTaylor, 2009, van der Raaij andBeyer, 2015).The inferred tritium ages were considered robust because of their well-defined input function (close proximity of our sampling sites to the high-resolution Kaitoke monitoring station) and because of long time series data in groundwater (Table 1).To enable a relatively comprehensive assessment of the potential of Halon-1301 as a groundwater age tracer, groundwater samples previously dated with tritium, SF 6 and CFCs covering a wide range of mean residence times and including anoxic and oxic samples and samples with apparent contamination/degradation of CFCs were chosen.We analysed 35 groundwater samples from 17 different sites in the Wellington region from 3 different aquifer systems (Lower Hutt groundwater zone (LHGWZ), Wairarapa groundwater system and Wainuiomata aquifer) and 8 river and equilibrated tap water samples for Halon-1301 and SF 6 , simultaneously.Groundwater samples in the LHGWZ and river water samples were collected as triplicates, of which two were analysed directly after sampling and one was analysed after 7 weeks storage at 14 • C. One river water sample was analysed after 1.23 years of storage at 14 • C. The location of the sampling sites and aquifer systems is shown in Fig. 2. The sampling sites, number of samples taken and corresponding aquifer systems are summarised in Table 1.Table 1 also includes the concentration of dissolved oxygen (DO), previously determined recharge temperature and amount of excess air (determined by Ar and N 2 analysis) (Jones and Gyopari, 2006;Stewart and Morgenstern, 2001;Tidswell et al., 2012) and the number of previously taken CFC (CFC-11 and CFC-12), SF 6 and tritium measurements.The groundwater systems are briefly described in the following.
Both the Wairarapa and the LHGWZ have formed in alluvial basins filled with greywacke gravel and marine deposits during glacial and interglacial periods.The Wairarapa is unconfined and is recharged both by rain and river infiltration while the LHGWZ is mostly confined and mainly river recharged.More detailed descriptions of the Lower Hutt groundwater system can be found in Grant-Taylor (1967), Reynolds (1993) and Gyopari (2014) and of the Wairarapa groundwater system in Begg et al. (2005) and Jones and Gyopari (2006).The Wainuiomata aquifer is a shallow, unconfined aquifer, which has formed in an alluvial valley filled with alluvial gravel and sand (Jones and Barker, 2005;WRC, 1993).
For determination of Halon-1301, a sampling procedure similar to the standard procedure for determination of water soluble gaseous tracers, such as SF 6 and CFCs was followed.To ensure the sampling of fresh unexposed ground- water (i.e.not the water stagnating in the dead volume of the well), the well was flushed at least 3 times of its volume and until conductivity, pH and DO stabilised.(Daughney et al., 2006).To avoid alteration of Halon-1301 concentrations with UV light and contamination or adsorption of the gas tracers (Halon-1301 and SF 6 ) from/onto the sampling material, only brown borosilicate glass bottles and nylon tubing were used and the use of PTFE/Teflon or other fluorine baring plastics was avoided (Reynolds et al., 1990;van der Raaij and Beyer, 2015).To avoid contamination of the samples with modern air, sampling was carried out under rigorous exclusion of air by inserting a nylon tube to the bottom of the sampling bottle and filling it from the bottom.Then the bottle was allowed to overflow, so that the water volume was replaced by several bottle volumes.The bottle was quickly capped and checked for presence of bubbles and if necessary the sampling process was repeated until no bubble is present.River water and a variety of equilibrated (at close to constant temperature) tap water samples were taken as representative modern water sample and to verify solubility data.This method presupposed no contamination by SF 6 or Halon-1301 from the air within our facilities, surrounding environment or river sampling locations, which seemed reasonable, due to the lack of sources of these compounds in the surrounding areas.Air samples were regularly analysed to confirm the lack of elevated SF 6 and Halon-1301 concentrations in our facilities.

Analytical system
The water samples were purged with ultra-pure (analytical grade) nitrogen gas in a vacuum sparge chamber (Busenberg and Plummer, 2000).Purging with nitrogen at a flow rate of 70 mL min −1 for 18 min was carried out to ensure complete degassing of the water sample in regards to removal of SF 6 and Halon-1301.The stripped gas then passed through a drying column (NaOH coated silica) to remove residual moisture and CO 2 to avoid interference in the detection system.To ensure consistent amounts of water samples were purged, the sparge chamber was filled until the filling mark (0.955 L) or the weight of the water sample was determined.If applicable, temperature and headspace volume were determined.Standard gas samples were pushed through a loop of known volume (9.97 ± 0.02 ml or 0.502 ± 0.001 mL, in the following referred to as 10 and 0.5 mL, respectively) and the temperature and pressure were recorded to determine the amount of standard gas analysed.
The samples (standard gas and purged gas from water samples) were then simultaneously analysed for Halon-1301 and SF 6 using a gas chromatograph with attached electron capture detector (GC/ECD) set-up including two cryogenic traps for pre-concentration (Busenberg and Plummer, 2008;Beyer et al., 2014b).The analytical set-up also allowed for simultaneous determination of CFC-12 (Busenberg and Plummer, 2008;Beyer et al., 2014b).However, an appropriately concentrated standard gas is needed to establish its calibra-tion curve.CFC-12 concentrations and inferred CFC-12 ages were therefore not determined in the study.
In the following the determination of Halon-1301 and SF 6 concentrations in water samples and resulting recharge year are described, which involves the determination of a calibration curve, solubility and where required excess air and headspace correction.

Calibration
The amount of Halon-1301 and SF 6 in all groundwater samples were determined by establishing a calibration curve (least-square fit, forced through 0/01 ) with approximately 10 mL certified air standard at various pressures.The certified air standard contained 3.27 ± 1.55 ppt Halon-1301 and 7.53 ± 0.81 ppt SF 6 among other gases (supplied by the Scripps Institution of Oceanography in 2011).A calibration curve was established every day before measurement commenced, since the performance of the GC/ECD can change from day to day, due to fluctuations in the environment (e.g.temperature) or aging of the material (e.g.column fill).Because Halon-1301 concentrations in 10 mL calibrated air standard did not sufficiently cover concentrations obtained in modern water samples, another standard gas containing 3.16 ± 0.3 ppb Halon-1301 and 1.02 ± 0.1 ppb SF 6 (prepared by New Zealand Industrial Gases (NZIG)) was used in a smaller standard loop of approximately 0.5 mL at various pressures.Additionally, tap water samples ranging from 1 to 15 L volume and 10 mL modern air samples at pressures from 1 to 3.5 bar were analysed to assess the linearity of the ECD signal towards Halon-1301 concentrations in the concentration range obtained in old to modern 1 L water samples.If linearity was found, then previously determined calibration curves (using the calibrated air standard) were linearly upscaled to estimate Halon-1301 concentrations in water.This was relevant for all groundwater samples for which calibration curves have been established at the time of measurement with calibrated air only.We were aware that this introduced additional uncertainty which we took into account (see Results section).
After determination of the molar amount of Halon-1301 (and SF 6 ) in a 1 L water sample purged in the vacuum sparge chamber, its equivalent atmospheric molar ratio at time of equilibrium (for groundwater samples at recharge) was determined using the solubility relationship (Henrys law, de-scribed in Supplement Sect.S1).In contrast to the solubility of SF 6 , which has been well studied and directly measured (Bullister, 2002;Wilhelm et al., 1977, Table 2), the solubility parameters of Halon-1301 have only been estimated by Deeds (2008) using the solubility estimation methods of Meylan and Howard (1991) and Meylan et al. (1996).Actual solubility measurements of Halon-1301 are not available in literature (according to our searches and further backed up by personal communication with Daniel Deeds, 6 March 2015).We used modern (equilibrated tap and river) water to estimate solubility to validate the solubility estimates.If applicable, the amount of Halon-1301 (and SF 6 ) in the water sample was corrected for headspace and/or excess air (previously determined by dissolved Ar and N 2 determination; Heaton and Vogel, 1981), also described in detail in Supplement Sect.S1.

Determination of recharge year
To infer the recharge year or residence time of the groundwater, the equivalent partial pressure of Halon-1301 and SF 6 in the atmosphere at time of recharge (determined as described above) was compared to their historic atmospheric records (illustrated in Fig. 1).Southern hemispheric atmospheric SF 6 records (Cape Grim station) are available at the NOAA (National Oceanic and Atmospheric Administration) (http://www.esrl.noaa.gov/gmd/;Thompson et al., 2004) and CDIAC (Carbon Dioxide Information Analysis Center) websites (Miller et al., 2008); data from 1973 to 1995 have been reconstructed by Maiss and Brenninkmeijer (1998).Southern hemispheric (Cape Grim) atmospheric Halon-1301 concentrations have been summarised and smoothed by Newland et al. (2013).Data from 1969 to 1977 have been reconstructed by Butler et al. (1999).We assumed that Halon-1301 concentrations are well mixed across the atmosphere of the Southern Hemisphere as suggested by Montzka and Fraser (2003) and Butler et al. (1998) and local sources of Halon-1301 are lacking as indicated by regular analysis of local air in this study, so that southern hemispheric atmospheric concentrations could be used to estimate concentrations of Halon-1301 in recharge.
Although a comprehensive analysis of potential local sources has not yet been carried out, studies such as that by Barletta (2011) in Los Angeles, USA, have not found local enhancement of Halon-1301 in city environments.We are aware of only one study that has found unusual fluctuations of Halon-1301 in the atmosphere: in two stations in Poland, at Krakow and Kasprowy Wierch stations.The research group is still investigating reasons, but speculate it may be attributed to local sources from close-by city/industry environments (Bartyzel, 2015).
In simple terms the recharge year can be found when observed (equivalent) atmospheric concentrations match historic atmospheric concentrations.This can be done using a simple "lookup" table to infer the piston flow recharge year.However, misleading age interpretations can be obtained  when using piston flow assumptions, which do not take account of mixing processes of groundwater in the aquifer or during sampling (e.g.Eberts et al., 2012).Therefore, lumped parameter modelling is often used to infer an age distribution and with it the MRT of the groundwater samples from tracer observations (Małoszewski and Zuber, 1982;Jurgens et al., 2012).In this study we adopt the commonly used exponential piston flow modelling (EPM), which had previously been found to best represent tritium (time series) and SF 6 observations in the studied groundwater.EP modelling was carried out using TracerLPM software (USGS) (Jurgens et al., 2012).
For 1 point tracer observations, as obtained for Halon-1301 and SF 6 in this study, a range of EPMs with various exponential to total flow ratios (referred to as 1/n; n has been defined as ratio of total to exponential flow by Małoszewski and Zuber, 1982) could be fit to the tracer observation.Since the mixing parameter could not be adequately constrained with a 1 point measurement of Halon-1301 and SF 6 , we constrained their 1/n ratio to the 1/n ratio previously inferred from tritium (time series) observations.We assumed this approach was adequate under the assumption of steady state at each sampling location, which has been indicated by assessment of time series hydrochemistry data (using trend and seasonality analysis).MRTs (using EPM or piston flow -PM) inferred from SF 6 and Halon-1301 concentrations were subsequently compared to previously determined MRTs inferred from tritium.We also commented on observed Halon-1301 concentrations in regards to previously observed degradation or contamination with CFCs (CFC-12 and CFC-11) in these wells.

Analytical uncertainty
Due to uncertainties related to the analytical procedure (calibration, analysis, etc.), the inferred recharge year and mean residence time (from Halon-1301 and SF 6 concentrations) can only be constrained to an age range.To determine the overall relative uncertainty, the Eurachem /CITAC (Cooperation on International Traceability in Analytical Chemistry) Guide CG4 (Ellison and Williams, 2012) was followed.This guide recommends the method described in Kragten (1994), which also implies a sensitivity analysis.The standard measurement error was determined as the total of the following (independent) uncertainties: where u1 is uncertainty from least-square regression (calibration curve), u2 uncertainty in standard gas concentration, u3 repeatability error from relative standard deviation of replicates, u4 uncertainty related to correction for headspace, u5 uncertainty related to correction for excess air, u6 uncertainty in recharge temperature and u7 uncertainty in solubility.
Replicate samples were analysed to determine the repeatability of the analysis.The absolute standard deviation is defined as where a i − xi is the difference between the concentrations obtained for one of the replicate samples a i with overall mean value xi for n samples and i number of replicates.The overall relative standard deviation can then be determined as median of all replicate samples: The limit of detection (LOD) and limit of quantification (LOQ) were determined by taking into account the slope and standard deviation (SD) of the calibration curve (Shrivastava and Gupta, 2011): Nomenclature: in the following the various forms of modern water (river and equilibrated tap water) are summarised and referred to as one sample type, namely modern water.Hence, all comparisons are made in relation to a total of 18 (17 groundwater + 1 modern water) samples.The term age or recharge year refers to an age or recharge year distribution, which is a function of MRT and mixing parameter (e.g.ratio of exponential to total flow for the EPM).

Calibration curve
Figure 3 illustrates the calibration curves of Halon-1301 obtained with the calibrated air standard (Scripps) and highly concentrated Halon-1301 standard (NZIG) with a nearly linear response of the ECD towards Halon-1301 concentration in the concentration range obtained for groundwater samples (signal up to 30 mV min −1 for modern water).Additional analysis of modern air at pressures ranging from 1 to 3.5 bar and analysis of water samples of 3 to 15 L (Fig. 4) confirmed the nearly linear response of the ECD towards Halon-1301 concentrations in this concentration range.Only for very high amounts of Halon-1301 (signals of approximately 1 order of magnitude higher than obtained in modern water) did the quadratic regression fit slightly better than the linear regression.Given this evidence of a linear signal response up to concentrations obtained in modern water, we linearly upscaled the calibration curve of Halon-1301 obtained with the calibrated air standard to estimate concentrations of Halon-1301 in all groundwater samples.Using this approach we introduced additional uncertainty, which we took account of during discussion of the inferred MRTs (for further detail see Sect.3.4: "Assessment of inferred Halon-1301 ages" and Supplement Sect.S2 "Assessment of elevated Halon-1301 ages").

Uncertainty
The analysis allowed for an average repeatability of 3.6 % for Halon-1301 (2.8 % for SF 6 ) and 9.8 % (6.9 % for SF 6 ) average standard deviation of the calibration curve.On average the overall analytical uncertainty in an average 2 New Zealand groundwater sample was 4.7 % for Halon-1301 (9.0 % for SF 6 ).This led to a larger uncertainty in inferred piston flow age for waters recharged before 1975 and after 2 A detailed study in New Zealand has shown groundwater samples have on average a recharge temperature of 12.1 ± 1.8 • C, 2.9 ± 1 mL (STP) kg −1 excess air and a headspace volume of 0.5 ± 0.05 mL (van der Raaij and Beyer, 2015).about 2000 when using Halon-1301, due to its characteristic S-shaped input function (Fig. 5).The LOD and LOQ of the analytical set-up was 0.32 and 0.98 fmol L −1 for Halon-1301, respectively (and 0.23 and 0.69 fmol L −1 for SF 6 , respectively).The LOQ was equivalent to a recharge year of 1975 for Halon-1301, at average recharge temperature (12.1 • C), 10 m elevation and lack of excess air and headspace.
Sensitivity analysis showed that the most significant contributors to the overall uncertainty were uncertainties related to the calibration curve, repeatability, excess air and headspace correction for Halon-1301 and SF 6 .Without considering headspace and excess air, the total uncertainty became only marginally smaller for Halon-1301 (4.4 instead of 4.6 %), but significantly smaller for SF 6 (3.2 instead of 9.0 %).Detailed determined uncertainties for each groundwater sample are shown in Figs. 6 and 7 and Table 3.
We note if SF 6 alone was analysed using a different GC column it could be more accurately resolved with 4.5 % overall uncertainty (van der Raaij and Beyer, 2015).However, our aim here was to simultaneously determine the two gaseous tracers SF 6 and Halon-1301 with a particular focus on resolving the Halon-1301 signal accurately.The higher uncertainty in SF 6 determination when using our approach may be resolved by adjustment of the column or ECD conditions or application of signal processing.
Please note that the analytical set-up also allows for simultaneous determination of CFC-12.This three way simultaneous determination of SF 6 , Halon-1301 and CFC-12 may allow for more robust groundwater dating, due to the ability to identify issues related to the limited application range of the individual tracers.These are contact with air during sampling (indicated by an increased concentration of all three gas tracers), degradation/contamination (indicated by a reduced/increased concentration of one or more of the gas tracers, respectively) or unsaturated zone processes such as diffusion (lag time) or retardation (indicated by a reduced concentration of all or one or more of the gas tracers in comparison to tritium ages).

Solubility
To test the reported solubility of Halon-1301, we determined the Henry coefficient (Eq.S1 in the Supplement) in equilibrated tap and river water samples and in relatively young groundwater (< 2 years MRT).These modern waters were collected for estimation of the solubility of Halon-1301.To estimate the robustness of the estimated Halon-1301 solubility, the solubility of SF 6 was also determined in these samples with the same method and compared to literature data.
Figure 6 shows the inferred solubility (ln K H ) of SF 6 and Halon-1301 in modern groundwater and equilibrated tap and river water compared to solubilities estimated by Deeds (2008) and Bullister et al. (2002) for Halon-1301 and SF 6 , respectively.Table 1 contains solubility parameters inferred from the found relationship in Fig. 6 along with previously reported solubility parameters.As can be seen, inferred solubility of SF 6 agreed well with its reported solubility, which indicated that our approach should give relatively robust Halon-1301 solubility estimates.Inferred solubility of Halon-1301 was significantly lower than estimated by Deeds (2008).When using the Deeds (2008) estimated solubility parameters, Halon-1301 concentrations were obtained which resulted in significantly older inferred Halon-1301 ages compared to tritium and SF 6 ages with an average discrepancy of +12 years in equilibrated tap and river water.This offset was removed when using our estimated Halon-1301 solubility parameters.
Due to absence of robust solubility data of Halon-1301, we used the solubility parameters estimated in this study (Ta- ble 3) to infer equivalent atmospheric Halon-1301 concentrations and with that infer Halon-1301 ages.Accurate measurement of the solubility of Halon-1301 is beyond the scope of this study.Due to the extremely low solubility of Halon-1301, specialised equipment is required.The estimated solubility had a relatively large uncertainty of 9.8 % (estimated for a regression analysis in Fig. 6), due to scatter in the data which may have been caused by uncertainty in recharge temperature, unaccounted heterogeneity, mixing of water, etc.The uncertainty in solubility added to the analytical uncertainty in equivalent atmospheric Halon-1301 concentration (estimated in the previous section), so that the overall uncertainty increased from 4.7 to 9.7 %.This increased uncertainty in turn affected the uncertainty in inferred Halon-1301 age as discussed in the following.

Overall
In the following we assessed inferred Halon-1301 mean ages in comparison to inferred SF 6 and previously inferred tritium and CFC mean ages.We considered elevated concentrations of Halon-1301, SF 6 or CFCs (> 10 %) as "potentially contaminated" and highly elevated concentrations (> 25 %) as "highly contaminated".Details on individual piston and exponential piston flow model MRTs inferred from Halon-1301 and SF 6 (in this study) and tritium (from previous studies) are listed in Table 3.
Inferred PM SF 6 and Halon-1301 ages (illustrated in Fig. 7) showed that Halon-1301 ages were on average 5.4 years higher than inferred SF 6 ages (over the entire age range), caused by reduced concentrations of Halon-1301 compared to SF 6 .However, piston flow ages are unrealistic, as they neglect mixing of water of different age in the subsur-   face or during sampling (e.g.Małoszewski and Zuber, 1982), also indicated by previously determined EPM ages inferred from tritium and SF 6 (e.g.Morgenstern and Taylor, 2009).In the following we applied an EPM and inferred MRTs from Halon-1301 and SF 6 concentrations.The choice of lumped parameter model significantly affected the age interpretation with Halon-1301, due to its S-shaped input function, which is skewed due to mixing processes (depending on the lumped parameter model choice).This highlighted the importance of considering mixing processes for inferring groundwater age from Halon-1301 observations.For SF 6 , this was less of a problem, due to its nearly linear atmospheric input since the late 1980s.The sensitivity of Halon-1301 concentrations towards mixing of groundwater of different age also implied that groundwater dating with Halon-1301 may allow better constraining of the mixing parameters compared to SF 6 .However, time series Halon-1301 data are necessary to confirm this supposition.

Consistency of inferred Halon-1301 ages with inferred tritium and SF 6 ages using the EPM
When using the EPM, inferred Halon-1301 and SF 6 MRTs agreed for the majority of sites (for 12 out of 18 sites) as summarised in Table 3. Inferred MRTs were considered as agreeing (i.e.insignificantly different) when their uncertainty bounds of 1 SD (except for one site where we accepted 1.1 SD) overlapped.The remaining sites indicated higher MRTs inferred from Halon-1301 compared to SF 6 .To assess whether these differences had been caused by processes affecting both gas tracers (such as lag time in the unsaturated zone) or only Halon-1301 (such as potential degradation or sorption which does not occur for SF 6 ), inferred Halon-1301 and SF 6 MRTs were compared to previously inferred tritium MRTs in Fig. 8.Where present, samples exhibiting probable CFC degradation/contamination are highlighted in Fig. 8.Comparison to inferred CFC ages could not be made, because all samples (with available CFC data) were degraded and/or contaminated in one or both of CFC-11 and CFC-12.At one of the 18 sites, both gases and tritium were close to the LOD, but evidence of slight contamination with modern air during sampling was found, indicated by elevated concentrations of both SF 6 and Halon-1301 which were incompatible with their low tritium concentrations.Evaluation of the performance of Halon-1301 as an age tracer in comparison to SF 6 and tritium was not possible for this sample, which was therefore excluded for the overall comparison.For the majority of the remaining 17 groundwater samples, inferred SF 6 ages agreed well with previously determined tritium ages, which indicated that unsaturated zone processes were not significant in this study.
Inferred Halon-1301 MRTs of 12 out of 17 sites were in agreement with inferred tritium and/or SF 6 MRTs (within an uncertainty of ∼ 1 SD).This included four older groundwater sites, which showed concentrations at or close to LOD of tritium and SF 6 , and were also free of Halon-1301 (Fig. 9 and Table 3).For the remaining waters (all relatively old and anoxic), inferred Halon-1301 ages were higher compared to tritium/SF 6 ages.The reasons for this offset are discussed in the following subsection.
As can be seen in Table 3, the relatively large uncertainty in estimated solubility led to additional uncertainty in inferred Halon-1301 ages (compared to estimates assuming only a 1 % uncertainty in solubility, for demonstration purposes).We found up to 16 years higher uncertainty in in- ferred Halon-1301 MRTs when accounting for the current uncertainty in solubility.Inferred Halon-1301 ages can potentially be better constrained with a more accurate solubility estimate.This also means the full potential of Halon-1301 as an age tracer cannot yet be realised due to absence of accurate/robust solubility data.

Conservativeness of Halon-1301
No significantly elevated Halon-1301 concentrations were found, despite that the sites cover situations of land use and well construction that result in CFC contamination.A significantly elevated Halon-1301 concentration was only found for one site in concert with an elevated SF 6 concentration, suggesting that the sample was contaminated with air during sampling (Fig. 8 and Table 3).This indicated no issues related to contamination for Halon-1301 from local sources at the studied sites.This has to be evaluated further, e.g. in groundwater recharged close to airports, where Halon-1301 is still in use as a fire suppressant during fuelling of planes.The lack of elevated Halon-1301 concentrations may also indicate that interference of the Halon-1301 signal with CFC-13 or other co-eluting compounds (as has been assessed in Beyer et al., 2014b) was not an issue in the studied groundwater samples.However, this needs to be assessed further in groundwater with elevated concentrations of CFC-13 or other potentially co-eluting compounds.
Significantly higher Halon-1301 MRTs to tritium and SF 6 MRTs (over an age range from 2.5 to 40 years MRT) were found in 5 of the 17 groundwater samples, where direct age comparison could be made.These samples also showed evidence of significant (even higher) degradation and/or contamination of one or both of CFC-11 and CFC-12.There are several possibilities for higher inferred Halon-1301 groundwater ages as a result of reduced Halon-1301 concentrations, which we assessed in detail in Supplement Sect.S2.A summary is presented in the following.Our assessment showed we could exclude degassing into headspace created by denitrification, production of methane or when groundwater is brought to the ground surface, since this would have affected all determined gas tracers, to the highest extend the least water soluble SF 6 , which we did not find in any of our samples.We could also exclude lag time in the unsaturated zone, because this would have also affected all gas tracers, dependent on their diffusion coefficient (Gooddy et al., 2006) and we did not find decreased concentrations of both Halon-1301 and SF 6 in any of our samples.Assuming that Halon-1301 behaved similarly to CFCs in regards to sorption to specific materials, we also considered the risk of sorption to well casing/sampling material was minimal as we followed a robust sampling procedure established for CFCs and SF 6 (using only borosilicate glass, stainless steel equipment and nylon tubing).
Potential degradation of Halon-1301 during storage was assessed by analysis of six groundwater samples from different sites (covering an age range from modern (< 1 year) to over 100 years MRT, and oxic to anoxic waters) stored for 7 weeks.The simultaneous determination of SF 6 and Halon-1301 allowed us to isolate Halon-1301 degradation, since SF 6 is not known to degrade in oxic or anoxic environments.Hence, an isolated reduced concentration in Halon-1301 would indicate Halon-1301 degradation, in contrast to a combined (Halon-1301 and SF 6 ) reduced concentration that would indicate, e.g., escaping of gas into headspace.Figure 9 illustrates concentrations determined before and after storage were within statistical uncertainty, indicating that Halon-1301 was stable in oxic and anoxic groundwater during storage for over 1 month at 14 • C. The concentration of Halon-1301 in one sample (river water) stored for over 1.2 years was also not significantly reduced compared to SF 6 .
The remaining possibilities for reduced Halon-1301 concentrations (i.e.increased inferred ages) were the following: (AGAGE, 2014), which made it more difficult to constrain the age of younger waters.
2. Increased inferred Halon-1301 ages in the remaining, particularly older samples with a MRT above 15 years were likely caused by (a) degradation, which is only likely to occur under anoxic/anoxic conditions (all affected samples are anoxic); (b) sorption to organic material in the aquifer (could not be excluded for any of the sites).
Further studies are needed to confirm whether Halon-1301 is degradable or reduced concentrations are a result of sorption/retardation in the aquifer.This can be studied by determination of Halon-1301 in relatively old (MRT of > 5 years) oxic groundwater and/or relatively young (MRT < 5 years) anoxic groundwater.Reduced concentrations of Halon-1301 in relatively old oxic water could confirm sorption/retardation, since degradation is likely only occurring in anoxic water.Similarly, analysis of relatively young anoxic/anoxic groundwater where sorption/degradation has not likely affected the concentration of Halon-1301 (due to a relatively short travel time in the aquifer and the currently levelling out atmospheric trend) could confirm degradation of Halon-1301 if reduced concentrations are found.

Summary and conclusion
This paper provided an insight into the suitability of the gaseous, water soluble compound Halon-1301 as groundwater age tracer.We demonstrated the capability of the analytical set-up for robust simultaneous determination of the gas tracers Halon-1301 and SF 6 (and CFC-12) on the same 1 L water sample, which provided an immense potential for more robust age interpretation of relatively young groundwater (recharged < 100 years).We estimated solubility, which is required to convert measured concentrations in water into atmospheric concentrations, from a range of equilibrated waters and relatively modern, oxic groundwater.We found that the solubility of Halon-1301 found in this study did not match its reported solubility.Uncertainties arising from this estimation approach, led to higher uncertainty in inferred MRT up to 16 years.More accurate determination of the solubility of Halon-1301 is required for better utilisation of its potential as age tracer.
We used piston and exponential piston flow modelling (PM and EPM) to infer age from Halon-1301 (and SF 6 ) concentrations in groundwater.Significantly different age interpretations were found with both modelling approaches.Halon-1301 was particularly sensitive to the choice of LPM due to its S-shaped input function, which is considerably skewed during mixing processes in contrast to SF 6 with a nearly linear atmospheric record.This indicated that the determination of Halon-1301 may allow for a better constraint of the mixing model.However, further study is needed to support this supposition with time series Halon-1301 data.Previously inferred CFC, SF 6 and tritium ages in the studied groundwater sites allowed us to compare the performance of Halon-1301 as an age tracer compared to other tracers.
Twelve of 17 groundwater samples, where direct comparison of inferred ages could be made, showed matching Halon-1301, SF 6 and/or tritium ages within an uncertainty of ∼ 1 SD.We found no significantly increased Halon-1301 concentrations in any of the analysed groundwater samples which indicated no apparent sources of contamination of Halon-1301 in our study, despite the fact that the sites included different land use environments and well construction that resulted in CFC contamination.This also indicated that interference with other co-eluting compounds was not an issue, since this would have led to increased concentrations of Halon-1301 determined in water.
Analysis of stored groundwater samples indicated that Halon-1301 was stable in oxic to anoxic water stored up to 7 weeks at 14 • C. Reduced concentration of Halon-1301 (along with significantly even further reduced concentration of CFC-12 and -11) at 5 of 17 sites needs to be assessed further.It is unclear if reduced concentrations were caused by degradation or retardation of Halon-1301 in the aquifer.
Despite these not fully understood reduced concentrations, we showed that Halon-1301 has strong potential as a complementary groundwater age tracer.If used in combination with other established tracers, it is likely to aid in reducing the ambiguity in groundwater age interpretations obtained though tritium, SF 6 and fading out CFC concentrations, and improve constraining mixing models.Since Halon-1301 is a gaseous tracer, it has additional potential to be used to assess unsaturated as well as saturated zone processes, especially with respect to the simultaneous determination of CFC-12 and SF 6 on the proposed analytical set-up.Due to its S-shaped, fading out atmospheric input and analytical detection limits, we suggest the appropriate application range for inference of groundwater age from Halon-1301 is for waters recharged between 1980 and 2005/2008.Higher uncertainty will be present in age estimates for waters of earlier (from 1970s) or more modern recharge.The uncertainty in inferred Halon-1301 age can be reduced by more accurate determination of its solubility.
To confirm the absence of local contamination sources, Halon-1301 needs to be assessed further at sites with higher risk of local sources (e.g.close to airports).To assess whether reduced Halon-1301 concentrations in older anoxic waters are a result of degradation or sorption, Halon-131 needs to be assessed in anoxic waters (preferably young -MRT < 5 years) that have been influenced by different compositions of bacteria and/or aquifer material, and/or in relatively old oxic sites (MRT > 5 years) with high organic content.Even if Halon-1301 is affected by degradation/sorption and/or contamination is occurring in specific areas, Halon-1301 is likely to be a more reliable groundwater age tracer than CFCs, which face issues regarding their reliability to in-fer groundwater age due to (anthropogenic) contamination and degradation in anoxic waters, as we observed in this study.Concentrations in the atmosphere are also fading out, which will make CFCs even less reliable in the future.
We suggest that Halon-1301 (or any other tracer) is used complementarily together with other tracers, to compensate for individual tracer limitations.We do not suggest that Halon-1301 is used as a stand-alone tracer (although in our study area it was significantly more reliable than CFCs, which are commonly used alone in the literature).Specifically, we recommend the simultaneous determination of Halon-1301 with SF 6 and CFC-12, using the cost-effective method presented in this study.This allows for the determination of three complementary age tracers in the same water sample, which may enable more precise determination of groundwater age (and mixing), assessment of unsaturated zone processes and increase robustness as the three tracers together allow identification and exclusion of problem samples, e.g.where contact with air has occurred during sampling, or where degradation of one or more of the age tracers has occurred.
The Supplement related to this article is available online at doi:10.5194/hess-19-2775-2015-supplement.

Figure 2 .
Figure 2. Groundwater wells and sampling locations in the Wellington region, New Zealand, are displayed as points; the black outlines represent the two catchments Hutt Valley (left catchment) and Wairarapa (right catchment).

Figure 4 .
Figure 4. Assessment of linearity of the ECD signal towards Halon-1301 using 10 mL modern air at different pressures (LEFT) and water at different volumes (RIGHT) showing an almost linear signal to pressure/volume (UPPER) and acceptable residuals (LOWER), lines in upper graphs represent the best least-square fit, fit with standard deviation of slope and 95 % confidence interval.

Figure 5 .
Figure 5.Effect of relative analytical uncertainty on inferred piston flow recharge year for SF 6 and Halon-1301.

Figure 6 .
Figure 6.Estimated solubility of Halon-1301 and SF 6 in equilibrated tap water, river water and oxic young groundwater in comparison to reported solubility data; * using data from Deeds (2008) for Halon-1301 and Bullister et al. (2002) for SF 6 .

Figure 8 .
Figure 8. Summary of mean residence time including error bars (±1 SD analytical uncertainty including uncertainty in solubility) inferred from Halon-1301, SF 6 vs. mean residence times inferred from tritium using the exponential piston flow model, data points are highlighted according to CFC-12/CFC-11 contamination/degradation (see legend); Halon-1301 and SF 6 were determined in this study, tritium and the CFCs were determined in previous study(s); the abbreviations "c" and "d" in the legend refer to contaminated and degraded in one or both CFCs, respectively; c/d refers to contamination and degradation was observed for either CFC-12 or CFC-11; "n/a" refers to no available CFC data.

Figure 9 .
Figure 9.Comparison of Halon-1301 concentration in 1 L water samples analysed directly after sampling (2 of 3) and after 7 weeks (1.2 years for Hutt River water sample) storage at 14 • C (1 of 3); * anoxic water samples.

Table 1 .
Beyer, 2015)ater samples analysed in this study: site name, amount of duplicates analysed, associated groundwater (GW) system, recharge temperature and excess air determined from noble gas analysis, dissolved oxygen (DO) and number of available CFC, tritium and SF 6 data.If no data are available for this site, the average NZ recharge temperature of 12.1 ± 1.8 • C and/or average NZ excess air 2.9 ± 1 mL (STP) L −1 (van der Raaij andBeyer, 2015)are used; LHGWZ + Lower Hutt groundwater zone; b groundwater shows considerable amount of methane and is considered as anoxic, despite relatively high oxygen concentration; n/a stands for not applicable. a
* Solubility parameters for Halon-1301 estimated in this study with an uncertainty of 10 %.