Journal of Hydrology 466–467 (2012) 11–22

Contents lists available at SciVerse ScienceDirect

Journal of Hydrology

journal homepage: www.elsevier .com/ locate / jhydrol

Quantifying groundwater discharge from different sourcesinto a Mediterranean wetland by using 222Rn and Ra isotopes

Valentí Rodellas a, Jordi Garcia-Orellana a,b,⇑, Ester Garcia-Solsona c, Pere Masqué a,b,Jose Antonio Domínguez d, Bruno J. Ballesteros d, Miguel Mejías d, Mario Zarroca e

a Institut de Ciència i Tecnologia Ambientals (ICTA), Universitat Autònoma de Barcelona, Campus UAB, 08193 Bellaterra, Spainb Departament de Física, Universitat Autònoma de Barcelona, Campus UAB, 08193 Bellaterra, Spainc Laboratoire d’Etudes en Géophysique et Océanographie Spatiales (LEGOS/OMP), 14 Av. Edouard Belin, 31400 Toulouse, Franced Instituto Geológico y Minero de España (IGME), Ríos Rosas 23, 28003 Madrid, Spaine Departament de Geologia, Universitat Autònoma de Barcelona, Campus UAB, 08193 Bellaterra, Spain

a r t i c l e i n f o

Article history:Received 23 December 2011Received in revised form 2 July 2012Accepted 5 July 2012Available online 15 July 2012This manuscript was handled by PhilippeBaveye, Editor-in-Chief, with the assistanceof Magdeline Laba, Associate Editor

Keywords:Ra isotopes222RnCoastal wetlandGroundwater dischargePeníscola marsh

0022-1694/$ - see front matter � 2012 Elsevier B.V. Ahttp://dx.doi.org/10.1016/j.jhydrol.2012.07.005

⇑ Corresponding author at: Departament de FísicBarcelona, Campus UAB, 08193 Bellaterra, Spain. Tel.:581 21 55.

E-mail addresses: Valenti.Rodellas@uab.cat (V. Ro(J. Garcia-Orellana), ester.garcia@legos.obs-mip.fr (E.@uab.cat (P. Masqué), ja.dominguez@igme.es (J.A.igme.es (B.J. Ballesteros), m.mejias@igme.es (M. Mejíasuab.cat (M. Zarroca).

s u m m a r y

Groundwater discharge constitutes the main water inflow of many coastal wetlands. Despite the poten-tial of Ra isotopes and 222Rn as tracers of groundwater discharge, the use of these radionuclides to quan-tify the groundwater inflow in coastal wetlands has been only scarcely addressed in the literature. Themain goal of this study is to evaluate the use of 222Rn and Ra isotopes to estimate the contribution of dis-tinct groundwater sources into a Mediterranean coastal wetland (the Peníscola marsh, Castelló, Spain).The Peníscola marsh is a small shallow wetland nourished by groundwater coming from four differentflowpaths: (i) a deep flow from the regional carbonate aquifer of El Maestrat, (ii) a shallow flow and(iii) an intermediate flow, both from the Irta Range and the detritic Vinaròs-Peníscola aquifer, and (iv)seawater intrusion. Data on 226Ra, 222Rn and salinity obtained in summer 2007 revealed that the deepgroundwater contribution was 15% of the total water inflow, whereas the shallow and intermediate flowpaths represented 32% and 48%, respectively. Seawater accounted only for the remaining 5% inputs to thewetland. Ra isotopes also allowed estimating the marsh water age in 1.2 days. Both the groundwater con-tributions derived from 222Rn measurements and the Ra-derived marsh water age agreed well with thedirect measurements obtained using propeller flow meters, evidencing the effectiveness of the usedmethods. An interannual comparison between the estimated groundwater inflow and the precipitationrevealed that shallow groundwater flows respond to local precipitation, whereas the deep groundwaterflow from the carbonate aquifer is dominated by a constant baseflow.

� 2012 Elsevier B.V. All rights reserved.

1. Introduction

The Mediterranean coast is characterized by numerous wet-lands, which are among the most productive environments of theworld and support high biodiversity of flora and fauna (Pearceand Crivelli, 1994). Unfortunately, the increase of anthropogenicpressure in coastal areas has led to the decline or even disappear-ance of large areas of coastal wetlands in recent decades, exceeding50% in many countries (De Stefano, 2004). This decline togetherwith an increasing appreciation of the essential functions provided

ll rights reserved.

a, Universitat Autònoma de+34 93 586 82 85; fax: +34 93

dellas), Jordi.Garcia@uab.catGarcia-Solsona), Pere.MasqueDomínguez), b.ballesteros@), Mario.Zarroca.Hernandez@

by wetlands prompted the development of policies and laws ad-dressed to its preservation and restoration. Since groundwater in-flow is a major component of the water budget in many coastalwetlands, the development of policies for the sustainable manage-ment of wetlands requires a precise understanding of its interac-tion with surrounding groundwater systems. However, thegroundwater inputs into a wetland are difficult to quantify becauseof the spatiotemporal heterogeneities that are characteristic of theaquifers (Krabbenhoft et al., 1990) and the potentially diversegroundwater sources that a wetland may receive (Schot and Was-sen, 1993). In that sense, Darcian estimations, hydrologic modelsand direct seepage measurements are hampered by errors anduncertainties (e.g. determining the hydraulic conductivity ofaquifers, the other water balance components, the spatial seepagevariations) (Cook et al., 2008).

Several geochemical tracers have been used to define thegroundwater inputs to a wetland, mainly because the waters inte-grate the tracers coming into the system via groundwater

12 V. Rodellas et al. / Journal of Hydrology 466–467 (2012) 11–22

pathways (Burnett et al., 2006). An ideal tracer of groundwater dis-charge should (i) be enriched in groundwater relative to wetlandswater, (ii) behave conservatively once released in the environment,and, in case of a radioactive tracer, (iii) have a half-life comparableto the water residence time (Corbett et al., 1997). The combineduse of the naturally occurring Ra isotopes (223Ra, 224Ra, 226Ra,228Ra) and 222Rn allows fulfilling all these requirements and thusare instrumental to quantify the groundwater discharge (Ramaand Moore, 1996) and evaluate the residence time of waters(Moore, 2000; Moore et al., 2006).

Ra isotopes and 222Rn have been widely used as tracers ofgroundwater discharge in several environments, including lakes(e.g. Corbett et al., 1997; Kluge et al., 2007), streams (e.g. Burnettet al., 2010), coastal lagoons (Garcia-Solsona et al., 2008a; Santoset al., 2008) or wetlands (e.g. Charette et al., 2003; Cook et al.,2008; de Weys et al., 2011). However, most of these works calcu-lated overall groundwater discharge rates and only a few studiesassessed the contribution of different groundwater sources tocoastal wetlands (e.g. Charette, 2007; Young et al., 2008).

The main objective of this study is to estimate the relative con-tribution of groundwater from four distinct sources into a Mediter-ranean wetland by using Ra isotopes (223Ra, 224Ra, 226Ra and 228Ra)and 222Rn, coupled with salinity and temperature measurements.In addition, we also aim to test the effectiveness of these radionuc-lides by comparing the results obtained with direct measurementsof marsh water flows, and to evaluate the temporal variability ofgroundwater discharges in an interannual basis.

2. Methods

2.1. Study site: Peníscola marsh

The Peníscola marsh is a relatively small (105 ha), shallow(<1 m), brackish-water wetland, located in the Spanish Mediterra-nean coast (Fig. 1). Although this wetland is separated from the seaby a coastal sandy barrier, the Peníscola wetland was artificiallychannelized to the sea by three main channels that converge intoa single one before flowing into the Mediterranean Sea through anarrow outlet (Fig. 1).

Fig. 1. (a) Location of the Peníscola marsh in the Mediterranean Sea Basin. (b) Schematpools. The flow direction of channelized waters is also indicated. The different systemsgroundwater flow paths inflowing into the wetland. The solid arrows represent shalldischarging through marsh sediments.

Since there is no surface water inflow, the only water sources tothe Peníscola marsh are rainfall (average annual precipitation of450 L m�2; AEMET) and groundwater. Several groundwatersources converge into the wetland from four different pathways(Fig. 1): (i) a shallow and horizontal flow path of fresh groundwaterfrom the detrital aquifer of the Vinaròs-Peníscola coastal plain andthe Irta Range, (ii) an intermediate (mid-depth) flow of groundwa-ter from the Vinaròs-Peníscola aquifer and the Irta Range that in-flows to the wetland seeping through marsh sediments, (iii) adeep groundwater flow from the regional carbonate Jurassic aqui-fer of El Maestrat, also inflowing through marsh sediments, and (iv)intruded seawater that mixes with inflowing groundwaters eitherin the sediments or in the coastal aquifer. Although the Irta Rangebelongs to the hydrogeological unit of El Maestrat, here we differ-entiate this system because it refers to local shallow karstic flowsof groundwater with its own hydrochemical signal. The regionalcarbonate Jurassic aquifer of El Maestrat is the most importantgroundwater reservoir in the area (recharge rates from 3.7 � 108

to 4.2 � 108 m3 y�1; Ballesteros et al., 2007) and is strongly karsti-fied, mostly draining to the sea through coastal springs in Penísco-la, Alcossebre and Badum (Garcia-Solsona et al., 2010; Mejías et al.,2008, 2012). A portion of this groundwater is advected to thePeníscola marsh ultimately passing through marsh sediments,where it converges with saline intrusion and groundwater fromthe mid-depth shallow systems and seeps into the Peníscola marshas brackish groundwater. The upward-seeping of brackish ground-water through marsh sediments is mainly located in the centralpart of the wetland, developing several spring pools (locally namedullals) in topographical depressed areas.

Water outflows from the Peníscola marsh essentially occurthrough the surficial outlet to the Mediterranean Sea (evaporationcan be neglected due to the short residence time of marsh waters).The micro-tidal conditions of the Mediterranean Sea (10–20 cm)preclude observing any tidal modulation in the Peníscola marsh.

2.2. Sampling

Water sampling was conducted in the Peníscola marsh in August2007, in February 2011 and between April 2007 and February 2008

ic map of the Peníscola marsh area displaying the wetland channels and the spring(El Maestrat, Vinaròs-Peníscola and Irta Range) are represented together with the

ow horizontal flows, whereas the dashed ones illustrate upflowing groundwater

V. Rodellas et al. / Journal of Hydrology 466–467 (2012) 11–22 13

in a monthly basis. During the first campaign, 16 surface water sam-ples from wetland channels were collected for Ra isotopes and 222Rnanalysis (Fig. 2). The samples were pumped from 30–50 cm belowthe water surface, taken as representative of the water column.Direct flow measurements were conducted in the marsh outlet usinga propeller flow meter (A.OTT Kempten, Z-30). The section of theoutlet was accurately determined in order to calculate the volumet-ric flow. End-member sampling for salinity, Ra isotopes and 222Rnconsisted in water collection from two wells (W1, W2) representa-tive of shallow groundwater flows from the Vinaròs-Peníscolaaquifer and the Irta Range, respectively, as well as a seawater samplefrom the Peníscola coastal area (SW). A submerged spring (ST32-D5)

Fig. 2. Sampling stations in the studied area: marsh water samples (stars) and end-mexperiment (triangle), samples for suspended particles (crosses) and direct flow measurwere sampled in August 2007. Stations ST7, ST32-D5 and ST47 were also sampled in Febr

detected in marsh waters via temperature (>28 �C) and salinity (>10)measurements was also sampled by pumping water from the deep-est part of the water column. A 60-m deep standpipe piezometer wasinstalled at the central part of the wetland (Pz1) in February 2012.This piezometer had a slotted section at the bottom of the pipe tocharacterize deep groundwater from the Maestrat aquifer.Temperature and salinity were measured in all water samples usingan YSI-556 handheld probe.

For Ra analyses, large volumes of water (10–50 L) were passedthrough columns loaded with MnO2-impregnated acrylic fiber(hereafter Mn-fibers) at a flow rate <1 L min�1 to quantitativelyextract Ra isotopes (Moore and Reid, 1973). 222Rn samples were

embers (rings), sediment, soil and sand samples (white circles), the incubationements (blue lines). The channels are emphasized with white lines. All the stationsuary 2011 and station ST47 was sampled monthly from April 2007 to February 2008.

14 V. Rodellas et al. / Journal of Hydrology 466–467 (2012) 11–22

collected into 250 mL bottles to be analyzed with a RAD-H2O system(Durridge Inc.). Although the RAD-H2O system is unsuitable fordetermining 222Rn concentrations in most surface waters, the high222Rn activities measured here allowed us to use this equipment.To minimize the contact of the 222Rn samples with air, water waspumped directly through a tube ending at the bottom of the bottleand leaving the water overflow the bottle for several seconds.

On February 2011 a survey was carried out to compare ground-water discharge estimations with the flows obtained through di-rect measurements of marsh waters. The stations ST7, ST32-D5(the submerged spring) and ST47 were re-sampled concurrentlywith direct marsh water flow measurements (four marsh sections;Fig. 2). Ra isotopes, 222Rn, temperature and salinity were measuredin all samples. A propeller flow meter was used in section F4, whilein sections F1, F2 and F3 the low flow and the amount of vegetationonly allowed to measure the surficial flow by means of the floatmethod. In order to evaluate the temporal variability of groundwa-ter discharges, from April 2007 to February 2008 a monthly mea-surement of 222Rn and 226Ra concentrations and salinity inoutflowing marsh waters (ST47) was carried out, coupled with di-rect flow measurements.

A total of 38 soil, sediment and sand samples were collected toobtain the distribution of long-lived Ra isotopes throughout thePeníscola marsh. Diffusive fluxes of Ra and 222Rn from sedimentswere estimated from one shallow (15–20 cm), large (25 cm in diam-eter) sediment core. The core was collected in August 2009 at thestation Inc_I (Fig. 2), where high 226Ra concentrations in sedimentshad been previously measured. To determine the amount of sus-pended particles in marsh waters a volume of 3 L of water from se-ven stations (Fig. 2) was passed through pre-dried-weighed filters of1 lm. The filters were dried and weighed again after sampling tocalculate a net concentration of suspended particulate matter.

2.3. Analytical methods

The portable 222Rn-in-air alpha spectrometer RAD7 (DurridgeInc.) was used for the determination of 222Rn activities in marshwaters (Burnett et al., 2001). To adapt this detector for water mea-surements, we used the RAD-H2O system (Durridge Inc.), which al-lowed the direct determination of 222Rn from bottles of 250 mL ofwater. All 222Rn activities were decay corrected for the time ofcollection.

Regarding Ra isotopes, the Mn-fibers were rinsed with radium-free deionized water, partially dried (Sun and Torgersen, 1998),and placed in a Radium Delay Coincidence Counter (RaDeCC) toquantify the short-lived radium isotopes (223Ra and 224Ra) as de-scribed by Moore and Arnold (1996). Uncertainties in concentra-tions of 223Ra and 224Ra were estimated following Garcia-Solsonaet al. (2008b). After the 223Ra and 224Ra measurements, the Mn-fi-bers were ashed (820 �C, 16 h) and transferred to hermeticallysealed counting vials (Charette et al., 2001) to determine thelong-lived Ra isotopes (226Ra and 228Ra) by gamma spectrometryusing a well-type Ge detector. To quantify 226Ra and 228Ra in soils,sediments and sands, samples were dried, crushed and hermeti-cally sealed in calibrated geometries, to be analyzed using a coaxialGe gamma spectrometer. Both Mn-fiber and soil gamma measure-ments were carried out after aging the samples a minimum of3 weeks to ensure the equilibrium between 226Ra and its daugh-ters. 226Ra and 228Ra were determined using the 214Pb and 228Acphotopeaks at 352 and 911 keV respectively. All radium activitieswere corrected for radioactive decay to sampling time.

2.4. Ra and 222Rn diffusive flux experiments

The sediment core collected for the incubation experiment(Inc_I) was placed in a bin filled with water to prevent artificially

impressed flow along the core. The overlying water was extracted,and the tube was refilled with Ra-free water (2–3 L) from the samemarsh location. Overlying waters were constantly aerated to pre-vent changes in redox conditions (Beck et al., 2007).

To estimate the diffusive flux of Ra, a closed water loop was cre-ated (two tubes were connected to the core) and interconnected toa column containing Mn-fiber (Fig. 3a). We used a peristaltic pump(constant rate of 0.2 L min�1) to recirculate the overlying watersthrough the Mn-fiber to quantitatively extract the Ra isotopesand bring the filtered water (free of Ra) back to the incubationchamber. Without interrupting the water flow, the Mn-fiber wasreplaced by a new one after progressively longer time periods(12, 24, 36, 48, 72 h). The Ra diffusive fluxes were estimated fromthe slope of the best fit of the Ra activity per core surface area(dpm m�2) plotted against incubation time (h).

A similar experiment was conducted to determine the 222Rn dif-fusive flux (Fig. 3b). After the Ra experiment, the incubation cham-ber was sealed to avoid 222Rn losses to the atmosphere. A waterflow from the free-air core was circulated to the RAD-Aqua com-mercial equilibrator (Durridge Inc.) at a rate of 0.5 L min�1 (0.03m3 h�1), where water was sprayed in a cylinder allowing the222Rn exhalation. The extracted 222Rn was pumped through aclosed air loop interconnected to the RAD7 alpha-detector (Dur-ridge Inc.), what allows the 222Rn quantification after correctingfor temperature (continuously monitored) (Burnett et al., 2001).After the water sample had been sprayed, it was derived to a car-boy where it was permanently aerated to eliminate the remaining222Rn before being introduced again to the incubation chamber.Considering the high 226Ra activities of marsh sediments, the timethat recirculated water remains in contact with sediments is largeenough to allow the quantification of the 222Rn concentrations inoutflowing waters using the RAD7 + RAD-Aqua system if measure-ments are integrated for 1 h. We estimated the 222Rn diffusive fluxfrom the hourly-averaged concentrations over 48 h (dpm m�3),multiplied by the water circulation rate (m3 h�1) and divided bythe core surface area (m2).

2.5. Desorbable Ra experiments

To determine the surface-bound desorbable Ra from suspendedparticles in marsh waters, about 10 g of sediments collected fromseven different stations were added to 10 L of filtered, Ra-freebrackish marsh water (salinity �6.7, the highest measured inmarsh waters). The sediments were collected in stations wherethe load of suspended particles was quantified (Fig. 2). Each samplewas stirred several times during 24 h and decanted to another con-tainer. The decanted water was filtered through a column contain-ing clean acrylic fiber and Mn-fiber to remove the remainingsuspended particles and extract all the Ra that had been desorbed,respectively. The content of Ra isotopes in the Mn-fiber was deter-mined as described above. The desorbable Ra was determined fromthe load of suspended particles (g L�1) and the experimental Radesorption (dpm g�1).

3. Results

The concentrations of 222Rn and Ra isotopes in water samplescollected in August 2007 are presented in Table 1, together withtheir activity ratios (AR) and data of salinity and temperature atthe time of collection. Data on potential end-members, includingwells (W1 and W2), seawater (SW), the piezometer (Pz1) and thesubmerged spring (ST32-D5), are also included in Table 1.

A salinity gradient was measured along the marsh waters, withlower salinities (S < 1) in the northern area, and higher salinitiestowards the marsh outlet (S � 5.7). The marsh water samples with

Fig. 3. Schematic representation of Ra (a) and 222Rn (b) diffusive flux experiments. (a) For Ra diffusive fluxes, the incubation chamber was aerated and the overlying waterwas continuously circulated through the Mn-fiber, which extracted the Ra isotopes from the water. The Mn-fiber was replaced at specific time intervals. (b) For 222Rn diffusivefluxes, the incubation chamber was sealed and the overlying water was circulated through the RAD-Aqua gas exchange cylinder, which was connected to the RAD7 detectorto quantify 222Rn. The sprayed water was reintroduced to the incubation chamber after it had been aereated to eliminate the remaining 222Rn.

V. Rodellas et al. / Journal of Hydrology 466–467 (2012) 11–22 15

higher salinities (�6–7) were located in stations near or withinsurficial spring pools (ST22, ST24 and ST38). The sample collecteddirectly from the submerged spring (ST32-D5) presented the high-est salinity (13.7) of marsh waters. Temperature measurementsshowed similar distributions, with the highest temperatures (upto 28 �C) associated with spring pools or focused groundwater dis-charges. The temperature results might be only used as a qualita-tive parameter, as temperature measurements were conductedalong the day, and thus, they are clearly influenced by diurnal tem-perature fluctuations.

A significant enrichment in Ra isotopes and 222Rn along themarsh was also observed (by a factor of 16, 21, 180, 15 and 410for 223Ra, 224Ra, 226Ra, 228Ra and 222Rn, respectively), with loweractivities in the northern part and increasing concentrations to-wards the wetland outlet. The activities of all Ra isotopes and222Rn were significantly higher at stations ST24 and ST38, and forthe long-lived Ra isotopes at ST22, all of them located close orwithin spring pools. These activities, particularly for 226Ra and222Rn, are among the highest activities previously measured inwetlands, coastal lagoons or similar environments (e.g. Charetteet al., 2003; Cook et al., 2008; Garcia-Solsona et al., 2008a; Schmidtet al., 2010). Activities of both short- and long-lived Ra isotopes,showed a strong positive correlation with salinity (Fig. 4). The226Ra activities were 1–3 orders of magnitude higher than theother isotopes, with 228Ra/226Ra activity ratios (AR) ranging from0.03 to 0.06 in the central part of the marsh. Most of the marshwater samples presented a 224Ra/228Ra AR ranging from 1 to 1.9.Enrichment of 224Ra relative to 228Ra have been reported for manyestuaries and salt marshes, and is attributed to the continuousleaching of Ra from sediments and the faster regeneration of224Ra (e.g. Charette et al., 2003; Hancock and Murray, 1996, Ramaand Moore, 1996). The sample from the submerged spring (ST32-D5) presented the highest activities measured in marsh watersfor all Ra isotopes and 222Rn (Table 1).

Regarding the direct groundwater measurements, samplesmeasured in shallow wells from the Vinaròs-Peníscola aquifer(W1) and the Irta Range (W2) had low salinities (<1) and the con-centrations of Ra isotopes and 222Rn were between 1 and 3 ordersof magnitude lower than those measured in waters of the marsh

channels. Water collected at 60 m in the deep piezometer (Pz1)presented higher temperatures (up to 45 �C), higher salinities (upto 20) and higher Ra activities (e.g. 2.7 � 104 dpm 100 L�1 for226Ra) than all the samples measured in marsh waters, but lower222Rn activities (7.4 � 105 dpm 100 L�1).

The data corresponding to the samples collected in February2011 survey are also included in Table 1. The highest Ra and222Rn activities, as well as temperature and salinity, were mea-sured at the submerged spring (F-ST32-D5). The 222Rn activitiesmeasured at this station were comparable to those determined inAugust 2007 and February 2011, whereas the salinity and Ra activ-ities were considerably lower in the latter survey.

The 226Ra activities in marsh sediments ranged from 13 to46.8 dpm g�1, and found to be high relative to world average con-centrations (2.1 dpm g�1: UNSCEAR, 2000) and soils collected inthe Western Mediterranean (0.5–3 dpm g�1: Garcia-Orellanaet al., 2006). Concentrations of 228Ra ranged from 0.6 to3.32 dpm g�1, comparable to the world average concentration(1.8 dpm g�1; UNSCEAR, 2000). The highest 226Ra concentrationswere observed in the central-western part of the marsh, resultingin 228Ra/226Ra AR ranging from 0.03 to 0.06.

The diffusive Ra fluxes from the sediments obtained from theincubation of marsh sediments were 3.2 ± 0.7, 122 ± 10,1850 ± 140 and 71 ± 10 dpm m�2 d�1 for 223Ra, 224Ra, 226Ra and228Ra, respectively. The 228Ra/226Ra flux ratio (0.038 ± 0.006) iscomparable with the 228Ra/226Ra AR of sediment from the marshcentral part (0.03–0.006), revealing that the flux differences amongRa isotopes derive from the different Ra concentrations in sedi-ments. The 222Rn flux obtained with this experiment was1200 ± 300 dpm m�2 h�1. It should be taken into account that Raand 222Rn diffusive fluxes are both upper limit estimates, not onlybecause we incubated a marsh sediment with one of the highest226Ra activities, but also because replacing continuously the over-lying enriched waters by water free of these radionuclides maxi-mizes the gradient between pore water and overlying waters,thus increasing the diffusion flux.

The desorption experiment conducted in seven marsh sedi-ment samples resulted in desorbable Ra activities of about0.010 ± 0.003 dpm g�1 for 223Ra, 0.25 ± 0.04 dpm g�1 for 224Ra,

Table 1Salinity, temperature, Ra and 222Rn activities and activity ratios (AR) of the marsh waters and end-members sampled in August 2007 and February 2011.

Sample Salinity Temperature(�C)

223Ra 224Ra 226Ra 228Ra 222Rn 224Ra/228Ra

(dpm 100 L�1) (103 dpm 100 L�1)

August 2007

Marsh samplesST7* 0.76 18.70 1.5 ± 0.2 19 ± 1 12.8 ± 0.6 11.5 ± 0.9 46 ± 9 1.63 ± 0.15ST8 1.14 24.58 1.7 ± 0.2 14 ± 1 74 ± 4 19 ± 3 21 ± 5 0.70 ± 0.11ST17 1.50 21.15 2.5 ± 0.5 60 ± 3 480 ± 30 48 ± 9 250 ± 20 1.2 ± 0.2ST18 1.92 21.95 4.0 ± 0.6 65 ± 3 820 ± 40 56 ± 4 550 ± 30 1.17 ± 0.11ST19 3.62 24.34 13.6 ± 1.2 210 ± 20 1490 ± 90 135 ± 9 460 ± 20 1.52 ± 0.15ST20 3.90 23.62 10.0 ± 1.5 104 ± 7 1940 ± 110 111 ± 11 270 ± 30 0.94 ± 0.11ST22 6.09 23.47 6.1 ± 1.3 130 ± 8 5000 ± 300 250 ± 30 44 ± 6 0.52 ± 0.06ST24 5.72 25.85 20 ± 2 360 ± 20 5300 ± 300 189 ± 13 3510 ± 100 1.9 ± 0.2ST25 3.01 23.72 6.1 ± 0.9 123 ± 7 1890 ± 110 101 ± 8 630 ± 20 1.21 ± 0.12ST31 1.72 22.34 2.8 ± 0.2 47 ± 3 410 ± 20 35 ± 3 980 ± 30 1.34 ± 0.14ST32-D1 4.69 17.20 18 ± 2 220 ± 20 2900 ± 200 170 ± 20 560 ± 20 1.3 ± 0.2ST36 4.41 25.05 14 ± 2 230 ± 20 2460 ± 150 164 ± 10 450 ± 40 1.41 ± 0.13ST38 6.73 28.62 20 ± 3 370 ± 20 8700 ± 500 330 ± 20 2970 ± 100 1.12 ± 0.09ST42 5.57 25.24 16 ± 2 260 ± 10 4900 ± 200 150 ± 9 1800 ± 50 1.74 ± 0.13ST43 5.68 25.78 21 ± 2 290 ± 20 4100 ± 200 173 ± 11 1680 ± 80 1.7 ± 0.2ST47 5.66 26.15 24 ± 2 270 ± 20 4100 ± 200 183 ± 10 1770 ± 90 1.5 ± 0.1

End-membersW1 0.83 19.52 1.5 ± 0.2 12.6 ± 1.0 22.8 ± 1.4 12 ± 1 4 ± 2 1.02 ± 0.10W2 0.30 n.a. n.a. n.a. 12.7 ± 0.4 n.a. 4.9 ± 1.1 –ST32-D5* 13.73 28.31 83 ± 6 1040 ± 60 17,100 ± 1100 580 ± 40 3700 ± 200 1.8 ± 0.2Pz1* 19.47 42.11 n.a. 3430 ± 110 27,100 ± 600 2500 ± 100 740 ± 20 2.1 ± 0.2SW* 38.05 n.a. 3.4 ± 0.4 44 ± 3 46.2 ± 0.8 17 ± 1 4.6 ± 0.9 2.5 ± 0.2

February2011F-ST7* 0.77 15.90 0.7 ± 0.2 16.2 ± 1.4 n.a. n.a. 50 ± 20 –F-ST32-D5* 11.79 33.55 41 ± 5 630 ± 50 7200 ± 400 420 ± 20 3990 ± 130 1.5 ± 0.2F-ST47 4.70 18.88 8.0 ± 1.2 194 ± 15 1690 ± 60 150 ± 9 1900 ± 70 1.30 ± 0.13

n.a. Not analyzed.* Samples used as end-members in the mixing models.

16 V. Rodellas et al. / Journal of Hydrology 466–467 (2012) 11–22

4.5 ± 0.5 dpm g�1 for 226Ra and 0.31 ± 0.21 dpm g�1 for 228Ra. Theseestimates are an upper limit because we used Ra-free water with thehighest salinity observed in marsh waters (6.7), which increased Radesorption. Concentrations of suspended particles in the sevenwater samples analyzed range from 0.3 to 2.1 g 100 L�1, with anarea-weighed average of 0.68 ± 0.07 g 100 L�1.

4. Discussion

The distribution of salinity and Ra and 222Rn concentrationsmeasured in marsh waters suggests the presence of two distinctgroundwater types discharging to the Peníscola marsh: (i) the dis-charge of fresh groundwater in the northeastern and eastern partsof the wetland characterized by low salinities (<1) and low activi-ties of Ra isotopes and 222Rn; and (ii) brackish groundwater dis-charging via springs or seeps distributed throughout the marsh,characterized by higher salinities, temperatures and Ra and 222Rnconcentrations. Considering the observed pattern and the hydro-geological functioning of the Peníscola marsh (Fig. 1), we assumethat: (i) groundwater inflowing into the northeastern and easternparts of the wetland accounts mainly for the shallow flow of freshgroundwater from the Vinaròs-Peníscola and the Irta Range sys-tems (GWshallow); (ii) Groundwater seeping through the marsh sed-iments (GWseep) is dominated by the deep flow of groundwaterfrom the carbonate Jurassic aquifer of El Maestrat (GWdeep), whichactually mixes with seawater intrusion (SW) and an intermediateflow of groundwater from the mentioned two local shallow sys-tems (GWmid) before discharging to the wetland (Fig. 1).

4.1. Evaluation of potential sources of Ra isotopes and 222Rn

In order to estimate the contribution of each groundwatersource by using Ra isotopes and 222Rn, the other potential inputs

of these radionuclides must first be evaluated. Aside from ground-water inputs, the other potential Ra and 222Rn sources that weidentified are desorption of surface-bound Ra from resuspendedparticles, diffusive fluxes of 222Rn and Ra from bottom sedimentsand ingrowth of 222Rn by decay of dissolved 226Ra.

To evaluate the importance of each possible source term weneed to estimate the average concentration of Ra isotopes and222Rn in marsh waters. Since (i) sampling stations were not hom-ogenously distributed throughout the marsh and (ii) each channelhad its own water flow, the integrated inventories should be calcu-lated weighting both considerations. However, since all marshwater outputs occur through the marsh outlet (station ST47), theconcentrations of Ra isotopes, 222Rn and salinities measured inmarsh outflowing waters (appropriately corrected for losses forradioactive decay and 222Rn evasion to the atmosphere) can beconsidered the best representation of integrated marsh waters. Inthe following subsections, each source term will be discussed in or-der to evaluate its relative contribution to the Ra and 222Rn inven-tories in marsh waters.

4.1.1. Diffusive fluxes from sedimentsConsidering an average marsh waters depth of 1.5 m and the

marsh water apparent age calculated in the Section 4.2.1.2.(TR = 1.2 d), the maximum experimental diffusive fluxes fromunderlying sediments would represent 1.1%, 3.8%, 3.7%, 0.1%, 0.1%of the total 223Ra, 224Ra, 226Ra, 228Ra and 222Rn concentrations,respectively (Table 2). Thus, even if the fluxes obtained in ourexperiment are maxima, the diffusion from sediments representsa negligible input of Ra and 222Rn into the marsh channels.

4.1.2. Desorption from resuspended particlesSince there are no surface water inputs carrying particles in sus-

pension, the suspended particles comprise the major source of

Fig. 4. Ra isotopes vs salinity for samples from the Peníscola marsh. Positive linear correlations are observed for all Ra isotopes.

V. Rodellas et al. / Journal of Hydrology 466–467 (2012) 11–22 17

desorbable Ra. The maximum desorption that could occur in marshwaters was on the order of 0.07 ± 0.02, 1.7 ± 0.2, 30.5 ± 0.3 and2.4 ± 1.4 dpm 100 L�1 for 223Ra, 224Ra, 226Ra and 228Ra, respectively(Table 2). These concentrations are also maxima estimates, as it isimplicitly considered that new sediments (with regenerated Ra)are resuspended at each time interval (marsh waters age, TR) (Becket al., 2007). Even so, the maximum desorption would only contrib-ute with less than 1% of the total Ra concentrations measured inoutflowing waters (ST47).

4.1.3. Production of 222Rn from dissolved 226Ra decay222Rn activities measured in marsh waters were 2–3 orders of

magnitude higher than 226Ra activities dissolved in marsh waters(Table 1). Thereby, 222Rn activities produced by decay of dissolved226Ra contribute to a very little fraction (�0.2%) of the total amountof 222Rn (Table 2).

4.1.4. Groundwater inputsSince the contribution of diffusion, desorption and production

of Ra isotopes and 222Rn account for less than 5% of the activitiesof these radionuclides in marsh waters (Table 2), the remainingsource (groundwater discharges) should account for the most partof Ra isotopes and 222Rn inputs to the Peníscola marsh. It should benoted that the contributions estimated in the previous subsectionscorrespond to upper limits, as the ST47 does not account for thelosses for radioactive decay and 222Rn evasion to the atmospherewhich had occurred all along the water pathway. Hence the

Table 2Supply (dpm 100 L�1) and relative importance (%) of each Ra isotope and 222Rn sourcgroundwater) are: diffusion from sediments, desorption from suspended particles and pro

223Ra%

22XRa–222Rn diffusion from sediments (dpm 100 L�1) 0.26 ± 0.06 1.1Mdif: modeled diffusion (dpm m�2 d�1) 3.2 ± 0.7TR: water age (d) 1.2hav: average water depth (m) 1.522XRa desorption from suspended particles (dpm 100 L�1) 0.07 ± 0.02 0.3Mdes: modeled desorption (dpm g�1) 0.010 ± 0.003Psus: suspended particles (g 100 L�1) 0.68 ± 0.08222Rn production from226Ra decay ([226Ra]ST47) (dpm 100 L�1) –

Total 1.4

Total 22XRa–222Rn (ST47) (dpm 100 L�1) 24 ± 2

contribution of groundwater discharges (>95%) represents a mini-mum estimate.

4.2. Groundwater contributions from each source

4.2.1. Mixing modelSince more than 95% of the total amount of Ra and 222Rn enter-

ing the wetland can be attributed to groundwater discharges,activities of these radionuclides in marsh waters should be ex-plained by a mixing of the shallow flow of fresh groundwater fromthe Vinaròs-Peníscola and the Irta Range systems (GWshallow) andgroundwater seeping through marsh sediments (GWseep). The mix-ing of these two groundwater types should also explain the salinityof marsh waters, as no other sources were identified and salinitybehaves conservatively. However, as evidenced from the relation-ship between 222Rn and salinity (Fig. 5), several marsh water sam-ples do not fall on the mixing line between the low 222Rn – lowsalinity waters (GWshallow) and the high 222Rn – high salinitywaters (GWseep), requiring the development of a more appropriatemixing model for the Peníscola marsh system.

The high Ra activities measured in the 60-m piezometer(2.7 � 104 dpm 100 L�1 for 226Ra at Pz1) suggest that groundwaterfrom the deep Jurassic aquifer of El Maestrat (GWdeep) could be themain source of Ra isotopes to the Peníscola marsh waters. How-ever, 222Rn activities measured in the piezometer (7 � 105

dpm 100 L�1) were 5–6 times lower than the 222Rn activities mea-sured in the submerged spring (4 � 106 dpm 100 L�1 at ST32-D5;

e to average marsh waters concentrations (ST47). Sources considered (aside fromduction for decay.

224Ra 226Ra 228Ra 222Rn% % % (103) %

10.1 ± 0.8 3.8 153 ± 12 3.7 0.24 ± 0.03 0.1 2.4 ± 0.6 0.1122 ± 10 1850 ± 140 71 ± 10 29 ± 71.2 1.2 1.2 1.21.5 1.5 1.5 1.51.69 ± 0.20 0.6 30.5 ± 0.3 0.7 2.1 ± 1.4 1.2 -0.25 ± 0.04 4.5 ± 0.5 0.31 ± 0.210.68 ± 0.08 0.68 ± 0.08 0.68 ± 0.08– – – 4.2 ± 0.2 0.2

4.4 4.5 1.3 0.4

270 ± 20 4200 ± 300 166 ± 11 1770 ± 90

18 V. Rodellas et al. / Journal of Hydrology 466–467 (2012) 11–22

Table 1). In addition, the maximum 222Rn groundwater activitiesmeasured directly from the local shallow aquifers, either fromwells close to the Peníscola marsh (4 � 103 and 5 � 103

dpm 100 L�1 at station W1 and W2, respectively; Table 1) or coast-al springs (103–104 dpm 100 L�1, unpublished data), are 2–3 ordersof magnitude lower than the 222Rn activities measured in themarsh submerged spring. The lack of any endmember with 222Rnactivities high enough to support the activities measured in marshwaters, and the high 226Ra activities measured in the Peníscolamarsh sediments (up to 47 dpm g�1), suggest that groundwaterdischarging via seeps or springs (GWseep) becomes enriched in222Rn during its advection through marsh sediments.

Considering that the primary source of Ra isotopes is the deepgroundwater flow (GWdeep), the mixing of the GWdeep with seawa-ter intrusion (SW) and intermediate groundwater flows from thelocal shallow systems (GWmid) prior to discharge to the wetlandresults in inflowing groundwaters (GWseep) having Ra activitiesand salinities depending on this three end-member mixing. UnlikeRa, as marsh sediments are the main source of 222Rn, the 222Rnactivity of the GWseep should be constant and independent of thismixing. These processes are summarized in a schematic plot(Fig. 6), where water fluxes are drawn together with the salinitiesand 226Ra and 222Rn activities measured at stations ST32-D5 andST24. Both samples were collected where seeping of the GWseep

seemed to be the only source of water: salinity (13.73) and temper-ature (28.3 �C) at station ST32-D5 indicated a point-source ground-water discharge, whereas station ST24 was located in a shallow(0.4 m depth) and narrow (0.5–1 m) channel without inputs otherthan seeping groundwater. The 222Rn activities measured in bothstations agree well ((3.7 ± 0.2) � 106 and (3.5 ± 0.1) � 106

dpm 100 L�1), whereas salinities (13.7 and 5.7 for ST32-D5 andST24, respectively) and Ra activities (e.g. (17.1 ± 1.1) � 103 and(5.3 ± 3) � 103 dpm 100 L�1 of 226Ra for ST32-D5 and ST24, respec-tively) present significant differences.

In summary, the Peníscola marsh system is described by a bin-ary mixing between the shallow flow of fresh groundwater(GWshallow) and groundwater inflowing through marsh sediments(GWseep). This latter fraction represents the mixing of the flowsof deep groundwater (GWdeep), mid-depth groundwater (GWmid)and seawater intrusion (SW) prior to inflow into the wetlandwaters. Considering the different enrichment processes for Ra iso-topes and 222Rn, the appropriate mixing model to describe the rel-ative contributions of distinct water types to the Peníscola marshcan be written as follows:

Fig. 5. 222Rn vs salinity for samples from the Peníscola marsh waters. The dashedline represents the mixing line between the low 222Rn – low salinity waters(GWshallow) and the high 222Rn – high salinity waters (GWseep).

fGWshal þ fGWdeep þ fGWmid þ fSW ¼ 1222RnGWshalfGWshal þ 222RnGWseepðfGWdeep þ fGWmid þ fSWÞ¼ 222RnSTxxeðkTþKT=hÞ

226RaGWshalfGWshal þ 226RaGWdeepfGWdeep þ 226RaGWmidfGWmid

þ 226RaSWfSW ¼ 226RaSTxx

SalGWshalfGWshal þ SalGWdeepfGWdeep þ SalGWmidfGWmid

þ SalSWfSW ¼ SalSTxx

ð1Þ

where fi, 222Rni, 226Rai and Sali are the relative fractions, the 222Rn,the 226Ra activities and the salinity, respectively, of each watersource: GWshallow, GWdeep, GWseep, SW and GWseep. The terms222RnSTxx, 226RaSTxx and SalSTxx represent the 222Rn, the 226Ra activi-ties and the salinity measured in a given sample; k is the decay con-stant of 222Rn (0.181 d�1), K is the 222Rn gas exchange velocity, h isthe area-weighed average of marsh water depth (1.5 m), and T is thetime elapsed since the water mass entered the system (marsh waterage). The term e(k�T+K�T/h) accounts for the 222Rn losses related to de-cay and evasion to the atmosphere along the water pathway beforethe water reached the sampling station. Although other Ra isotopescould be used, here we favored 226Ra because it is the most enrichedisotope in marsh waters and its radioactive decay can be neglected.The use of this model requires the characterization of all end-mem-bers (GWshallow, GWdeep, GWmid, SW and GWseep) and the estimationof the 222Rn gas exchange velocity and the apparent marsh waterage, all of them established in the next subsections.

4.2.1.1. Selection of end-members. Since the shallow flow of ground-water from the Vinaròs-Peníscola aquifer has the same character-istics as the shallow flow from the karstic Irta Range system (lowRa, 222Rn and salinities), we cannot distinguish both flows usingthese tracers. Either the sample from the north part of the easternchannel (ST7) or the samples from the wells nourished by ground-water from the Vinaròs-Peníscola littoral plain (W1) or the IrtaRange (W2) may be used to characterize the fresh groundwaterend-member from the local shallow aquifers (GWshallow). Salinitiesand Ra concentrations are similar at these stations, whereas 222Rnactivities are one order of magnitude lower in both wells. Thesedifferences can be explained by the shallowness of the wells (20–40 cm), which could enhance the 222Rn evasion to the atmosphere.For this reason, station ST7 is taken as the best representation ofthe GWshallow end-member. This station is also used to characterizethe GWmid, as this end-member refers to the deeper flow ofgroundwater from the same local shallow systems that dischargeto the wetland via seeps through marsh sediments.

The sample collected from the 60-m deep piezometer (Pz1) wasused to characterize the deep groundwater from the regional car-bonate Jurassic aquifer of El Maestrat (GWdeep). The high tempera-ture (>40 �C) measured in Pz1 reveals its hydrothermal origin, afeature of the deep Maestrat aquifer already documented (Sán-chez-Navarro et al., 2004). The salinity (�20) of sample Pz1evidences the influence of seawater intrusion, and its use as anend-member would lead to overestimate the calculated contribu-tion of deep groundwater. The station SW collected at the Peníscolacoastal sea is used to characterize the seawater end-member (SW).

Since the GWseep discharge to the marsh channels occursthrough submerged springs or seeps, the best approach to obtaina representative end-member for 222Rn associated to the GWseep

may be the direct collection of discharging groundwater seepingthrough the sediments of the marsh channels. The deepest samplecollected at station ST32 (ST32-D5), where a focused, point-sourced groundwater discharge was detected from salinity andtemperature measurements, may be considered the best represen-tation of the GWseep end-member. Notice that neither 226Ra norsalinities of the GWseep have to be used in our model, because

Fig. 6. Schematic representation of fluxes of waters to the Peníscola marsh; salinity, 226Ra and 222Rn concentrations (dpm 100 L�1) are also indicated. Data of stations ST24and ST32-D5 are also shown to emphasize the variations in salinity, 226Ra, 222Rn and 224Ra/228Ra of GWseep in different submerged springs.

V. Rodellas et al. / Journal of Hydrology 466–467 (2012) 11–22 19

variations in the mixing between the GWdeep, the GWmid and SWresults in GWseep having different salinities and Ra content, thuspreventing to establish a precise Ra- and salinity- end-membersfor the GWseep (Fig. 6).

A summary of the selected end-members is presented in Table 1and Fig. 6. As extremely distinct tracer signatures characterize allthe end-members, the results of the mixing model are insensitiveto small variations in the end-member values.

4.2.1.2. Estimation of 222Rn gas exchange rate. As 222Rn activitiesmeasured in marsh waters are 3–4 orders of magnitude higherthan 222Rn activities in air (�300 dpm 100 L�1) in Eq. (1), atmo-spheric outputs are considered to only depend on the 222Rn gas ex-change rate (K) and the 222Rn concentration in water (applying afactor h to normalize for depth) (Eq. (1)). We take K = 0.16 m d�1,as obtained by Cook et al. (2008) via an injected tracer experimentusing SF6 for a shallow (<1 m) wetland. This experimental 222Rngas exchange rate was preferred to empirical equations (e.g. Mac-intyre et al., 1995) because it was obtained from an environmentwith similar characteristics to our study area. Although the resultsof the model are highly sensitive to the 222Rn gas exchange rate (K),its variations would only influence the contribution of groundwa-ter from the local shallow systems (GWshallow and GWmid). Forinstance, had we used K = 0.24 m d�1 (i.e., a 50% higher), the frac-tions of the GWshallow and the GWmid would be reduced and in-creased, respectively, by a factor of 10–15%, whereas thecontribution of the GWdeep and the SW would remain constant.

4.2.1.3. Estimation of apparent marsh water age. The variation ofthe ratios between Ra isotopes of different half-lives can be usedto determine the apparent age of marsh waters. Moore et al.(2006) developed an approach that allows considering continu-ous additions of Ra throughout the system. Assuming that thereare no losses of Ra aside mixing or radioactive decay, the appar-ent marsh water age can be estimated by using the followingequation:

TR ¼Fð222Ra=228RaÞ � Ið222Ra=228RaÞ

Ið222Ra=228RaÞ � k224ð2Þ

where F(224Ra/228Ra) is the 224Ra/228Ra activity ratio (AR) of thegroundwater inputs into the system, I(224Ra/228Ra) is the224Ra/228Ra AR of a given marsh water sample, and k224 is the decayconstant of 224Ra (0.1894 d�1). Although other Ra isotopes may beused, here we used 224Ra/228Ra AR because the Peníscola marshtimescale is appropriate to the 224Ra half-life and both radionuc-lides belong to the same decay chain.

Given that the GWseep discharge is largely the main input of Raisotopes into the Peníscola marsh, we used the 224Ra/228Ra AR ofsample ST32-D5 (1.8 ± 0.2) as the best representation of inflowinggroundwaters (F(224Ra/228Ra)) into the system. The apparentmarsh water age calculated here represents the time elapsed sincethe GWseep enters into the wetland. Notice that, although the Raactivities of the GWseep depend on the mixing of the GWdeep withSW and the GWmid prior to discharge to the wetland, the activityratios between Ra isotopes are constant in the GWseep (Fig. 6). Con-sidering the wetland outflowing waters as the best representationof the Peníscola marsh waters, we used the 224Ra/228Ra AR at sta-tion ST47 (1.46 ± 0.13) to obtain an average marsh water age of1.2 ± 0.6 days.

The outflow of Peníscola marsh waters was measured using apropeller flow meter in August 2nd and September 13th 2007,with an average of 37,000 ± 4000 m3 d�1. Since the marsh watervolume is 64,000 m3, we can independently estimate a Peníscolamarsh waters residence time of 1.7 ± 0.2 days. The Ra-derivedapparent age refers only to the time elapsed since the GWseep

enters into the Peníscola marsh, whereas the residence time de-rived from the direct measurements accounts also for theGWshallow discharge, which mainly occurs at the northern partof the Peníscola marsh. Therefore, as flow meter-derived resi-dence time estimations are expected to be longer than Ra-basedapparent ages, we can consider that both estimates are in goodagreement.

Eq. (2) may be also used to determine the time that it takesfor the GWseep to reach a given station since it entered the wet-land. Calculated ages for the different stations ranged from 0 to12.9 days. The longer water ages correspond to samples collectedfrom sinkholes where water exchange with surrounding watersis limited, and thus a higher residence time is reasonable.

20 V. Rodellas et al. / Journal of Hydrology 466–467 (2012) 11–22

4.2.2. Relative contributions from different groundwater sourcesConsidering the outflowing marsh waters (ST47) as representa-

tive of the Peníscola marsh waters and the apparent marsh waterage of 1.2 d calculated above, the shallow flow of groundwaterdischarging from the local systems of Vinarós-Peníscola and IrtaRange (GWshallow) would account for 32% of the total water inputs,whereas the intermediate flow of groundwater seeping throughmarsh sediments from the same local systems (GWmid) would rep-resent a half (48%) of the total inputs. The deep flow of groundwa-ter from the regional Jurassic aquifer of El Maestrat (GWdeep)would represent 15% of the total water inputs and seawater contri-bution (SW) would only account for the remaining 5%. Applyingthese estimates to the direct measurements of the outflowingmarsh waters (37,000 m3 d�1), we can calculate a groundwaterdischarge from the regional Jurassic aquifer of El Maestrat (GWdeep)of 5600 m3 d�1 and a groundwater discharge from the local Vina-ròs-Peníscola and Irta Range systems (GWshallow + GWmid) of29,400 m3 d�1.

The outputs of the mixing model for all the stations are repre-sented in Fig. 7. The samples collected in the north part of themarsh are primarily (>80%) nourished by the shallow flow of freshgroundwater (GWshallow). When channelized marsh waters, thatflow southward, reach the central part of the wetland, their GWseep

component (including the GWdeep, the GWmid and the SW) in-creases suddenly, revealing the discharge of seeping groundwaterin this area. From this region southwards, the marsh waters retaina fairly constant component of GWdeep (�15%), suggesting the con-tinuous inputs of seeping groundwater along this marsh area. Arelatively small (<5%) contribution of seawater is observed innearly all the samples of the marsh.

Fig. 7. Relative contributions of the distinct groundwater flowpaths for each marshstation, corresponding to the August 2007 sampling: deep flow (GWdeep – black),shallow flow (GWshallow – white), intermediate flow (GWmid – light grey) andseawater (SW – dark grey).

4.3. Comparison between Ra and 222Rn-based groundwatercontribution and direct measurements

Following the mixing model described above (Section 4.2.1.),the relative contribution of each water source (i.e. GWshallow,GWmid, GWdeep, and SW) in the Peníscola marsh during the Febru-ary 2011 sampling may be obtained from the 222Rn, Ra isotopesand salinity measurements of the outflowing waters (F-ST47)and the end-members (Table 1). We used the results of 222Rn,226Ra and salinity obtained at stations F-ST7 as the GWshallow

and GWmid end-member and the F-ST32-D5 as the GWseep end-member. We used samples Pz1 and SW to characterize the GWdeep

and the SWseep end-members, respectively (temporal variations inthese end-members are expected to be minimal). Notice that the222Rn activities for the GWseep end-member do not present signif-icant differences between the two sampling campaigns, althoughthey were conducted in different seasons (summer 2007 and win-ter 2011). This reinforces the hypothesis that 222Rn becomes en-riched in GWseep during its advection through marsh sediments.Unlike 222Rn, as GWdeep is the main source of Ra isotopes to thewetland, a different mixing between GWdeep, SW and GWmid be-fore entering the wetland in February 2011 led to a diminutionof the Ra content of GWseep by a factor of about 2 relative tothe August 2007 sampling. Considering the 224Ra/228Ra AR of thegroundwater inputs into the system (1.5 ± 0.2) and the 224Ra/228RaAR of the outflowing marsh waters (1.30 ± 0.13), we appliedEq. (2) to derive a marsh water apparent age of 0.9 ± 0.7 daysfor February 2011. This Ra-derived apparent age is also consistentwith the residence time obtained from the direct measurementsof outflowing marsh waters (1.3 days). The mixing model for Feb-ruary 2011 results in relative contributions of 35, 48, 9 and 8 % forGWshallow, GWmid, GWdeep and SW, respectively (65% for GWseep,i.e., GWmid + GWdeep + SW).

These results can be compared with independent estimatesfrom direct flow measurements carried out in four sections of themarsh channels. Three of the measurements (F1–F3) were con-ducted in the initial part of the three channels that are mainlynourished by the shallow groundwater flow, allowing us to esti-mate the inflowing GWshallow. The section F4 was placed in themarsh outlet to the sea, in order to estimate the outflow of marshwaters. Assuming that there are no additional water inputs, the dif-ference of inflowing GWshallow (5000, 7200 and 5400 m3 d�1 mea-sured in sections F1, F2 and F3, respectively) and outflowingmarsh waters (48,000 m3 d�1 in section F4) represents the flowof water that enters the marsh from submerged seeps (i.e. theGWseep). We find that groundwater inputs from GWshallow wouldaccount for 37% of the outflowing marsh water, whereas GWseep

would represent the remaining 63%. These estimates are in goodagreement with the Ra and 222Rn derived groundwater fractions,evidencing the effectiveness of using these radionuclides, togetherwith salinity measurements, to estimate the contribution ofgroundwater from different aquifers to a wetland.

4.4. Temporal variability of groundwater contributions

The monthly measurements of 226Ra, 222Rn and salinity in out-flowing marsh waters (ST47), coupled with direct flow measure-ments, can be used to estimate the relative contributions ofgroundwater from each source over the annual cycle. Thesemonthly contributions were obtained by applying the mixing mod-el described in Eq. (1) to the monthly tracer measurements and themarsh water residence time derived from the monthly flow mea-surements (i.e. water volume divided by flow) (Table 3). The appli-cation of this mixing model implicitly assumes that salinities, 226Raand 222Rn concentrations in the end-members are constant overthe year.

Table 3Monthly measurements of 222Rn and 226Rn activities, salinity and flow carried out atstation ST47 from April 2007 to February 2008.

Sample Salinity 226Ra(dpm 100 L�1)

222Rn(103 dpm 100 L�1)

Flow(m3 d�1)

April 2007 4.8 2710 ± 110 1220 ± 30 59,800May 2007 5.0 3000 ± 400 1380 ± 80 73,100June 2007 6.0 3300 ± 140 1230 ± 50 66,800July 2007 6.0 3810 ± 150 1190 ± 40 47,000August 2007 5.6 4100 ± 200 1270 ± 60 39,800September

20075.7 3400 ± 200 1310 ± 30 34,100

October2007 5.4 3290 ± 140 1260 ± 50 38,800November

20075.2 3100 ± 300 1220 ± 70 50,900

December2007

5.1 3000 ± 300 1130 ± 30 46,300

January2008

5.3 2600 ± 200 1190 ± 70 82,900

February2008

5.7 3000 ± 200 1330 ± 40 47,100

V. Rodellas et al. / Journal of Hydrology 466–467 (2012) 11–22 21

Since temporal variations in groundwater discharge flows arelargely controlled by seasonal changes in precipitation over the an-nual cycle, we conducted a qualitative comparison between thegroundwater discharge derived from the mixing model and theprecipitation pattern. The relative contributions of each groundwa-ter source were converted to relative groundwater flows by multi-plying them by the flow measured in the marsh outlet. The data onprecipitation was obtained from the pluviometric stations of theSpanish Meteorological Agency (AEMET) at Alcalà de Xivert, lo-cated at 15 km from the study site. This comparison suggests thatestimated groundwater flows from the local shallow systems ofVinaròs-Peníscola and Irta Range (GWshallow + GWmid) respond tothe accumulated precipitation during the 3 months previous toeach sampling (Fig. 8). A similar lag between local recharge byrainfall and discharge, by 1–5 months, has been previously ob-served in other studies from coastal shallow aquifers (e.g. Chan-gnon et al., 1988; Garcia-Solsona et al., 2010; Smith et al., 2008).

Fig. 8. Monthly relative contributions of groundwater flows from the local shallow systeaquifer of El Maestrat (GWdeep – black) and seawater (SW – grey) to the Peníscola marsh fprevious to the sampling day is also represented.

On the other side, groundwater fluxes from the regional deepJurassic aquifer of El Maestrat (GWdeep) are relatively constant overthe year (6200 ± 1400 m3 d�1), showing minor variations in re-sponse to the local precipitation (Fig. 8). Indeed, the temporal vari-ations of deep groundwater discharge from the carbonate Jurassicaquifer of El Maestrat follows a hydrological functioning character-istic of most karstic aquifers (Padilla and Pulido Bosch, 1995): apermanent baseflow that softens the response to the rainfall(around 4500 m3 d�1) and a ‘‘quickflow’’ through the most trans-missive part of the aquifer, resulting in fast responses to therecharging events.

5. Conclusions

Ra isotopes and 222Rn have been used, coupled with salinitymeasurements, to estimate the contribution of groundwater fromdifferent sources to the Peníscola marsh (Spain), a coastal wetlandnourished by groundwater from the regional carbonate Jurassicaquifer of El Maestrat and the local systems of the Vinaròs-Penís-cola and the Irta Range. The combined use of 222Rn, Ra isotopesand salinity measurements proved to be a valuable tool to estimatethe relative contribution of different groundwater flows into a wet-land, confirmed by the comparison with a direct and independentmethod. Ra isotopes were also instrumental in providing withessential information of the wetland water age.

We also successfully monitored the evolution of groundwaterdischarges from two aquifers during a year by using 226Ra, 222Rn,salinity and direct measurements. Our results revealed that precip-itation is an important driver of groundwater discharges from thelocal shallow systems, whereas inputs from the regional carbonateaquifer are dominated by a permanent baseflow.

Acknowledgments

The authors gratefully acknowledge our colleagues at the Lab-oratori de Radioactivitat Ambiental for their help and assistanceduring field work. This project has been funded partially by the

ms of Vinaròs-Peníscola and Irta Range (GWshallow+mid – white), the deep carbonaterom April 2007 to February 2008. The accumulated precipitation during the 90 days

22 V. Rodellas et al. / Journal of Hydrology 466–467 (2012) 11–22

Spanish Government project EDASMAR (Ref. CGL2006-09274/HID)and the Consejo de Seguridad Nuclear project 2686-SRA. V.R.acknowledges financial support through a PhD fellowship(AP2008-03044) from MICINN (Spain). Support from a post-doc-toral fellowship to E.G.-S. (EX2009-0651; Plan Nacional de I-D+i2010–2012, Spain) is acknowledged. Support for the research ofP.M. was received through the prize ICREA Academia, funded bythe Generalitat de Catalunya.

References

Ballesteros, B.J., Marina, M., Mejías, M., Domínguez, J.A., 2007. Caracterizaciónhidroquímica del acuífero carbonatado profundo de El Maestrazgo (Castellón).[Hydrochemical characterisation of the deep, carbonated El Maestrazgo aquifer(Castellón)]. In: Coastal Aquifers: Challenges and Solutions. TIAC 07, vol. 23.Instituto Geológico y Minero de España – Hidrogeología y aguas subterráneas,pp. 549–564.

Beck, A.J., Rapaglia, J., Cochran, J.K., Bokuniewicz, H.J., 2007. Radium mass-balance inJamaica Bay, NY: evidence for a substantial flux of submarine groundwater.Mar. Chem. 106, 419–441.

Burnett, W.C., Kim, G., Lane-Smith, D., 2001. A continuous monitor for assessment of222Rn in the coastal ocean. J. Radioanal. Nucl. Chem. 69, 21–35.

Burnett, W.C., Aggarwal, P.K., Aureli, A., Bokuniewicz, H., Cable, J.E., Charette, M.A.,Kontar, E., Krupa, S., Kulkarni, K.M., Loveless, A., Moore, W.S., Oberdorfer, J.A.,Oliveira, J., Ozyurt, N., Povinec, P., Privitera, A.M.G., Rajar, R., Ramessur, R.T.,Scholten, J., Stieglitz, T., Taniguchi, M., Turner, J.V., 2006. Quantifying submarinegroundwater discharge in the coastal zone via multiple methods. Sci. Total.Environ. 367, 498–543.

Burnett, W.C., Peterson, R.N., Santos, I.R., Hicks, R.W., 2010. Use of automated radonmeasurements for rapid assessment of groundwater flow into Florida streams. J.Hydrol. 380, 298–304.

Changnon, S.A., Huff, F.A., Hsu, C.F., 1988. Relations between precipitation andshallow groundwater in Illinois. J. Clim. 1, 1239–1250.

Charette, M.A., 2007. Hydrologic forcing of submarine groundwater discharge:insight from a seasonal study of radium isotopes in a groundwater-dominatedsalt marsh estuary. Limnol. Oceanogr. 52, 230–239.

Charette, M.A., Buesseler, K.O., Andrews, J.E., 2001. Utility of radium isotopes forevaluating the input and transport of groundwater-derived nitrogen to a CapeCod estuary. Limnol. Oceanogr. 46, 465–470.

Charette, M.A., Splivallo, R., Herbold, C., Bollinger, M.S., Moore, W.S., 2003. Saltmarsh submarine groundwater discharge as traced by radium isotopes. Mar.Chem. 84, 113–121.

Cook, P.G., Wood, C., White, T., Simmons, C.T., Fass, T., Brunner, P., 2008.Groundwater inflow to a shallow, poorly-mixed wetland estimated from amass balance of radon. J. Hydrol. 354, 213–226.

Corbett, D.R., Burnett, W.C., Cable, P.H., Clark, S.B., 1997. Radon tracing ofgroundwater input into Par Pond, Savannah River Site. J. Hydrol. 203, 209–227.

De Stefano, L., 2004. Freshwater and Tourism in the Mediterranean. WWFMediterranean Programme Report, Rome.

De Weys, J., Santos, I.R., Eyre, B.D., 2011. Linking groundwater discharge to severeestuarine acidification during a flood in a modified wetland. Environ. Sci.Technol. 45, 3310–3316.

Garcia-Orellana, J., Sanchez-Cabeza, J.A., Masqué, P., Àvila, A., Costa, E., Loÿe-Pilot,M.D., Bruach-Menchén, J.M., 2006. Atmospheric fluxes of 210Pb to the westernMediterranean Sea and the Saharan dust influence. J. Geophys. Res. 111, 1–9.

Garcia-Solsona, E., Masqué, P., Garcia-Orellana, J., Rapglia, J., Beck, A.J., Cochran, J.K.,Bokuniewicz, H.J., Zaggia, L., Covallavini, F., 2008a. Estimating submarinegroundwater discharge around Isola La Cura, northern Venice Lagoon (Italy),by using the radium quartet. Mar. Chem. 109, 292–306.

Garcia-Solsona, E., Garcia-Orellana, J., Masqué, P., Dulaiova, H., 2008b. Uncertaintiesassociated with 223Ra and 224Ra measurements in water via DelayedCoincidence Counter (RaDeCC). Mar. Chem. 109, 198–219.

Garcia-Solsona, E., Garcia-Orellana, J., Masqué, P., Rodellas, V., Mejías, M.,Ballesteros, B., Domínguez, J.A., 2010. Groundwater and nutrient dischargethrough karstic coastal springs (Castelló, Spain). Biogeosciences 7, 2625–2638.

Hancock, G.J., Murray, A.S., 1996. Sources and distribution of dissolved radium inthe Bega River estuary, southeastern Australia. Earth Planet Sci. Lett. 138, 145–155.

Kluge, T., Ilmberger, J., von Rohden, C., Aeschbach-Hertig, W., 2007. Tracing andquantifying groundwater inflow into lakes using a simple method for radon-222 analysis. Hydrol. Earth Syst. Sci. 11, 1621–1631.

Krabbenhoft, D.P., Bowser, C.J., Anderson, M.P., Valley, J.W., 1990. Estimatinggroundwater exchange with lakes. (1) The stable isotope mass balance method.Water. Resour. Res. 26, 2445–2453.

Macintyre, S., Wanninkhof, R., Chanton, J.P., 1995. Trace gas exchange across the air-sea interface in freshwater and coastal marine environments. In: Matson, P.A.,Harris, R.C. (Eds.), Biogenic Trace Gases: Measuring Emissions from Soil andWater. Blackwell Science Ltd., pp. 52–97.

Mejías, M., Garcia-Orellana, J., Plata, J.L., Marina, M., Garcia- Solsona, E., Ballesteros,B., Masqué, P., López, J., Fernández-Arrojo, C., 2008. Methodology ofhydrogeological characterization of deep carbonate aquifers as potentialreservoirs of groundwater, Case of study: the Jurassic aquifer of ElMaestrazgo (Castellón, Spain). Environ. Geol. 54 (3), 521–536.

Mejías, M., Ballesteros, B.J., Antón-Pacheco, C., Domínguez, J.A., Garcia-Orellana, J.,Garcia- Solsona, E., Masqué, P., 2012. Methodological study of submarinegroundwater discharge from a karstic aquifer in the Western MediterraneanSea. J. Hydrol. http://dx.doi.org/10.1016/j.jhydrol.2012.06.020.

Moore, W.S., 2000. Ages of continental shelf waters determined from 223Ra and224Ra. J. Geophys. Res. 105, 117–122.

Moore, W.S., Arnold, R., 1996. Measurement of 223Ra and 224Ra in coastal watersusing a delayed coincidence counter. J. Geophys. Res. 101, 1321–1329.

Moore, W.S., Reid, D.F., 1973. Extraction of radium from natural waters usingmanganese-impregnated acrylic fibers. J. Geophys. Res. 79, 8880–8886.

Moore, W.S., Blanton, J.O., Joye, S.B., 2006. Estimates of flushing times, submarinegroundwater discharge, and nutrient fluxes to Okatee Estuary, South Carolina. J.Geophys. Res. 111, C09006.

Padilla, A., Pulido Bosch, A., 1995. Study of hydrographs of karstic aquifers by meansof correlation and cross-spectral analysis. J. Hydrol. 168, 73–89.

Pearce, F., Crivelli, A.J., 1994. Characteristics of Mediterranean Wetlands.Conservation of Mediterranean Wetlands – 1. MedWet Publications, Arles.

Rama, Moore, W.S., 1996. Using the radium quartet for evaluating groundwaterinput and water exchange in salt marshes. Geochim. Cosmochim. Acta 60,4645–4652.

Sánchez-Navarro, J.A., Coloma, P., Pérez-Garcia, A., 2004. Evaluation of geothermalflow at the springs in Aragón (Spain), and its relation to geologic structure.Hydrogeol. J. 12, 601–609.

Santos, I.R., Niencheski, F., Burnett, W.C., Peterson, R., Chanton, J., Andrade, C.F.F.,Milani, I.B., Schmidit, A., Knoeller, K., 2008. Tracing anthropogenically drivengroundwater discharge into a coastal lagoon from southern Brazil. J. Hydrol.353, 275–293.

Schmidt, A., Gibson, J.J., Santos, I.R., Schubert, M., 2010. The contribution ofgroundwater discharge to the overall water budget of two typical Boreal lakesin Alberta/Canada estimated from a radon mass balance. Hydrol. Earth. Syst. Sci.14, 79–89.

Schot, P.P., Wassen, M.J., 1993. Calcium concentrations in wetland groundwater inrelation to water sources and soil conditions in the recharge area. J. Hydrol. 141,197–217.

Smith, C.G., Cable, J.E., Martin, J.B., Roy, M., 2008. Evaluating the source andseasonality of submarine groundwater discharge using a radon-222 pore watertransport model. Earth Planet. Sci. Lett. 273 (3–4), 312–322.

Sun, Y., Torgersen, T., 1998. The effects of water content and Mn-fiber surfaceconditions on 224Ra measurement by 220Rn emanation. Mar. Chem. 62, 299–306.

United Nations Scientific Comité on the Effects of Atomic Radiation (UNSCEAR),2000. Sources and Effects of Ionizing Radiation, vol. I. Sources, United NationsPublications, New York.

Young, M.B., Gonneea, M.E., Fong, D.A., Moore, W.S., Herrera-Silveira, J., Paytan, A.,2008. Characterizing sources of groundwater to a tropical coastal lagoon in akarstic area using radium isotopes and water chemistry. Mar. Chem. 109, 377–394.