Scholarly article on topic 'Geochemical and hydrological processes controlling groundwater salinity of a large inland wetland of northwest Australia'

Geochemical and hydrological processes controlling groundwater salinity of a large inland wetland of northwest Australia Academic research paper on "Earth and related environmental sciences"

CC BY-NC-ND
0
0
Share paper
Academic journal
Chemical Geology
Keywords
{"Fortescue Marsh" / Pilbara / Salinity / Groundwater / "Stable isotopes" / Cl/Br}

Abstract of research paper on Earth and related environmental sciences, author of scientific article — Grzegorz Skrzypek, Shawan Dogramaci, Pauline F. Grierson

Abstract Understanding mechanisms of hydrochemical evolution of groundwater under saline and brine wetlands in arid and semiarid regions is necessary to assess how groundwater extraction or injection in large-scale basins may affect the natural interface between saline–fresh aquifers in those systems. We investigated the evolution of groundwater of the Fortescue Marsh, a large inland wetland of northwest Australia that is mainly supplied by floodwater from the upper Fortescue River catchment. The marsh is located in the Pilbara region, one of the world's major iron ore provinces, where mining activities often occur below water tables. Here, we quantified the stable isotope and chemical composition of groundwater, surface water and rainfall in and around the marsh to better understand how saline marshes and playa lakes function geochemically, hydrologically and ecologically. The deep groundwater (>50m depth) of the Fortescue Marsh is highly saline (>100gL−1), whilst shallow groundwater (~0–20m depth) and surface water are mainly fresh or brackish. Currently, the marsh is mainly recharged by occasional floodwater. Consequently, salt in the marsh is concentrated by evaporation of rainfall. We expected that the hydrochemical composition of surface and groundwater would therefore reflect the chemical composition of rainwater. We analysed 206 water samples for stable isotope composition (δ 2H, δ 18O) and water chemistry, including: pH, dissolved oxygen, electrical conductivity (EC), and total dissolved solids (TDS), as well as Na, Ca, Mg, K, Si, Fe, HCO3, SO4, Cl, Sr and Br. We then developed geochemical models and a salt inventory to estimate the geological time of salt accumulation and to decouple geochemical characteristics of salt from modern groundwater. We found that groundwater associated with the marsh can be divided into two distinct groups that are characterised by their stable isotope compositions; i) fresh and brackish groundwater (TDS <10gL−1; δ 18O −8.0±0.9‰) and ii) saline and brine groundwater (TDS >10gL−1, δ 18O varies from +2.5 to −7.2‰). Fresh groundwater was evaporated by <20% compared to rainwater. Brackish water mainly reflects modern recharge whilst saline and brine groundwater reflects mixing between modern rainfall, brackish water and relatively old groundwater. The Cl load in the groundwater originates from rainfall and accumulates over time as it is recycled due to precipitation of evaporates and re-dissolution on the marsh during subsequent flooding events. The stable isotope composition of the deeper brine groundwater also suggests a complex evolution, which cannot be explained by evaporation under current conditions from modern rainfall. We thus conclude that the deeper brine groundwater under the Fortescue Marsh developed under a different climatic regime and that the current salt in the marsh has accumulated over at least 40,000years, but could have been as long as 700,000years.

Academic research paper on topic "Geochemical and hydrological processes controlling groundwater salinity of a large inland wetland of northwest Australia"

CHEMICAL GEOLOGY

Geochemical and hydrological processes controlling groundwater (J)croSSMark

salinity of a large inland wetland of northwest Australia^

Grzegorz Skrzypek a,b'*, Shawan Dogramacib,c, Pauline F. Grierson a,b

a West Australian Biogeochemistry Centre, School of Plant Biology, The University of Western Australia, 35 Stirling Highway, Crawley, WA 6009, Australia b Ecosystems Research Group, School of Plant Biology, The University of Western Australia, 35 Stirling Highway, Crawley, WA 6009, Australia c Rio Tinto Iron Ore, 152-158 St George's Terrace, Perth, WA 6000, Australia

ARTICLE INFO ABSTRACT

Understanding mechanisms of hydrochemical evolution of groundwater under saline and brine wetlands in arid and semiarid regions is necessary to assess how groundwater extraction or injection in large-scale basins may affect the natural interface between saline-fresh aquifers in those systems. We investigated the evolution of groundwater of the Fortescue Marsh, a large inland wetland of northwest Australia that is mainly supplied by floodwater from the upper Fortescue River catchment. The marsh is located in the Pilbara region, one of the world's major iron ore provinces, where mining activities often occur below water tables. Here, we quantified the stable isotope and chemical composition of groundwater, surface water and rainfall in and around the marsh to better understand how saline marshes and playa lakes function geochemically, hydrologically and ecologically. The deep groundwater (>50 m depth) of the Fortescue Marsh is highly saline (> 100 g L-1), whilst shallow groundwater (-0-20 m depth) and surface water are mainly fresh or brackish. Currently, the marsh is mainly recharged by occasional floodwater. Consequently, salt in the marsh is concentrated by evaporation of rainfall. We expected that the hydrochemical composition of surface and groundwater would therefore reflect the chemical composition of rainwater. We analysed 206 water samples for stable isotope composition (S2H, S18O) and water chemistry, including: pH, dissolved oxygen, electrical conductivity (EC), and total dissolved solids (TDS), as well as Na, Ca, Mg, K, Si, Fe, HCO3, SO4, Cl, Sr and Br. We then developed geochemical models and a salt inventory to estimate the geological time of salt accumulation and to decouple geochemical characteristics of salt from modern groundwater. We found that groundwater associated with the marsh can be divided into two distinct groups that are characterised by their stable isotope compositions; i) fresh and brackish groundwater (TDS < 10 g L-1; S18O - 8.0 ± 0.9%») and ii) saline and brine groundwater (TDS > 10 g L-1, S18O varies from +2.5 to -7.2%). Fresh groundwater was evaporated by <20% compared to rainwater. Brackish water mainly reflects modern recharge whilst saline and brine groundwater reflects mixing between modern rainfall, brackish water and relatively old groundwater. The Cl load in the groundwater originates from rainfall and accumulates over time as it is recycled due to precipitation of evaporates and re-dissolution on the marsh during subsequent flooding events. The stable isotope composition of the deeper brine groundwater also suggests a complex evolution, which cannot be explained by evaporation under current conditions from modern rainfall. We thus conclude that the deeper brine groundwater under the Fortescue Marsh developed under a different climatic regime and that the current salt in the marsh has accumulated over at least 40,000 years, but could have been as long as 700,000 years.

© 2013 The Authors. Published by Elsevier B.V. All rights reserved.

Article history:

Received 30 January 2013

Received in revised form 19 August 2013

Accepted 21 August 2013

Available online 29 August 2013

Editor: David R. Hilton

Keywords:

Fortescue Marsh

Pilbara

Salinity

Groundwater

Stable isotopes

☆ This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial-No Derivative Works License, which permits non-commercial use, distribution, and reproduction in any medium, provided the original author and source are credited.

* Corresponding author at: School of Plant Biology M090, The University of Western Australia, 35 Stirling Highway, Crawley, WA 6009, Australia. Tel.: +61 8 6488 4584.

E-mail addresses: grzegorz.skrzypek@uwa.edu.au, gskrzypek@yahoo.com (G. Skrzypek), shawan.dogramaci@riotinto.com (S. Dogramaci), pauline.grierson@uwa.edu.au (P.F. Grierson).

1. Introduction

Wetland environments, including inland salt lakes and marshes, are important ecological systems, serving as specialist habitats that also store and filter water and dampen flood and storm effects. Wetlands are driven by their hydrologic regimes, which can be extreme in warm, arid environments as they are often subject to prolonged droughts interrupted by episodic rainfall events. In particular, wetlands in arid regions are susceptible to chemical transformations of their shallow groundwater in drought periods through hypersalinity or through mobilisation of metals and acidification associated with redox changes. Whilst there has been

0009-2541/$ - see front matter © 2013 The Authors. Published by Elsevier B.V. All rights reserved. http://dx.doi.Org/10.1016/j.chemgeo.2013.08.035

increasing attention given to surface water-groundwater interactions across a range of ecosystem types, particularly with respect to water fluxes, biogeochemical processes and ecology, the groundwater isotope geochemistry of inland wetland systems in more arid regions remains poorly understood (Lamontagne et al., 2005; Jolly et al., 2008).

Saline wetlands and lakes are common in the arid regions of inland Australia, where evaporation can greatly exceed rainfall. These saline ecosystems are usually formed in internally drained basins whenever the floors of such basins intersect the water table (Duffy and Al-Hassan, 1988; Yechiechi and Wood, 2002). The salinity of inland lakes is mainly driven by either evaporation of slowly discharging groundwater (groundwater dependent lakes), or progressive evaporation of rapidly delivered pulses of floodwaters that eventually seeps and mixes with the underlying groundwater (surface water dependent lakes). Most of the inland saline lakes and marshes across Australia are of the former type and considered "windows" to underlying highly evaporated groundwater (Rosen etal., 1994; Harrington et al., 2008). In such systems, the hydrochemistry of the lake water is mainly dependent on the degree of evaporation and on the mineralogy and chemical composition of the rocks through which the groundwater percolates. Generally, groundwater becomes more saline with increasing distance from recharge zones, where time for water-rock interaction is longer, resulting in aquifer mineral dissolution and exchange with groundwater in the discharge zones (Chebotarev, 1955). In groundwater fed systems, the discharge is also assumed to be relatively constant over time and frequently concentrations of major ions are close to saturation.

In contrast to these saline playa lake systems, large-scale wetlands mainly supplied by occasional floodwater are relatively uncommon in Australia's inland. Salt in these wetlands originates from evaporation of rainfall; surface and groundwater hydrochemical compositions are thus likely to reflect the chemical composition of rainwater inputs (Herczeg and Edmunds, 1999). Large volumes of fresh floodwater are delivered sporadically (on cycles of years to decades), such that the wetland is subject to rapid wetting periods when floodwater is rushing in to a terminal basin followed by a slow drying period as the water progressively evaporates. As a consequence, the salinity of the surface water varies over time as initially delivered water is fresh and becomes progressively more saline as it evaporates. In both types of systems, extensive evaporation often leads to accumulation of carbonates, gypsum and halite (Rosen et al., 1994; Harrington et al., 2008; Cartwright et al., 2010). Precipitation of calcium and magnesium carbonates can also result in formation of thick layers of calcrete, which in turn act as a "hard pan", restricting surface water infiltration and mixing of surface water with groundwater. However, the relative influence of these differing mechanisms ofgroundwater evolution on the natural interface between saline-fresh aquifers in these systems is not clear.

Geochemical modelling and quantification of the stable isotope and chemical composition of groundwater, surface water and rainfall may help in understanding how saline marshes and playa lakes of these types function hydrologically (and therefore ecologically) and what hydrochemical and hydrological regimes are responsible for their formation. A combination of stable isotope and biogeochemical analyses is particularly useful for determining evaporative losses and time of water retention in surface pools (Fellman et al., 2011), characterisation of local recharge conditions and unravelling reasons for increasing groundwater salinity (Tweed et al., 2011; Costelloe et al., 2012), as well as for calculation of the volume of rain required to induce recharge in arid zones (Harrington et al., 2002; Dogramaci et al., 2012). The chemical and stable isotope composition of groundwater can also be used to identify the dominant hydrological processes in semiarid basins and present day relationships between their hydrology and water chemistry (Acheampong and Hess, 2000; Herczeg et al., 2001; Harrington et al., 2008).

Stable isotopic mass balance models have also been successfully utilised in studies of various saline lake complexes (Herczeg et al., 1992; Gat and Levy, 1978). In addition, untangling processes that involve

groundwater-surface water interactions requires understanding of the effects of topography, geology and climate on groundwater flow regimes (Toth, 1963). This multidimensional approach allows reconstruction of geochemical processes leading to formation of saline and brine terminal lakes that are frequently associated with unique saline ecosystems (Bowler, 1986; Dutkiewicz et al., 2000; Yechiechi and Wood, 2002).

The Fortescue Marsh is the largest wetland of inland northwest Australia; however, the extent of surface water on the marsh varies dramatically from total dryness to > 1000 km2 depending on the frequency and intensity of rainfall. The marsh receives drainage from the upper catchment of the Fortescue River in the Hamersley Basin. The Lower Fortescue River is separated from the Fortescue Marsh and Upper Fortescue River by the Goodiadarrie Hills (Fig. 1A, B). Surface water overflow may occur only after very large cyclones, events that may occur only a few times every 100 years. Unlike surface water flow, groundwater horizontal flow is very slow and hydraulic conductivity ranges from 0.1 to 1 m/day (Rio Tinto, 2013). The degree to which there is groundwater discharge from the Fortescue Marsh to the Lower Fortescue River is currently unknown, although discharge, if any, is likely to be very slow given the horizontal flow rates across the marsh.

The deep groundwater (> 50 m b.g.l, below ground level) of the Fortescue Marsh is highly saline (> 100 g L-1), whilst shallow waters and surface water are fresh or brackish. This contrasts with the majority of saline wetlands and playa like systems in Australia, which are characterised by higher salinity in the shallow aquifers due to direct evaporation from exposed groundwater or through capillary action (De Dekker, 1988). Thus, the Fortescue Marsh is acting at least in part as a terminal basin recharged via episodic cyclones coupled with frequent and prolonged dry periods. This unusual combination has led to unique hydrochemical characteristics and relationships between fresh and saline waters.

Understanding the evolution of surface and groundwater hydro-chemistry, especially the balance between highly saline and fresh groundwater, is crucial for proper water management in arid zones in the face of both anthropogenic and climate driven changes. The Fortescue Marsh, similar to many arid zone wetlands around the world, is an area of high national and international conservation significance (Environmental Protection Authority, 2013). The marsh is also located in one ofthe world's major iron ore provinces; ore extraction has undergone rapid expansion in recent decades and mining activities often occur below water tables, which require new strategies and new knowledge for optimising water management (Environmental Protection Authority, 2013). Rainfall across the Pilbara region over recent decades has also been high relative to the preceding century or so (Cullen and Grierson, 2007), although future climate scenarios and the frequency of recharge events resulting from summer cyclones remain uncertain. Consequently, there is an urgent need to better understand the hydrologic functioning of key systems in order to assess the risk and potential impacts of changing hydrological regimes on groundwater resources across the region.

Here, we sought to understand groundwater flow regimes and to determine the predominant processes that contribute to the water and salt balance and evolution of hydrochemistry in the Fortescue Marsh catchment. We used a conventional mass balance model based on major ions, geochemical modelling as well as water stable isotopes (S2H and S18O), to distinguish the sources of water and salt in various geochemical types of water. We also estimated the time required for salt accumulation in the Fortescue Marsh and propose a model to explain how salt beneath the marsh accumulates and why brackish waters have stable isotope signatures similar to that of fresh waters.

2. Materials and methods

2.1. Hydrology and hydrogeology of the Fortescue Valley

The Fortescue Marsh occupies a trough between the Chichester and Hamersley ranges in the semi-arid Pilbara region of northwest Australia

Fig. 2. The conceptual geological model of the Fortescue Marsh area; pink and yellow points indicate sampled bores (the 3D model does not encompass the entire sampling area shown in Fig.1).

and occurs as a mid-section of the Fortescue River (Fig. 1A, B). The Fortescue River is approximately 760 km long, rising 30 km south of the town of Newman (Fig. 1A). The Fortescue River drains a 30,000 km2 catchment of the Hamersley Basin and contiguously flows only following exceptionally large flood events (www.bom.gov.au). Physiographically, the river is divided into two sections, as the Upper Fortescue River (UFR) is separated from Lower Fortescue River (LFR) by the Goodiadarrie Hills (Figs. 1,2). The LFR drains in a westerly direction towards the coast, whereas east of the hills the Fortescue Marsh receives drainage from the UFR. The scope of this study only covers the UFR from the headwaters to the Goodiadarrie Hills, which includes the marsh and a small section of the UFR (Figs. 1, 2).

In addition to the UFR, numerous ephemeral creeks on the southern and northern flanks of the Fortescue Valley discharge to the marsh directly. Runoff from rainfall on the flanks initially drains down gradient as overland flow before concentrating in a defined flow channel. In steep areas on both sides of the Fortescue Valley, the runoff is rather rapid with relatively low losses and defined drainage channels are typically in close proximity. In the lower slope areas of the marsh, runoff processes are rather slow with relatively higher evaporative losses and a greater distance between defined drainage channels. The surface area of the marsh is ~1000 km2 and elevation ranges from 400 to 405 m asl (above sea level) (Fig. 2). The flood storage level in the marsh would have to exceed the 410 m elevation mark in order to overspill into the LFR. Since 1981, the Fortescue River upstream of the marsh has been regulated by the Ophthalmia Dam (capacity 32 GL and surface area of 12.6 km2), which provides water to the township of Newman (Payne and Mitchell, 1999).

Following significant flood events, which are usually associated with large summer cyclones, water may cover > 1000 km2 of the marsh surface. Evaporation of surface water results in increasing salinity as the water recedes. Salt efflorescences have been periodically identified only on the dry surfaces of the marsh, particularly near the edges of pools after ponded rainwater and floodwater had evaporated; salt deposits are not observed in other parts of the Upper Fortescue River (Rio Tinto, 2013). The water table depth along the valley ranges from ~25 m at the base of the ranges to about 0 m (i.e., at surface) in the middle of the marsh. The water table depth along the Fortescue River varies

seasonally depending on rainfall frequency and flooding and also ranges from ~25 m to 0 m under the saturated Fortescue River bed.

Four major formations in the Hamersley Group comprise local aquifers: (1) Marra Mamba Iron Formation (iron oxides dominated), (2) Wittenoom Formation (dolomite and shale dominated), (3) Brockman Iron Formation (iron oxides dominated) and (4) Tertiary and Quaternary surficial sediments (carbonate dominated) (Hickman, 1983). The Fortescue Marsh is developed on Quaternary sedimentary deposits underlined by Wittenoom Formation, which is characterised by fractured dolomite and shale (Fig. 2). The Wittenoom Formation is frequently overlain by the Oakover Formation, a sequence of younger Tertiary lacustrine carbonate, silcrete and mudstone rocks deposited in the palaeodrainage of the Fortescue River Valley (Clout, 2011).

The main aquifers under the Fortescue Marsh are the colluvial and alluvial aquifers, which form a thin veneer of sediments over the upper flanks of the Chichester and Hamersley Ranges and thicken (up to ~50 m) towards the centre of the marsh, transitioning from colluvial slopes to alluvial outwash. These sediments consist mainly of unconsolidated and usually poorly sorted silt, sand and gravel. Sediment permeability is highly variable and generally low due to significant clay contents. Both alluvial and colluvial aquifers are locally confined by consolidated massive clays and calcrete and slicrete layers several metres thick. The complexity of the lithology is reflected by a wide range of water salinity across the marsh. However, in general the surface waters are fresh to brackish and deep waters saline to brine. In general, the direction of subsurface flow in alluvial aquifers is along the main marsh plain from the east to the west. The hydraulic gradient along the Fortescue River between Ophthalmia Dam and Goodiadarrie Hills is very low (~100 m/180 km), and is close to zero (~1 m/90 km) on the marsh between 14 Miles Pool and Goodiadarrie Hills. The hydraulic gradients along secondary tributaries, e.g. Weeli Wolli Creek are significantly greater (250 m/70 km); however, the bank storage and total discharge from surrounding creeks are very limited (Fig. 1B).

22. Rainfall patterns over the Fortescue Valley

The study area is characterised by a semiarid climate with extended drought periods, scattered localised showers and occasional large

Fig. 1. (A) Satellite image showing location of the study within the Fortescue Marsh Catchment (5/07/2004 Pilbara TM Mosaic 3/1). The river channel position is approximate as it changes from year-to-year. Pink and yellow points indicate sampled bores. The extent of the flooded area on the marsh, as of 11/02/2012, was assessed using a helicopter survey. The Block Model area is presented in Fig. 2. (B) Major pathways of the drainage in the study area. Note very low hydraulic gradient, between Ophthalmia Dam and Goodiadarrie Hills -100 m/180 km and 14 Miles Pool and Goodiadarrie Hills is close to zero (-1 m/90 km). The hydraulic conductivity is also low 0.1-1.0 m/yr.

cyclone events that may result in large-scale flash floods. The average annual pan evaporation is >3000 mm yr-1, which is almost ten-fold greater than the average rainfall (www.bom.gov.au). We summarise here the variability in rainfall volume and distribution across the Fortescue Valley based on the most recent decade of rainfall data (mean annual sum for period 2001-2011, www.bom.gov.au) from three weather stations located in different parts of the valley (Fig. 1a). Mean annual rainfall sums are different for the upper and lower parts of the catchment, reflecting both an orographic effect and distance from the ocean (Fig. 3). Mean annual rainfall is ~350yr-1 at the marsh (Marillana Weather Station) and similar in the upper part of the catchment (Newman Weather Station, upstream from the Fortescue Marsh, ~96 km SE from Marillana). In contrast, the average rainfall at the Wittenoom weather station (~ 119 km NW from Marillana and downstream from the marsh) is significantly higher at ~472 mmyr-1. Approximately 85% of the annual rainfall occurs in the austral summer months, mainly from December to March (Marillana Station, 1936-present).

The daily amounts of rain received by different weather stations across the catchment were also highly variable. Most of the small rainfall events were very localised and therefore not observed at all three stations. This observation is confirmed by the lack of correlation and the statistically insignificant relationship between daily sums recorded at the three stations (Fig. 3). However, these smaller events do not result in surface flows and are unlikely to contribute to groundwater recharge (Dogramaci et al., 2012). In contrast, most of the large rainfall events, which are associated with cyclones, were observed at all three stations. The correlations between each station's monthly sum are statistically significant (R2 = 0.69 p < 0.001 Marillana-Newman, R2 = 0.66 p < 0.001 Marillana-Wittenoom; and R2 = 0.50 p < 0.001 Newman-Wittenoom), confirming that the observed pattern in mean rainfall distribution was consistent for at least the last 10 years (Fig. 3).

2.3. Sampling methods

Groundwater from 61 production and monitoring bores at 48 locations covering most of the upper Fortescue Valley was sampled four times over a two-year period (Dec 2009-Dec 2011) (Figs. 1 & 2). These samples encompassed: 1) head waters upstream from the Fortescue Marsh, including the Ophthalmia Dam Lake; 2) groundwater and surface water from the marsh; and 3) groundwater from alluvial fans of major tributaries flowing into the marsh (Sandy, Weeli Wolli and Coondiner Creeks). Representative samples were collected shortly before and after the wet season; however, not all bores were sampled during each season owing to difficulties with access.

The sampling programme was designed to assess the influence of rainfall patterns on isotopic and chemical composition of groundwater in the alluvium aquifer. The first sampling in May 2010 was conducted after a dry period and only 155 mm of rain had occurred over the previous 12 months (~50% of mean annual sum). The last large flooding over the marsh prior to sampling occurred in 2006, when Marillana Station received 720 mm (double the mean annual precipitation) over five months of the wet austral summer (Dec 2005-April 2006). In contrast, a total of 743 mm of rainfall occurred over the entire three years subsequent to May 2006. The second sampling in Oct 2010 was conducted after an extended drought period (only 201 mm in the 12 months prior to sampling). The third sampling in May 2011 followed an average wet season with rainfall close to the long-term mean (308 mm between Jan 2010 and Apr 2011, with a total of 397 mm over the 12 months prior to sampling). The fourth and last sampling in Dec 2011 followed a dry winter period with only 106 mm (May-Nov 2011). The number of bores varied across the catchment and sampling was more restricted in the middle part of the marsh (Figs. 1, 2).

All groundwater samples were collected from production and monitoring bores after pumping out at least three casing volumes. Sample

Fig. 3. Distribution of rainfall for the three weather stations closest to the Fortescue Marsh (www.bom.gov.au).

EC, pH, Eh, dissolved oxygen and temperature were measured in situ using a portable multimeter (INSITU Multi parameter Water Quality TROLL® 9500). Water samples were collected in 0.5 L containers for major ion analyses and in 40 mL sealed glass vials for stable isotope analysis following filtration using a sterile syringe microfilter (0.21 |am).

2.4. Analytical methods

In total, 206 water samples were analysed for stable isotope composition (62H, 618O) and water chemistry, including: pH, dissolved oxygen, electrical conductivity (EC), and total dissolved solids (TDS), as well as Na, Ca, Mg, K, Si, Fe, HCO3, SO4, Cl, Sr and Br. The stable hydrogen and oxygen isotope compositions of water samples were determined utilising an Isotopic Liquid Water Analyser Picarro L1102-i (Picarro, Santa Clara, California, USA) at the West Australian Biogeochemistry Centre, The University of Western Australia. The 62H and 618O values were normalised to the VSMOW scale (Vienna Standard Mean Ocean Water), following a three-point normalisation (Paul et al., 2007) based on three laboratory standards, each replicated twice and reported in per mil (%) (Skrzypek et al., 2010). All laboratory standards were calibrated against international reference materials determining the VSMOW-SLAP scale (Coplen, 1996) provided by the International Atomic Energy Agency (62HVSMOW and 618OVSMOW equals and 0% and 62HSLAP = -428.0% and 618OSLAP = - 55.50%). The analytical uncertainty (one standard deviation) was determined as 1.0% for 62H and 0.1% for 618O.

The concentrations of major ions in water samples were analysed by SGS laboratories Australia Pty. Ltd. in Perth, Western Australia, utilising Inductively Coupled Plasma-Optical Emission Spectrometer (ICP-OES, method ref. APHA 3120B) for Ca, Mg, Na, and K cations, Ion Chromatography for Br- (method ref. APHA4110B), Discrete Analyser in Water by Aquakem DA for SO4- and Cl- anions (method ref. APHA 4500) and HCO- alkalinity by acid titration (method ref. APHA 2320). The water salinity used in this study follows the classification of Freeze and Cherry (1979) based on total dissolved solids (TDS): fresh < 1 g L-1; brackish > 1 and < 10 g L-1; saline water > 10 and < 100 g L-1 and brine > 100 g L-1.

2.5. Data analysis

The cumulative water loss by evaporation from surface water pools across the marsh was calculated using stable isotope mass balance equation re-formulated for a non-steady state model: f = 1 - ((6L - 6*) / (6P - 6*))(1/ m), where 6L is the observed 618O value in surface or groundwater and 6P is the mean 618O value of precipitation (Dogramaci et al., 2012). The 6 value represented the local limiting isotope enrichment from evaporation as 6* = (h6A + s) / (h - s / 1000), reflecting local climate conditions, whilst m reflects fractionation due to air temperature and humidity as previously defined by Gat (1981); m = (h - s / 1000) / (1 - h + sk / 1000). For details relating to the derivation of the equation, refer to Gibson and Reid (2010). Relationships between 62H, 618O, d-excess (deuterium excess as defined by Dansgaard (1964) is d = 62H - 8 x 618O) and Cl concentration were analysed using linear regression and statistical significance was evaluated based on the p-value for the F-test as a part of a one-way ANOVA analysis (p < 0.05).

3. Results

The salinity (TDS) of the unconfined alluvium aquifers across the Fortescue Valley ranged from fresh (0.25 g L-1) to brine (190 g L-1). Based on chemical compositions, groundwater in the Fortescue Valley can be divided into two distinct groups; i) fresh and brackish groundwater (FBG) (TDS < 10 g L-1) and ii) saline and brine groundwater (SBG) beneath the Fortescue Marsh (TDS > 10 g L-1, Figs. 1 & 4). The chemistry of the FBG is dominated by Mg-Ca and HCO3 ions,

whereas the SBG is dominated by Na and Cl ions. The Na and Cl concentrations constitute > 80% of major cations and anions of SBG. The ratio of Na and Cl relative to other dissolved cations and anions increases as the water becomes more saline. The Cl/Br ratio of the FBG is similar or slightly higher than that of seawater ratio (288), suggesting these ions are of marine origin. In contrast, most of the SBG groundwater is characterised by Cl/Br ratios of between ~800 and ~ 12,000, suggesting modification of the original marine signature (Cartwright et al., 2006). Ratios of Sr/Ca, SO4/Ca, and Mg/Ca increased with increasing Cl concentrations for all groundwater across the Fortescue Valley (Fig.4).The progressive increase in the concentration of Sr, Mg and SO4 relative to Ca concentration (e.g. an increasing Sr/Ca ratio) may reflect carbonate mineral precipitation and the removal of Ca and HCO3 at the higher ends of the salinity spectrum (Fig. 4). Consequently, the Sr/Ca ratio increases with increasing salinity, which in turn is reflected by higher ô18O (Fig. 4F).

The stable isotope compositions of the fresh-brackish groundwater (FBG) in the Fortescue Valley are characterised by a relatively narrow range of ô18O - 8.0 ± 0.9% and ô2H - 56.2 ± 5.3% (Fig. 5). Both ô18O and ô2H values fall slightly to the right of the Local Meteoric Water Line (LMWL ô2H = 7.03ô18O+4.78), indicating partial evaporation prior to recharge (Fig. 5). The slope of the regression line for ô18O and ô2H relationship for FBG (ô2H = 5.79 ± 0.18 x ô18O - 9.63 ± 1.44, R2 = 0.88, p < 0.001) is similar to that obtained previously for groundwater of the Hamersley Basin (ô2H = 6.13 ± 0.16 x ô18O

— 6.47 ± 1.28; Dogramaci et al., 2012), but the intercept was lower (Fig. 5). In contrast to FBG, the ô2H and ô18O values for the saline-brine groundwater (SBG; TDS > 10 g L-1) under the Fortescue Marsh are characterised by more positive ô-values; ô18O varies from +2.5 to

- 7.2% whilst ô2H varies from -10.3 to - 51.6%. Based on the regression models for ô2H and ô18O, the relationship for the SBG (ô2H = 3.51 ± 0.12 x ô18O - 26.56 ± 0.47; R2 = 0.94 p < 0.001) is significantly different to the FBG ( Fig. 4); the linear regression line describing the relationship for the SBG also falls to the far right of LMWL and Local Evaporation Line (LEL ô2H = 5.16 x ô18O -14.37; Fig. 5).

The stable isotope composition and major ion concentrations in the SBG vary both spatially and with depth. In general, the Cl concentration in shallow groundwater under the marsh is relatively low and ranges from 33 g L-1 to 56 g L-1 compared to deeper groundwater (which has a Cl concentration up to 96 g L-1). The increase in Cl concentration with depth in the SBG also corresponds to a gradual enrichment in 18O. Seasonal variation in major ions and in the stable isotope composition varies and depends on the bore depth and location. Whilst the ô18O of the SBG varies in a very narrow range, usually close to analytical resolution (±0.1%), the Cl concentration varies significantly between different sampling periods and can be up to ± 1 g L-1 in saline ground-water under the marsh.

4. Discussion

4.1. Spatial variation in Cl, 81sO and elemental ratios

Elevation, hydraulic gradient and location in the landscape, as well as groundwater flow path, determine groundwater salinity and its stable isotope composition across the Fortescue Marsh. The distribution of ô18O values, from a minimum of - 9.7% for the FBG to a maximum of + 2.5% for SBG, follows distinct topographical features along the valley (Fig. 6). Permanent and semi-permanent water bodies such as Coondiner or 14 Miles Pools (Fig. 1A) do not have a significant impact on groundwater chemistry of the marsh. Groundwater under the upper Fortescue River, small tributaries and the alluvial fans adjacent to the marsh is always fresh and has isotope signatures similar to that of large volume rainfall events. In contrast, the groundwater under floodplains and the marsh itself is saline and more enriched in 18O and 2H, reflecting partial evaporation and the prevalence of recharge from vertical infiltration. The chemical composition of groundwater in this

ÎOOOO.O

13 O m cq

1000.0

ca Z 100.0

m CQ 10.0

CO -Z.

100,000

10,000

A t y

3 m » ✓ /

// s <

A s* "lofi 0o<Sc 0 •

0.1 1.0 10.0 100.0 1000.0 10000.0 Cl (mmol/L)

10 0.1

B c )0

o Cb © O 'a ? ho

D_ JB9

of 3 ^

^—Sf îa water

60 50 40

D o

o0( §

-CK > p < b

0.030 0.025 0.020

y 0.015 (/1

0.010 0.005

1.0 10.0 100.0 1000.0 10000.0 Cl (mmol/L)

10.0 100.0 1000.0 10000.0 Cl (mmol/L)

o °oc ° Ç gtio o 10 3)

0.1 1.0 10.0 1 00.0 1000.0 10000.0 Cl (mmol/L)

<8 o D O 9

Of ) ^ 8

0.1 1.0 10.0 100.0 1000.0 10000.0 Cl (mmol/L)

0.000 „

° 8° o

_ ° Q>f *B

"-15 -10 -5 0 5

ô18O[%0, VSMOW]

Fig. 4. Selected dissolved ion distribution in groundwater beneath the Fortescue Marsh. The Br concentration for the saline and brine groundwater (SBG) falls below the evaporation line suggesting the removal of Br from groundwater (A, B). Increased Mg, SO4 and Sr relative to Ca ions with increasing Cl ions suggest the removal of Ca by carbonate precipitation (C, D, E). The relatively high Sr/Ca coupled with high 618O values suggests evaporation and carbonate precipitation as two major factors controlling groundwater geochemistry (F).

part of the catchment is driven by evaporation and is therefore frequently saturated in carbonate minerals and gypsum.

Ophthalmia Dam intercepts the headwaters of the Fortescue River (Fig. 1A, B). The influence of seepage from the dam on groundwater in the alluvial aquifer underlying the Fortescue River down gradient is also insignificant. The high evaporation from the large open surface of Ophthalmia Dam (12.6 km2) leads to seasonal variation in surface water chemistry and stable isotope composition (Fig. 7). However, there is no detectable change in the S18O value of alluvium groundwater, even within a few kilometres downstream of the

dam (Fig. 6). We measured ô18O values of +12.0%» in dam water in Oct 2010 following three years of drought (Fig. 7). However, the ô18O value of groundwater downstream of the dam never exceeded — 6.8%, even after prolonged drought (Fig. 6). The stable isotope composition of the alluvium aquifer only reflects partial evaporation of large rainfall events (ô18O —12% to —11%; Dogramaci et al., 2012). Thus, the groundwater in the alluvial aquifer below Ophthalmia Dam is likely to consist of local rainfall or infiltration of runoff, which also suggests the dominance of vertical flow over horizontal groundwater throughflow.

Fig. 5. Local Ground Water Lines (LGWL) for the FBG (full symbols) and for the SBG (open symbols) calculated from four sampling seasons May 2010-Nov 2011. Local Meteoric Water Line (LMWL) and Local Evaporation Line (LEL) are as calculated by Dogramaci etal. (2012).

In contrast, the S18O values of groundwater of the alluvial and deep fractured aquifers under the Fortescue Marsh ranged from — 1.6%» to + 2.5%o. The shallow saline groundwater under the marsh extends > 80 km along the Fortescue River valley downstream from 14 Miles Pool to the Goodiadarrie Hills (Fig. 1A, B). The largest pool on the northeast part of the marsh, 14 Mile Pool, is the expression of the alluvium aquifer water table in the deeper sections of the Fortescue River channel. The salinity and stable isotope composition of 14 Mile Pool reflects progressive evaporation, as deciphered from transient sampling at the end of extended drought (Oct 2010: S18O 16.3%, S2H 72.7%, Cl 2.92 g L—1) and wet season (May 2011: S18O — 2.3%, S2H — 28.8%, Cl 0.04 g L—1). The stable isotope composition of surface water at 14 Miles Pool also falls on the LEL, providing further evidence that the origin of the pool water is rainfall and perhaps highly evaporated fresh water infiltrating down gradient in the alluvium aquifer along the Fortescue River. The observed changes in the stable isotope composition suggest that the pool water is not significantly diluted nor flushed by groundwater, as it is in other much smaller through-flow pools in upper parts of the secondary tributaries (Fellman et al., 2011). Therefore, 14 Miles Pool acts as both recharge and discharge zones depending on the underlying groundwater gradient. However, the low, hydraulic gradient (0.0001) suggests that the seepage volume from surrounding groundwater is rather limited. This results in progressive increase in Cl concentration and increase of S18O value by evaporation during long dry spells (Fig. 1B). Similarly, groundwater downstream from the Fortescue Marsh (west of the Goodiadarrie Hills) is relatively fresh (Cl < 1 g L—1) with S18O values of between — 6.8 and — 8.7%, which is close to those observed for groundwater across the Hamersley Basin ( — 8.02 ± 0.83%; Dogramaci et al., 2012). The outflow of surface water from the marsh at the western end is restricted by the elevated Goodiadarrie Hills (Fig. 1B). However, groundwater on both sides of the hills has similar stable isotope composition and salinity (e.g., FOR054 and FOR069, Fig. 6a), suggesting limited influence of the topographic high on the groundwater flow and salinity or the dominance of vertical flow (recharge) over horizontal groundwater throughflow.

The groundwater under the alluvial fans of Coondiner, Weeli Wolli and Sandy Creeks flowing from the Hamersley and Chichester Ranges

down to the marsh is also fresh (Cl <0.2 g L—1), with negative S18O values — 9.5% that are characteristic of relatively high rainfall >20 mm events (Fig. 6; Dogramaci et al., 2012). This finding suggests that the salinity and S18O values increase down gradient from the valley flanks and that both these parameters are related to topographic lows in the valley. In steep areas on both sides of the Fortescue Valley, runoff is rapid with relatively low evaporation losses that result in relatively low Cl concentrations. In contrast, runoff from lower slopes is comparatively slow with relatively higher evaporative losses over distance, which results in more saline recharge. The progressive increase in groundwater Cl from the surrounding ranges towards the marsh is shown in Figs. 6 and 8. The Ca, Mg, HCO3 and SO4 ions dominate the chemical composition of fresh water at the margins of the valley. The major ion/Cl ratio except for Na decreases from the surrounding hills towards the middle of the Fortescue Marsh and ultimately evolving to the saline and brine (SBG) groundwater under the marsh.

The two main initial conditions required and observed for the evolution and the formation of groundwater high in both Na and Cl are a Ca/ HCO3 mole ratio >0.5 and a SO4/Ca mole ratio »1 (Fig. 4; Herczeg and Lyons, 1991). The progressive precipitation of calcite and gypsum allows the concentration of SO4 to increase whilst HCO3 diminishes (Fig. 4). The relative decrease in the concentration of HCO3 in the alluvium groundwater at a Cl concentration of ~10 mmol L—1 suggests precipitation of carbonate (Fig. 4C; Hardie and Eugster, 1970). Because the Ca/HCO3 ratio is greater than one and the stoichiometry of calcite precipitation requires the removal of two mole of HCO3 for each mole of Ca, the species originally present in the highest concentration, in this case Ca, continues to build up in solution whilst HCO3 diminishes relative Ca. However, the precipitation of calcite alone does not explain the gradual increase in Sr/Ca and Mg/Ca ratios observed in the SBG groundwater (Fig. 4E, F). The only process that can result in the progressive increase of these ratios is continuous precipitation and redissolution of carbonate minerals and gypsum due to wetting and drying processes (Drever and Smith, 1978).

42. Seasonal variation in Cl, S18O and elemental ratios of groundwater and surface water

Whilst S18O values and Cl concentrations of groundwater varied across the valley, there was little seasonal variation. The groundwater in bores intercepting the alluvium aquifer had remarkably consistent S18O values, confirming negligible evaporation from the saturated zone even after prolonged drought. The lack of seasonal change in S18O is particularly apparent in deep bores and the shallow bores that are located on the alluvial fans of both the Hamersley and Chichester Ranges. As might be expected, the highest seasonal variation in S18O was observed in shallow bores (at ~20 m depth) on the Fortescue Marsh (Fig. 6). In May 2011, after a relatively wet season, stable isotope signatures of alluvial groundwater under the marsh were slightly more negative owing to rapid infiltration of rainwater from larger rainfalls with significantly negative S18O values. Addition of fresh rainwater diluted more saline groundwater resulting in lower Cl concentrations after the wet season in most of the alluvium bores surrounding the Fortescue Marsh. However, groundwater Cl concentration slightly increased at 11 bores located in the lowest elevation section of the marsh. This observation can be explained by dissolution of salt that was left on the surface and in the unsaturated zone after evaporation of water from previous flooding events. The limited unsaturated storage under the Fortescue Marsh (maximum water table depth is 3 m) and the progressive evaporation in ponded areas tend to produce highly localised patches of relatively saline water.

4.3. Evaporation versus mixing deduced from Cl and S18O composition

The stable isotope composition and hydrochemistry of collected samples confirms that both water and salt in groundwater from the

Fig. 6. Transects along (A-A') and across (B-B', C-C') the Fortescue Valley showing a gradual change of isotope composition and salinity (see Fig. 1 for transect locations and relation to topography).

Fig. 7. Seasonal changes in Cl concentration and 618O of surface water of the Ophthalmia Dam.

Fortescue Marsh originated from rainfall (Fig. 4A). Based on the isotopic and hydrochemical data, groundwater recharge is very rapid (Fellman etal., 2011; Dogramaci etal., 2012) and evaporative losses are relatively low (<20%). However, the shallow brackish waters have a characteristically negative S18O signature that is similar to the FBG, coupled with a high concentration of Cl ions. These two contrasting attributes cannot be attained by evaporation of rainwater alone and must be the result of additional processes such as the dissolution of Cl salts by flood or rainwater infiltrating to the alluvial aquifer. The deep SBG on the other hand, is characterised by a highly enriched 18O and 2H stable isotope composition that also cannot be explained by evaporation of modern rainfall under current climate conditions.

Using the observed enrichment of up to — 2% of S18O in the FBG (mean S18O — 9.0 to — 7.0%) compared to high volume rainfall events (mean S18O —12% to —11%), the percentage of evaporative loss prior to recharge can be calculated. Assuming non-steady-state conditions for evaporation, a mean annual air temperature of 24 °C and relative humidity of 31% (Newman weather station; www.bom.gov.au), the evaporate loss calculated from Gordon-Craig model ranges from 13% to 20%. The observed small evaporative rate compared to the potential pan

Fig. 8. Piper plot illustrating the chemistry of the groundwater in the Fortescue Marsh area (sampling in May 2011). The red point represents typical chemical composition of seawater, and the green point represents mean ion concentrations for rainfall for the Hamersley Basin as reported by Dogramaci et al. (2012).

evaporation (~3000 mm yr-1, Wittenoom station) can only occur due to rapid recharge. Moreover, the slope of regression line of S2H and S18O relationship for FBG is only slightly higher than that of the general LEL for the Hamersley Basin (Fig. 5) and identical to the LEL for Coondiner Creek catchment, which is up gradient of the Fortescue Valley (LEL, S2H = 5.1 S18O -19.7, Fellman et al., 2011).

In contrast, the slope and intercept of the regression line for the S2H-S18O relationship for the SBG are lower (S2H = 3.51 ± 0.12 x S18O - 26.56 ± 0.47) than that of both the FBG and the LEL, and intersect the LMWL at more positive S-values (S2H - 8.9% and S18O — 57.8%). The stable isotope composition of source water prior to evaporation determines the intercept, but climate conditions influence the slope of evaporation line. In general, higher relative humidity (RH) and higher air temperature (Tair) correspond to a relatively lower slope (Fig. 5). We modelled hypothetical regression lines to test whether evaporation of current rainfall under various climatic conditions can produce a slope similar to the observed regression line for SBG. The two most likely isotope compositions of the source water could be: i) high volume rain events >20 mm with S18O = -11.6% and S2H = - 76.9%, or ii) FBG with S2H - 8.9% and S18O - 57.8%. The simulation of climate encompassed the temperature range 0 and 100 °C and relative humidity 0 to 99.9%. The results suggest that even under such extreme climatic conditions it is not possible to produce a slope <4.2 from rainfall with a modern isotope composition. This suggests that the calculated regression line for SBG is not a LEL, but reflects mixing between two end members representing deep brine SBG and fresh rainfall. The stable isotope composition of the deep SBG also suggests a complex evolution, which cannot be explained by evaporation under current conditions from modern rainfall. The dynamic fraction-ation from brine is very different to that in fresh and brackish waters (Horita, 1989; Dutkiewicz et al., 2000).

The observed salinity and S18O values may perhaps be explained by progressive evaporation from a highly saline lake that existed in

the past, although further investigations are needed to confirm this hypothesis.

4.4. Groundwater mixing calculation

A concomitant increase in Cl concentration and S18O usually reflects the progress of evaporation. However, the rate of water loss calculated from the Cl and S18O mass balance is significantly different. The 1% change in S18O reflects 1-3% of volume loss, assuming evaporation occurs at temperature between 15 °C and 30 °C and humidity 20-80% for the range of S18O, as observed for groundwater in the Fortescue Valley. The loss of 1 -3% of water volume would only result in Cl concentration increase of 1-3 g L-1 (assuming initial Cl concentration of 48 g L-1), which is much lower than the observed change in depth profiles (9 to 21 Cl g L-1 per 1%) (Figs. 9,10). This suggests evaporation alone is not the process responsible for the variation in salinity increase with depth. Consequently, mixing between a relatively fresher ground-water at the top of the water column with deeper and higher salinity groundwater is the only explanation for the observed salinity depth profile under the Fortescue Marsh (Simmons and Narayan, 1997). The mixing process of two end members is corroborated by the difference in chemical and isotopic characteristics of the two groups of groundwater, which is also explained by the relationship between their respective S18O and d-excess (Fig. 11). Two well-separated groups of samples are identified, each representing a group of aquifers with distinct salinity ranges and S2H-S18O regression line. The correlation between S18O and d-excess for both groups is statistically significant (p < 0.001); however, the relationship is stronger for saline and brine waters (R2 = 0.96) than for fresh and brackish waters (R2 = 0.52). As expected, the fresh and brackish samples exhibit d-excess values that are characteristic of modern precipitation (d = +13.9 to +0.4%). In contrast, d-excess for the saline and brine samples ranges from + 6.8 to - 33.2%, reflecting the mixing of waters with different isotopic

5180 [%o, VSMOW]

Fig. 9. Correlation between depth 618O and Cl concentration in multi-piezometer bores at the northern fringe of the Fortescue Marsh, p-value <0.001, data for May 2011.

compositions. As identified above, the brine end-member from deep bores with the highest observed salinity also had the most negative d-excess values ( — 30%, Fig. 11).

The large variation in Cl concentration coupled with the narrow range of ô2Handô18O values of the FBG (R2 = 0.01) suggests that evaporation prior to recharge is small and other processes such as mixing are also affecting groundwater chemistry. If there was evaporation we would expect an increase in Cl concentration to be coupled with enrichment in 18O. In contrast, ô18O and Cl values of the SBG are strongly correlated (R2 = 0.88, p < 0.001). Cl concentrations in rainwater were 0.0014 g L-1, which would require evaporation of 99.999% of water to concentrate Cl to levels observed in deep SBG (96 g L-1). Such extensive evaporation would result in an increase in ô18O to the maximum enrichment value of +51% (Fellman et al., 2011). Therefore, the only possible process to explain our observed ô18O values of + 2.5 to — 7.2% coupled with an extremely high Cl concentration in the SBG is mixing of groundwater with a relatively negative ô18O value, such as

FBG or rainwater originating from large volume events, or alternatively dissolution of dry salt. The regression line for the ô2H-ô18O relationship for SBG can be explained as a mixing line between two end members (Fig. 5). The isotope and Cl mixing model were used to test if the measured range of Cl and isotope composition for the brackish and saline groundwater can be obtained using defined end members (Eqs. (1) and (2)). The stable isotope composition of one end member is defined by the value of the intersection between regression line for FBG and SBG (Ô18O — 7.4%) with Cl concentration of ~0.7 g L—1 (median for FBG). The second end member is the water sample with the highest measured Cl concentration of 96 g L—1 and ô18O+2.5%», (FOR104 - 100 mdeep). In order to confirm validity of this approach, we calculated the contribution of the two end members for an arbitrary sample (FOR061, Figs. 9 & 10) with Cl concentration of 40 g L—1 and Ô18O —3.3%. The calculated results from both models (based on Cl and ô18O) are identical (±1%) suggesting the contribution of 59% of fresh water to brine groundwater. Accordingly, the brine groundwater is evolved in two steps: i) evaporation of ~ 17% of rainwater during retention on ground surface prior to recharge (ô18O change from —11.6% to — 7.4%) and ii) mixing of this partially evaporated rainwater in differing proportions with the existing saline groundwater (ô18O +2.4% as in 100 m deep bores). These results suggest at least some degree of hydraulic conductivity between deep and shallow groundwater on the marsh, despite several locally impermeable horizons of clay and calcrete.

FFBG = (SmppOR061 EndSBG )/(EndFBG —EndSBG )

fsbg — 1 ffï

Fig. 10. Distribution of salinity and 618O in multi-piezometer bores at the northern fringe of the Fortescue Marsh.

where FFBG and FSBG are fractions of fresh and brine end members, respectively. The SmpFOR061, EndSBG and EndFBG denote S18O and Cl concentration (g L—1) for sample, fresh and brine end-members, respectively.

4.5. Major ion evolution

The Cl/Br ratio for groundwater of the Fortescue Valley shows a distinct trend with increasing salinity. The origin of the Cl and Br ions in rainfall across the Pilbara is sea spray suspended from the oceans along the coast of north and northwest Australia (Dogramaci et al., 2012). The bulk of Cl and Br ions is then carried inland by winds and their concentration decreases along the wind path with increasing

S1tiO [%0 VSMOW] 8180 [%0 VSMOW]

Fig. 11. Two distinct types of groundwater: fresh and brackish (closed symbols) and saline and brine (open symbols).

distance from the ocean, resulting in higher Cl and Br deposition rates near the coast than inland but similar Cl/Br ratios to seawater (~288). Consequently, if evaporation alone were the dominant process resulting in higher Cl and Br concentrations, the Cl/Br ratios would have remained similar to the original rainfall (288). Because of the high solubility of Br and the absence of Cl bearing minerals in the aquifers underlying the Fortescue Valley, the most likely process that explains the increasing Cl/Br ratio is the removal of Br through cyclic drying and wetting due to extreme drought and subsequent wet climatic conditions (Braitsch, 1971; Drever and Smith, 1978). The evaporation of rainfall in numerous surface water pools along the creek line of the surrounding ridges to the east and west of the Fortescue Valley to dryness results in sequential precipitation of calcite, gypsum, magnesium salts such as starkeyite and eventually halite. During precipitation of halite, the Br ions are excluded from halite structure due to its low distribution coefficient (Herrmann, 1980). In contrast, after rainfall in the following wet season these minerals dissolve in a reverse order (i.e. halite back to calcite) resulting in changes in ion/Cl ratios. The Cl/Br ratio increases in dissolving water because of a relatively higher Cl/Br ratio in halite. The continuous drying and wetting processes are supported by the progressive increase in Sr/Ca and Mg/Ca ratios (Fig. 4). The dissolution of primary carbonate minerals such as high Mg-calcite and dolomite results in release of Mg and Sr as well as Ca and HCO3 to the liquid phase. Re-precipitation of calcite from evaporated water results in the exclusion of Mg and Sr ions into calcite structure resulting in continuous increase in both Mg/Ca and Sr/Ca ratios (Fig. 4E & F). This is particularly evident for the deeper brine groundwater, where chemical evolution of the ions occurs due to the dissolution of salt efflorescence precipitated from the previous season with ponded floodwater and the gradual recharge to the underlying aquifer under the Fortescue Marsh.

4.6. Salt inventory under the Fortescue Marsh

The load of salt observed in the Fortescue Marsh was mainly delivered via rainfall over several tens to hundreds of thousand years. We estimated the time required to accumulate the observed Cl load in the Fortescue Marsh based on the Cl concentration in rainfall and ground-water, and a salt balance model. We based our model on the field observations and measurements: 1) flooding in January 2012 resulted in inundation of around 512 km2; 2) the average Cl concentration in flood-water from ten sites in February 2012 was 50 mg L-1;and3) measured Cl concentrations in groundwater under the marsh ranged from 4 to 96 g L-1. We thus made the following assumptions: 1) the total area of the marsh that is inundated during cyclonic events ranges from 500 km2 to 1000 km2; 2) the average depth of water in the flooded valley is 1 m, resulting in a total surface water volume of 0.5 to 1.0 km3 (from DEM model); 3) the average frequency of cyclones is 4.2 years

(based on Wittenoom, Newman and Marillana weather stations); 4) the volume of water in the SBG aquifer is 10 km3 (1000 km2 x thickness of 100 m, porosity 10%); 5) Cl in rainwater is 0.0014 g L-1 (as measured); 6) Cl in flood water varies between 0.0014 and 0.050 g L-1; and 7) mean annual precipitation 350 mm. In our model, we estimated the time that would be required to attain half of the observed maximal observed concentration of Cl in the SBG (48 g L-1).

We estimated a range of40,000 years to 700,000 years would be required to accumulate sufficient Cl in the SBG, based on the maximum and minimum possible ranges of flooded area and Cl concentrations. This time frame is within the range calculated for salt inventories from the Goldfields region in southern Western Australia (Johnson et al., 1999). The mechanism of salt accumulation in the Fortescue Marsh can be divided into three stages. First, regional rainfall or cyclone events deliver substantial amounts of floodwater to the marsh and dissolve salt accumulated on the dry surface (Fig. 12a). The floodwater partially evaporates (up to 20%) and recharges the underlying aquifer raising the water level and filling depressions. Second, the ponded water progressively evaporates but recharge of the underlying aquifer is minimal due to its saturation (Fig. 13b). Third, the ponded water in depressions dries out, followed by precipitation of efflorescent salt on the surface and in sediments, which will be dissolved in subsequent flooding (Fig. 12c). The precipitation and dissolution of salt on seasonal or decadal time scales result in the development of a shallow brine groundwater deficient in all major ions but Cl and Na, which is then entrained in the more recent inflow water, thereby producing brine with a relatively depleted isotopic signatures (Fig. 12d). Groundwater flow is thus dominated by vertical flow rather than horizontal throughflow across the valley.

5. Conclusions

Whilst several hydrological riddles in this area remain unsolved, our findings reveal new insights in to the functioning of this large and unusual inland wetland. The hydrology of the Fortescue Marsh is driven by infrequent large cyclonic events. Recharge occurs during these events from the alluvial plain to the underlying aquifers. The highly saline nature of groundwater coupled with a low hydraulic gradient suggests that horizontal groundwater flows are very slow (< 5.0 m/yr). The chemical and isotopic signatures of groundwater under the marsh suggest the dominance of vertical flow processes.

Our set of combined chemical and stable isotope analyses of surface water and groundwater allowed decoupling of water and salt evolution in groundwater over time. We could also characterise groundwater flow and hydraulic connections, and identify interactions between aquifers using this approach. With the exception of deep brine groundwater under the Fortescue Marsh, most of the groundwater reflects modern

PDE008764 7v3

Fig. 12. Conceptual diagram of saline water evolution on the Fortescue Marsh (refer to discussion in Section 4.6).

recharge. However, the salt inventory reflects accumulation of salt over millennia. As consequence, frequently even brackish and saline waters have the stable isotope signature of modern precipitation, suggesting modern recharge and dissolution of older salt recycled in repetitive wetting-drying cycles. The Cl mass balance calculations suggest between 40,000 and 700,000 years would be required to accumulate the observed salt load in the Fortescue Marsh. It is therefore not surprising that the isotopic composition of groundwater in the deep brine aquifer under the Fortescue Marsh reflects recharge under a different climatic regime and that waters are a mixture of modern recharge and older, deep groundwater.

Acknowledgements

This research was supported by the Australian Research Council (ARC) in partnership with Rio Tinto (LP120100310) and an ARC Future Fellowship awarded to G. Skrzypek. We thank several members of our teams at RTIO and UWA whose work made this study possible, especially Glenn Kirkpatrick and Paul Hedley for assistance in the field, Timothy Brunner for setup of the database and Maree Swebbs for help with the map and the cross-section preparation from RTIO, Douglas Ford and Ela Skrzypek at the West Australian Biogeochemistry Centre at UWA for help with sample analyses and Alexandra Rouillard for assistance during field work. We also acknowledge the kind support of Sue and Lee Bickell (Marillana Station); Victor and Larissa Glison (Mulga Station) and Barry and Bella Grett (Ethel Creek Station). We thank Fortescue Metal Group Ltd. for assistance in groundwater sampling. We are also grateful to Dr Phil Commander (Department of Water, Perth) for providing useful comments on the early version of the

manuscript and to Jordin A. Barclay (Fortescue Metal Group Ltd.) for a critical reading of the final version of the manuscript.

References

Acheampong, S.Y., Hess, J.W., 2000. Origin of the shallow groundwater system in the southern Voltaian Sedimentary Basin of Ghana: an isotopic approach. J. Hydrol. 233, 37-53.

Bowler, J.M., 1986. Spatial variability and hydrologic evolution of Australian lake basins: analogue for Pleistocene hydrologic change and evaporite formation. Palaeogeogr. Palaeoclimatol. Palaeoecol. 54, 21 -41.

Braitsch, O., 1971. Salt Deposits, Their Origin and Composition. Springer-Verlag, Berlin, New York.

Cartwright, I., Weaver, T.R., Fifield, L.K., 2006. Cl/Br ratios and environmental isotopes as indicators of recharge variability and groundwater flow: an example from the southeast Murray Basin, Australia. Chem. Geol. 231, 38-56.

Cartwright, I., Weaver, T., Cendon, D.I., Swane, I., 2010. Environmental isotopes as indicators of inter-aquifer mixing, Wimmera region, Murray Basin, Southeast Australia. Chem. Geol. 277, 214-226.

Chebotarev, I.I., 1955. Metamorphism of natural waters in the crust of weathering. Geochim. Cosmochim. Acta 8, 22-28.

Clout, J.M.F., 2011. The Roy Hill Project. AisIMM 3,52-65.

Coplen, T.B., 1996. New guidelines for reporting stable hydrogen, carbon, and oxygen isotope-ratio data. Geochim. Cosmochim. Acta 60, 3359-3360.

Costelloe, J.F., Irvine, E.C., Western, A.W., Tyler, M., 2012. Identifying fluvial recharge and artesian upwards leakage contributions to arid zone shallow, unconfined groundwa-ter. Chem. Geol. 326-327,189-200.

Cullen, L.E., Grierson, P.F., 2007. A stable oxygen, but not carbon, isotope chronology of Callitris columellaris reflects recent climate change in north-western Australia. Clim. Chang. 85, 213-229.

Dansgaard, W., 1964. Stable isotopes in precipitation. Tellus 16,436-468.

de Dekker, P., 1988. Biological and sedimentary facies of Australian salt lakes. Palaeogeogr. Palaeoclimatol. Palaeoecol. 62,237-270.

Dogramaci, S., Skrzypek G., Dodson, W., Grierson, P.F., 2012. Stable isotope and hydrochemical evolution of groundwater in the semi-arid Hamersley Basin of subtropical northwest Australia. J. Hydrol. 475, 281-293.

Drever, J.I., Smith, C.L., 1978. Cyclic wetting and drying of the soil zone as an influence on the chemistry of groundwater in arid terrains. Am. J. Sci. 278,1448-1454.

Duffy, C., Al-Hassan, S., 1988. Groundwater circulation in a closed desert basin: topographic scaling and climatic forcing. Water Resour. Res. 24,1675-1688.

Dutkiewicz, A., Herczeg, A.L., Dighton, J.C., 2000. Past changes to isotopic and solute balances in a continental playa: clues from stable isotopes of lacustrine carbonates. Chem. Geol. 165, 309-329.

Environmental Protection Authority, 2013. Environmental and water assessments relating to mining and mining-related activities in the Fortescue Marsh management area. Report and Recommendations of the Environmental Protection Authority No 1484.The Government of Western Australia, Perth (51 pp.).

Fellman,J., Dogramaci, S., Skrzypek, G., Dodson, W., Grierson, P., 2011. Hydrologic control of dissolved organic matter biogeochemistry in pools of a subtropical dryland river, Western Australia. Water Resour. Res. 47, W06501.

Freeze, R.A., Cherry, JA., 1979. Groundwater. Prentice-Hall, Englewood Cliffs (604 pp.).

Gat, J.R., 1981. Groundwater stable isotope hydrology. Deuterium and Oxygen-18 in the Water Cycle, IAEA Technical Report Series 210,223-240.

Gat, J.R., Levy, Y., 1978. Isotope Hydrology of Inland Sabkhas in the Bardawil Area, Sinai. Limnol. Oceanogr. 23, 841-850.

Gibson, J.J., Reid, R., 2010. Stable isotope fingerprint of open-water evaporation losses and effective drainage area fluctuations in a subarctic shield watershed. J. Hydrol. 381, 142-150.

Hardie, L.A., Eugster, H.P., 1970. The evolution of closed-basin brines. Mineralogical Society of America Special Paper, 3, pp. 273-290.

Harrington, G.A., Cook, P.G., Herczeg, A.L., 2002. Spatial and temporal variability of groundwater recharge in central Australia: a tracer approach. Ground Water 40,518-528.

Harrington, N.M., Herczeg, A.L., Le Gal La Salle, C., 2008. Hydrological and geochemical processes controlling variations in Na+-Mg2+-Cl—-SO4 groundwater brines, southeastern Australia. Chem. Geol. 251, 8-19.

Herczeg, A.L., Edmunds, W., 1999. Inorganic ions as tracers. In: Cook, P., Herczeg, A. (Eds.), Environmental Tracers in Subsurface Hydrology. Kluwer Academic Press, Boston, pp. 31-78.

Herczeg, A.L., Lyons, W.B., 1991. A chemical model for the evolution of Australian sodium chloride lake brines. Palaeogeogr. Palaeoclimatol. Palaeoecol. 84,43-53.

Herczeg, A.L., Barnes, C.J., Macumber, P.G., Olley, J.M., 1992. A stable isotope investigation of groundwater-surface water interactions at Lake Tyrrell, Victoria, Australia. Chem. Geol. 96,19-32.

Herczeg, A.L., Dogramaci, S.S., Leaney, F.W.J., 2001. Origin of dissolved salts in a large, semi-arid groundwater system: Murray Basin, Australia. Mar. Freshw. Res. 52,41 -52.

Herrmann, A.G., 1980. Bromide distribution between halite and nac1-saturated seawater. Chem. Geol. 28,171-177.

Hickman, A.H., 1983. Geology of the Pilbara Block and its environs. Bull. Geol. Surv. West. Aust. 127,1-268.

Horita, J., 1989. Analytical aspects of stable isotopes in brines. Chem. Geol. 79,107-112.

Johnson, S.J., Commander, D.P., O'Boy, CA, 1999. Groundwater resources of the Northern Goldfields, Western Australia. Water and River Commission Hydrogeological Record Series, Report HG 2.Government of Western Australia, Perth.

Jolly, I.D., McEwan, K.L., Holland, K.L., 2008. A review of groundwater-surface water interactions in arid/semi-arid wetlands and the consequences of salinity for wetland ecology. Ecohydrology 1,43-58.

Lamontagne, S., Leaney, F.W., Herczeg, A.L., 2005. Groundwater-surface water interactions in a large semi-arid floodplain: implications for salinity management. Hydrol. Process. 19, 3063-3080.

Paul, D., Skrzypek G., Forizs, I., 2007. Normalization of measured stable isotope composition to isotope reference scale — a review. Rapid Commun. Mass Spectrom. 21,3006-3014.

Payne, A.L., Mitchell, A.A., 1999. An assessment of the impact of Ophthalmia Dam on the floodplains of the Fortescue River on Ethel Creek and Roy Hill stations. Resource Management Technical Report No.124.0729-3135Government of Western Australia.

Rio Tinto, 2013. Koodaideri Iron Ore Mine and Infrastructure Project, Public Environmental Review, July 2013, Assessment No. 1933, Rio Tinto Iron Ore, Perth. (http://www. riotintoironore.com/ENG/media/337.asp Accessed 25/07/2013).

Rosen, M.R., Miser, D.E., Starcher, M.A., Warren, J.K., 1994. Formation of dolomite in the Coorong region, South Australia. Geochim. Cosmochim. Acta 53, 661-669.

Simmons, C.T., Narayan, K.A., 1997. Mixed convection processes below a saline disposal basin. J. Hydrol. 194, 263-285.

Skrzypek G., Sadler, R., Paul, D., 2010. Error propagation in normalization of stable isotope data: a Monte Carlo analysis. Rapid Commun. Mass Spectrom. 24, 2697-2705.

Toth, J., 1963. A theoretical analysis of groundwater flow on small drainage basin. J. Geophys. Res. 68,4795-4812.

Tweed, S., Leblanc, M., Cartwright, I., Favreau, G., Leduc, C., 2011. Arid zone groundwater recharge and salinisation processes; an example from the Lake Eyre Basin, Australia. J. Hydrol. 408, 257-275.

Yechiechi, Y., Wood, W.W., 2002. Hydrogeologic processes in saline systems: playas, sabkhas and saline lakes. Earth Sci. Rev. 58, 343-365.