The use of groundwater geochemistry to explore for Au mineralization has faced some challenges. For this type of mineralization, the best pathfinder element is Au itself. However, the analytical technique for precious metals including Au and Ag has a relatively low sensitivity, in particular for brackish to saline waters. Medium- to high-salinity waters need to be diluted before analyses, leading to even lower Au concentrations and more samples below detection thresholds. Here, we present methods where a pre-concentration step is added by using activated carbon sachets that adsorb and concentrate the precious metals of interest from the water sample. This approach was used during a regional survey in the Northern Yilgarn Craton, as well as on several smaller case studies, to determine the potential of groundwater geochemistry in identifying Au mineralization at depth.

The Northern Yilgarn Craton is a particularly well-suited area for groundwater hydrogeochemistry surveys as it has an abundance of sampling sites at windmills and bores. Groundwater analysis of the Northern Yilgarn Craton was performed on >5000 samples following a semi-regular grid of 4–10 km spacing. This study covers an area of c. 315 000 km2 including numerous minesites and buried ore deposits.

The regional study revealed many widespread groundwater Au anomalies when samples were collected <5 km of a Au deposit. The primary pathfinder elements are As and Ag, and the oxy-anions Mo and W are shown to be closely related to Au mineralization and are effective pathfinders. Each of the pathfinder elements shows potential but cannot be used as a single element to reveal the mines and deposits with accuracy. This is mostly related to the diversity of orebody geochemistry. However, these pathfinder elements can highlight areas of interest and have value when used in combination with other elements and within a regional context. Yet, the oxy-anions Mo and W stand out. When detected in groundwater these elements are closely related to Au mineralization and are effective pathfinders. The multi-element index AuMin can solve some of these individual elements’ limitations and be used to highlight areas of interest from the groundwater hydrogeochemistry. AuMin anomalies are more widespread than single Au anomalies but it is useful to focus the exploration effort.

The ‘background’ baseline data provided by the regional sampling are crucial when interpreting the data acquired around minesites and deposits. It revealed differences between east and west Northern Yilgarn Craton greenstone hosted samples. In the eastern area we focused on two case studies on different style deposits: orogenic Au at Agnew and secondary and orogenic Au at Garden Well. Groundwater geochemistry at Agnew returned a high concentration of many pathfinder elements (As, Ag, Co, Sb, W), which strongly reflects the pathfinder assemblages found in individual orebodies along the mineralized corridor. In contrast, groundwater sampled at Garden Well had only a few samples with detectable or significant concentrations in pathfinder elements (Ag and minor As).

We showed that Au in groundwater is a powerful technique to identify areas of Au mineralization undercover, in particular when detected in association with pathfinder elements. Individual pathfinder elements in isolation are challenging to interpret as they can be associated with other mineral systems (e.g. Ni or volcanic hosted massive sulfides deposits).

Overall, the application of groundwater geochemistry to Au exploration in the Northern Yilgarn Craton is promising, as long as the data are put into a regional geo-environmental context and used for large-scale exploration such as tenement selection rather than deposit delineation.

Supplementary material: Additional site data and geological background legends are available at

Thematic collection: This article is part of the Hydrochemistry related to exploration and environmental issues collection available at:

Since the Australian gold rush of the 1850s, the discovery and mining of gold (Au) is intertwined with Australian history, both economically and socially. This is particularly true for Western Australia, which has been the major Australian Au producer since the 1890s (Mudd 2007). The early days of Au exploration when explorers could locate mineralization at surface are long gone. The presence of a thick, exotic cover over a large portion of the Australian territory (up to 80%; Schodde 2014) is a major challenge for Au mineral exploration. As such, in recent years, the cost and difficulty of mineral exploration has increased. Alternative exploration approaches e.g. biogeochemistry (e.g. Brooks 1983), hydrogeochemistry (e.g. Gray et al. 2016a), Ultrafine+ (e.g. Noble et al. 2019), partial soil extractions and soil gases (e.g. Klusman 1993; Noble et al. 2018) are being developed to highlight new and deeper exploration targets and as such a greater emphasis is being placed on exploring through deep (>30 m) transported cover.

In Australia, there is continued and growing interest in groundwater geochemistry for mineral exploration (Wallace et al. 2018; Reid et al. 2020, 2021; Schroder et al. 2020).When groundwater is in contact with mineralized rocks, it interacts with them by leaching metals, which creates a geochemical halo and thus can reveal areas of mineralization. In Australia, groundwater geochemistry has been used to highlight U (Noble et al. 2011), volcanic hosted massive sulfides (VHMS; Gray et al. 2018) and Pb–Zn (de Caritat et al. 2005) mineralization. But this approach has also been used on a global scale revealing mineralization in North and South America andAfrica (Cameron and Leybourne 2005; Leybourne and Cameron 2006, 2010; Kidder et al. 2022; Rissmann et al. 2022).

The use of groundwater geochemistry to explore for Au mineralization has faced some challenges. For this type of mineralization, the best pathfinder element is Au itself; however, the sensitivity of the technique remains to be improved for Au and other precious metals including silver (Ag). For example, at the time of sample collection Au in groundwater is typically below inductively coupled plasma mass spectrometry (ICP-MS) detection limits (0.1 μg /l–1) and concentrations further decrease with increasing distance from mineralization – i.e. at regional scale (Gray et al. 2009).

Where brackish to saline groundwater is the norm, many studies have investigated and used novel ways to overcome the issue of salinity (Cameron 1978; de Caritat et al. 2005; Leybourne and Cameron 2008, 2010; Eppinger et al. 2012). For instance, many Australian studies have focused on improving the use of brackish to saline groundwater for mineral exploration (Mann and Deutscher 1978; Giblin and Mazzucchelli 1997; Gray 2001; Pirlo and Giblin 2004; de Caritat et al. 2005; Gray and Noble 2006; Noble et al. 2011). One of the solutions is to add a pre-concentration step by using activated carbon sachets that adsorb and concentrate the precious metals of interest (Gray et al. 2009).

In Australia, finding easy sampling access to groundwater can prove to be difficult, so to gain a good picture of an area it is important to identify the location and accessibility of groundwater. The Northern Yilgarn Craton has been identified as a particularly well-suited area for groundwater hydrogeochemistry surveys as it has an abundance of windmills and bores giving continuous and regular watering points for cattle and sheep. As cattle generally require water every 5–10 km and sheep every 3–5 km, this provides a semi-regular grid of water access. More than 5300 water bores and wells were sampled between 2007 and 2013; these provide access to spatially unbiased groundwater samples over an area that covers 700 × 450 km2, with numerous known mineral deposits including several Au deposits which had higher-density sampling carried out to determine the mineral footprint in groundwater.

Through this study, we provide data associated with unpublished case studies as well as compile the results of the regional sampling campaigns (previously published in internal reports and data releases; Gray et al. 2009, 2014, 2016a, b; Gray and Bardwell 2016). These regional data are also processed and interpreted within the broader Australia-wide hydrogeochemical context. In addition, we recalculated indices and ratio values using a novel software application (XT HydroTM) ensuring homogeneity of calculations.

In particular, we aim to evaluate the potential of dissolved Au in groundwater for identifying Au deposits and identify what may impact the detection and footprint of Au systems within the groundwater chemistry of this region. Two case studies with closer spaced sampling are also presented to illustrate the groundwater signatures of two mineralization types: primary orogenic Au mineralization (Agnew) and secondary Au mineralization (Garden Well).

The climate of the Northern Yilgarn Craton is semi-arid to arid, with hot summers and cool winters and a mean annual rainfall between 200 and 300 mm (Australian Bureau of Meteorology 2020). The landforms are generally gently undulating with low relief. The elevation across the sampling area gradually decreases toward the south from 600 to 350 m above sea level. The dominant landforms of the Northern Yilgarn Craton are sandplains, plateaus, breakaways, colluvial and alluvial plains, outcropping bedrock, salt lakes with feeder drainage channels, and sand dunes (Anand and Paine 2002). These landforms and the relatively flat surface conceals the complex underlying regolith that has been exposed to a variety of climatic conditions leading to important events of weathering, erosion and deposition (Anand and Paine 2002).

The Northern Yilgarn Craton is characterized by relatively shallow, fresh to brackish groundwater (Commander 1989; Gray et al. 2009, 2014), with pH values between 7 and 9 (Gray et al. 2014). The vegetation profile of the Northern Yilgarn Craton is open woodland dominated by Acacia aneura (mulga) across most landscape settings (Moore 2005; Meissner and Coppen 2013). Triodia species (spinifex) are common across sand plains and sand dunes. Callitris glaucophylla (white cypress pine) is present along breakaways and gullies throughout the region (McQueen et al. 2021). Soils are dominantly acid–sandy soils with red–brown hardpans (Anand and Paine 2002).

In the Northern Yilgarn, the shallow water table and groundwater flow generally reflect the surface landscape features. It is therefore relatively easy to determine the flow direction of the groundwaters that have been sampled for this work as they follow topography and drain into the palaeodrainage systems (which are also the contemporary drainage). This is a significant advantage for the interpretation of the data in terms of possible source direction of Au (or associated elements) anomalies.

The Yilgarn geology is marked by an east–west transition represented by greenstone belts with a north–south general orientation (Fig. 1). The Yilgarn Craton is subdivided into six terranes; in the east, Burtville Terrane, Kurnalpi Terrane, Kalgoorlie Terrane, which together form the Eastern Goldfield Superterrane; in the west, the Youanmi Terrane, the SW Terrane and the Narryer Terrane in the NW (Cassidy et al. 2006). Our sampling area covers five of these terranes, but is largely concentrated on the Youanmi, the Kurnalpi and Karlgoorlie Terranes. The Yilgarn Craton is of Archean age and is mostly composed of granites (up to 65% of the craton; Huston et al. 2012) as well as greenstone belts. Most of the greenstone belts are composed of ultramafic and mafic rocks – i.e. komatiite and basalt flows – and are largely located within the Kurnalpi, Kalgoorlie and Youanmi Terranes. Abundant ore deposits, dominantly Au and Ni, have been discovered in association with the greenstones, but Au is also hosted by granitic rocks and metasediments (e.g. Cassidy et al. 1998; Witt and Vanderhor 1998).

Sampling of groundwaters

The sampling of groundwater followed the protocol described in detail in Gray et al. (2009). Approximately 5330 groundwater samples were collected by CSIRO in this region (Fig. 1) as part of several campaigns which have been compiled into a consistent data set. The regional sampling campaigns were collected over a period of five years in three projects, with multiple samples assayed and re-sampled in each campaign to check consistency across seasons, batches and laboratories. These groundwaters were sampled from wells and bores used for livestock and human consumption. Minesite and exploration drill holes were also sampled where accessible. Most samples derived from water tables within 10 m of the surface, except for those concentrated around active minesites which can be affected by mine dewatering. The most effective collection method (c. 50% of samples) was direct sampling from actively pumping farm bores (i.e. windmill, solar). To avoid any possible in-train contamination or precipitation within metal pipeworks, samples were taken as close to the pumping stem as possible. The remaining sites were sampled using a flow-through bailer fitted with one-way valves (Gray et al. 2009). Bailers were used in bores that were abandoned or had access difficulties. Where possible, bailed samples were collected at least 5 m below the water table.

Deposit samples were from exploration drill holes rather than stock bores, but apart from mineralization related elements there was little significant differences between the populations’ chemistries and so they could be compared together.

Field measurements, laboratory preparation and analyses

Parameters including pH and oxidation–reduction potential (ORP), electrical conductivity and temperature were directly measured in the field using a TPS WP81 meter and corresponding TPS sensors. These sensors need regular calibration, which was performed twice weekly for all except the pH sensor. The calibration of pH was performed daily using three buffered solutions with pH of 4, 7 and 10; the ORP electrode calibration used a Ag/AgCl reference electrode and ZoBell's check solution (ZoBell 1946) which allows conversion of field ORP to Eh. A 2.76 and 58 mS cm–1 standard solution was used to calibrate the conductivity sensor. All electrodes were stored in saturated KCl (3.5 M) solution.

Sub-samples were collected for laboratory analyses of cations, anions, alkalinity and Au/ PGE analysis. An unfiltered sample was collected for alkalinity analysis by titration. Approximately 100 ml was filtered using a handpump with a 0.45 μm filter for anion (Cl, SO42, Br, F and NO3) analysis by ion chromatography at CSIRO Perth Laboratories, with a sample split sent to CSIRO Waite Laboratory in Adelaide for analysis of dissolved organic C and PO4. A subsequent 100 ml was filtered in the field and then later acidified (0.2% HNO3 added from 69% high-purity analytical grade concentrated nitric acid) for cation analysis. Major elements (Al, B, Ca, Cr, Cu, Fe, K, Li, Mg, Mn, Na, P, S, Si, Sr and Zn) were analysed by inductively coupled plasma optical emission spectroscopy at Geoscience Australia in Canberra and CSIRO in Adelaide, and trace elements (Ag, As, Ba, Cd, Ce, Co, Cr, Cu, Dy, Er, Eu, Ga, Hf, Ho, La, Lu, Mo, Nb, Nd, Ni, Pb, Pr, Rb, Sb, Sc, Sm, Sn, Sr, Ta, Th, U, V, W, Y, Yb, Zn and Zr) were analysed by ICP-MS also at Geoscience Australia and CSIRO. Estimated analytical precisions can be found in Gray et al. (2014) and are reported in Table 1.

In groundwater Au, Ag, Pt and Pd are often below detection limits of the current analytical techniques. To overcome this issue, a pre-concentration step was added to the sampling protocol (Carey et al. 2003; Buskard et al. 2020). This pre-concentration step relies on the leaching of Au, Ag, Pt and Pd from the water and their adsorption onto activated carbon (Gray et al. 2016a). Carbon sachets are currently manufactured by CSIRO Laboratories in Perth. The sachet was made by sealing three sides of nylon mesh by heating with a soldering iron. Exactly 1 g of pre-washed activated carbon grains was placed in the mesh before sealing the fourth side by heating. The sachet was placed in a 1 l bottle with 10 g NaCl to provide an approximate ionic balance between samples irrespective of initial water sample salinity.

The 1 l bottles were filled with unfiltered water, which has the benefit of saving significant time in the field and has the potential to allow capture of colloidal Au particles (Hamilton et al. 1983) which would be lost with a 0.45 μm filtering step and has been shown in the past to produce consistent and comparable results to filtered samples (Carey et al. 2003). For the pre-concentration step, the 1 l bottles containing activated carbon sachets and groundwater samples were rolled at room temperature for at least four days to ensure all potential Au ions were adsorbed by the activated carbon. After this, the volume of water was measured in a graduated cylinder and recorded. The carbon sachet was removed (nylon mesh sieve), rinsed with deionized water (which would remove any potential particles from the unfiltered water or potential Au nuggets) and air dried in clean conditions for two days. The sample was then placed in a sealed plastic bag, labelled and sent for ICP-MS analysis at the commercial laboratory Ultra Trace Pty. Ltd. (now Bureau Veritas Minerals, Canning Vale, WA). There, the carbon was isolated from the sachet and ashed at a temperature of 600°C. The ashed carbon was then dissolved in Aqua-Regia and the resulting solution analysed by ICP-MS. Extensive duplicate and blank tests indicated a more realistic detection limit/accuracy of c. 2 ng l–1 for Au, Pt and Pd and 20 ng l–1 for Ag. All the regional results are publicly available on the CSIRO data access portal (Gray and Bardwell 2016; Gray et al. 2016b) while the newly presented case study results are in the Supplementary data.

For quality control, laboratory standards, blanks and duplicate samples were used at a ratio of 1 every 20 for internal and external analyses. Quality control is particularly important since the sampling of the North Yilgarn Craton was carried out over a five year period. Duplicates between the different laboratory analyses, between methods and between instruments were also required to prevent batch effects.

Using duplicates, blanks and standards results, the errors for each element were calculated as % difference errors, half relative difference errors and 95% confidence errors on the batch (Table 1). The errors are defined as:
where 1.96 is the 95th percentile of a normal distribution with a mean value of 1, σ is the standard deviation of all assay1–assay2 values, and n is the number of duplicates.

Increased testing has been implemented in the case of the Agnew case study samples, as some of the results were orders of magnitude higher than the other data sets. In this case duplicates of all samples were assayed for Au by both ICP-MS and neutron activation analysis (INAA), from two separate carbon sachets in two separate bottles.

Data treatment – XT HydroTM

XT HydroTM is a new software solution developed by CSIRO for producing multi-element indices such as AuMin, and element ratios such as Mg:Ca. Producing these indices was historically a slow process involving a series of complicated Microsoft Excel spreadsheets (Gray et al. 2009) which obfuscated a lot of the process and required significant time verifying results to ensure human error had not occurred. The novel XT HydroTM delivers an automated, easy access to these analysis methods, produces data in line with the FAIR Guiding Principles (findable, accessible, interoperable, reusable) (Wilkinson et al. 2016), and provides useful visualizations of the resulting data for faster interpretation.

The development of this software involved extracting the formulae from the Microsoft Excel spreadsheets originally used for producing the indices with all parameters and constants traced and documented. The process was reproduced using Python code which provides a more robust and configurable method of producing the indices. The resulting output data set adheres to a standard format allowing for data to be more easily combined and the indices calculated in a consistent manner across the different data sources.

Au deposits can be revealed by using direct indicators (Au and often Ag) as well as other associated pathfinder elements (e.g. Sb, As, Bi, W, Mo). These are commonly found in association with Au and thus can play a role in revealing Au mineralization under cover. The results of individual elements are shown in Gray et al. (2016b; link: and some are discussed below. While some pathfinder elements can be measured directly in the water, as is the case for As, Sb, Mo and W, precious metals (Au, Ag, Pd and Pt) need to be pre-concentrated following the protocol described above. Here, we report the elemental composition of the groundwaters for the elements of interest as well as the AuMin index. As part of this study, we examine the results of the regional-based groundwater sampling and the anomalies associated with Au systems. All samples are integrated within the national groundwater geochemistry database. This database contains >250 000 datapoints across the Australian continent and provides a robust statistical background for the interpretation of the regional study and in particular for the anomalous samples.


In the Northern Yilgarn Craton, Au concentrations in the regional groundwater vary from less than 10 ng l–1 in ‘background’ samples to up to 200  ng l–1 in waters near mines (generally <5 km spaced, Fig. 2. A record concentration of 19 312  ng l–1 was measured in groundwaters within the close spaced sampling at the Agnew Au mine and between two of the existing mine pits (Gray et al. 2014). Overall, the greatest Au concentrations reported herein comes from areas of higher sampling density (<5 km sample spacing) over known mineralization (Fig. 3). Although these elements (Au, As, Ag, Sb, Mo, W) can be associated with other mineralization styles, Au in particular has much higher concentrations around the Au systems than the other types of mineralization.


Regional As concentrations in the Northern Yilgarn are generally below 10  ng l–1 with elevated concentrations mostly found in samples close (<5 km) to known mineralization (30–2900  ng l–1; Gray et al. 2014; Fig. 4). Similar to Au, extreme As concentrations as high as 15 100  ng l–1 were reported in close spatial relationship with deposits of the Agnew Au mine region, where sample density is greater (<1 km sample spacing). Overall, groundwater As shows a similar distribution to Au in the Northern Yilgarn (Fig. 4).


Regional Ag concentrations in the Northern Yilgarn are generally below 150  ng l–1 with elevated concentrations (600–9400  ng l–1; Gray et al. 2014) detected in samples close (<5 km) to known Au and/or VHMS mineralization. In many locations of the Northern Yilgarn, Ag is also well correlated with Au and As but this association is not systematic (Fig. 5). Overall Ag displays less spatial correlation to Au orebodies than Au or As.

Other pathfinders (Sb, Mo, W)

Other pathfinder elements (Sb, Mo, W) can be associated with some, or many, of the Au systems in the Northern Yilgarn (e.g. Anand et al. 2019). Sb and W were not analysed for all samples but where they were analysed, the results are valuable for the discussion of dispersion around Au and other mineralized systems. The Sb results (Fig. 6) do show elevated concentrations around several of the known deposits (including the Agnew case study).

In the Northern Yilgarn Craton, Mo has a large, multi-point anomaly in the Cue area (which has no known source and needs further investigation), but single point anomalies in other regions, many of which do not relate to known mineralization (Fig. 7). Most of the known Au deposits do have elevated Mo concentrations associated with them, but not all, making it difficult to include them as part of an exploration index.

W has a very similar spatial pattern to Mo, displaying anomalous concentrations associated with some of the Au systems in the region (Agnew, Thunderbox, Sunrise Dam, Wiluna). Yet not every sample associated with Au systems is revealed by anomalous W values. Overall, there are more coherent W anomalies around the numerous Ni sulfide systems in the region (Fig. 8).

Shallow aquifers were sampled during this campaign for two reasons: (1) there is an abundance of bores and wells in the Northern Yilgarn due to the extensive cattle and sheep farming in this region which provides an ideal, semi-regular grid of sample sites, and (2) groundwater flow direction can be inferred from the landscape and surface geology and enables easier interpretation compared to complex deeper aquifer studies. The Southern Yilgarn region was not sampled during this campaign as the groundwaters are generally hypersaline and acidic and as such do not host regular bores for sampling. The detailed comparison between sampling strategies (bailing and pumping) and their validity for downstream analyses are discussed elsewhere (Gray et al. 2009, 2014).

The primary outcome from the results is that Au concentration is the best direct indicator for Au systems in the Northern Yilgarn, and more sampling is encouraged. Elevated Au concentration highlights many of the known Au mineral systems in the region where sampling is within 5 km of the mine/deposit. Gold can also be associated with other mineral systems but in general have lower concentrations than those of Au systems. From the Geochemist's Workbench® v10 modelling, it is predicted that Au should be present as AuCl2 but other papers (Vlassopoulos et al. 1990; Vlassopoulos and Wood 1990; Gray 1998; Williams-Jones et al. 2009) show that Au can also be present as Au(0), thiocyanate, thiosulfate, cyanate, iodide, bromide or other organo-complexes in natural solutions in the various pH:Eh conditions covered by these data sets. However, the consistency of results across all regions in the national data set and the similarity of concentrations from the multiple tested deposits indicate that speciation does not matter for the direct detection of Au using carbon sachets.

As is one of the primary orogenic Au pathfinder elements that seems to have high concentrations correlated with most of the known Au systems in the region compared to other pathfinder elements (Fig. 4). One reason for this may be that As in oxidized, neutral pH waters exists as arsenate (HAsO42) which is very stable and may preserve the signature of mineralization over a longer period of time and longer distance than other pathfinders which exist as cations (Fig. 9).

The discrepancy of Ag being associated with some Au systems and also high in areas with no Au is attributed to the presence of various other mineral systems in this region which also contain Ag. For example, VHMS systems can be a source for Ag in groundwater unrelated to Au systems (Gray et al. 2018). Therefore, although Ag cannot be used as an accurate standalone element for the detection tool for Au mineral systems, it can be useful in association with the other indicator elements (i.e. Au, As).

Where the oxyanions Mo and W (in addition to As) are detectable in groundwater they are effective pathfinders for Au exploration (Noble and Gray 2010). The differences in individual orebody composition (Fisher et al. 2011) impacts the groundwater chemistry. Both Mo and W in groundwater have also shown anomalous concentrations that are unrelated to Au concentration and any known Au systems. Therefore, Mo and W alone cannot be considered as direct indicators for Au systems but where they are present in association with elevated Au and other pathfinders they should increase the likelihood of targeting Au mineralization. In the waters of the northern Yilgarn most of the enriched Mo and W samples (and also Ni and Co and Cr) are around, or on the margins of, the greenstones (which at the scale of sampling are a mix of felsic volcanics, mafic and ultramafic rocks). It is possible that the anomalies are inherently related to the greenstone belts within the Northern Yilgarn, but the highest concentrations are seen to be associated with mineral systems.

Multi-element indices: the AuMin index

To limit ambiguous results, such as those related to the individual elements described above, which can lead to false-positive or false-negative samples, a multi-element index, known as the AuMin index (‘goldmin’), was developed (Gray et al. 2009). The AuMin index relies on normalization of each of the three elements Au, As and Ag using a Box–Cox transformation. Once rescaled, the new values are summed and provide values comprised between 0 and 1 (where the concentrations converted to a value of 1 is the 98th percentile of the regional data). The higher values correspond to geochemical anomalies whereas the lower values represent background. The AuMin index is then the summation of each of Au, As and Ag divided by 3, which was calculated using the XT HydroTM application and was tested over large areas including the Northern Yilgarn (Gray et al. 2014, Fig. 10) and the Capricorn Orogen area (Thorne et al. 2018). While it was suggested that AuMin index improves the detection of orebodies (Gray et al. 2014), the present study reprocessed the data and found that, in the study area, Au concentrations alone were as precise, or better, in revealing the known mineral systems.

The areas shown with moderate AuMin Index scores (0.35–0.9) occur in the regions that host the majority of the Au deposits in the Northern Yilgarn (Fig. 10). The highest values of AuMin correspond to the larger Au camps of Agnew and Sunrise Dam. The AuMin index can be used to highlight broad areas of interest from the groundwater hydrogeochemistry, whereas the single Au concentration is useful to focus the exploration effort. The results are akin to a Au prospectivity map of the Northern Yilgarn.

Difference in the greenstone belts

The groundwater concentration of single-element Au, associated pathfinder elements or the use of exploration indices (AuMin) were all successful in highlighting many known mines and deposits as well as areas of interest for future Au exploration at different scales. The eastern greenstone belts (Norseman–Wiluna, Duketon, Yandal Belts) are easily highlighted by these methods, which are characterized by elevated Au concentrations in samples near Agnew, Garden Well, Thunderbox, Wiluna and Sunrise Dam Au mines as well as many minor prospects and deposits (Figs 2 and 11). In contrast, the western greenstone belt groundwater analyses fail to highlight known deposits and mines located along the western Mt Magnet–Meekatharra greenstone belt, with overall lower concentrations of Au; the only exception to this is a single sample with higher Au concentration (76 ng l–1) located c. 45 km North of Cue.

Multiple factors can be responsible for this difference in groundwater signature between western and eastern greenstone belts: (i) while the geology of both greenstone belts is similar there are some differences such as the lack of komatiitic rocks in the western belts; (ii) the sample spacing in the western area of the Northern Yilgarn does not include many samples within <5 km of the known mines or deposits (which is approximately the distal footprint of Au in groundwater shown in the other belts); (iii) the orebodies in the Mt Magnet–Meekatharra Belt are generally smaller (Witt and Vanderhor 1998) than in the eastern belts. Other factors such as (iv) differing depths of weathering (how much interaction the current groundwater has with ore), (v) the amount of sulfide within the ore (i.e. more dispersion than nuggetty Au), (vi) the size and orientation of the fault systems (Witt and Vanderhor 1998) and (vii) groundwater flow (how connected the aquifers are around those faults) may each play a role in downgrading the concentration and spatial extent of Au in groundwater in this region.

In the Yilgarn Craton, different styles of Au deposits have distinct elemental signatures within the ore (Fisher et al. 2011) these distinctions can also be reflected in the groundwater geochemistry. While there are significant differences in the size of the Agnew and Garden Well systems, multiple water samples were collected over each mineralization and are ideal case studies to show the utility of the regional samples as a baseline for comparison. Of these two systems, one was already being mined (Agnew), whereas the other (Garden Well) had been discovered in 2009 (Anand et al. 2021), only two years prior to the sampling campaign (2011) and mining operations began soon after.

At Agnew samples were collected along the mineralizing corridor whereas at Garden Well samples were collected before operation and directly over mineralization. The regolith environments for both sites are different, and this is likely causing differences in the expression of mineralization-related elements within the groundwater (Figs 12, 13). Agnew mineralization is made of a series of orogenic (primary) Au orebodies along a large fault system (Fisher et al. 2011), located on a local topographical high with mainly weathered in situ regolith materials above a shallow primary mineralization. By contrast, Garden Well has thicker (c. 30 m), transported palaeochannel sediments above primary mineralization; the base of the palaeochannel sediments also hosts well-developed secondary Au enrichment within Fe-rich pisoliths (Anand et al. 2021). Both Agnew and Garden Well sites have extremely elevated concentrations of Au in groundwater, with most samples being within 1 km of the orebody (Figs 14, 15).

These sites both have multiple samples showing high Au concentrations related to the orebodies. Moreover, the different ‘styles’ of Au mineralization are of interest to evaluate the versatility of the groundwater geochemistry approach.

Agnew case study

The Agnew study area lies c. 300 km NNW of Kalgoorlie. The region hosts numerous Au and Ni deposits and has produced 6.6 MOz of Au (Czarnota et al. 2010). It is adjacent to the Mt Ida shear zone and has a very complex architecture (Fisher et al. 2011). The site is located on a local rise, at higher elevations (c. 520 m asl) than the surrounding regional areas (c. 470 m asl). As a result, the depth to groundwater is often greater than commonly occurs in the Northern Yilgarn Craton (Agnew area water table depth = 30–60 m v. 10–20 m in the rest of the Northern Yilgarn). However, although higher elevation is responsible for much of this difference, the water table depth is probably also increased by draw-down from dewatering of the adjacent mines. At Agnew the Au deposits are primary orogenic deposits; however, this is a complex system where the mineralogy and geochemistry of each orebody varies (Fisher et al. 2011; Thébaud et al. 2018), which in turn impact the groundwater geochemistry. The deposits are hosted by the Agnew–Wiluna greenstone belt. Au mineralization dominantly occurs in quartz breccia lodes, quartz tension-veins and is disseminated with arsenopyrite–pyrite–biotite alteration assemblages. These mineralization styles are principally controlled by competency contrasting lithological contacts.

Groundwater samples collected in the Agnew region were collected from a mixture of diamond, air core and RAB drill holes. Figure 14 shows the contemporary drainage channels with flow direction marked to show general trends for where anomalism could disperse down gradient. There is a large tailings dam in the centre of the sampling area (marked on Fig. 14).

In general, some of the more common Au associated elements (potential pathfinders) are Bi, Te, W, Sb, As and/or Ag. However, the Agnew system is host to many individual orebodies (Fisher et al. 2011), with each one of them possessing a different elemental assemblage in the ore with the only constant being Au. These changes are reflected within the groundwater chemical data – e.g. Sb (Fig. 6) and Mo (Fig. 7).

Prominent structural and mineralogical changes occur in the vicinity of Agnew. In the groundwater, these are delineated by variations in a range of geochemical parameters and elements such as pH, Sr:Ca and Mg:Ca ratios, HCO3-, Cr, V, Ba and Si (Supplementary. Table 1). The variations are mainly due to the changes in the complex geology and orebody composition. The lithology-related elements do not appear to have a broad spatial distribution and do not seem to blend or mix, allowing hydrogeochemistry to be employed here as a potential lithology mapping tool at this scale. The above parameters highlight lithological changes between the Agnew ultramafic unit in the east to the Jones Creek Conglomerate unit in the west, but do not distinguish the mineralization (Noble and Gray 2010; Duuring et al. 2012).

Various trace metals are valuable pathfinders or target elements for mineralization in the Agnew site. Dissolved Au, Ag, As, W, Co and Mn all show a high contrast (and in many cases some of the highest concentrations for each element in the whole data set for the region) across the mineralized corridor, which comprises both orebodies and fluid pathway (faulting), and most of these elements are recognized as being present within the Au ores (Fisher et al. 2011). Due to the high Au concentrations reported from the carbon sachets (>10 000 ng l–1), a second sample was sent for INAA at Becquerel Laboratories, Canada, with comparable high concentrations also being reported through this method.

If more drill holes were available towards the west of the mineralizing corridor it would be possible to determine the true extent of element migration along the flow pathways. The samples to the north indicate that Au dispersion is not carried on down the channels but restricted to c. 1 km around the orebodies. If that holds true throughout the region then several of the anomalous Au concentrations could indicate other orebodies that have not been mined (not discovered or not economic).

Both single-element data (Au, As and Ag) and the combined AuMin index show the region as highly anomalous for Au prospectivity (Fig. 14). In the area surrounding Agnew the AuMin ranges from 0.5 to 1.6, and most of those values are higher than any found within the regional data set (>98th percentile). In addition, the Au concentrations are much higher (by 2–5 orders of magnitude) than regional backgrounds (Fig. 14; Noble et al. 2010). There are very few samples within the Agnew mineralized corridor that are of ‘low’ Au concentration and all samples show a sharp contrast between background (often below detection) and anomaly. Regional background data are crucial to interpret and provide context for the groundwater data of the near-mine samples.

Several explanations may account for the strong anomalies in this region: (i) the presence of a large mineral system, which extends along an active fluid pathway (fault system) and is connected all the way through to the shallow water table; (ii) the local topographic high would limit the dilution of the mineralization signature by external – i.e. meteoric– water sources; and (iii) being on a rise means that the transported regolith in this region is relatively shallow to non-existent (Barnes et al. 2014) and the deeper water table is more likely to be interacting with weathered mineralization.

The cluster of points downstream from the tailings could be contaminated. However, there is low dissolved organic C, and no total N that is not correlated with NO3 which makes it unlikely that there is a contribution from cyanide leakage. The pH of these samples is not lower than that of other samples across the corridor. Samples are also oxidizing which would not suggest the breakdown of organics from tailings. The highest concentrations for Au in particular are not from the samples nearest and downstream from the tailings. Other key elements like Sb and W are not elevated in all samples to the south of the tailings.

Garden well case study

The Garden Well Au Mine is located c. 330 km NNE of Kalgoorlie within the Duketon Greenstone Belt (Fig. 2). The Garden Well mineral system is made up of two distinct, vertically stacked Au orebodies. The deepest Au accumulation is a primary, Archean-aged orogenic Au deposit, that has been subjected to deep weathering and is concealed by Eocene palaeochannel sediments (up to 32 m thick; Anand et al. 2021). The shallowest Au ore was a secondary Au accumulation within the Eocene palaeochannel (formed on and within Fe-rich pisoliths which acted as a redox and sorption trap), above the deeper primary Au deposit (Anand et al. 2021; Fig. 13). The secondary nature of the Au is evidenced by its location directly above the primary mineralization. There are also no other known sources for the Au up slope from the palaeochannel, and the Au within the pisoliths contain very little Ag, suggesting leaching of Au and Ag from a primary source with the Ag staying in solution and not re-precipitating with the Au (Stewart and Anand 2014; Anand et al. 2021).

The primary Au ore is associated with pyrite and arsenopyrite and is enriched in As, Cu, Bi, Se, Sn, Mo, In and S (Anand et al. 2021). The major gangue minerals are carbonates, chlorite and fuchsite. The secondary Au is associated with ferruginous materials and associated pathfinder elements such as As, Cu and S (Anand et al. 2021).

The groundwaters sampled at Garden Well were mostly from uncased drill holes (a mix of diamond, rotary air blast and air core), taken before the mining operation started. The water table was shallow (c. 2–5 m depth) and hosted within the palaeochannel. Therefore, these samples and the data provided herein represent a shallow groundwater system, in contact with secondary Au in pisoliths. Further information is required to determine whether there is an aquifer in contact with the deeper primary mineralization. The only major element change compared to other sites in the region (and related to the palaeochannel) is an increase in salinity and elements associated with an increase in salt (SO42, Br, Cl, Na).

The groundwater signature for mineralization in this system is dominantly Au, with very high concentrations (100–600 ng l–1) in many of the selected boreholes, with no significant As and only two points with elevated Ag concentrations (Fig. 15). This indicates that the groundwater signature mostly reflects secondary Au. The modelling for As speciation (Fig. 9) shows that all samples fit within the same stability field for As(V) and there is no significant changes in redox or As speciation evident which might explain the lack of As. The increase in salinity associated with the palaeochannel suggests that any Au in solution would likely be as a Au–chloride complex rather than as an organo-complex, which would potentially help preserve the Au in solution. Observations from many sites within the Yilgarn (North and South), show that the absolute concentrations of Au are very similar for each Au system regardless of the major groundwater parameters. This indicates that the form of the Au should have no bearing on it being detected through the use of the carbon sachet methods.

When other pathfinder elements, e.g. Ni and Cr, are also present it suggests that the borehole has a strong connectivity between primary and secondary systems. The Ni and Cr in this case are not found in high concentrations within the channel waters but are most likely sourced from ultramafic rocks below, which are the host for the primary mineralization. Within the regional Northern Yilgarn data set, the only times that Ni and Cr are above detection are either near Ni mineral systems or ultramafic rocks (komatiites; Gray et al. 2009). Also highlighting connectivity between shallow and subsurface mineralization, there is also moderately anomalous SO4:Cl, minor W and depleted NO3 which typically reflect the weathering of primary sulfides (Supplementary Table 1).

Further sampling would be required to determine the extent of the Au dispersion downstream from the orebody; the drill holes available at the time did not continue down the palaeochannel and the next samples (part of the regional data set) encountered down this channel (c. 10 km away) had concentrations of all mineralization related elements at below detection, in which case the dispersion halo is most likely c. 1 km but is definitely <10 km.

Comparison of case studies

Groundwater geochemistry at Agnew returned a high concentration of many pathfinder elements (As, Ag, Co, Sb, W), which strongly reflects the pathfinder assemblages found in individual orebodies along the corridor. This indicates that the groundwater was sufficiently exposed to the ore-bearing sulfides to mobilize and highlight the ore-related elements, even in a context of mine dewatering.

In contrast, for the groundwater sampled at Garden Well only a few samples show detectable or significant concentrations in pathfinder elements (Ag and minor As), whereas Au consistently highlights the mineralization. This can be explained by a groundwater interacting mostly with the shallower secondary Au enrichment, with no, or very limited, interaction with the deeper, primary mineralization which hosts more pathfinder elements. Another contributing factor could also be related to a dilution effect: more water sits within the palaeochannel (where secondary Au occurs) than is upwelling from a hypothetical deeper fractured rock aquifer that may be in contact with the primary Au. To verify this latter hypothesis, more detailed sampling at specific depths would be required.

The only major difference between the major ion chemistry of the two sites, and the greater regional data set as a whole, is that the waters from Garden Well have increased salinity and hence all the major ions are significantly higher in most of the Garden Well samples compared to those in the Agnew sampling. However, the relative abundance of each of the major ions remains similar between the two sites and the regional data (Fig. 16) showing that all waters in the region are predominately Na–Cl based with varying proportions of SO42, HCO3, Mg and Ca. Agnew is more enriched in Mg compared to Garden Well, which reflects the changes in the underlying lithology but this is unrelated to Au distribution and ore weathering products.

At Garden Well the Au dispersion is <1 km, which may reflect that not enough samples were taken down the palaeochannel to determine the full extent of dispersion, and further testing would be recommended. Waters from the palaeochannel would be expected to flow further and faster than those outside the channel.

At Agnew, if each cluster of high Au concentrations is considered to be related to an orebody then the dispersion is likely to also be c. 1 km. However, due to the difficulty in establishing what is connected or independent, the broader ‘halo’ of the Au mineral system is likely closer to c. 5 km.

The regional, and closer spaced, case studies on the use of groundwater geochemistry for Au exploration in the Northern Yilgarn Craton have highlighted the potential of the technique and some of its limitations based on east–west Yilgarn comparison and case studies. These can be summarized as follows.

  • Groundwater geochemistry is a powerful method for Au exploration through cover using an environment-friendly procedure where boreholes are available.

  • Sampling for Au using the 1l unfiltered water protocol, with activated C extraction, is easy, effective and very robust in the Northern Yilgarn Craton environment.

  • In groundwater, Au concentration is the most reliable indicator for Au mineralization, out of the individual elements and multi-element index that were investigated herein.

  • As and Ag (± Sb, W, Mo) may also be utilized to detect primary orogenic Au systems; however, there is the chance to miss secondary or deeper systems, and many of the pathfinder elements can also be hosted within other styles of mineralization such as VHMS.

  • Regional context is crucial to evaluate the absolute anomalism of samples over a deposit or tenement. The use of hydrogeochemical mapping to define prospective areas in the Northern Yilgarn was successful particularly for Au by sampling groundwater from farm bores and windmills at the 4–10 km scale.

  • Hydrogeochemical exploration for Au delineated the two world-class Au camps (Agnew and Sunrise Dam) as well as several of the other major mines/deposits in the region, and can give prospectivity estimates for various greenstone belts.

The techniques used herein are showing great promise for future Au exploration through cover, yet further study is needed to better understand the potential and limitations related to the type/size of deposits and accessibility of sampling bores. Overall groundwater geochemistry for Au exploration is shown to be useful at tenement scale because the widespread dispersion haloes associated with groundwater anomalies do not appear suitable for smaller-scale exploration such as delineation of the orebody. To successfully highlight regional anomalism, sample spacings need to be <5 km scale and ideally <1 km scale to reveal a clear mineral system signature. Variations in sample spacings can explain why some of the individual mineral systems have not been directly detected with Au. In addition, other variable parameters such as smaller orebody sizes and depths and/or changes in the depth of weathering (i.e. depth of water table) could also be of significance. Strong spatial correlations with Au were observed for As (Fig. 4) and Ag (Fig. 5), and some correlations with Sb (Fig. 6), Mo (Fig. 7) and W (Fig. 8). As is an effective pathfinder for many Au systems. However, some systems, most likely secondary Au accumulations or vein Au with low sulfide presence, do not show correlation between As and Au. Ag also correlates well with most Au systems, but is also strongly present within VHMS systems (Gray et al. 2018) which can lead to false positives when exploring for Au systems. Other pathfinder elements (such as Bi, Te, Sb, W) are hampered by being at the limits of detection for the current techniques and also unreliable adsorption by carbon sachets. This can lead to them being irregularly assayed and detected across the region. Where they are detected, there is direct correlation between these pathfinders and Au, and other mineral systems. The multi-element exploration index, AuMin, can increase the size of the regional footprint from hundreds of metres to 1–5 km but it also has the potential to pick out VHMS providing false positives for Au systems.

Provided that access to groundwater exists, hydrogeochemical exploration provides a tool to enhance prospectivity and improve exploration success in areas of cover and difficult terrain. We encourage more industry and government partners to analyse for Au as part of the ‘normal’ analytical suite.

We would like to thank and acknowledge David Gray for all the dedication and time spent on the inception through to completion of all the sampling and starting the data compilation that forms the basis of this study and more to come. David was an international leader in exploration hydrogeochemistry, always enthusiastic about his science and he was always willing to share his knowledge and mentor the next generation of scientists and explorers. We hope that we have done all his work justice and are successfully continuing his legacy.

The authors would like to thank the many people who contributed to the collection of this data over several years, as well as the laboratory assistance/field preparation from Tenten Pinchand and Derek Winchester. Sam Bradley and Sandra Occhipinti are also thanked for their contribution through the Exploration Toolkit (XT) project.

We also thank and appreciate the reviewers and editors for their efforts and assistance.

NR: conceptualization (lead), data curation (equal), formal analysis (lead), investigation (equal), methodology (equal), validation (lead), writing – original draft (lead), writing – review & editing (equal); CP: investigation (supporting), validation (supporting), writing – original draft (supporting), writing – review & editing (equal); RLT: conceptualization (supporting), data curation (supporting), methodology (supporting), software (supporting), validation (supporting), visualization (supporting), writing – review & editing (supporting); AH: data curation (supporting), formal analysis (supporting), investigation (supporting), methodology (supporting), software (lead), validation (equal), visualization (lead), writing – review & editing (supporting); JH: data curation (supporting), methodology (supporting), software (equal), validation (equal), visualization (equal), writing – review & editing (supporting); RRPN: data curation (equal), formal analysis (equal), funding acquisition (equal), investigation (equal), methodology (equal), project administration (equal), resources (equal), supervision (lead), validation (equal), writing – original draft (supporting), writing – review & editing (supporting); DJG: conceptualization (equal), data curation (equal), formal analysis (equal), funding acquisition (equal), investigation (equal), methodology (equal), project administration (equal), resources (equal), supervision (equal)

This work was funded by the Minerals and Energy Research Institute of Western Australia (M414 and M407), AMIRA (P778A) and CSIRO Mineral Resources through the Exploration Toolkit Project.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

The data sets generated during and/or analysed during the current study are available in the CSIRO Data Access Portal repository, or in the supplementary table.

This is an Open Access article distributed under the terms of the Creative Commons Attribution 4.0 License (