Integrating portable X-ray fluorescence site survey and ArcGIS models for rapid risk assessment and mitigation strategies at an abandoned arsenic mine site: a case study

ABSTRACT

Australia’s metalliferous abandoned mine sites (MAMSs), pose tangible threats to the environment and human health. To address these concerns, our study utilised state-of-the-art handheld XRF technology to conduct a real-time assessment of the Mole River arsenic mine site. The data revealed notably elevated levels of arsenic and manganese, with the southeast corner of the site identified as a contaminant hotspot. We used a tiered risk assessment approach to compare the detected contaminant concentrations to the Australian health investigation levels (tier 1). This led us to a broader examination of erosion vulnerabilities and the potential migration of contaminants (tier 2). Further, a hydrological assessment (tier 3) identified significant erosion in southern regions, indicating the potential for contaminants to be transported off-site through surface water runoff to Sam’s Creek and Mole River. The proximity of a reservoir to these runoff pathways brought forth additional challenges, especially during heavy rainfall events. Subsequent laboratory analysis of water samples reinforced our findings, as they confirmed heightened arsenic concentrations in Mole River downstream, accentuating the potential risks to ecosystems and human health. By integrating the XRF contour map and erosion assessment with the RUSLE model, valuable insights are gained into critical hotspots with high contamination and erosion potential. By directing rehabilitation efforts towards critical hotspots, resources can be allocated more efficiently and cost-effectively.

GRAPHICAL ABSTRACT

KEYWORDS:

• Abandoned mine sites
• rapid risk assessment
• X-ray fluorescence walkover survey
• ArcGIS models
• soil erosion modelling

Introduction

The presence of numerous metalliferous abandoned mine sites (MAMSs) in Australia can have significant implications for both human health and the environment. These sites often contain hazardous substances, including heavy metals such as lead, arsenic and mercury, which can persist in the soil and groundwater for extended periods [1]. This contamination poses risks to plants, animals and humans who come into contact with these sites. Moreover, the proximity of contaminated mine sites to water sources like rivers and groundwater systems increases the potential for the contamination to spread, potentially affecting drinking water supplies [2]. Areas with shallow water tables or natural springs are particularly vulnerable to mining-related contamination. Abandoned mining sites can also cause long-term environmental degradation, such as soil erosion, land subsidence and depletion of groundwater resources [3].

Public safety and human health impacts from contamination are primary concerns of legacy mine sites. However, due to their remote location nature with limited human activity, MAMSs often have low priority for remediation. It is important to note that the mere presence of contaminants does not automatically imply a high risk to human health and the environment [4]. As MAMSs are typically inactive and ‘abandoned’, the on-site risks of exposure to hazards associated with these sites are relatively low. This is contingent upon the proper implementation of measures such as erecting fences to restrict public access to contaminated areas. The primary risk assessment focus of abandoned mine sites should revolve around the potential migration of contaminants from the site to areas beyond its boundaries. This migration raises concerns as the contaminants can pose risks to vulnerable environmental receptors and human populations if they are transported to nearby regions including through surface water runoff and other pathways of dispersion (such as dust and groundwater transport). By focusing on contaminated areas that are prone to off-site transport through erosion, rehabilitation efforts can be more targeted, efficient, and cost-effective [5].

Efficient and accurate assessment of legacy mine sites is crucial to ensuring effective remediation and risk management strategies. One promising technology that has emerged for this purpose is handheld or portable X-ray fluorescence (pXRF) spectrometry, which enables rapid and non-destructive analysis of soil, sediment and water samples. Recently, Chakraborty et al. [6] applied a handheld XRF spectrometry for the rapid assessment of soil contamination by heavy metals in the Baia Mare area, Romania. The methodology included the use of the XRF in conjunction with non-parametric indicator kriging for soil pollution hotspot mapping. A similar study by Kim et al. [7] explored the integration of pXRF with Geographic Information Systems (GIS) to map concentrations of heavy metals, specifically Pb and Zn in beach sands. Besides heavy metals, pXRF has also been applied to measure soil organic matter with the Cubist algorithms [8]. A recent review of portable XRF applications was conducted by Ravansari et al. [9] discusses the utility of portable pXRF in environmental soil assessments, highlighting its rapid, mobile, and cost-effective nature. The review suggests that with proper calibration and consideration of influencing factors, pXRF can provide highly accurate elemental assessments. By employing pXRF, site assessments can be conducted more efficiently, saving valuable time and resources, while providing detailed information on contaminant distribution and concentration. Beyond pXRF, it is worth noting that other portable techniques such as Laser-Induced Breakdown Spectroscopy (LIBS) with advanced modelling, e.g. machine learning, also offer effective, real-time soil heavy metal analysis in mining areas, emphasising elements, such as chromium [10].

It is also worth noting that the soil sample preparation significantly impacts the measurements taken using pXRF. The study conducted by Goff et al. [11] explored three sample preparation methods: direct scans from field-moist soils, scans after drying and grinding followed by loose packing in plastic cups, and scans from pressed-powder pellets. The results indicated that soil preparation affects the data quality from pXRF measurements, with pressed powder pellets often providing the best correlation to standard benchtop XRF data. However, the accuracy of pXRF data can vary based on the element or compound being analyzed. According to Zhou et al. [12, 13], soil moisture content can significantly impact the accuracy of pXRF measurements, especially when analyzing targeted elements at low concentrations, which is close to the detection limit. In addition, sample thickness can significantly impact the accuracy of XRF measurements. According to the study conducted by Zhou et al. [14], inaccuracies in XRF analysis can arise from samples being too thin, failing to adequately absorb primary X-rays or emitting sufficient secondary fluorescence for precise detection, or from samples being overly thick, which may cause absorption enhancement effects that skew the results. According to the pXRF measurement study conducted by Zhu et al. [15], the soil samples were prepared at a thickness of around 2 cm (20 mm) to ensure complete attenuation of the X-ray beam by the soil.

The RUSLE (Revised Universal Soil Loss Equation) model can be applied to enhance the assessment of erosion vulnerability and aid in identifying areas most prone to soil loss [16]. It helps in identifying regions most susceptible to soil loss by considering factors such as rainfall, soil type, topography, crop system, and management practices. We have previously utilised the RUSLE model to predict downward infiltration and soil loss in areas contaminated with petroleum hydrocarbons due to rainfall and run-off [17]. The RUSLE model can be implemented using a combination of GIS and other publicly accessible remote sensing data to assess factors such as soil erosion potential, slope gradient, and land cover to predict the soil erosion rate of a certain area [18]. For instance, Abdelsamie et al. [19] applied the RUSLE model in conjunction with multispectral satellite data from Landsat to predict soil erosion in the El-Minia region of Egypt. Similar applications include Mohammed et al. [20] conducted an assessment of soil erosion risk in southern Syria using the RUSLE model integrated with geoinformatics approaches, focusing on the eastern part of the Yarmouk Basin in Al-Swida. It utilised publicly accessible climatic data, field sampling data, and Landsat satellite imagery to estimate the RUSLE model parameters.

In addressing the complex challenges presented by abandoned metalliferous mine sites, the study endeavours to harness the synergistic potential of pXRF technology and GIS alongside the RUSLE model. This integrated approach aims to deliver a comprehensive and nuanced understanding of both contamination and erosion risks, thereby offering a robust framework for targeted environmental remediation and conservation efforts. By implementing this approach at the Mole River arsenic mine site as a case study, the study leverages the capabilities of pXRF technology for quick on-site assessment of soil samples. These samples display elevated concentrations of heavy metals, especially arsenic, enabling the rapid identification of contamination hotspots through GIS. The RUSLE model is utilised to assess erosion vulnerabilities within the site, identifying regions at high risk of soil loss. This assessment is crucial for understanding how erosion could potentially facilitate the off-site transport of contaminants, exacerbating environmental and health risks. The study seeks to establish a tiered framework for risk assessment that incorporates both the intensity of contamination (Tier 1) and the potential for contaminant migration due to erosion (Tier 2). Furthermore, Tier 3 assesses potential pathways for off-site contaminant transport, aiming to develop targeted mitigation strategies to prevent environmental degradation and protect downstream ecosystems and human populations. This holistic view allows for a more strategic allocation of remediation resources, ensuring that efforts are focused where they can have the greatest impact.

Materials and methods

Site description

The Mole River arsenic mine site, approx. 400 m by 320 m, is the case study area which is demonstrated in Figure 1. The Mole River arsenic mine site is located approximately 70 km north of Glenn Innes in New South Wales, Australia. The mine was operational in the early 1900s and was closed in the 1930s. Mining activities excavated arsenopyrite-rich veins in the area, which were then processed to extract arsenic, primarily for use in insecticides, herbicides and wood preservatives. The site is known to have high concentrations of arsenic, with minor occurrences of antimony, manganese and zinc [21]. According to the previous study conducted by Ashley and Lottermoser [22], soil types within the mining vicinity vary, extending from shallow, skeletal forms with occasional rock outcrops to depths of 25–30 cm, and predominantly consist of sandy loam, silt loam, and sandy clay loam compositions. The site is situated at an altitude of about 450 meters above sea level and has a relatively gentle slope.

Figure 1. The Mole River arsenic mine site (Satellite map obtained in 2023).

The site contains numerous ruins of buildings, identified as ‘BR’ in Figure 1, which are believed to have been used for arsenic processing. Additionally, there are four waste dumping zones, marked as ‘WD’ in Figure 1. The Mole River is situated close by, with distances ranging from approximately 100–200 meters from the site. Originating from the southern hills of the mine, Sams Creek runs alongside the waste dumping areas, leading to substantial contamination due to effluent discharge. Agricultural communities located both upstream and downstream from the site are at risk, heightening concerns about the possible health impacts of arsenic exposure.

According to the previous study [22], the site has been subject to numerous environmental studies and clean-up efforts, aiming to mitigate its impact on both human health and the environment. For example, the most impacted area has been fenced off to prevent trespassing, and the bare soil has largely been revegetated. Three water reserves, labelled ‘WR’ in the satellite image, are designed to intercept runoff water and prevent further contamination of the Mole River. Based on observations, WR1 and WR2 are designed to intercept potential contamination from the dumping areas WD1 and WD3, while WR3 is intended for WD2.

On-site measurement

A handheld XRF, Olympus Delta DP 6000, was applied for metal and metalloid measurement in the field. The Olympus Delta DP 6000, dedicated to mining exploration, is equipped with a Silicon Drift Detector (SDD) and a Rhodium (Rh) X-ray tube with a maximum power of 40 kV and 4W. The detector has an active area of 30 mm² and a resolution of 140 eV. For metal and metalloid measurements, the instrument uses a combination of fundamental parameters and empirical calibration to calculate concentrations of the elements of interest, including lead, arsenic, copper, zinc, cadmium, mercury and nickel, with a limit of detection (LOD) at ppm levels for most elements of interest.

The pXRF instrument, conducts scans for 30 s per beam, across two beams, totaling 60 s for each field scan. Calibration was conducted daily before field testing, utilising a silica blank and certified reference materials (CRMs), namely NIST 2711a and NIST 2710a, which were obtained from Sigma-Aldrich. An acceptable error margin has been established at within ±10% when compared against the values of these CRMs. Additionally, the instrument underwent hourly recalibrations with 316 stainless-steel chips during field operations.

On the mine sites, soil samples were gathered using a 10 by 10 (total 100 locations) grid-based sampling method at various locations and were analysed on-site. The distance between each sample was maintained at about 50 metres using GPS navigation (Garmin). As shown in Figure 2, the sampling grid with more than a hundred sampling locations covered the entire mining area. To mitigate the impact of soil moisture variation, the site measurement was conducted during a dry week in the dry, warm season (early Autumn), when we confirmed that there had been no recorded rainfall within the preceding three days. Before measurement, we ensured that the soil was in a dry condition. The soil sample collection process involved removing the top layer of soil, which contained significant organic matter, and taking soil samples from a depth of approximately 3 cm. A 2smm stainless steel soil sieve from Sigma-Aldrich was used to further eliminate any remaining gravel and organic matter. To meet this requirement of soil sample thickness around 20 mm [15], the collected soil samples were placed into Petri dishes, 100 mm in diameter and 20 mm in depth, sourced from Sigma-Aldrich. The soil samples were distributed evenly within the Petri dishes, ensuring that the surface of the soil was level with or above the rim of the dishes at the measurement points.

Figure 2. Arsenic results comparison between pXRF and ICP-OES.

Approximately 500 grams of each soil sample were collected in zip-locked bags for further laboratory analysis. The metals and metalloids were examined via Inductively Coupled Plasma Optical Emission Spectroscopy (ICP-OES) with the microwave acid digestion method [23]. Surface water samples were collected from locations affected by erosion and runoff at the mine sites. To evaluate the extent of contamination, water samples were systematically collected from various locations, including the confluence of Sam’s Creek and the Mole River, as well as upstream and downstream areas of the Mole River. The water samples were filtered through 0.45 µm filters in the field and then transferred to 50 mL centrifuge tubes. To maintain a pH level below 2, 1 mL of 65% nitric acid was added to each tube for preservation. These samples were also analyzed using ICP-OES, and the analysis results are shown in Table 1.

Table 1. Arsenic concentration in different water locations.

Locations	Arsenic concentration
Sam’s Creek upstream (470 m away from the site)	19 μg/L
Sam’s Creek near location WR3	1.78 mg/L
Sam’s Creek near MR intersection (20 m)	1.21 mg/L
Mole River upstream (610 m away from the site)	2 μg/L
Mole River downstream (200 m away from the site)	101 μg/L
HIL for drinking water	10 μg/L
HIL for freshwater	24 μg/L

The RUSLE erosion model

The RUSLE, introduced by Tim et al. in 1992, estimates long-term average annual soil loss (A) in tonnes per hectare per year, calculated as A = R × K × L × S × C × P. In this formula, R represents the rainfall and runoff factor, K stands for soil erodibility, L and S denote slope length and steepness, respectively, which influence erosion risk due to water flow and runoff velocity, and C and P are land management factors, indicating the effect of canopy, ground cover, and soil properties on erosion potential. To determine the RUSLE parameters, the following steps were taken:

R factor – using the equation R = 17.67 log(MAR) – 27.24 [24], with the average annual rainfall (MAR) from Tenterfield, Mole Station being 730.2 mm over the last decade (2013–2022) [25], the calculated R factor, which denotes rainfall and runoff impact, is 23.4.

K factor – was derived from soil texture data via the eSPADE database, with an estimated value of 0.06 [26].

L S factor – The spatial analysis function in ArcGIS was utilised to compute both the L and S factors, using the digital elevation model, Elvis, with a 5-meter resolution, sourced from Geoscience Australia’s National Elevation Data Framework [27].

C factor – The C factor, indicating vegetation cover and land management from the latest Google satellite imagery of the mine site, ranges from 0.01 for full vegetation to 1.0 for bare soil, per NRCS standards [28]. The P factor assesses erosion control effectiveness from practices like terracing. ArcMap Maximum Likelihood analysis classified land cover, setting C factors: 0.2 for shrubs, 0.05 for grass, 0.01 for woodlands, and 1.0 for bare soil, showing varying erosion protection levels.

P factor – The P factor assessment used ArcMap to create a slope map, indicating how slope gradient impacts erosion control. ‘Grass strips’ were implemented as a conservation method, effective in slowing water flow and reducing runoff [29]. The P factor varies with slope: 0.27 for flat areas, 0.3 for slopes of 7.0–11.3 degrees, 0.4 for 11.3–17.6 degrees, 0.45 for 17.6–26.8 degrees, and 0.5 for slopes over 26.8 degrees, indicating increased erosion potential with steeper slopes.

Results and discussion

Accuracy of pXRF

The mean of relative errors for the pXRF measurements compared to the conventional ICP-OES method was approximately 30%, varying with the concentration levels of the metals and metalloids. Figure 2 demonstrates the arsenic measurements using the two methods. It is observed that the measurement results between the two methods are closer when arsenic concentrations are elevated, reaching thousands of mg/kg (see Figure 2a). However, higher differences were observed when the concentrations were lower (see Figure 2b). This may be due to the significant influence of soil moisture and organic matter on pXRF accuracy, particularly at lower concentrations, as mentioned in the introduction [12, 13]. According to eSPADE, the area's soil typically contains over 2% organic matter. Despite sieving, smaller organic matter particles may remain, potentially impacting pXRF measurements. Additionally, our measurements indicate significant iron levels, reaching thousands of mg/kg, in the soil samples. According to Rosin et al. [30], these elements can modify X-ray absorption and scattering, leading to discrepancies in detected concentrations. Yet, at higher concentrations, these effects diminish, aligning pXRF results more closely with ICP-OES. High concentrations reduce the influence of sample variations, unlike at lower levels, where minor composition differences significantly affect pXRF accuracy compared to the averaging effect of ICP-OES’s sample digestion.

It is worth noting that pXRF technology has been extensively utilised for field measurement activities due to its portability and ease of use. The results from pXRF measurements are considered reasonably sensitive and accurate, especially since we ensure instrument checks with the CRM both before and during field measurements. Furthermore, the primary goal of these measurements is to identify contamination hotspots, primarily arsenic in our case swiftly. Typically, the contaminants at these hotspot locations are present at relatively high concentrations, often exceeding thousands of mg/Kg in the soil samples. Consequently, pXRF results exhibiting reasonable sensitivities and accuracies are deemed acceptable for our purposes of rapid field measurement applications.

Soil mapping with pXRF

In accordance with the Australian health investigation levels (HILs) for commercial and industrial land recommended by the National Environment Protection (Assessment of Site Contamination) Measure (NEPM), Table 2 provides the HIL values for the elements under consideration [31]. Handheld XRF measurements at the abandoned mine site detected elevated levels of arsenic (As) and manganese (Mn), both of which exceeded the Australian Health Investigation Levels (HILs). The soil in certain areas of the site is enriched with titanium (Ti) and iron (Fe). In some spots, Ti marginally exceeds the HIL, but this slight excess is not deemed significant. On the other hand, Fe concentrations remain below the HIL. Given these findings, both Ti and Fe can largely be categorised as existing at ‘background levels’ in the soil. From the HIL values, a notable observation is that the arsenic concentration, despite being half that of iron (Fe), significantly exceeds the HIL. The distribution patterns of these four elements (As, Mn, Ti and Fe) are depicted in Figure 3.

Figure 3. Distribution of the four primary elements detected by the handheld XRF.

Table 2. The top concentration level of the elements versus their health investigation levels (HILs).

Element	Top concentration (g/Kg)	HIL* (g/Kg)	Concentration ÷ HIL
Arsenic	21.26	3.0	6.67
Manganese	16.8	2.0	8.4
Titanium	8.48	8.0	1.05
Iron	40.44	60.0	0.67

*HIL is referenced from NEPM for Commercial/industrial land D [31].

The south-eastern corner emerged as a significant area of concern due to its elevated concentrations of Mn, surpassing 1.6 g/kg. According to Figure 1 and the report from 1999 [22], this area was not linked to significant mining activities such as ore processing or waste disposal. Satellite imagery shows that this region, located outside the fenced boundaries of the mining area, is densely populated with trees and vegetation that can withstand harsh conditions. Therefore, the background soil in this area can be considered naturally rich in manganese. This finding is also confirmed by the report [22].

Specific spots within the location showed heightened levels of arsenic. Arsenic hotspots, using a threshold of 2%, were identified at locations 10 and 12, as well as in the vicinity of locations 29, 31, and 63. Notably, locations 10 and 63 displayed the highest concentrations of titanium among these. The high level of arsenic in the soil at locations 10 and 63 can be attributed to their historical use as waste dumping areas (WD 1 and 2) for mining activities, as indicated in the 1999 report [22]. Typically, waste materials from mining operations, especially those involving arsenic-rich ores, can contain significant quantities of this toxic element. Over time, arsenic from the dumped waste can leach into the soil, leading to elevated levels and potentially creating environmental and health hazards. Therefore, the high arsenic concentrations found in the soil at these locations are likely a legacy of past mining operations, reflecting the environmental impact of these activities on the site’s soil quality.

Tier 1 – contamination hotspot identification

To prioritise the distribution of environmental risk associated with each contaminant, one approach is to calculate the ratio of the concentration of each concerned element and its respective HIL. This ratio helps determine the significance of each contaminant’s distribution and its potential impact on the environment and human health. By comparing the ratio values, locations with higher ratios indicate a greater deviation from the HIL and hence pose a higher human health risk.

To simplify the prioritisation process and obtain a more comprehensive assessment of the overall contamination level at each location, the sum of the ratios of the elements can be used to provide a more holistic and straightforward picture of the overall contamination level, taking into account the cumulative impact of multiple contaminants. By considering the combined effect of all elements, a more accurate understanding of the severity of the contamination at each location can be achieved.

Hence, for each location (j), the tier 1 value in this risk-based land assessment is demonstrated with equation (1). (1) $\mathrm{Tier}{1}_j=\sum {\displaystyle \frac{\mathrm{Concentraio}{\mathrm{n}}_i}{\mathrm{HI}{\mathrm{L}}_i}}$(1) Where: ‘i’ presents each of the concerned elements (e.g. As and Mn).

The distribution of tier 1 values is illustrated in Figure 4, which provides an assessment of contamination severity based on the Australian HIL for each element. The areas with higher cumulative ratios, tier 1 value, indicate a higher overall contamination level and should be given priority for further investigation, remediation, and risk management strategies. Among the measurement locations, No. 10, 31 and 63 exhibited the highest tier 1 values primarily due to excessive concentrations of arsenic in the waste dumping areas. The southeast corner of the site was designated as the second priority area based on the Tier 1 values. This particular region exhibited the highest levels of manganese.

Figure 4. Risk-based assessment tier 1 distribution.

The tier 1 value provides a conservative indication of the potential risk to both human and environmental receptors at the contaminated site. However, the risk primarily affects personnel who spend a significant amount of time on the site and may be exposed to the contaminants through dust inhalation, digestion, or dermal absorption, where the onsite human health risk assessment can be further investigated. As the mines have been abandoned, environmental protection authorities can establish legislation or install barriers to prevent further human activities on the contaminated sites. Hence, the tier 1 value does not provide an indication of the prioritisation for remediation, where the risks of contaminants being transported off the site to receptors downstream must be considered.

Soil erosion investigations

The analysis of soil loss distribution using RUSLE, as depicted in Figure 5, reveals significant erosion occurring in the southern and southwestern regions of the mining site. The highest soil loss rates, reaching approximately 25 tons per hectare per year, are observed in these areas. This indicates a heightened vulnerability to erosion and emphasises the need for effective erosion control measures in these specific locations. According to the digital elevation, it is evident that rainwater runoff follows a predominant flow path from the southwest portion of the mining site towards the Mole River located in the northeast. This natural drainage pattern can contribute to the transportation of eroded soil particles, sediment and potentially associated contaminants towards the Mole River.

Figure 5. Soil loss distribution.

The effectiveness of the RUSLE model in relatively small areas is subject to debate, given its design for broader landscapes. In our specific application, while the soil loss values derived from the RUSLE model may lack high precision, the model’s principles remain effective for ranking erosion severity across different sections of the mining site, based on landscape and soil vegetation levels. This method facilitates targeted identification of areas at higher risk of contaminant transport, thereby guiding focused mitigation strategies, even if the exact soil loss figures are not precisely calibrated for our investigation’s scale. This prioritisation can inform conservation efforts and soil preservation tactics, adapting large-scale erosion models to address localised environmental challenges.

It’s worth noting that there are many open sources, e.g. eSPADE (Australia), that provide erosion data mapping at a relatively low resolution, which might overlook variations in land use, such as bare soil, roads, and changes in vegetation, which might not be suitable in smaller areas like those in our case study site. Accurate and detailed land use information is crucial for our site investigation, especially as we aim to identify potential environmental risk hotspots with high soil mobility. The presence or absence of vegetation significantly influences our assessment, impacting soil stability and the potential spread of contaminants. By employing satellite imaging, we can classify the vegetation cover of the land in greater detail, thereby enhancing the accuracy and resolution of our erosion investigation.

Tier 2 – environmental risk hotspot identification

With the tier 2 risk assessment, we aimed to obtain a more comprehensive understanding of the overall risk at each location by incorporating both the tier 1 value and soil loss data. This approach takes into consideration both the severity of contamination and the potential for off-site transport of contaminants, providing a more holistic view of the risk landscape. To calculate the tier 2 value for each location, the tier 1 value was multiplied by the normalised (N) soil loss value, as shown in equation (2)(2) $\mathrm{Tier}2=(\mathrm{Tier}1)\times N(\mathrm{Soil}\mathrm{Loss})$(2) . (2) $\mathrm{Tier}2=(\mathrm{Tier}1)\times N(\mathrm{Soil}\mathrm{Loss})$(2) The tier 1 value represents the level of contamination at each location, while the normalised soil loss value reflects the potential for contaminants to be transported beyond the site boundaries. Locations with tier 2 values exceeding 1 are classified as high off-site risk areas. By incorporating these factors, the tier 2 value serves as a comprehensive indicator of the overall risk level at each location. Locations with higher contamination levels and a greater potential for off-site transport will have higher tier 2 values, indicating a higher level of hazardous conditions. On the other hand, locations with lower soil loss risks and lower potential for off-site transport will have lower tier 2 values.

Based on the tier 2 distribution depicted in Figure 6, it is evident that a significant portion of the area exhibits tier 2 values lower than 1 when compared to their corresponding tier 1 values. However, certain areas, particularly locations 10 in WD1 and 31 near WR3, continue to be classified as high-risk.

Figure 6. Risk-based assessment tier 2 distribution.

By focusing on locations with higher tier 2 values, resources can be strategically allocated and targeted measures can be implemented where they are most needed. This approach ensures that mitigation efforts are directed towards the locations that pose the greatest risk to nearby ecosystems and water bodies, optimising the effectiveness of risk management strategies.

Tier 3 – off-site transportation pathway assessment

In tier 3 of the risk assessment, the focus shifts to identifying and analyzing potential runoff pathways, or water channels, for the off-site transport of contaminants. This is so that a remediation strategy can be effectively conducted to mitigate their potential impacts on downstream receptors. These may include the construction of barriers, diversion channels or containment systems to intercept and redirect runoff, as well as the implementation of erosion control measures to minimise soil erosion and the detachment of contaminated particles.

Using advanced techniques and tools such as hydrological modelling and spatial analysis in ArcGIS, the potential runoff pathways can be effectively identified. One of the key methods employed is ArcGIS flow accumulation analysis, which utilises DEM data to predict stream channels and drainage basins. By analysing the topography of the site, including the elevation data and slope characteristics, ArcGIS can generate a flow accumulation map, which illustrates the accumulation of water and can help pinpoint the most likely pathways for runoff (Figure 7).

Figure 7. Runoff pathways for arsenic off-site transportation.

From Figure 7, it is evident that stream channel CH1 originates from location 10, passes through the waste dumping areas WD1 and WD3, and eventually flows into the Mole River. The satellite imagery in Figure 6 clearly shows that water reservoirs WR1 and WR2 are situated to intersect this identified water pathway. The presence of the reservoirs in the vicinity of the runoff pathway introduces an important consideration in the assessment of off-site transportation pathways. The reservoir serves as a barrier against contaminated water flow into nearby water bodies, but its effectiveness can diminish during heavy rain. If runoff exceeds the reservoir’s capacity, overflow may occur, allowing water to bypass the barrier and reach the Mole River. It’s crucial to consider overflow potential during heavy rains to devise effective mitigation strategies. Authorities should evaluate the reservoir’s capacity, particularly WR1, to manage increased runoff in extreme weather, factoring in size, depth, structural integrity, and overflow mechanisms. To reduce off-site contaminant transport during overflows, secondary containment, such as extra retention ponds or berms along the estimated channel may be needed.

The presence of Sam’s Creek, a notable water channel (CH2) on the eastern side of the legacy mining site, holds significant importance. The existence of Sam’s Creek highlights the potential connectivity between the contaminated site and nearby surface water bodies, raising concerns for environmental receptors reliant on these water resources, where the risk is also reported in Ashely’s study [22]. This channel cuts across the eastern boundary of a notable hotspot, specifically location 31, and continues towards the Mole River. As shown in Figure 7, there is also a water reservoir, WR3, intended to intersect this water pathway. However, Figure 4 reveals that the site contamination level (tier 1) and the surrounding soil near WR3, particularly at locations 31 and 63, might not be effectively intersected by WR3 and could therefore be transported off-site via CH2. Furthermore, another subchannel that traverses WD2 (near locations 31 and 63) does not intersect with WR3 either.

To evaluate the extent of contamination, water samples were systematically collected from various locations, including the confluence of Sam’s Creek and the Mole River, as well as upstream and downstream areas of the Mole River. Laboratory analysis revealed notable variations in the concentrations of arsenic among the different sampling points (Table 1). The concentration of arsenic in Sam’s Creek was found to be 1.21 mg/L, indicating a substantial contamination level. Downstream samples from the Mole River exhibited a concentration of 101 μg/L, which is significantly higher than the regulated limit. In contrast, the upstream section of the Mole River demonstrated a concentration of 2 μg/L, which falls within the acceptable range. The concentrations recorded at the confluence of Sam’s Creek and the Mole River, as well as at the downstream section of the Mole River, surpassed the maximum contaminant level for arsenic in freshwater and drinking water established by NEPM, which is set at 24 and 10 μg/L for As (III), respectively [31]. Particularly concerning is the fact that the concentration in the downstream water was approximately 50 times higher than the regulated limit.

It is important to highlight that the risk assessment conducted in this study focused primarily on the surface water and runoff pathways and did not consider the transport of contaminants through groundwater. The exclusion of groundwater transport in this risk assessment is based on the assumption that the mining activities did not significantly impact the deep subsurface and the natural flow of groundwater. However, it is worth noting that during the mining activity, external contaminants such as hydrocarbons and acids may have been introduced to the site. These contaminants are typically associated with mineral processing activities and are concentrated in specific areas within the site. While this risk assessment does not directly address the potential groundwater contamination, it is important for the authority to monitor and assess the quality of groundwater in and around the mining site to ensure its long-term protection and prevent any unforeseen impacts on local water resources.

Conclusion

The handheld XRF technology revealed significant contamination levels at the Mole River arsenic mine site, especially elevated arsenic and manganese concentrations, with the southeast corner being notably concerning. By comparing with Australian health standards, certain areas were identified with alarming contamination. Additionally, the site’s southern regions showed vulnerability to erosion and potential contaminant spread. Hydrological modelling pinpointed Sam’s Creek and Mole River as potential contaminant transport routes, and the nearby reservoir added complexities, especially during heavy rainfall. Lab tests confirmed high arsenic levels in Mole River, suggesting risks to aquatic life and humans. By combining the XRF contour map with the RUSLE model, we identified critical hotspots of contamination and erosion. Focusing rehabilitation on these areas ensures efficient resource use and higher restoration success, while also minimising further environmental harm.

Data available statement

The data that support the findings of this study are available from the corresponding author, Wang L., upon reasonable request.

Acknowledgments

The authors extend their appreciation to the Cooperative Research Centre for Contamination Assessment and Remediation of the Environment (CRC CARE), the Department of Defence, Australia, and the University of Newcastle for their invaluable support. This research was financially backed by CRC CARE Pty Ltd and conducted within the Global Centre for Environmental Remediation (GCER) laboratories at the University of Newcastle.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by Cooperative Research Centre for Contamination Assessment and Remediation of the Environment.