Science of the Total Environment 640–641 (2018) 1259–1271

Contents lists available at ScienceDirect

Science of the Total Environment

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

Mapping magnesium sulfate salts from saline mine discharge with

airborne hyperspectral data
Kirrilly S. Pfitzner ⁎, Andrew J. Harford, Timothy G. Whiteside, Renee E. Bartolo
Department of the Environment and Energy, Supervising Scientist Branch, PO Box 461, Darwin, NT 0801, Australia

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• Composition and extent of MgSO4 efflo-

rescence mapped with hyperspectral
data
• Hexahydrite and starkeyite were the
dominant salt mineralogy identified by
XRD.
• Elevated uranium detected by a spectral
absorption at 0.44 μm confirmed by XRD
• A constrained energy minimisation
technique matched field data to
airborne survey
• Remote method detected known
locations of efflorescence from saline
drainage.

a r t i c l e i n f o a b s t r a c t

Article history: Managing saline water discharges from mining operations is a global environmental challenge. Measuring the lo-
Received 19 February 2018 cation and extent of surface efflorescence can indicate solute movement before changes in electrical conductivity
Received in revised form 30 May 2018 (EC) are detected in waterways. We hypothesised through the use of a case study that ground-based reflectance
Accepted 31 May 2018
spectrometry and airborne hyperspectral (450–2500 nm) analysis of surface efflorescence could be a rapid
Available online 7 June 2018
method for monitoring large regions of the surrounding environment, including downstream of remote mines.
Editor: D. Barcelo X-ray diffraction and X-ray fluorescence were used to determine mineralogy and elemental composition of sur-
face salts around a uranium mine. Salt samples were found to be mixtures of magnesium sulfate. The reflectance
Keywords: of field spectra varied depending on the hydration of the mineral, mainly hexahydrite and starkeyite. A
Ranger uranium mine constrained energy minimisation technique was used to match the field reflectance spectra to the airborne
Kakadu national park data. Airborne matches were confirmed at the field sampling sites and surrounds. Salts were also detected at
Remote sensing lower matches at mine water irrigation areas where excess mine water had previously been applied. Hence,
Spectral reflectance hyperspectral remote sensing is a potentially rapid and sensitive method for mapping magnesium sulfates
Absorption spectra
over large areas in operating and rehabilitated mines. It was successfully demonstrated as a tool for monitoring
and assessment of efflorescence as a result of saline processes.
Crown Copyright © 2018 Published by Elsevier B.V. All rights reserved.

Abbreviations: CEM, constrained energy minimization; CR, continuum removal; EC,
electrical conductivity; GCT2, Gulungul Creek tributary 2; HyMap, airborne
hyperspectral mapper; LAA, land application area; MNF, minimum noise fraction; SWIR,
shortwave infrared; VNIR, visible to near-infrared; XRD, X-ray diffraction; XRF, X-ray
fluorescence.
⁎ Corresponding author.
E-mail address: kirrilly.pfitzner@environment.gov.au (K.S. Pfitzner).
1. Introduction
Much attention is given to acid rock drainage, often referred to as
acid mine drainage, in the mining industry, but saline drainage without
acid rock drainage, as well as neutral or alkaline drainage, may also pro-
duce detrimental environmental effects (Lottermoser, 2010; Nordstrom

https://doi.org/10.1016/j.scitotenv.2018.05.396
0048-9697/Crown Copyright © 2018 Published by Elsevier B.V. All rights reserved.
1260 K.S. Pfitzner et al. / Science of the Total Environment 640–641 (2018) 1259–1271
et al., 2015). Salinity issues in the mining industry may occur during
mining or post-mining and arise through the dissolution and
mobilisation of metals, non-metals, and metalloids (Nordstrom et al.,
2015). Saline drainage is characterised by neutral to alkaline pH mine-
waters (The International Network fur Acid Prevention (INAP), 2009;
Verburg et al., 2009). Local carbonate resources as well as high sulfate
concentrations are required for saline drainage processes to occur. Sa-
line drainage can result in salinisation of freshwater creeks, rivers and
groundwater and the modification of geochemistry that can have im-
pacts on biota, including aquatic and terrestrial plants and animals
(Canedo-Arguelles et al., 2016). Suspended clay particles flocculate in
saline water and increased sunlight penetration can result in algal
blooms (Sholkovitz, 1976). Soil erosion may occur if terrestrial vegeta-
tion cover is altered by being scalded or intolerant to saline drainage.
Mining spoils that produce saline drainage include coal (Kurbiel et al.,
1996), uranium (Lottermoser and Ashley, 2005) and diamond mining
(Nordstrom et al., 2015; van Dam et al., 2013). Surface mineral salts,
or efflorescence, are indicative of saline drainage. Remotely sensed de-
tection and mapping of efflorescence around mines and downstream
overtime is possible due to the bright white reflectance of salt minerals.
However, other bright white coloured features may confound the re-
sults, such as alluvial sands, white pipes, concrete, white metallic roof
tops and mining tanks.
Selected sulfate, carbonate and chloride salts have been
characterised in the controlled laboratory environment by reflectance
spectroscopy (Hunt et al., 1971; Hunt and Ashley, 1979; Clark et al.,
1990; Crowley, 1991; Grove et al., 1992). Most published salinity studies
using remote sensing investigate bare soils or vegetation signals as an
indirect indicator to salinity as a result of land degradation or vegetation
dieback from saline processes (e.g. Farifteh et al., 2008; Rao et al., 1995;
Verma et al., 1994; Dehaan and Taylor, 2002; Belluco et al., 2006; Bilgili
et al., 2011). Examples of mapping salt mineralogy using hyperspectral
remote sensing in acid mine drainage affected mine sites (Swayze et al.,
2000; Bierwirth and Pfitzner, 2001; Pfitzner and Clifton, 2006a) in-
volved historical acid mine drainage pyrite mines, with mining not un-
dertaken to current environmental standards, and where sulfidic host
minerals caused hectares of efflorescence (e.g. copiapite, jarosite,
melanterite) co-precipitating with oxidised iron minerals (ferrihydrite,
Fig. 1. The location of Ranger Uranium Mine within the Ranger Project Area.
shwertmannite, goethite, hematite). These metal-rich minerals have
transition metals such as iron or copper which contain unique absorp-
tion features due to electronic processes that dominate and characterise
the visible-near infrared portion of the spectral signature thereby en-
abling their spectral identification and discrimination (Hunt et al.,
1971; Hunt and Ashley, 1979). By contrast, the efflorescence of saline
drainage show absorption features due to vibrational processes, namely
molecular substitution and water content. These patterns of spectral
variation in magnesium sulfates are not well documented. Research
into the reflectance spectra of magnesium sulfates has focused mostly
on salts on Planet Mars where they are constituents of the Martian
soil, and played an important role in the past hydrologic cycle. These
studies have included data from Mars missions but also laboratory stud-
ies under Martian atmospheric scenarios, often using finely-ground,
synthesised minerals with no impurities (Cloutis et al., 2006; Cloutis
et al., 2007; Chou and Seal, 2007; Chou et al., 2013; Bishop and
Pieters, 1995; Bishop et al., 2005; Bonello et al., 2005; Brown and
Cudahy, 2006; Leftwich et al., 2013; McCord et al., 1998; McCord et al.,
1999; Roach et al., 2009; Vaniman et al., 2004; Vaniman and Chipera,
2006; Dalton, 2003). A few isolated examples of magnesium sulfate
spectra exist in public domain spectral libraries. The USGS spectral li-
brary (Clark et al., 2007; Kokaly et al., 2017) contains one spectrum
each of epsomite (as “commercial epsomite salts”) and starkeyite (as
an “intimate mixture of starkeyite plus possible other salts”). The
ASTER spectral library (Baldridge et al., 2009) contains two spectra of
epsomite (one of commercially produced epsomite salts and the other
a synthesised product).
Small, patchy locations of surface efflorescence observations, com-
bined with high electrical conductivity (EC) levels in creek water adja-
cent to a uranium mine prompted this field and remote sensing study.
We hypothesised that airborne hyperspectral analysis of efflorescence
could be a rapid method for monitoring the surrounding environment
of remote mines. Salt samples were taken for X-ray diffraction (XRD)
to characterise mineralogy, X-ray fluorescence (XRF) for elemental
composition of potential contaminants (such as Mg, SO, U, Mn) and
ground-based reflectance spectrometry as an input to match the air-
borne measurements. To cover this extent, an airborne hyperspectral
survey (HyMap) utilising 124 bands across the reflective solar
 K.S. Pfitzner et al. / Science of the Total Environment 640–641 (2018) 1259–1271
wavelength region of 450–2500 nm with contiguous spectral coverage
at 3.1 m ground sample distance was commissioned. The aim was to
use these data to identify the extent of any salt affected areas off-site
of the mining operations including areas upstream of the mine site to
the downstream wetlands.
While MgSO4 minerals have been mapped on Mars, or as assem-
blages in acid mine drainage environments, here we apply a study to di-
rectly map minerals from saline drainage without acid mine drainage. A
similar application has not been published before.
2. Materials and methods
2.1. Location
The case study was the Ranger Uranium Mine, approximately
260 km east of Darwin, Northern Territory, Australia, surrounded by
the World Heritage Kakadu National Park (Fig. 1), which is jointly man-
aged by Traditional Custodians. The mine is adjacent to Ramsar-listed
wetlands. The climate is wet-dry tropical with approximately
1700 mm mean annual rainfall falling mostly between October and
May.
Fig. 2. Ranger Uranium Mine features including the Tailings Storage Facility (TSF), Gulungul Creek Tributary 2 (GCT2), Gulungul Creek, Land Application Areas (LAA), Borrow pit and
Tailings West Wall Sump (TWWS).
1261
Ranger Uranium Mine was an open cut mine with two pits (pit 1 and
pit 3) and currently processing is continuing from stockpiled material
(see Fig. 2). Features of the mine important here are shown in Fig. 2
and include the Tailings Storage Facility (TSF), the Tailings West Wall
Sump (TWWS), a borrow pit, numerous Land Application Areas (LAA)
and Gulungul Creek Tributary 2 (GCT2) that flows into Gulungul Creek
during the wet season.
Field spectrometry and salt samples for XRD and XRF analyses were
collected to the western side of the tailings storage facility, at the tailings
west wall sump, along GCT2 and the borrow pit (refer to Fig. 3). The
samples visually appeared variable in crystallinity, texture and mois-
ture. Field sample collection sites were obtained at point 1, from an
area offsite known as Gulungul Creek Tributary 2 (GCT2), and tracked
towards the confluence of Gulungul Creek at point 5. Measurements
were also taken at a borrow pit to the SW of the tailings storage facility
(points 6–9), at the tailings west wall sump (points 10–11) and along
the track from the mine towards GCT2 (points 12–13). Sites 14–16
were co-located at site 2, clearly visible as precipitating iron-
oxyhydorixides. Representative photos are provided in Fig. 4.
Known locations of salt that had been observed along the Gulungul
Creek Tributary and in the borrow area were targeted for fieldwork as
1262 K.S. Pfitzner et al. / Science of the Total Environment 640–641 (2018) 1259–1271
Fig. 3. Location of field sample sites in relation to the tailings storage facility (TSF).
input signatures to the hyperspectral airborne data in order to identify
unknown patches of salt, should they exist. To identify the mineralogy
and elementary composition of the salts measured spectrally, these
sites were also sampled by XRD and XRF.
2.2. Field-based spectrometry
It was essential to measure reflectance spectra in the field as the
mineral assemblages were physically fragile and not sufficiently chem-
ically stable for transport to the laboratory (Crowley, 1991). Spectral
data were acquired using an ASD FieldSpec4 spectrometer (www.asdi.
com) covering 350–2500 nm with three separate detectors: visible to
near-infrared (VNIR) (350–1050 nm), shortwave infrared (SWIR) 1
(1000–1800 nm), and SWIR2 (1800–2500 nm). Sampling intervals
were 1.4 nm for the VNIR region and 2 nm for the SWIR with a spectral
resolution of 3 and 10 nm, respectively (Hatchell, 2002). The surface
salts were often patchy in size and extent, and to gain pure in situ spec-
tral signatures, salts were measured at a distance of 10 cm from the salt
surface. Measurements were made at sensor zenith of approximately
45° with a 25° fore-optic providing around 4.6 cm diameter ground
field of view. A standard sampling procedure (Pfitzner et al., 2011)
was otherwise adopted. For each spectral measurement, the location
was provided by a GPS receiver and a photograph was taken. The
weather conditions were hot (up to 39 °C), humid (~ 90% relative hu-
midity) and clear (apart from b5% cirrus cloud cover).
The field spectral signatures and XRD and XRF were sampled in Au-
gust and the hyperspectral image data collection was nearly one month
later. There was no rainfall during this period, preserving the salt crys-
tals on the ground. The September timing of the airborne data capture
was opportunistic. The capture coincided with the late dry season in
the monsoonal environment, which is an ideal time to collect data be-
cause crystals have evaporated from solution over time but not yet af-
fected by rainfall.
The raw spectra recorded were presented as reflectance and
resampled to remove atmospheric water vapour. Data were imported
into a salt spectral library and endmembers viewed using a continuum
removal (CR) to calculate absorption feature position, depth and
width (Clark and Roush, 1984). Absorption features were then identi-
fied and compared.
2.3. X-ray diffraction and X-ray fluorescence
Samples of evaporites were scraped from the surface. Care was taken
to avoid collecting underlying sediment. The sample was first placed in
a snap lock bag and then in a portable cooler. The XRD and XRF analyses
were undertaken by The Commonwealth Scientific Industrial and Re-
source Organisation (CSIRO).
For the XRD measurements, 1.5 g of the samples were ground for
10 min in a micronizing mill (McCrone) under ethanol. The resulting
slurries were oven dried at 60 °C then thoroughly mixed in an agate
mortar and pestle before being lightly back pressed into stainless steel
sample holders for XRD analysis. XRD patterns were recorded using Fe
filtered Co Ka radiation, an automatic divergence slit, a 2° anti-scatter
slit and a fast X'Celerator Si strip detector. The diffraction patterns
were recorded from 3° to 80° in steps of 0.017° 2 theta with a 0.5 s
counting time per step for an overall counting time of approximately
35 min. The diffractometer was a PANalytical X'Pert Pro (www.
panalytical.com). Qualitative analysis was performed on the XRD data
using CSIRO's in-house XPLOT and HighScore Plus (from PANalytical)
search/match software. Quantitative analysis on the soil samples were
performed on the XRD data using the commercial package
SIROQUANT from Sietronics Pty Ltd. The results were normalized to
100% and did not include estimates of unidentified or amorphous mate-
rials (Mark Raven, CSIRO, pers comm).
The samples were also characterised using XRF with ~1 g of each
oven dried (105 °C) soil sample accurately weighed with 4 g of 12–22
lithium borate flux. The mixtures were heated to 105 °C in a Pt/Au cru-
cible for 20 min to completely dissolve the sample then poured into a
32 mm Pt/Au mould heated to a similar temperature. The melt was
cooled rapidly over a compressed air stream and the resulting glass
disks were analysed on a PANalytical Axios Advanced wavelength dis-
persive XRF system using a CSIRO in-house calibration (Mark Raven,
CSIRO, pers comm).
Trace elements were measured using pressed powders. Approxi-
mately 4 g of each oven dried sample (105 °C) was accurately weighed
with 1 g of Licowax binder and mixed well. The mixtures were pressed
in a 32 mm die at 12 tons pressure and the resulting pellets were
analysed on the same system (Mark Raven, CSIRO, pers comm).
2.4. Hyperspectral airborne HyMap data and processing
HyMap, airborne hyperspectral mapper data with 124 bands from
456 to 2496 nm were flown during the late dry season (September)
for six flight-lines at 3.1 m ground sampling distance giving a mosaic
area of ~8.5 × 22 km. The raw data were pre-processed by converting
the radiance data to surface reflectance using HyCorr (Hyperspectral
Correction) and EFFORT (Empirical Flat Field Optimal Reflectance Trans-
formation) (Boardman, 1998). The HyMap sensor is a gyro-stabilised
platform with empheris data (GPS/internal navigation system informa-
tion) collected during flight overpass to allow precision georeferencing
to map coordinates. In this case, Shuttle Radar Topography Mission
(SRTM) digital elevation model (DEM) data was used in the process.
Geometric and cross track correction were applied and then the six
flight lines were mosaicked into a data cube by HyVista. Information
on the HyMap sensor is available (Cocks et al., 1998). An overlay of
mine site GIS layers confirmed precise georeferencing.
 K.S. Pfitzner et al. / Science of the Total Environment 640–641 (2018) 1259–1271 1263

Site 1 Site 2 Site 3

Site 4 Site 6 Site 7

Site 8 Site 10

Site 11 Site 12 Site 13

Fig. 4. Examples of surface expressions at the sampled sites.

The VNIR and SWIR minimum noise fraction (MNF) colour compos-
ites and decorrelation stretch images were used to visualise maximum
information for the 124 bands in a RGB three band display. A mask of
the mining footprint was made to exclude on-site mine analysis. The
processed field spectra were used as endmembers in a constrained en-
ergy minimisation (CEM) algorithm. The CEM is a whitened version of
the matched filter detector (Johnson, 2002). A finite impulse response
filter was passed through each field spectra while minimising output
energy resulting from a background other than the field spectral targets.
A covariance matrix, calculated using all of the image pixels, was used to
characterise the composite unknown background data. Results were
presented as a series of grey scale images for each field endmember.
The CEM image data were classified into 5% intervals. Pixel locations
of spectral matches were exported into a GIS environment and the loca-
tion of surface salt mapped. Fig. 5 summarises the method as a flow
chart.
1264 K.S. Pfitzner et al. / Science of the Total Environment 640–641 (2018) 1259–1271
Fig. 5. Flow chart of methodology used to identify salt from the airborne hyperspectral data.
3. Results
3.1. In situ visual descriptions
At site 1, situated at the GCT2 bore, salts were powdery, shallow and
occurred along the banks of the dry tributary bed. At site 2, downslope
from site 1, iron rich minerals were observed. Site 3 was more heavily
vegetated and sampled at the confluence of GCT2 with the “Radon
Springs Track” where a spring expressed wet ferrous material. Salts at
this site occurred along the vehicle track and were mainly visible
along the ridge between the tyre ruts. Site 4, situated at the confluence
of GCT2 with Gulungul Creek, showed crystalline salts seeded onto
twigs and leaf litter on the ground. The salts at sites 6–9 within an old
borrow pit appeared dry and expressed linearly downslope (to the SE)
in metre wide patches. The salts at sites 10–11 (tailings west wall
sump or TWWS) were crystalline, probably evaporating in situ from
the tailings wall runoff material that had collected in the sump. Sites
12–15 were measured near the tailings west wall sump “onsite” of the
mining lease but not associated with the water from tailings west wall
sump. Sites 12–15 appeared powdery.
3.2. X-ray diffraction analyses
Mineral mixtures were dominated by magnesium sulfates (Table 1).
Samples 1–13 were of salt minerals and samples 14–17 were of iron-
rich minerals. Dominant (D), codominant (CD), sub-dominant (SD),
minor (M) and trace (T) content of minerals were: N 60%, b 60%,
20–60%, 5–20% and b 5% content, respectively. The salt samples were
magnesium sulfate mixtures, either hexahydrite (MgSO4·6H2O) domi-
nant, starkeyite (MgSO4·4H2O) dominant or co-dominant of the two
minerals. Some of the samples analysed also contained trace or minor
levels of sodium/magnesium sulfates including the minerals, konyaite
[Na2Mg(SO4)2·5H2O] or bloedite [Na2Mg(SO4)2·4H2O].
Less than 5% of gypsum (CaSO4·2H2O) content was found in all sam-
ples. Some samples had up to 20% gypsum and were also characterised
by lepidocrocite (FeO OH). Magnesium sulfates can be replaced by
sodium magnesium sulfates (Sanches-Moral et al., 1998), with higher
temperatures favouring magnesium salt precipitation over sodium
phases as magnesium salts are less soluble between 30 and 45 °C
(Leduc, 2010). The relative mobility of magnesium, which is greater
than calcium, will also determine the concentration of solutions, and re-
search specifically at Ranger Uranium Mine has shown that calcium
ameliorates magnesium toxicity for some aquatic species (van Dam
et al., 2010).
The iron-rich minerals were characterised by lepidocrocite, with
minor-trace levels of goethite, starkeyite, hexahydrite, gypsum, bloedite
and konyaite. Goethite (alpha FeOOH) is yellow-brown in colour and is
a common weathering product of silicate minerals in rocks and soils. In
more acidic environments, goethite forms where oxidation is slow
enough to allow crystal growth from solution (Schwertmann et al.,
1985) and is stable above pH 4 (Van Breemen, 1988). Lepidocrocite
(gamma FeOOH) is typically bright orange and often forms associations
with goethite. It is found in soils as a result of reduction of Fe-oxide soils
due to O2 deficiency and subsequent re-oxidation (Schwertmann,
1988). The formation of goethite and lepidocrocite is favoured in solu-
tions with pH from 5 to 7 and lower carbonate ions in solution favour
lepidocrocite over goethite formation (Schwertmann, 1988; Cornell
and Schwertmann, 1966). Clay minerals were evidence of sub-surface
contamination. The sub-surface clays were not measured in the surface
spectra.
3.3. X-ray florescence analyses
Twenty eight elements were measured by XRF, with sodium the
lightest element measured and uranium the heaviest (Table 2). The fol-
lowing elements were measured but not detected, either as elements
existing but under detection limit of the used instrument, or not
existing: Co, Nb, Sb, Pr, Pb and Th. The salts contained up to 23% MgO
and up to 54% SO3, whereas the iron-rich sediments contained b6%
MgO and between 3.2 and 16.6% SO3. The SiO2 content correlated to
 K.S. Pfitzner et al. / Science of the Total Environment 640–641 (2018) 1259–1271 1265

Table 1
XRD results of the studied seventeen reflectance spectroscopy samples of saline minerals and iron-rich minerals.

Lepidocrocite
Hexahydrite

Starkeyite

Konyaite

Goethite
Gypsum

Bloedite

Chlorite
Sample
number

Quartz

Kaolin

Mica
1 M CD CD T T

2 M D M T T

3 SD D T T

4 M D SD T

5 SD D M T

6 SD M D T T T T T

7 SD CD CD T T T

8 T D T T T T

9 T D T T

10 T CD CD T M T

11 D SD T T

12 M T D T T T T

13 T D T T T

14 CD M M T CD M

15 SD T M M T D T M

16 D T M T M T

17 CD T CD M

Dominant (D) ≥ 60%, co-dominant (CD) sum N 60%, sub-dominant (SD) = 20–60%, minor (M) = 5–20%, trace (T) ≤ 5%.

the amount of quartz. The sodium and calcium content correlated with
bloedite and gypsum, respectively. Apart from sample 10, with minor
bloedite and a Na2O content of 4%, the salts had only trace amounts of
bloedite and b 2% Na2O, and there was no spatial difference in occur-
rence with distance to the mine. Minor to trace amounts of gypsum
(b20%) were detected in all samples and all had b3% CaO. There was
no evidence of a change in calcium content with distance from the
mine. Low levels of P2O5, Cl, Ti, and Sr were detected. Some samples
contained small amounts of V, Cr, Mn, Cu, Zn, Ga, As, Br and Rb and
these were present in some salts and iron-rich samples. The iron-rich
sediments contained higher aluminium oxide content corresponding
to the presence of clays, particularly kaolin in samples 15, 16 and 17.
The iron-rich materials (including sample 2, with trace lepidocrocite),
contained the only detected rare earth elements: Y; Ce; I; and Nd. Iden-
tifying and locating secondary iron-rich minerals was significant as they
stabilise iron and can fix potentially high concentrations of toxic trace
metals, and combinations of clay mineral associations may maximise
the bioavailability of metal contaminants (Schwertmann, 1988). Sam-
ples 10, 11 and 12 contained U concentrations of 0.324, 0.025 and
0.019%, respectively. The U content of Sample 10 was equivalent to
grade 5 ore, and samples 11 and 12 equivalent to waste rock. Sample
10 also had the highest proportions of sodium oxide (4%), chlorine
(0.36%), potassium oxide (0.49%) and nickel (0.006%), and was repre-
sentative of the saline precipitation from the tailings storage facility
contained in the sump of the tailings west wall sump that collects pro-
cess water.
Vanadium was only detected in samples 6 and 7, which were col-
lected at the borrow pit to the SW of the tailings storage facility. Prior
to 2007, the milling and mining process at Ranger Uranium Mine intro-
duced vanadium from catalyst material used in the preparation of sulfu-
ric acid from elemental sulfur (Baily (1984) in Noller (1991)), and may
represent process water. Vanadium is also common in uraniferous
sandstone, so may also be associated with the natural minerals.
3.4. Spectra
The spectra are graphed by the dominant mineral groups, confirmed
by XRD analysis: hexahydrite dominant, starkeyite dominant, co-
dominant hexahydrite and starkeyite spectra, and mixed spectra
(Fig. 6). Iron-rich spectra are not discussed further here due to the
focus on salts. The data are presented as reflectance (0–1) and as contin-
uum removed (CR) to highlight the absorption features and their posi-
tion, depth and width.
The spectra representing hexahydrite (samples 3, 8 and 9) in Fig. 6
show similar shapes. There were absorptions across 1130–1300 nm,
and strong water absorptions at ~ 1450 and 1940 nm. Stronger absorp-
tions occured in sample 8 and the shallowest in sample 3. Reflectance
increases from the blue region to the near infrared (NIR). Maximum re-
flectance (up to 60%) occurs at 1080 nm with a characteristic
hexahydrite “bump” with small absorptions on either side of the peak
(Crowley, 1991; Cloutis et al., 2007; Dalton, 2003). The starkeyite spec-
tra (samples 6, 12 and 13) showed similar spectral shapes within the
1266 K.S. Pfitzner et al. / Science of the Total Environment 640–641 (2018) 1259–1271

group but were different to hexahydrite. Small absorption features were

0.324
0.025
0.019
present at 490 nm and water absortpions at ~ 1450 and 1940 nm. The

U
uranium content of sample 12, detected at 0.019% by XRF, was too

0.019

0.024
small to show in the field-based spectra. Co-dominant minerals (sam-
Nd ples 1, 7 and 10) also showed water absorptions at ~1450 and

0.036

0.034
0.047
0.021
0.019
1950 nm. Water absorptions were widest in Sample 10. Sample 10
Ce

displayed the 450 nm uranyl absorption, with 80% reflectance in the

NIR. There was a small absorption at 960 nm, a peak at 1070 nm and an-

0.062
0.057
0.038
other absorption at 1200 nm, maintaining the hexahydrite bump in the
I

co-dominant spectra. The mixed spectra also maintain the hexahydrite

0.003
0.004
0.005
0.003
0.006
0.010
0.003

0.003

0.006

0.010
0.013
0.022
0.014
bump and while these samples are mixtures, hexahydrite was the dom-
Zr

inant mineral. The small absorption was at 970 nm, increasing to

0.004
0.003
0.001
~1000 nm and the right wing of the absorption was at 1180 nm. The
1400 nm water absorption was sharp and double in sample 5 and
Y

broad in sample 11, whereas the 1900 nm absorption feature was

0.003
0.008
0.001
0.002
0.002
0.003
0.002

0.002
0.002
0.001
0.005
0.008
0.007
0.004
0.002
0.002
sharp in sample 4 and broad in sample 11. There was a very small
Sr

450 nm absorption in sample 11, as a function of the charge transfer

0.001

0.004

0.002

0.004
0.003
0.003
transitions between uranium and oxygen atoms (0.025% uranium de-
Rb

tected in XRF).
0.003

0.001

0.001

0.005
0.006
0.002
0.002
Most of the NIR features seen in saline drainage minerals are vibra-
Br

tional in origin, and are representative of molecular groups, particularly

XRF results (% composition) for the same samples examined – in situ- by reflectance spectrometry. Elements with increasing atomic number from sodium to uranium.

hydroxyl and water molecules. Most of the absorption features seen in

0.003
0.002

the evaporite mineral spectra were due to water molecules that are es-
As

sential components in each mineral structure (Crowley, 1991; Dalton,

0.002

0.002

2003). While matching absorption features is a powerful technique for

Ga

mineral identification, absorption bands attributable to water are usu-

ally limited in utility because a wide range of minerals exhibit features
0.002
0.003
0.002

0.003

0.001

0.003

0.004
Zn

in this region (Cloutis et al., 2007) and atmospheric water maximises

the airborne signal. The absorption features in the ground-based spectra
0.005
0.005
0.003

0.004
0.003
0.003

0.005

near 1900 nm become progressively shallower, consistent with a reduc-

Cu

tion in water content (Dalton, 2003) with sharper, deeper absorptions

0.006

0.003

being attributable to an increase in water content. The spectral charac-

Ni

ter of these hydration features is highly dependent on the sample envi-

ronment and on the nature of water and hydroxyl (Bishop and Pieters,
Fe2O3

17.00

42.70
47.40
15.40
38.70
0.52

0.91
0.34
0.21
2.61
2.65
0.18
0.07
0.62
0.18
1.97
1.08

1995). Crystallisation structure also affects the reflectance spectra of hy-

drated magnesium sulfates (Bonello et al., 2005). A reduced intensity of
0.014
0.050
0.012
0.018
0.058
0.006
0.011

0.008
0.006

0.019
0.016
0.020
0.019
0.013
0.012

the absorption features and slight increase in overall reflectance were

Mn

consistent with a reduction in water but could also be achieved by re-

0.004

0.006

0.007
0.006

ducing the grain size or otherwise increasing scattering (Dalton,

Cr

2003). More hydrated magnesium sulfate minerals generally show

broader absorption features (Crowley, 1991; Cloutis et al., 2007) but
0.004
0.006

soil moisture, colour and roughness also affect reflectance (Drake,

V

1995).
0.06
0.06
0.11
0.10
0.09
0.18
0.05
0.02

0.04
0.02
0.10
0.01
0.11
0.18
0.22
0.15

The ultraviolet-blue region between 350 and 460 nm were lost

Ti

when the field spectra were resampled to the HyMap spectral range.
1.58
2.58
0.57
0.79
0.49
0.91
0.84
1.05
0.41
0.27
0.14
2.07
1.21
2.04
0.84
0.13
0.29
CaO

This was significant for the spectrum of sample 10 but was not relevant
to magnesium sulfate mapping.
0.10
0.18
0.07
0.06
0.06
0.12
0.16
0.34
0.15
0.49
0.08
0.25
0.13
0.29
0.37
0.25
0.14
K2O

3.5. Image analysis

0.04
0.08
0.07
0.03
0.05
0.18
0.09
0.12
0.07
0.36
0.05
0.03
0.06
0.08
0.10
0.03
0.06
Cl

Given the spectral diversity of the hydrous magnesium dominated

49.20
36.70
41.90
44.40
41.70
40.50
43.10
52.50
50.10
48.30
50.20
46.10
53.40
16.60
10.50
3.21
5.00

sulfate mineral assemblages, no attempt was made to map minerals,

SO3

but rather to identify the location and extent of surface magnesium sul-
P2O5

fate dominated salts. The Ranger Uranium Mine was visible against a
0.01
0.03
0.01
0.01
0.01

0.01
0.00
0.01
0.02
0.01
0.03
0.01
0.05
0.06
0.05
0.07

natural landscape in the red-green-blue (RGB) (bands 14, 7, 1) true col-

our image (Fig. 7a). The false colour infrared-red-green (IR, R, G) (bands
16.50
16.80
21.00
15.40
15.20

11.30
39.10
25.20
SiO2

7.23
6.37

1.66
0.36
3.31
0.43
8.15
0.74
8.21

21, 14, 7; Fig. 7b) and normalized difference vegetation index (NDVI)
images (Fig. 7c) highlighted the green vegetation of the fluvial systems
Al2O3

0.89
1.25
1.49
1.44
1.27
3.32
1.19
0.54
0.13
1.35
0.19
2.59
0.28
3.02
4.58
5.12
2.93

and the footprint of the mine. A decorrelation stretch of bandwidths

similar to Landsat TM bands 7, 5, and 2 highlighted geological features
21.20
15.40
16.20
18.20
16.80
17.70
18.00
21.30
19.10
21.20
20.40
19.70
23.00
% XRF composition

MgO

(Fig. 7d). The MNF images of the VNIR (Fig. 7e) and SWIR (Fig. 7f)
5.54
3.32
0.77
1.68

data showed the variation in information content available in the air-

borne hyperspectral data.
Na2O

0.75
0.60
0.66
0.50
0.62
0.20
0.18
1.12
1.26
4.03
0.40
0.64
1.35
0.53
0.48
0.13
0.25

The “background” or non-salt signatures were calculated and the

Table 2

statistics based on the covariance matrix produced results that

10
11
12
13
14
15
16
17
1
2
3
4
5
6
7
8
9

corresponded with field observations. For the intended purpose here,

 K.S. Pfitzner et al. / Science of the Total Environment 640–641 (2018) 1259–1271 1267

Hexahydrite dominant reflectance spectra (left) and CR (right). Sample numbers 3,8,9

Starkeyite dominant reflectance spectra (left) and CR (right). Sample numbers 6,12,13

Co-dominant reflectance spectra (left) and CR (right). Sample numbers 1,7,10

Mixed minerals reflectance spectra (left) and CR (right). Sample numbers 4,5,11
Fig. 6. Salt reflectance spectra and with continuum removed (CR) for sample sites and confirmed by XRD (hexahydrite and starkeyite dominant). Sample numbers given in the upper right
of each panel.

we were interested in identifying and discriminating salt spectra from matches. The top matches, 95–100% (Fig. 8a) show salts at the tailings
other background spectra excluding the operating mine site. Fig. 8 west wall sump, GCT2, in an area to the SW of the tailings storage facility
shows the top spectral match at 5% increments to the top 80% of and the borrow area to the SSW of the tailings storage facility. These
1268 K.S. Pfitzner et al. / Science of the Total Environment 640–641 (2018) 1259–1271

Fig. 7. HyMap airborne data (a) red-green-blue (RGB) true colour, (b) false colour infrared-red-green (IR, R, G), (c) normalized difference vegetation index (NDV)I (d) Decorrelation
stretch of bandwidths similar to Landsat Thematic Mapper ™ bands 7, 5, and 2 to highlight geological features, (e) visible-near-infrared (VNIR) minimum noise fraction (MNF),
(f) short-wave infrared (SWIR) MNF.

sites were confirmed in the field. The 90–100% match (Fig. 8b) included
more pixels of the same sites as the top spectral match, as well as at a
controlled release point from the retention pond number 1 known as
Djalkmarra release point. Fig. 8c highlights these main areas of salts
mapped around the mine. Un-verified locations were spectrally
matched at tolerances b90% of the top matches. These areas include
other sites to the west of the tailings storage facility, within seasonal bil-
labongs (Corridor Billabong) and along the Gulungul tributary that runs
south of the mine tailings storage facility, which has previously shown
elevated EC. At thresholds b80% of maximum, saline minerals were indi-
cated in other tributaries in proximity to the mine including Corridor
and Magela Creeks. Away from the mine, pixels were also highlighted
north of the mine in Magela Creek towards the populated small commu-
nity, Mudjinberri (Fig. 8d). Other matches at lower thresholds in the re-
gion were shown but not mapped.
4. Discussion
Mine sites are often remote, reducing the ability for them to be mon-
itored at suitable spatial and temporal scales. Analysis of remotely
sensed hyperspectral data offers a means to monitor saline drainage dis-
charges both onsite and for large regions offsite. This can provide the
spatial location of surface expression of salts prior to the potential detec-
tion of elevated EC readings in creeks and tributaries, enabling the
potential for management intervention preceding potential aquatic
contamination.
Magnesium sulfate efflorescence on the ground, such as hexahydrite
and starkeyite, can be identified and delineated with airborne
hyperspectral technology. To determine the mineralogy and elemental
composition of the surface efflorescence, XRD and XRF were required al-
though, for many minerals, spectroscopy is not only more rapid but also
more sensitive to lower concentrations than XRD (Crowley, 1991;
Drake, 1995). The main spectral features in the signatures are due to
water and the crystalline structure. Although the magnesium sulfate
minerals do not give their own unique VNIR-SWIR signature due to var-
iation in physical and chemical attributes, they are unique in the spec-
tral space and can be identified and detected from hyperspectral data.
Matching absorption features is a powerful technique for mineral iden-
tification in hyperspectral data. Field-based spectrometry was required
to characterise the variation in spectra due to factors such as mineral-
ogy, water content, crystallinity and the presence of accessory elements
to build an in situ spectral knowledge prior to being able to identify
these signatures from airborne data. Public domain spectral libraries
and pure mineral spectra obtained from the laboratory are currently un-
suitable for magnesium sulfates. For future use of airborne
hyperspectral monitoring, it may be that XRD and XRF are not required
as a localised magnesium sulfate spectral library is developed at Ranger
Uranium Mine. Here we provide magnesium sulfate dominant spectra
indicative of this tropical environment, unlike the Mars reference
 K.S. Pfitzner et al. / Science of the Total Environment 640–641 (2018) 1259–1271 1269

Fig. 8. Map of the location of salts obtained from matching field spectra to the HyMap data (a) Top 95%, (b) top 90%, (c) top 80% and (d) subset of main area. Green pixels have not been field
verified. Image base layer is 2315 nm (a-c) is band 2315 nm of the HyMap data, Image base layer (figure d) is an aerial photograph to show the detailed location of salts.

spectra that is often synthesised and measured at temperatures and/or
pressures of the Martian environment (Cloutis et al., 2006; Cloutis
et al., 2007; Chou and Seal, 2007; Chou et al., 2013; Bishop and
Pieters, 1995; Bishop et al., 2005; Bonello et al., 2005; Brown and
Cudahy, 2006; Leftwich et al., 2013; McCord et al., 1998; McCord et al.,
1999; Roach et al., 2009; Vaniman et al., 2004; Vaniman and Chipera,
2006; Dalton, 2003).
Previous studies identified the salt along the Gulungul Tributary as
epsomite (Brazier and Tayler, 2008; Milnes, 1986; Fitzpatrick, 1986),
but epsomite was not detected in the XRD results. The absence of
epsomite could be due to a change in hydration during transport to
the laboratory for X-ray analyses, however, the high VNIR spectral sig-
nature observed in epsomite in published public domain spectra
(Clark et al., 2007; Kokaly et al., 2017; Baldridge et al., 2009) were not
1270 K.S. Pfitzner et al. / Science of the Total Environment 640–641 (2018) 1259–1271
present during in situ spectral measurement, but rather the lower re-
flectance magnitude of hexahydrite and starkeyite were observed.
The mine site needed to be masked out as the inclusion of the mine
saturated the spectral results. The mine site is known to express magne-
sium sulfate salts as a result of the waste rock mineralogy (Brazier and
Tayler, 2008; Milnes, 1986; Fitzpatrick, 1986; Milnes et al., 1986). The
inclusion of the operating mine site dominated the results with signa-
tures attributable to Pit 1 (which was being infilled with tailings at the
time of airborne overpass) and waste rock. Therefore, to better investi-
gate salts off-site the mine site needed to be masked out.
The field spectra were used for spectral matching. A number of algo-
rithms were trialled including Matched Filtering, Constrained Energy
Minimisation (CEM), Adaptive Coherence Estimator and Spectral
Angle Mapper, although these method comparisons are not outlined
here. The data were also spectrally subset into VNIR, SWIR and full
datasets prior to spectral matching processing, and PCA and MNF trans-
formations were trialled. The best results obtained were with the CEM
algorithm. CEM has previously been used to successfully map waste
products of mining (Farrand and Harsanyi, 1997; Elbakary and Alam,
2008) and was deemed to be the most appropriate algorithm here.
A challenge with mapping magnesium sulfate salts from saline
drainage with hyperspectral data is that the process of crystallisation
is on short time scales of evaporative concentration and hydration/de-
hydration. These sulfate minerals precipitate from solution with the hy-
dration states depending on factors of temperature, relative humidity
and the chemistry concentration from which they form (Crowley,
1991; Chou and Seal, 2007; Chou et al., 2013; Vaniman et al., 2004).
Timing of capture is best suited before the onset of rain. In the mon-
soonal tropics, there is a regular and predictable window to obtain
data before the wet season. With repeat surveys, patches of efflores-
cence that were previously unknown may be identified, pulses can be
tracked and, areas of previous concern may be monitored for changes.
The HyMap system has recently been upgraded and repeat surveys
can co-acquire HyCam data (high resolution digital aerial photography)
at 15 cm ground spatial distance and 1 m digital terrain model. The
HyCam data can provide further context and assist in interpreting the
HyMap data.
Using ground-based spectrometry, we could detect salts from pro-
cess water due to a uranyl absorption feature, which was verified as ura-
nium in the XRF results. Uranium cannot be detected using airborne
hyperspectral data alone due to the atmospheric interference in the ul-
traviolet region of the spectrum, but hyperspectral data can be com-
bined with airborne gamma spectrometry data for assessing
environmental issues from uranium mining (Pfitzner and Martin,
2003; Pfitzner and Clifton, 2006b). Field-based spectrometry is a quick
and useful measurement for detecting uranium in minerals, even at
concentrations as low as 0.3% content.
While hyperspectral instruments can be mounted on remotely
piloted air systems, their spectral and spatial coverage is currently lim-
ited compared to commercial airborne sensors such as HyMap. It is
likely in the future that hyperspectral data will be collected from re-
motely piloted air systems as range capabilities of the technology in-
creases. The parallel development of hyperspectral sensing and
remotely piloted air system technologies will provide wide-range op-
portunities for the monitoring of environmental degradation from salin-
ity impacts. For example, a similar technique could be applied to coal
mines with a sodium-based salt investigation. Dryland salinity has had
massive environmental impacts across the globe and where it occurs,
there is potential to map direct salt efflorescence rather than indicators
of saline drainage, which would help with targeted management and
early-intervention. Acid mine drainage environments have already
been mentioned as suitable for mineral mapping by hyperspectral
data and the minerals mapped can be associated by their relevant pH
precipitation range. The information could not be gained from multi-
spectral data because limited spectral resolution inhibits differentiating
salts from sands and anthropogenic infrastructure (white pipes,
concrete, white metallic roof tops and mining tanks) that also have
bright white reflectance signatures. Also, multispectral satellite data
that includes the SWIR region such as WorldView3 data is susceptible
to cloud over, especially in September at the study site. In the
hyperspectral data, the salt minerals were spectrally identifiable due
to polyhedral water and moisture in the spectral features that were ab-
sent from the bright white non-salt materials.
Note that a number of in situ and laboratory toxicity tests were con-
ducted in Gulungul Creek during elevated EC events arising from the
contaminated areas identified to the west of the tailings storage facility,
and showed no significant response from exposure to increased con-
ductivity water and that there was no effect on human health. The
source of the efflorescence was most likely the waste rock walls of the
tailings storage facility and an inception trench and sumps have since
been engineered to collect run-off (Energy Resources of Australia Ltd,
2017).
5. Conclusion
Application of ground-based spectrometry, XRD, XRF and airborne
hyperspectral remote sensing data to map magnesium sulfates over a
large area offers a quick solution for management intervention in an op-
erating or rehabilitated mine when monitoring and assessing efflores-
cence as a result of saline drainage. Large areas of land affected by
mining and remote mine areas can be routinely monitored for efflores-
cence using hyperspectral airborne data when combined with an initial
ground survey. Here we have shown the identification and discrimina-
tion of mineral efflorescence as a result of mining from hyperspectral
data is not limited to acid mine drainage environments but is also appli-
cable to saline drainage environments. With the advancing technologies
of remotely piloted air systems, it is likely that this technology will be
more widely used for routine detection and monitoring in the future.
Acknowledgements
John Miller, Kate Turner, Sean Fagan (Supervising Scientist Branch)
and Ben McTavish (previously of Energy Resources of Australia) assisted
with the fieldwork. John Miller provided comments to this paper. Mark
Raven (CSIRO) performed XRD and XRF on the field samples.
References
Baily, P.A., August 1984. Some Design and Operating Aspects of the Ranger Uranium Mine
Treatment Plant. Proceedings AusIMM Conference, Darwin, NT, Parkville, Victoria,
pp. 183–191.
Baldridge, A.M., Hook, S.J., Grove, C.I., Rivera, G., 2009. The ASTER spectral library version
2.0. Remote Sens. Environ. 113, 711–715.
Belluco, E., Camuffo, M., Ferrari, S., Modenese, L., Silvestri, S., Marani, A., Marani, M., 2006.
Mapping salt-marsh vegetation by multispectral and hyperspectral remote sensing.
Remote Sens. Environ. 105 (1), 54–67.
Bierwirth, P.N., Pfitzner, K.S., 2001. Identifying Acid Mine Drainage Pollution at Captains
Flat, NSW, Using HyMap Data. Proceedings IEEE. International Geoscience and Re-
mote Sensing Symposium, 9–13 July 2001. University of NSW, Sydney, Australia.
Bilgili, A.V., Cullu, M.A., van Es, Harold, Aydemir, A., Aydemir, S., 2011. The use of
hyperspectral visible and near infrared reflectance spectroscopy for the characteriza-
tion of salt-affected soils in the Harran Plain, Turkey. Arid Land Res. Manag. 25 (1),
19–37.
Bishop, J.L., Dyar, M.D., Lane, M.D., Banfield, J.F., 2005. Spectral identification of hydrated
sulfates on Mars and comparison with acidic environments on earth. Int.
J. Astrobiol. 3, 275–285.
Bishop, J.L., Pieters, C.M., 1995. Low-temperature and low atmospheric pressure infrared
reflectance spectroscopy of Mars soil analog materials. J. Geophys. Res. 100 (E3),
5369–5379.
Boardman, J.W., 1998. Post-ATREM polishing of AVIRIS apparent reflectance data using
EFFORT: a lesson in accuracy versus precision. Summaries of the 7th JPL Airborne
Earth Science Workshop, JPL. Publica 97-21, 53.
Bonello, G., Berthet, P., D'Hendercourt, L., 2005. Identification of magnesium sulfate hy-
dration state derived from NIR reflectance spectroscopy. Lunar and Planetary Science
XXXVI.
Brazier, J., Tayler, K., 2008. Characterisation of Evaporates in a Drainage Channel Outside
the Fence on the Western Side of the Ranger TSF. Office of the Supervising Scientist.
 K.S. Pfitzner et al. / Science of the Total Environment 640–641 (2018) 1259–1271
Brown, A.J., Cudahy, T.J., 2006. Hyperspectral Imaging of Sulfate Evaporate Deposits in
Western Australia and on Mars. Electronic Imaging. International Society for Optics
and Photonics.
Canedo-Arguelles, M., Hawkins, C.P., Kefford, B.J., Schaffer, R.B., Dyack, B.J., Brucet, S.,
Buchwalter, D., Dunlop, J., Frör, O., Lazorchak, J., Coring, E., Fernandez, H.R.,
Goodfellow, W., González Achem, A.L., Hatfield-Dodds, S., Karimov, B.K., Mensah, P.,
Olson, J.R., Piscart, C., Prat, N., Ponsá, S., Schulz, C.J., Timpano, A.J., 2016. Saving fresh-
water from salts: ion-specific standards are needed to protect biodiversity. Science
351 (6276):914–916. https://doi.org/10.1126/science.aad3488.
Chou, I., Seal, R.R., 2007. Magnesium and calcium sulfate stabilities and the water budget
of Mars. J. Geophys. Res.: Planets 112 (E11) (1991–2012).
Chou, I.M., Seal, R.R., Wang, A., 2013. The stability of sulfate and hydrated sulfate minerals
near ambient conditions and their significance in environmental and planetary sci-
ences. J. Asian Earth Sci. 62, 734–758.
Clark, R.N., King, T.V.V., Klejwa, M., Swayze, G.A., Vergo, N., 1990. High spectral resolution
reflectance spectroscopy of minerals. J. Geophys. Res. 95, 12653–12680.
Clark, R.N., Roush, T.L., 1984. Reflectance spectroscopy: quantitative analysis techniques
for remote sensing applications. J. Geophys. Res. 89 (B7), 6329–6340.
Clark, R.N., Swayze, G.A., Wise, R., Livo, K.E., Hoefen, T.M., Kokaly, R.F., Sutley, S.J., 2007.
USGS Digital Spectral Library spli06a US Geological Survey, Data Series 231.
Cloutis, E.A., Craig, M.A., Mustard, J.F., Kruzelecky, R.V., Jamroz, W.R., Scott, A., Bish, D.L.,
Poulet, F., Bibring, J.-P., King, P.L., 2007. Stability of hydrated minerals on Mars.
Geophys. Res. Lett. 34 (20).
Cloutis, E.A., Hawthorne, F.C., Mertzman, S.A., Krenn, K., Craig, M.A., Marcino, D., Methot,
M., Strong, J., Mustard, J.F., Blaney, D.L., Bell III, J.F., Vila, F., 2006. Detection and dis-
crimination of sulfate minerals using reflectance spectroscopy. Icarus 184 (1),
121–157.
Cocks, T., Jenssen, R., Stewart, A., Wilson, I., Shields, T., October 1998. The HyMapTM Air-
borne Hyperspectral Sensor: The System, Calibration and Performance. 1st Earsel
Workshop on Imaging Spectroscopy, Zurich (6 pages).
Cornell, R.M., Schwertmann, U., 1966. The Iron Oxides. Structure, Properties, Reactions,
Occurrence and Uses. VCH, Weinheim, Germany, p. 573.
Crowley, J.K., 1991. Visible and near-infrared (0.4-2.5 μm) reflectance spectra of playa
evaporite minerals. J. Geophys. Res. 96 (B10:16) (231-16, 240, September 10, 1991).
Dalton, J.B., 2003. Spectral behaviour of hydrated sulfate salts: implications for Europa
mission spectrometer design. Astrobiology 3 (4), 771–784.
van Dam, R.A., Hogan, A.C., McCullough, C.D., Houston, M.A., Humphrey, C.L., Harford, A.J.,
2010. Aquatic toxicity of magnesium sulfate, and the influence of calcium, in very low
ionic concentration water. Environ. Toxicol. Chem. 29:410–421. https://doi.org/
10.1002/etc.56.
van Dam, R.A., Humphrey, C.L., Harford, A.J., Sinclair, A., Jones, D.R., Davies, S., Storey, A.W.,
2013. Site-specific water quality guidelines: 1. Derivation approaches based on phys-
icochemical, ecotoxicological and ecological data. Environ. Sci. Pollut. Res. 21 (1),
118–130.
Dehaan, R.L., Taylor, G.R., 2002. Field-derived spectra of salinized soils and vegetation as
indicators of irrigation-induced soil salinization. Remote Sens. Environ. 80 (3):
406–417. https://doi.org/10.1016/S0034-4257(01)00321-2.
Drake, N.A., 1995. Reflectance spectra of evaporite minerals (400–2500 nm): applications
for remote sensing. Int. J. Remote Sens. 16 (14), 2555–2571.
Elbakary, M.I., Alam, M.S., March 17, 2008. Mine detection in multispectral imagery data
using constrained energy minimization. Proceedings of SPIE - The International Soci-
ety for Optical Engineering. Optical Pattern Recognition XIX, 697707 https://doi.org/
10.1117/12.781735.
Energy Resources of Australia Ltd, 26 August 2017. Ranger Wet Season Report 2016/17.
Farifteh, J., van der Meer, F., van der Meijde, M., Atzberger, C., 2008. Spectral characteris-
tics of salt-affected soils: a laboratory experiment. Geoderma 145, 196–206.
Farrand, W.H., Harsanyi, J.C., 1997. Mapping the distribution of mine tailings in the Coeur
d'Alene river valley, Idaho, through the use of a constrained energy minimization
technique. Remote Sens. Environ. 59 (1), 64–76.
Fitzpatrick, R., 1986. Chapter 2. Morphology, Classification and Genesis of “Mine Soils” on
the Rock Dumps in: Rehabilitation of Waste Rock Dumps Ranger Mine. CSIRO Divi-
sion of Soils, Adelaide.
Grove, C.I., Hook, S.J., Paylor III, E.D., 1992. Laboratory Reflectance Spectra of 160 Minerals,
0.4 to 2.5 Micrometres. Jet Propulsion Laboratory Publication. 92-2, Pasadena, CA,
United States.
Hatchell, D.C., 2002. Analytical Spectral Devices. 3rd Edition. Analytical Spectral Devices
(140 pages).
Hunt, G.R., Ashley, R.P., 1979. Spectra of altered rocks in the visible and near infrared.
Econ. Geol. 74, 1613–1629.
Hunt, G.R., Salisbury, J.W., Lenhoff, C.J., 1971. Visible and near-infrared spectra of minerals
and rocks. VI. Sulfides and sulfates. Mod. Geol. 3, 1–14.
Johnson, S., 2002. Constrained energy minimization and the target-constrained
interference-minimised filter. Opt. Eng. 42 (6), 1850–1854.
Kokaly, R.F., Clark, R.N., Swayze, G.A., Livo, K.E., Hoefen, T.M., Pearson, N.C., Wise, R.A.,
Benzel, W.M., Lowers, H.A., Driscoll, R.L., Klein, A.J., 2017. USGS Spectral Library Ver-
sion 7. U.S. Geological Survey, Data Series 1035.
1271
Kurbiel, J., Balcerzak, W., Rybicki, S.M., Swist, K., 1996. Selection of the best desalinisation
technology for highly saline drainage water from coal mines in southern Poland.
Desalinisation 106, 415–418.
Leduc, E.M.S., 2010. Hydrated Sodium-Magnesium Sulfate Minerals Associated with In-
land Saline Systems. (Master's Thesis). Queens University, Canada.
Leftwich, K., Bish, D.L., Chen, C.H., 2013. Crystal structure and hydration/dehydration be-
havior of Na2Mg (SO4)2·16H2O: a new hydrate phase observed under Mars-relevant
conditions. Am. Mineral. 98 (10), 1772–1778.
Lottermoser, B., 2010. Mine Wastes: Characterization, Treatment and Environmental Im-
pacts. Springer-Verlag, Berlin, Germany.
Lottermoser, B.G., Ashley, P.M., 2005. Tailings dam seepage at the rehabilitated Mary
Kathleen uranium mine, Australia. J. Geochem. Explor. 85 (3), 119–137.
McCord, T.B., Hansen, G.B., Fanale, F.P., Carlson, R.W., Matson, D.L., Johnston, T.V., Smythe,
W.D., Crowley, J.K., Martin, P.D., Ocampo, A., Hibbitts, C.A., Granahan, J.C., 1998. Salts
on Europa's surface detected by Galileo near infrared mapping spectrometer. Science
280 (5367), 1242–1245.
McCord, T.B., Hansen, G.B., Matson, D.L., Johnston, T.V., Crowley, J.K., Fanale, F.P., Carlson,
R.W., Smythe, W.D., Martin, P.D., Hibbitts, C.A., 1999. Hydrated salt minerals on
Europa's surface from the Galileo near-infrared mapping spectrometer (NIMS) inves-
tigation. Journal of Geophysical Research: Planets 104 (E5), 11827–11851
(1991–2012).
Milnes, A.R., 1986. Rock Weathering in the Waste Dumps at the Ranger Project Area.
Focus Report 3 to Ranger Uranium Mines Pty Ltd.
Milnes, A.R., Riley, G.G., Raven, M.D., 1986. Chapter 1: Rock weathering, landscape devel-
opment and the fate of uranium in waste rock dumps and the low grade stockpile.
Rehabilitation of Waste Rock Dumps Ranger Mine. CSIRO Division of Soils, Adelaide.
Noller, B.N., 1991. Non-radiological contaminants from uranium mining and milling at
ranger, Jabiru, Northern Territory. Environ. Monit. Assess. 19, 383–400.
Nordstrom, D.K., Blowes, D.W., Ptacek, C.J., 2015. Hydrogeochemistry and microbiology of
mine drainage: an update. Appl. Geochem. 57, 3–16.
Pfitzner, K.S., Bartolo, R., Carr, G., Esparon, A., Bollhoefer, A., 2011. Standards for Reflec-
tance Spectral Measurement of Temporal Vegetation Plots. Department of Sustain-
ability, Environment, Water, Populations and Communities. Supervising Scientist
Report.
Pfitzner, K., Clifton, R., 2006a. Integration of airborne CASI and gamma ray data for mine
site characterisation. J. Spat. Sci. 51 (2).
Pfitzner, K., Clifton, R., 2006b. Integration of airborne CASI and gamma ray data for mine
site characterisation. J. Spat. Sci. 51 (2).
Pfitzner, K., Martin, P., 2003. Mapping the South Alligator River Valley Using Integrated
Datasets. Internal Report 444. Australian Government, Department of the Environ-
ment and Heritage, Supervising Scientist, Canberra.
Rao, B., Sharma, R.C., Ravi Sankar, T., Das, S.N., Dwivedi, R.S., Thammappa, S.S.,
Venkataratnam, L., 1995. Spectral behaviour of salt-affected soils. Int. J. Remote
Sens. 16 (12), 2125–2136.
Roach, L.H., Mustard, J.F., Murchie, S.L., Bibring, J.P., Forget, F., Lewis, K.W., Aharonson, O.,
Vincendon, M., Bishop, J.L., 2009. Testing evidence of recent hydration state change in
sulfates on Mars. J. Geophys. Res.: Planets 114 (E2) (1991–2012).
Sanches-Moral, S., Salvador, O., Garcia-Del-Cura, M.A., Canaveras, J.C., 1998.
Penecontemporaneous diagenesis in continental saline sediments: bloeditization in
Quero playa lake (La Mancha, Central Spain). Chem. Geol. 149, 189–207.
Schwertmann, U., 1988. Some properties of soil and synthetic iron oxides. In: Stucki, J.W.,
Goodmann, B.A., Schwertmann, U. (Eds.), Iron in Soils and Clay Minerals. 217. Nato
Advanced Institute. Dordrecht, Neth.: Reidel Publishing Company, pp. 203–205.
Schwertmann, U., Cambier, P., Murad, E., 1985. Properties of goethites of varying crystal-
linity. Clay Clay Miner. 33, 369–378.
Sholkovitz, E., 1976. Flocculation of dissolved organic and inorganic matter during the
mixing of river water and seawater. Geochim. Cosmochim. Acta 40 (7), 831–845.
Swayze, G.A., Smith, K.S., Clark, R.N., Sutley, S.J., Pearson, R.M., Vance, S.J., Hageman, P.I.,
Briggs, P.H., Meier, A.L., Singleton, M.J., Roth, S., 2000. Using imaging spectroscopy
to map acidic mine waste. Environ. Sci. Technol. 34, 47–54.
The International Network fur Acid Prevention (INAP), 2009. Global acid rock drainage
guide (GARD guide). http://www.gardguide.com.
Van Breemen, N., 1988. Redox processes of iron and sulfur involved in the formation of
acid sulfate soils. In: Stucki, J.W., Goodman, B.A., Schwertmann, U.D. (Eds.), Iron in
soils and clay minerals. Reidel Publishing Co., Dordrecht, The Netherlands.
Vaniman, D.T., Bish, D.L., Chipera, S.J., Fialips, C.I., Carey, J.W., Feldman, W.C., 2004. Magne-
sium sulfate salts and the history of water on Mars. Nature 431, 663–665.
Vaniman, D.T., Chipera, S.J., 2006. Transformations of Mg- and Ca sulfate hydrates in Mars
regolith. Am. Mineral. 91, 1628–1642.
Verburg, R., Bezuidenhout, N., Chatwin, T., 2009. The global acid rock drainage guide
(GARD guide). Mine Water Environ. 28, 305.
Verma, K.S., Saxena, R.K., Barthwal, A.K., Deshmukh, S.N., 1994. Remote sensing technique
for mapping salt affected soils. Int. J. Remote Sens. 15 (9), 1901–1914.