Atmospheric Pollution Research xxx (2015) 1e7

H O S T E D BY

Contents lists available at ScienceDirect

Atmospheric Pollution Research

journal homepage: http://www.journals.elsevier.com/locate/apr

Original article

Integrating dispersion modelling and lichen sampling to assess

harmful heavy metal pollution around the Karabash copper smelter,
Russian Federation
Adam S. Pollard a, b, c, *, Ben J. Williamson d, Mark Taylor e, William O. Purvis d,
Marc Goossens f, Stefan Reis g, Pavel Aminov h, Valery Udachin h, Nicholas J. Osborne c, i, j
a

Royal Cornwall Hospitals NHS Trust, Truro, Cornwall, TR1 3LJ, UK

Pollard Systems Ltd, Mevagissey, Cornwall, PL26 6TL, UK
European Centre for the Environment and Human Health, University of Exeter Medical School, Truro, TR1 3HD, UK
d
University of Exeter, Camborne School of Mines, Penryn, Cornwall, TR10 9EZ, UK
e
Public Health Dept, Fakulta Zdravotnictva a Socialnej Prace, University of Trnava, Slovakia
f
Geosense, Den Ham, The Netherlands
g
NERC Centre for Ecology & Hydrology, Penicuik, Midlothian, EH26 0QB, UK
h
Institute of Mineralogy, Russian Academy of Science, Miass, Russia
i
Department of Paediatrics, University of Melbourne, Flemington Road, Parkville, Melbourne, Australia
j
Department of Clinical Pharmacology, Sydney Medical School, University of Sydney, Sydney, NSW, Australia
b
c

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 16 November 2014
Received in revised form
15 April 2015
Accepted 17 April 2015
Available online xxx

Cu smelting has had a severe impact on the environment around the town of Karabash, Russia.
Dispersion modelling has been carried out to estimate atmospheric fallout of metals and metalloids
(henceforth termed metals for brevity) from the copper smelter, calibrated and ground-truthed using
metal concentrations in transplanted and naturally growing lichens.
Lichens (Hypogymnia physodes) were transplanted in June 2011 from a relatively little impacted
reference site to stations in NEeSW and WeE transects centred on the smelter at Karabash. The
transplants were removed during September 2011 and then analysed for As, Cu and Pb. The results were
compared with deposition loads estimated using TAPM modelling which was carried out for particles of
various sizes, and with simple chemistries, and with the smelter conceptualised as a continuously
emitting point source.
Variation in the ratio of lichen divided by modelled concentrations was lowest for Pb, ranging from
30.3 to 939.9 and 4.9e107.8 for PM2.5 and PM10, respectively, across eight sample points.
The TAPM modelling is in agreement with previous studies that smelter emissions are the major
source of environmental Pb deposition around Karabash. Further modelling will be required to determine whether the Pb in the lichens is largely sourced from current smelter airborne emissions or
windblown soil particles containing historic additions of Pb.
Copyright 2015 Turkish National Committee for Air Pollution Research and Control. Production and
hosting by Elsevier B.V. All rights reserved.

Keywords:
Air pollution
Deposition
Heavy metals
Lichens
Environmental impact assessment

1. Introduction

* Corresponding author. Research Development and Innovation, Royal Cornwall

Hospitals NHS Trust, Truro, TR1 3HD, UK. Tel.: 44 (0) 1872 258163; fax: 44 (0)
1872 256420.
E-mail address: Adam.pollard@rcht.cornwall.nhs.uk (A.S. Pollard).
Peer review under responsibility of Turkish National Committee for Air Pollution Research and Control.

Copper smelting, particularly where pollution control measures

are inadequate, results in emissions to air of SO2 (Newhook et al.,
2003; Williamson et al., 2008) and particulate matter containing
a range of metals such as As (Zhao et al., 2010), Pb (Swarup and
Dwivedi, 2012), Cu, Cd (Cuypers et al., 2012) and Cr (Yeganeh
et al., 2012). These are potentially harmful to human respiratory
health and may accumulate through the food chain (Garrow et al.,
2000). Copper-bearing and other metallic compounds have been

http://dx.doi.org/10.1016/j.apr.2015.04.003
1309-1042/Copyright 2015 Turkish National Committee for Air Pollution Research and Control. Production and hosting by Elsevier B.V. All rights reserved.

Please cite this article in press as: Pollard, A.S., et al., Integrating dispersion modelling and lichen sampling to assess harmful heavy metal
pollution around the Karabash copper smelter, Russian Federation, Atmospheric Pollution Research (2015), http://dx.doi.org/10.1016/
j.apr.2015.04.003

A.S. Pollard et al. / Atmospheric Pollution Research xxx (2015) 1e7

measured at potentially harmful levels in cultivated crops (Yang

et al., 2002) and in commonly consumed wild foodstuffs such as

rek
fungi collected close to smelters (Svoboda et al., 2000; Koma
et al., 2007). The human ingestion of heavy metals such as Pb and
Cd has been shown to cause a wide variety of health problems
m et al., 2012),
which can include damage to the skeletal (Engstro
renal and reproductive systems (World Health Organization, 1996).
Problems associated with Pb, and other heavy metal exposures, can
arise at all stages of life, from foetal development onwards (Kantola
et al., 2000; Araya et al., 2006).
An area well documented as experiencing the environmental
effects of Cu smelting is Karabash in the Ural Mountains of Russia,
where large scale smelting has been carried out since the 1830s
(Spiro et al., 2004; Russia Channel, 2013). The spatial pattern of
airborne particulate fallout from the Karabash smelter has been
assessed from the chemical analysis of lichens found growing
naturally and transplanted to and retrieved from locations at distances of up to 33 km from the smelter (Williamson et al., 2004;
Spiro et al., 2012). Further to evidence being drawn depicting
gross contamination for up to 10 km from the smelter (Tote et al.,
2014), anomalies in public health have caused concern, e.g. ndings
of very high levels of Pb within hair sampled from local children,
and As, Pb and Zn in locally grown vegetables signicantly above
Russian maximum allowable concentrations (Revich, 2010).
The use of lichens in biomonitoring is well established for a
variety of pollutants including fallout of heavy metals from
smelters (Conti and Cecchetti, 2001). It is comparatively low cost
and is discrete which is ideal for forensic applications and may
avoid issues of equipment tampering. Its application is particularly
favourable in spatially extensive surveys of remote regions where it
is impractical or too costly to employ large numbers of recording/
sampling devices (Nimis et al., 2002). Beyond simple mapping of
pollutant levels in the atmosphere, their fallout and impacts, researchers have shown correlations between lichen bioindicators
and locally elevated levels of serious illnesses including cancer
(Cislaghi and Nimis, 1997). Various methods of lichen biomonitoring may be applied including assessments of species
abundance, diversity and health, and chemical and isotopic
composition. Where lichens are not naturally present (e.g. the
environment is too contaminated), transplants may be used. These
are sampled from a background or reference site and transplanted to monitoring stations for a xed exposure period. The
lichen transplants can then be analysed, and their compositions
compared with samples from the reference site to determine and/
or map relative levels and sources of metals and other chemical
species incorporated over a known time period.
To our knowledge this is the rst time that lichen biomonitoring
techniques have been combined with contemporary computer
modelling. The benets of this combined approach are that computer modelling, once ground-truthed, is relatively rapid and
cheap, can be carried out remotely, avoiding political and sociological sensitivities (Purvis et al., 2013), problems with site accessibility and unfavourable weather conditions. Lichen and other
similar biomonitoring methodologies may be compromised by
heavy snow or rain, snow effectively excluding lichen surfaces from
atmospheric deposition and heavy rain potentially washing particulate off lichen surfaces (Purvis et al., 2013). Lichen monitoring
also ideally requires standard conditions at lichen monitoring stations for data comparability, which are not always present,
including broadly similar habitats (woodland density and type etc.),
phytogeographical zones and transplant station/tree characteristics
(tree species, height and girth). In addition, lichens should not be
regarded as inert deposimeters or passive lters of airborne
pollutants, and lichen biomonitoring is not quantitative as there are
a number of factors which can affect particle deposition and

retention of metals and other species within lichen tissues and on

their surfaces (Purvis et al., 2013).
For the computer modelling aspect of the study, we have applied
the atmospheric dispersion model TAPM [CSIRO, Canberra,
Australia, (Hurley et al., 2005)] in order to predict the magnitude of
dry deposition from emissions of the Karabash smelter. With the
exception of mercury, heavy metals emitted from industrial production and combustion processes are typically bound to particulate matter. Dispersion and deposition of PM10 and PM2.5 have been
used as an analogue for metal dispersion and deposition rates. This
approach was successfully used in a recent study which found high
levels of metals in the ne and ultrane airborne particle fractions

lez-Castanedo
around a copper smelter near Huelva, Spain (Gonza
et al., 2014).
The aim of this study was to test the TAPM model for predicting
the spatial distribution of particulate metal fallout around the
Karabash smelter, Russian Federation. This was achieved by
comparing the results with measured As, Cu and Pb contents of
naturally growing and transplanted lichens (Hypogymnia physodes),
collected at various distances from the smelter.
2. Materials and methods
We used TAPM, an atmospheric transport model, to simulate
the near-source atmospheric transport and dry deposition of particulate matter-bound metal emissions around the Karabash
smelter (54 42 N, 52 34 E) (Fig. 1). Meteorological data, which
underpins the TAPM model, was examined for the period 10th June
to 16th September 2011. A modelling framework evolved which
used local weather data and, in the absence of reliable and accurate
emissions data for the smelter, an estimated emissions rate, based
on production data (DKL Engineering, 2012), rened to

Fig. 1. Sites at which lichens were transplanted around the smelter at Karabash and
which lay within the solution grid (grid spacing 2 km). The smelter lies at the centre of
this grid depicted by the yellow square (source: TAPM, CSIRO). The green areas
represent land whereas bodies of surface water are shown as blue. This graphic is
taken directly from TAPM. Overlain is a radial plot of the wind direction, (in three
gure compass bearings and rounded to the nearest 22.5 ) relative to the smelter
taken hourly throughout the model period. Wind direction is predominantly from the
NW.

Please cite this article in press as: Pollard, A.S., et al., Integrating dispersion modelling and lichen sampling to assess harmful heavy metal
pollution around the Karabash copper smelter, Russian Federation, Atmospheric Pollution Research (2015), http://dx.doi.org/10.1016/
j.apr.2015.04.003

A.S. Pollard et al. / Atmospheric Pollution Research xxx (2015) 1e7

approximate the smelter's expulsion velocity and rate at which

particulate was expelled.
Metal concentrations in the lichens (Table S1) were compared
with modelled values for dry deposition at each lichen sampling
point. Proportionality between the two measures was determined
across the points, and for the metal for which this proportionality
was strongest, the modelled deposition was converted into a
model lichen metal concentration using a published relationship.
This was performed for only one metal, purely to determine
whether the model was t for the purpose of determining its
source and whether it might be applicable to other fallout. If the
differential between modelled and actual metal lichen concentrations was consistent across several sites, this would allow inferences to be made as to the composition of the material expelled
by the smelter. This could then be feed into a rerun of TAPM
potentially providing context to the environmental impacts of the
smelter's operation.

undertaken by carrying out 5 replicate sample dissolutions and

analyses of the lichen reference material BCR-482. The average
measured concentration of Pb was 44.4 ppm (s.d. 2.5), compared
with a certied value of 40.9 ppm (s.d. 1.4), for As an average
measured concentration of 0.8 ppm (s.d. 0.2), compared with a
certied value of 0.85 (s.d. 0.07), and for Cu an average measured
concentration of 7.2 ppm (s.d. 0.4), compared with a certied
value of 7.03 (s.d. 0.19 ppm), with accuracy therefore within
10% of published values. Duplicate samples of lichen transplants
from the reference site U0 showed variations in As, Cu and Pb
concentrations of up to 40%. This is a product of natural variability
of fallout of particles onto lichen surfaces, even in lichens at close
(tens of m) proximity, which emphasises the qualitative nature of
lichen biomonitoring. Levels of Pb, Cu and As in lichen sample
solutions were always at least two orders of magnitude higher
than theoretical detection limits for ICP-MS.
2.2. Statistical analysis

2.1. Lichen transplant methodology and chemical analysis

Over 150 twigs, each colonised by at least 5 Hypogymnia physodes thalli, were transplanted during June 2011 from Betula at
Nyazminsky Ridge, the reference site (U0), to monitoring stations in
a NEeSW transect centred on the Karabash smelter, from Severnye
Peche (approximately 25 km south of Karabash, U1) to Kyshtym
(approximately 33 km northeast of Karabash, U8) (Fig. 1). This
reference site was selected as it was deemed to be relatively little
impacted by smelter emissions due to an abundance of Hypogymnia, the presence of a rich diversity of lichen species and a
vibrant understorey. The twigs were sampled using powder-free
gloves and stored in paper capsules prior to transplantation. All
transplant monitoring stations were located between 280 and
695 m in elevation above sea level, apart from site U0 at Nyazminsky Ridge (800 m), and more than 150 m from roads, apart from
site U11 (Novoandreevka). At each monitoring station, 6 Betula
trees were selected, each growing in open situations and having
similar physical characteristics (height, girth), to which two Hypogymnia-colonised twigs were tied securely to the base. All transplanted Hypogymnia-covered twigs were collected after a three
month exposure period and stored in clean polythene bags.
Samples of naturally growing Hypogymnia physodes thalli were
also collected (June and September 2011) from where sufcient
material was available for chemical analysis (ideally >200 thalli),
which was generally at sites >10 km from the Karabash smelter, i.e.
beyond the most heavily impacted zone. Naturally growing lichens
were also collected from four stations (U18-21), 3 lying WNW
(<20 km) and 1 E (<38 km) of Karabash in the Forest-Steppe Zone.
In the laboratories of the University of Exeter, Hypogymnia thalli
were removed from both the naturally growing and transplanted
twigs from each site using a stainless steel knife and gloves to
avoid contamination. Samples were bulked from each tree at each
site. The sample materials were sorted under a binocular microscope to remove extraneous materials including bark, twigs,
leaves, and lichens of species other than Hypogymnia physodes. The
sorting was carried out in a clean cabinet using gloves and
ceramic forceps. Up to 4 g of air dried lichen was ground in an
agate mortar and pestle under liquid nitrogen. The resulting
powder was mixed and then dried in a desiccator for at least 24 h,
noting the average weight loss. Aliqouts of about 0.2 g were
digested in 5 ml of high purity concentrated HNO3 and 0.5 ml high
purity HF in a closed vessel microwave digestion system and
diluted as appropriate with doubly deionised water. Multi-element
analysis was performed via inductively coupled plasma mass
spectrometry (ICP-MS) calibrated using commercially available
standard solutions. An assessment of accuracy and precision was

Modelled deposition was compared against the concentration of

As, Cu and Pb in the transplanted lichens by calculating the ratio
between the two quanta at each sample point. Where the lichen
samples obtained were not in locations that could be predicted by
the TAPM model, concentrations were estimated via linear interpolation using levels at the four closest predicted sites (called
solution nodes) and the lichen sample's distance from each of the
four nodes.
Model output was downloaded to MS Excel and analysed using
arithmetic, graphical and statistical analysis, the latter involving the
Bland-Altman technique or difference plot (Bland and Altman,
1986). Following this, evidence was sought as to whether inferences could be made about the actual nature of the exhaust
emission strength and the longer term effect of the smelter.
2.3. Estimating the stack output
As an addendum to this study, using the evidence in this
research, the inferred emission rate of the smelter was estimated
calculating back from observed lichen concentrations. This value
was then fed into TAPM which was run over a one year, 2011, to
estimate the annual deposition of Pb per year around the Karabash
smelter. These values were then compared with known annual Pb
deposition rates experienced in the UK in order to produce an estimate of levels of Pb deposition.
3. Theory and calculation
Eulerian and the mixed Eulerian/Lagrangian approaches were
used to simulate metal deposition within the lichen with the results
compared against actual lichen concentrations. Simulating the
spread of the smelter plume using the Eulerian approach is done
throughout a network of points in a region using equations which
represent the governing physical processes involved. The
Lagrangian approach assumes that the polluting source produces a
number of particles, for which trajectories are simulated subject to
the modelled meteorological forces.
The dry deposition scheme used in TAPM for gaseous pollutants
is based on Wesely's approach (Wesely, 1989) whilst the deposition
velocity for ne matter, APM (Airborne Polluting Matter) and FPM
(Fine Polluting Matter) is calculated based on a methodology
documented by Seinfeld and Pandis (2012), including the effect of
particle settling velocity. A full description of the TAPM model
formulation is available in Hurley et al. (2005).
The conguration of TAPM for this study together with how the
results were employed is described below.

Please cite this article in press as: Pollard, A.S., et al., Integrating dispersion modelling and lichen sampling to assess harmful heavy metal
pollution around the Karabash copper smelter, Russian Federation, Atmospheric Pollution Research (2015), http://dx.doi.org/10.1016/
j.apr.2015.04.003

A.S. Pollard et al. / Atmospheric Pollution Research xxx (2015) 1e7

3.1. Solution grid

A computerised map e the TAPM solution grid - was constructed with the smelter at its centre (Fig. 1). The spacing of the
grid was chosen in order to make best use of the program's ability
to process the data obtained from the lichen sample points. The
solution nodes within the nested grids had spacings which ranged
from 2000 m (inside grid or pollution grid, Fig. 1) to a 20 km grid
spacing for the most distal grid, large enough to handle meteorological modelling. Each grid comprised 25 by 25 solution points,
spaced within 25 vertical layers, stretching from the earth's surface
up to 8 km into the atmosphere.
3.2. Model period
The model was run between the 10th June to 16th September
2011, reecting the time period during which the lichen transplants
were exposed at each monitoring station before removal and
analysis. Data relating to the local weather during this interval was
fed into the model.
3.3. Initial conditions
Metal concentrations in transplanted lichens were measured at
the time they were transplanted, and this value subtracted from the
concentrations in the lichen when they were nally removed and
re-tested. At the time of transplantation, the lichens contained an
average of 3 ppm As, 18 ppm Cu and 72 ppm Pb. Zero air pollution
was assumed at the start of the modelling period and simulated
levels of dry deposition across the modelling period fed into the
estimation of actual concentrations within the lichen.
3.4. Boundary conditions
The height of the smelter stack from the ground was estimated
to be 127 m (personal communication) and the chimney's radius
approximately 2 m (using photographic evidence together with the
assumed smelter height). From these gures, and information on
rates of production (DKL Engineering, 2012), an estimate of
5.08 m s1 was made for the velocity at which emissions left the
stack, and 145.4 g s1 for the particulate expulsion rate. These data
allow TAPM to model the plume of exhaust from the chimney stack.
In turn, the exit velocity is derived from the volume of matter being
dispelled and the cross-sectional area of the exhaust outlet.
3.5. Modelled quantum
Selection of the pollutants modelled was informed by previous
work that has empirically linked the dry deposition of metals with
the contamination of lichens sampled from around a smelter in
Greenland (Sondergaard et al., 2012). The quanta modelled within
TAPM concerned classications of particle size; PM2.5 and PM10
(nominally particulate matter with diameters less than 2.5 mm and
10 mm, respectively) and APM (comprising dissolvable particles of
size less than 10 mm). Each of these quanta produces dry deposition
as an output. Selected results from the modelling of these particles
are presented in this paper for comparison.
3.6. Calibration
Predicted deposition levels were converted into estimates of
deposition of metals in the transplanted lichens using an empirical
relationship (Sondergaard et al., 2012). In turn, these estimates
were compared against actual metal levels determined in lichens
collected three months after transplantation. Sources of variance

between the two measures were determined. The empirical relationship that holds for Pb deposition is:

BcC

(3.1)

Where
B is the dry deposition across a modelling period in mg m2.
-C is the increase in element concentrations in transplanted lichens during the same sampling period in ppm.
The parameter c is an element specic coefcient. For example, c
is 1.01 for Pb.
Pb concentrations modelled using TAPM and equation (3.1)
were compared with actual levels within each transplanted
lichen, providing inferences as to the emission strength of the
smelter.
4. Results
Across all of the sample points, the t between concentrations of
As, Cu and Pb found in the lichens and the modelled dry deposition
was closest for Pb (Table S2 and Table S3) in all but one of the model
subcategories. Consequently, the analysis presented in this study
focuses on the suitability of TAPM to predict Pb deposition around
the Karabash smelter.
The size of the modelled smelter plume is affected by a variety of
factors including wind direction and strength. An analysis of wind
conditions revealed that the wind blew from a direction between
West and North-North-East for 80% of the time during the
modelling period (Fig. 1). In addition to their direction relative to
the smelter, the distance of each point from the smelter is presented in Table S2. With respect to the smelter, sites NJU3, NJU11,
TSU4 and TSU5 lie downwind whilst the remaining sites lie in a
direction more orthogonal to the prevailing wind. Across all of
these sites, there is some correlation between Pb concentrations
found in transplanted lichens (Figure S1), but different relationships appear when the points are subdivided between those
downwind of the smelter and otherwise (Figure S2).
Actual Pb concentrations found in the transplanted lichens were
compared with modelled concentrations of the three particle sizes
(Table 1). This was undertaken to illustrate the sensitivity in model
output to model quantum used and proximity to the smelter, and to
demonstrate the range in accuracy between the points of the model
framework employed. Without exception, the actual concentrations within the lichens were higher than modelled. Factors
Table 1
Pb concentrations in lichen at each site throughout the modelling period (Lichen);
modelled accumulations of particulate matter of three types and rened using
Sondergaard's ndings; a coefcient of regression between modelled and actual
levels; and the ratio of the actual to modelled levels. Results were obtained using the
combined Eulerian and Lagrangian approach.
Lichen (ppm)a

Modelled levels (ppm)b

Lichen/Modelled
(ratio)c

Pb

PM2.5

PM10

AAPM

PM2.5

PM10

AAPM

17.0
75.4
160.0
768.9
145.1
44.5
24.2
30.4

0.6
0.4
0.8
0.8
0.3
0.2
0.1
0.1
0.39

3.5
3.7
6.8
7.1
2.7
1.3
0.8
0.6
0.51

15.9
15.4
26.2
31.6
14.3
10.2
10.4
10.2
0.70

30.3
181.9
192.9
939.9
417.1
251.8
227.5
361.8

4.9
20.3
23.6
107.8
52.9
34.2
30.0
47.9

1.1
4.9
6.1
24.3
10.2
4.4
2.3
3.0

Site

NJU3
NJU11
TSU4
TSU5
TSU6
TSU7
NJU9
NJU10
R2
a

The concentration of Pb found in transplanted lichen retrieved from the eld.

The concentration of Pb in lichen as modelled using TAPM and Sondergaard's
relationship.
c
The actual lichen concentration divided by modelled.
b

Please cite this article in press as: Pollard, A.S., et al., Integrating dispersion modelling and lichen sampling to assess harmful heavy metal
pollution around the Karabash copper smelter, Russian Federation, Atmospheric Pollution Research (2015), http://dx.doi.org/10.1016/
j.apr.2015.04.003

A.S. Pollard et al. / Atmospheric Pollution Research xxx (2015) 1e7

Table 2
The range in ratio of observed Pb concentrations to modelled concentrations at sample points proportioned by their position relative to the smelter and the direction of the
prevailing wind. For each set of points, the number in square brackets is the maximum ratio divided by the minimum.
Quantum

All points (8)

All points less NJU3 and TSU5 (6)

Downwind (4 points)

Lateral (4 points)

PM2.5

30.3e939.9
[31.0]
4.9e107.8
[22.0]
1.1e24.3
[22.1]

181.9e417.1
[2.29]
20.3e52.9
[2.61]
2.3e10.2
[4.43]

30.3e192.9
[6.37]
4.9e23.6
[4.82]
1.1e6.1
[5.55]

227.5e417.1
[1.83]
30.0e52.9
[1.76]
2.3e10.2
[4.43]

PM10
APM

affecting the relationship between the modelled and sampled Pb

levels included the particulate modelled, ie. PM2.5, PM10 or APM,
and the direction of each point from the smelter. Apart from the
closest point to the smelter, the modelling framework employed
here underestimated the lichen Pb concentrations by a lesser degree for points downwind from the smelter than for the others.
The t between the sampled Pb concentration and modelled
dry deposition is shown in Table 2. The variability in the ratio of the
dry deposition modelled using TAPM and the metal contents of the
lichens varies signicantly between the points. This diversity is
greater within points on the downwind arc, as opposed to between
those located laterally to the direction of the prevailing wind. From
Table 2, when all points are included, the maximum variation occurs when modelling PM2.5, with the least variation obtained when
modelling with PM10. An analysis demonstrated that the actual and
modelled lichen concentrations diverge at the closest point to the
smelter using Bland-Altman analysis (Figure S3). For the set of six
points (which excludes the downwind points closest and furthest
away from the smelter, i.e. NJU3 and TSU5) dry deposition ranged
from 640 to 6852 mg m2.
In addition to dry deposition, TAPM can simulate the wet
deposition of a polluting species. From Table S4, the variability in
t of modelled deposition (the sum of wet plus dry) is least for
the same six points and using PM10. Across these six points, the
variation in ratio of the modelled deposition model and Pb sampled
from the lichens ranged from 41.4 to 69.3 across the six points (with
the maximum ratio being a factor of 1.7 greater than the minimum).
4.1. Predicting stack emission strength
To gauge the potential longer term effect of the smelter on the
local environment, TAPM was run across 2011 [using PM10 to
represent smelter particulate emissions as this size class was found
to have the best t to Pb concentrations in the lichen samples
(Table 2)]. In this second run, an exhaust strength of 2951 g s1 was
selected (increased by a factor of 20.3 compared to the earlier run)
for the following reasons:
1. Within Table 2 the smallest level of disagreement occurred
when modelling using PM10 and for points lateral to the
Table 3
Modelled Pb deposited (wet and dry) at the points within this study (expressed as a
range across the points excluding points TSU5 and NJU3). Results are from a TAPM
simulation of the Karabash smelter which is assumed to expel Pb at the following
rates: Rate 0 (calculated using the smelter's production of 0.65 MT yr1 equating to
emissions of 45.3 g s1); Rate 1 (derived from material dening the smelter's waste
output: 145.4 g s1); Rate 2 (calculated assuming no re-suspension from topsoil and
all bioaccumulations due to exhaust smelter of strength 2952 g s1). The results
presented here are set against background Pb deposition rates for the UK as a
comparator.
Deposition
1

1

Dry (kg ha yr )
Wet (kg ha1 yr1)

Rate 0

Rate 1

Rate 2

UK

48.7e69.9
336e1207

156e224
1046e3870

3173e4551
21 233e78 571

<2.5 e >50
<5 e >100

prevailing wind. The most conservative ratio of observed Pb

concentrations to modelled within this set of points is 30.0.
2. Including two additional downwind points to this set of four
only modestly increased the range of ratios, with the maximum
ratio divided by the minimum ratio equalling 2.61.
Of these six points the most conservative degree of underestimation was 20.3. Using this scale factor, a new model was then run,
for PM10 deposited over a 12 month period. A zero pollution initial
condition was assumed; again a conservative assumption if the
impact on health of smelting around Karabash is to be modelled.
By increasing the exhaust strength by a factor of 20.3 (to
2951 g s1), the dry deposition of Pb in the environment around the
smelter increased to between 0.369 and 0.455 kg m2 yr1. The
combined wet and dry Pb deposition at these six points throughout
2011 was modelled as between 2.5 and 8.2 kg m2, equivalent to
25 430 to 82 270 kg ha1 (Table 3).
5. Discussion
5.1. Model construction and assumptions
A conceptual and numerical model was constructed depicting
particle fallout from the copper smelter at Karabash. In the absence
of emissions data, the model was run over the same time period as
the lichen transplant monitoring. Metal concentrations in lichens
were found to be proportional to air pollution modelled around the
smelter, with agreement best along the transect orthogonal to the
prevailing wind direction.
TAPM takes into account surface topology of an area, including
rock formations (orography) and vegetation cover in addition to
regional meteorology. TAPM also assumes uniform background air
pollutant concentrations in simulating the contaminant transport
and deposition throughout a model domain. When setting up a
mathematical model of any physical process it is necessary to
employ both initial conditions and boundary conditions. Here we
employ a zero contaminant (or clean) initial condition
throughout the model domain which is obviously not consistent
with the historically polluted nature of the environment around
Karabash. Consequently, the variance in accuracy when modelling
the concentrations within the lichens might be contributed to by
the model not including the re-suspension of contaminated soils
close to the smelter.
The boundary condition set within this model is that of the
smelter itself and the expulsion of pollutants into the atmosphere.
This emission rate is assumed to be constant within the modelling
framework presented, thereby failing to account for the likely
variability in smelter emission rates. Furthermore, diffuse emissions from smelter operations and other industrial activities on the
smelter site, as well as further away, may contribute to deposition
rates. This increases the difculty of making direct and accurate
assessments of source strength and the degree of contribution of
the individual source (the smelter) to observed deposition. This

Please cite this article in press as: Pollard, A.S., et al., Integrating dispersion modelling and lichen sampling to assess harmful heavy metal
pollution around the Karabash copper smelter, Russian Federation, Atmospheric Pollution Research (2015), http://dx.doi.org/10.1016/
j.apr.2015.04.003

A.S. Pollard et al. / Atmospheric Pollution Research xxx (2015) 1e7

challenging aspect of atmospheric deposition modelling was

recently discussed by Dore et al. (Dore et al., 2014) for scenarios in
the UK.
TAPM has been used commercially and in a number of published
academic studies, which have been collated and reviewed by CSIRO
(Commonwealth Scientic and Industrial Research Organisation,
Australia) (Hurley et al., 2008). However, there are other available
air pollution simulation models, some of which have been
compared with TAPM. A study conducted in New Zealand complemented TAPM for its ability to simulate the drivers of air
pollution (Gimson et al., 2007). The capability of TAPM to model air
quality within an industrial region with a complex topography was
successfully demonstrated for PM10 in Romania (Grigoras et al.,
2012).
5.2. Delineating the effect of current and historical metal deposition
around Karabash

particulate could occur by wet deposition. Certainly, rainfall will

impact on the capacity for the lichens to incorporate and retain
metals, for reasons explained in the Introduction. The variability in
t between the modelled and observed concentrations might be
partially accounted for if wet deposition is considered. Specically,
wet deposition would include the particles which are expelled from
the smelter and fall to the ground with rainfall. Figure S3 illustrates
the ratio of the addition of wet and dry deposition at each point to
the observed concentrations, split by their position relative to the
smelter. Although the ratios are comparable to those presented in
Table 2 for points in directions orthogonal to the prevailing wind
from the smelter, for those in the direction of the prevailing wind
the ratios deteriorate when wet deposition is included as well as
dry. Therefore, the variability in the ratios between modelled and
observed concentrations in the lichen might not be explained by
the lack of consideration of wet deposition.
5.4. Processes within the lichens

The modelling framework required calibration prior to use with

respect to the long-term effects of the smelter's operation. To
achieve this, model output was compared against the levels of Pb,
As and Cu found in lichens around the smelter. A sensitivity analysis
then ensued which identied Pb as the element for which there
was best proportionality between modelled dry deposition and
lichen composition. Modelled dry deposition was then converted to
model lichen Pb concentration using a previously determined
empirical relationship. The resulting estimates of Pb content in the
lichens were then compared with actual values and again, proportionality was found to be strongest in the orthogonal transect.
Furthermore, the degree to which the model underestimated the
actual strength of the smelter exhaust was quantied. After
adjusting the assumed exhaust strength according to this ratio, the
model was re-run over a 12 month period to estimate the impact of
the smelter on the local environment. Although resulting Pb
deposition could be higher than that in the UK, it is likely that a
proportion of this would be due to capturing of windblown soil
particles contaminated by historical smelting or from residual Pb
from the use of leaded fuels, banned in Russia in 2003 (UNEP,
2008).
The inuence of historical and current smelting on present
environmental quality is backed by the Pb levels in soil sampled
around the smelter which were between 1.5 and 2 mg kg1 (Revich,
2010). The potential inuence of contaminated soils is suggested
further from Table 1, with the ratio of actual to modelled concentrations being lowest at the point furthest downwind from the
smelter, NJU3. A cause of increasing soil Pb concentrations towards
the smelter might be that larger and more dense particles fall closer
to their source (Williamson et al., 2008). Additional variation in the
accuracy of results will be caused by the likely mixed particle sizes
expelled from the smelter, which will travel to varying heights and
distances. Therefore, the model assumptions that particles will be
homogenously sized might be awed.
5.3. The impact of weather and meteorological data
A further source of error when comparing the modelled and
actual levels of Pb in lichens is the adequacy of the weather data.
The meteorological datasets available for the TAPM modelling only
contain hourly data, necessitating the use of 1 h as a time step for
the modelling framework. To reduce the uncertainty in modelling
the plume, weather data available on a more frequent basis would
be required.
In addition to this, the model so far only assumes dry deposition,
where Sondergaard's relationship links dry deposition with Pb, Zn
and Cd levels in lichens, although of course signicate fallout of

An additional source of error between predicted and actual

lichen concentrations is likely to be introduced around the prior
linear relationship assumed between dry deposition and metals
incorporation within lichens. The lichens transplanted across
transects around Karabash was of the type Hypogymnia physodes.
However, in converting the modelled dry deposition into estimated
Pb, Zn and Cd concentrations in lichen, a relationship was used
shown to hold for the lichen Flavocetraria nivalis (Sondergaard
et al., 2012). Not only has this relationship not been proven to
hold for Hypogymnia physodes but its derivation involved exposing
the lichen to only 7 weeks of pollution at 1.8 m above ground level.
There was further variability in t between observation and
estimations due to the modelling approach selected. This is borne
out from Figure S1, where the ratio of lichen concentrations to dry
deposition is most sensitive to the modelling solution methods
employed at the lichen locations closest to the smelter.
For all model runs undertaken to date, the largest degree of
underestimation of metal deposition in lichens occurs at the extremity of the model domain downwind. This suggests other processes are involved in metal fallout from the smelter to the
downwind point. One scenario could be that this fallout accumulates in the topsoil from where it can be re-suspended and transported, for example blown with the wind. In addition, although
rainfall may deposit metal contaminants as in wet deposition,
rainfall can also wash a contaminant through a lichen, thereby
having the opposite effect of reducing a contaminant's presents in a
lichen (Purvis et al., 2013).
The lichen themselves will be sensitive to background levels of
contamination (Spiro et al., 2004). In the UK, the background levels
of Pb deposition range, for dry deposition, from below 2500 to over
50 000 g ha1 yr1 (DEFRA, 2008a) and for wet deposition, from
5000 to over 100 000 g ha1 yr1 (DEFRA, 2008b) depending on
location. Therefore, background levels of metal contaminants like
Pb can vary enormously, especially around the Karabash smelter
due to the re-suspension of soils contaminated by historical
smelting. This potentially disconnects the modelled effect of current emissions from the accumulation of metalloid concentrations
found in lichens sampled from the local environment.
6. Conclusion
To assess the current role of the smelter on the rate of uptake of
environmental Pb within the lichen, the area's mining history remains inuential due to the dispersion of historically deposited soil
metals. Assuming the fugitive emission rate is applicable to Karabash, a proportion of the Pb taken up by lichens around the

Please cite this article in press as: Pollard, A.S., et al., Integrating dispersion modelling and lichen sampling to assess harmful heavy metal
pollution around the Karabash copper smelter, Russian Federation, Atmospheric Pollution Research (2015), http://dx.doi.org/10.1016/
j.apr.2015.04.003

A.S. Pollard et al. / Atmospheric Pollution Research xxx (2015) 1e7

Karabash smelter is likely to be caused by the re-suspension of such

contaminated soils.
Lichen transplant experiments in the Karabash area of Russia
indicate high levels of atmospheric fallout of heavy metals, specically those known to be emitted from the smelter. By combining
the data concerning metal levels in these lichens with the TAPM
computer modelling software, we were able to show the inuence
of prevailing winds on the deposition of such pollutants, and predict which areas would suffer the highest levels of deposition. Being able to model the deposition of environmental contaminants in
this way has implications for the responsible siting of accommodation close to smelting facilities, and for identifying the areas most
likely to need remediation in order to achieve lower levels of metals
in the environment.
Conict of interest
The authors declared that they have no conicts of interest to
this work.
Acknowledgements
We gratefully acknowledge funding from a 3 year (2010e2012)
EU FP7 contract Impact Monitoring of Mineral Resources Exploitation (ImpactMin, Contract: 244166). Dr David Bellis is thanked
for carrying out the preparation and ICP-MS analysis of the native
and transplanted lichens.
The European Centre for Environment and Human Health (part
of the University of Exeter Medical School) is supported by investment from the European Regional Development Fund 2007 to
2013 (Contract: 500020) and European Social Fund Convergence
Programme for Cornwall and the Isles of Scilly.
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://
dx.doi.org/10.1016/j.apr.2015.04.003.
References
Araya, M., Pizarro, F., Olivares, M., Arredondo, M., Gonzalez, M., Mendez, M., 2006.
Understanding copper homeostasis in humans and copper effects on health.
Biol. Res. 39, 183e187.
Bland, J.M., Altman, D., 1986. Statistical methods for assessing agreement between
two methods of clinical measurement. Lancet 327, 307e310.
Cislaghi, C., Nimis, P.L., 1997. Lichens, air pollution and lung cancer. Nature 367,
463e464.
Conti, M.E., Cecchetti, G., 2001. Biological monitoring: lichens as bioindicators of air
pollution assessment d a review. Environ. Pollut. 114, 471e492.
Cuypers, A., Keunen, E., Bohler, S., Jozefczak, M., Opdenakker, K., Gielen, H.,
Vercampt, H., Bielen, A., Schellingen, K., Vangronsveld, J., 2012. Cadmium and
copper stress induce a cellular oxidative challenge leading to damage versus
signalling. Metal Toxic. Plants Percept. Signal. Remediat. 65e90.
Department for Environment, Food and Rural Affairs, 2008a. http://www.
pollutantdeposition.defra.gov.uk/node/357 (accessed 25.07.13.).
Department for Environment, Food and Rural Affairs, 2008b. http://www.
pollutantdeposition.defra.gov.uk/node/356 (accessed 25.07.13.).
DKL Engineering, 2012. http://www.sulphuric-acid.com/sulphuric-acid-on-theweb/Acid%20Plants/ZAO%20Karabashmed%20-%20Karabash.htm
(accessed
22.04.14.).
Dore, A.J., Hallsworth, S., McDonald, A.G., Werner, M., Kryza, M., Abbot, J., Nemitz, E.,
Dore, C.J., Malcolm, H., Vieno, M., Reis, S., Fowler, D., 2014. Quantifying missing
annual emission sources of heavy metals in the United Kingdom with an atmospheric transport model. Sci. Total Environ. 479e480, 171e180.
m, A., Michae
lsson, K., Vahter, M., Julin, B., Wolk, A., kesson, A., 2012.
Engstro
Associations between dietary cadmium exposure and bone mineral density and
risk of osteoporosis and fractures among women. Bone 50, 1372e1378.
Garrow, J.S., James, P.T., Ralph, A., 2000. Human Nutrition and Dietetics. Elsevier,
p. 2000.

Gimson, M., Olivares, G., Khan, B., Zawar-Reza, P., 2007. Part 1-assessment of
Meteorological Models, Dispersion Modelling in New Zealand. Golder associates, FRST programme protecting New Zealand's Clean Air.

lez-Castanedo, Y., Moreno, T., Ferna

ndez-Camacho, R., de la Campa, A.M.S.,
Gonza
Alastuey, A., Querol, X., de la Rosa, J., 2014. Size distribution and chemical
composition of particulate matter stack emissions in and around a copper
smelter. Atmos. Environ. 98, 271e282.
Grigoras, G., Cuculeanu, V., Ene, G., Mocioaca, G., Deneanu, A., 2012. Air pollution
dispersion modeling in a polluted industrial area of complex terrain from
Romania. Romanian Rep. Phys. 64, 173e186.
Hurley, P.J., Physick, W.L., Luhar, A.K., 2005. TAPM: a practical approach to prognostic meteorological and air pollution modelling. Environ. Model. Softw. 20,
737e752.
Hurley, P.J., Edwards, M., Luhar, A., 2008. TAPM v4. Part 2: Summary of Some
Verication Studies, CSIRO Marine and Atmospheric Research Paper. CSIRO.
ger, P., Tooming, A., Juravskaja, J., Pasanen, M.,
Kantola, M., Purkunen, R., Kro
Saarikoski, S., Vartiainen, T., 2000. Accumulation of cadmium, zinc, and copper
in maternal blood and developmental placental tissue: differences between
Finland, Estonia, and St. Petersburg. Environ. Res. 83, 54e66.


, J., 2007. Metal/metalloid contamination and
Kom

arek, M., Chrastn, V., Stchov
a
isotopic composition of lead in edible mushrooms and forest soils originating
from a smelting area. Environ. Int. 33, 677e684.
Newhook, R., Hirtle, H., Byrne, K., Meek, M.E., 2003. Releases from copper smelters
and reneries and zinc plants in Canada: human health exposure and risk
characterization. Sci. Total Environ. 301, 23e41.
Nimis, P., Scheidegger, C., Wolseley, P., 2002. Monitoring with Lichensemonitoring
Lichens: NATO Science Series iv: Earth and Environmental Sciences, vol. 7.
Kluwer Academic Publishers, NATO Science Ser.
Purvis, O.W., Williamson, B.J., Spiro, B., Udachin, V., Mikhailova, I.N., Dolgopolova, A.,
2013. Lichen monitoring as a potential tool in environmental forensics: case
study of the Cu smelter and former mining town of Karabash, Russia. Geol. Soc.
Lond. Spec. Publ. 384, 133e146.
Revich, B., 2010. Social Inequality and Environmental Health in the Russian Federation, Environment and Health Risks: of Social Inequalities, p. 238.
Seinfeld, J.H., Pandis, S.N., 2012. Atmospheric Chemistry and Physics: from Air
Pollution to Climate Change. John Wiley & Sons.
Sondergaard, J., Bach, L., Asmund, G., 2012. Modelling atmospheric bulk deposition
of Pb, Zn and Cd near a former Pb-Zn mine in West Greenland using transplanted Flavocetraria Rivalis lichens. Chemosphere 90, 2549e2556.
Spiro, B., Weiss, D., Purvis, O., Mikhailova, I., Williamson, B., Coles, B., Udachin, V.,
2004. Lead isotopes in lichen transplants around a Cu smelter in Russia
determined by MC-ICP-MS reveal transient records of multiple sources. Environ. Sci. Technol. 38, 6522e6528.
Spiro, B., Udachin, V., Williamson, B., Purvis, O.W., Tessalina, S., Weiss, D., 2012.
Lacustrine sediments and lichen transplants: two contrasting and complimentary environmental archives of natural and anthropogenic lead in the South
Urals, Russia. Aquat. Sci. 1e14.

, K., Kala
Svoboda, L., Zimmermannova
c, P., 2000. Concentrations of mercury, cadmium, lead and copper in fruiting bodies of edible mushrooms in an emission
area of a copper smelter and a mercury smelter. Sci. Total Environ. 246, 61e67.
Swarup, D., Dwivedi, S., 2012. Research on effects of pollution in livestock. Indian J.
Animal Sci. 68.
The Russia Channel, 2013. http://www.russia-channel.com/urals/karabash/
(accessed: 07.02.13.).
Tote, C., Delalieux, S., Goossens, M., Williamson, B.J., Swinnen, E., 2014. Monitoring
environmental health using spot-vegetation-derived and eld-measured
spectral indices in Karabash, Russia. Int. J. Remote Sens. 35, 2516e2533.
United Nations Environment Programme, 2008. http://www.lead.org.au/PCFV/
PCFV_Lead_Matrix-CEE-EECCA_200808_%28Pb&S%29.pdf (accessed 04.02.15.).
Wesely, M.L., 1989. Parameterization of surface resistances to gaseous dry deposition in regional-scale numerical models. Atmos. Environ. (1967) 23,
1293e1304.
Williamson, B.J., Mikhailova, I., Purvis, O.W., Udachin, V., 2004. SEM-EDX analysis in
the source apportionment of particulate matter on Hypogymnia Physodes
lichen transplants around the Cu smelter and former mining town of Karabash,
South Urals, Russia. Sci. Total Environ. 322, 139e154.
Williamson, B.J., Purvis, O.W., Mikhailova, I.N., Spiro, B., Udachin, V., 2008. The
lichen transplant methodology in the source apportionment of metal deposition around a copper smelter in the former mining town of Karabash, Russia.
Environ. Monit. Assess. 141, 227e236.
World Health Organization, 1996. Trace Elements in Human Nutrition and Health.
Yang, X.-E., Long, X.-X., Ni, W.-Z., Ye, Z.-Q., He, Z.-L., Stoffella, P.J., Calvert, D.V.,
2002. Assessing copper thresholds for phytotoxicity and potential dietary
toxicity in selected vegetable crops. J. Environ. Sci. Health, Part B 37,
625e635.
Yeganeh, M., Afyuni, M., Khoshgoftarmanesh, A.-H., Sofanian, A.-R., Schulin, R.,
2012. Health risks of metals in soil, water, and major food crops in Hamedan
Province, Iran. Hum. Ecol. Risk Assess. An Int. J. 18, 547e568.
Zhao, F.-J., McGrath, S.P., Meharg, A.A., 2010. Arsenic as a food chain contaminant:
mechanisms of plant uptake and metabolism and mitigation strategies. Annu.
Rev. plant Biol. 61, 535e559.

Please cite this article in press as: Pollard, A.S., et al., Integrating dispersion modelling and lichen sampling to assess harmful heavy metal
pollution around the Karabash copper smelter, Russian Federation, Atmospheric Pollution Research (2015), http://dx.doi.org/10.1016/
j.apr.2015.04.003