View
4
Download
0
Category
Preview:
Citation preview
ACTA
UNIVERSITATIS
UPSALIENSIS
UPPSALA
2006
Digital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 200
Geochemical Landscape Analysisfor the Risk Assessment of AcidMine Drainage in a WetlandEnvironment
ANDREA SZUCS
ISSN 1651-6214ISBN 91-554-6601-Xurn:nbn:se:uu:diva-6992
���������� �������� �� ������ �������� � �� �������� ������� � ��������������������� ������� �������� ��������� ��� ���� �� �� �� !� �"� ������ ! ���� !#"����"�$ �"� �������� %��� �� ������� � &����"$
��������
�'���� ($ ����$ ���"������ )������� (������ !� �"� *��+ (�������� ! (��� ,��������� � � -����� &������$ (��� ����������� ���������$ ������� ��� � ���� ����� � � ������� ���� ������� �� �� ������� � ��� �� ��� � ������ ���$ ./ ��$ ������$ 0�12 /�344.3����35$
(������� ! ���� ��� ������� 6(,�7 ������ �������� !�� ������ ���� ��%���� ��+ ����� �� ����������� � �"�� �"���� �� ��8������ � ������ �%���� %"��� ���� �������"��� "�� ��� ����"����� !� ��� 9� ����� �� �"� �������� %����� ������ �� ����$*��+ ��������� ! (,� �� ������� ������� ���� ��:����� � "������ ������" �"�� �� ���� ������ �"� ��������� ! ������ ������� �� �������� ���������$ 0 �"�� �"���� � ��+���%�� ���"������ ������� !��� ������� �� �������� ������� �� �������� !�(,� ���+ ���������$ ( ������ ���"������ �������� ������� �� �� �������� � �������� ���� ���"������ ��������� ���"������ ��������� ���"������ !�% ������� �����"������ �������� � �"� ������� ������ �� ���� ��������$ ������� ������ �����"������ �������� ��� �����!��� ���� �������� ���"������ �� �0� ���"��$ (��:������ �"������ �������� �������� �� ���� � ���������� !������ %"��" ��� �������� ���� �� ������� ��+� ! �"� ��� ������� ������ 6;�� <�� ,� 2�� #� �� =7 � �"� ��������$<� ���� ������� ����� ����������� ���"�� ! &�������� ���� (������ ��� ���� � ���������� ������ ��'�� %��" ��������� �"������� �� ������ ������$ �"� ������� ����������� !����� ������ � �"� �������� �� ������� �� ������������ ���� ������� ���"��$ *�������"% �"�� �"� �������� ������ ���"������ �������� ������� ���"� �� �� �����!!������� !� �"� ���+ ��������� ! ���� ��� �������� � �"� ������ ��������%����� ��������$ 0� �� ��������� �� �"� ������� �"�� �"� ������� ���"������ ������� � �"������� �������� �� �� ��!!������� �"����������� �� �"� ���������� ! <� !������$ (� �"�(,� ����"���� ����� ����� ����"��� !����� �� ����� ������ ������� ��������� ������ � �"� ����$ ,���� ��� ���� �������� � �"���� � "��������� ������ ������� ! �"� �������� ����� � ���� �� "��� �"� ������� ! ������� ������ � �"��������$
� ������ ���� ��� �������� ��������� ���"�������� ��������� ���+ ����������������� %�����
���� � ����� � ���� �� � !���� ��� �� � "�����# $% ������� ���� ����� �!&'()*%������� �� � �
> (���� �'��� ����
0��2 ��4�3���.0�12 /�344.3����35�� � �� �� ����3�//� 6"��� ??��$+�$��?������@��A�� � �� �� ����3�//�7
To my sons.
List of Publications
This thesis is based on the following papers that are referred to by their Ro-
man numerals. The publishers kindly gave their permission for reproduction
of papers.
I Szucs, A. Jordan, G. and Qvarfort, U., 2002. Geochemical model-
ling of acid mine drainage impact on a wetland stream using landscape
geochemistry, GIS and statistical methods. In: A. G. Fabbri, G. Gaal
and R. B. McCammon (eds), Deposit and Geoenvironmental Models
for Resource Exploitation and Environmental Security, NATO Science
Series, 2. Environmental Security, 80, pp. 425-460. Kluwer Academic
Publishers, Dordrecht. II Jordan, G., Szucs, A., Qvarfort, U. and Szekely, B., 1997. Evalua-
tion of metal retention in a wetland receiving acid mine drainage. In:
X. Xuejin (editor), Geochemistry, Proceedings of the 30th IGC, Vol.
19, pp. 189-206. VSP Publisher, Utrecht. III Szucs, A. and Jordan, G., 2005. Geochemical landscape analysis:
development and application to the risk assessment of acid mine
drainage. A case study in Central Sweden. Science of the Total Envi-
ronment (submitted).
IV Jordan, G. and Szucs, A., 2002. Environmental Mapping of Geo-
chemical Systems. In: P. T. Bobrowsky (editor), Geo-environmental
Mapping: Theory, Methods and Applications, pp. 57-74. A. A.
Balkema, Lisse.
APPENDIX
V Jordan, G. and Szucs, A., 1997. The role and future of geology in
modern environmental research and decision support. In: H. Wang, D.
F. Branagan, Z. Ouyang, X. Wang (eds), Theory of Geology, Proceed-
ings of the 30th IGC, Vol. 26, pp. 237-249. VSP Publisher, Utrecht.
Other publications published as result of PhD research.
1. Szucs, A., Jordan, G. and Qvarfort, U., 2006. Geochemical landscape
analysis: development and application to the risk assessment of acid mine
drainage. Case study in Central Sweden and Hungary. Abstracts, 2nd Na-
tional Conference on Landscape Ecology, Debrecen, Hungary.
2. Jordan, G. and Szucs, A., 1999. Environmental Mapping of Geochemical
Systems. Abstracts, International Conference and Annual Meeting of
FOREGS on European Geochemical Mapping, Vilnius, Lithuania (oral
presentation).
3. Jordan, G. and Szucs, A., 1998. Environmental Mapping of Geochemical
Systems. Abstracts, NATO Advanced Studies Institute Meeting, Matra-
haza, Hungary (oral presentation).
4. Szucs, A. and Jordan, G., 1998. Integrated modelling of acid mine drainage
impact on a wetland using landscape geochemistry, GIS technology and
statistical methods. Abstracts, NATO Advanced Studies Institute Meeting,
Matrahaza, Hungary (oral presentation).
5. Szucs, A. and Jordan, G., 1997. Modelling of acid mine drainage impact on
a wetland using landscape geochemistry, GIS technology and statistical
methods. Abstracts, IAMG „97. Int‟l. Symp. on Geomathematics, Barce-
lona, (oral presentation).
6. Jordan, G. and Szucs, A., 1997. Appliacation of landscape geochemical
methods and GIS technology for geoenvironmental mapping. Abstracts,
GAC Special Symposium on Geoenvironmental Mapping, Ottawa (poster).
7. Jordan, G., Szucs, A., Qvarfort, U. and Szekely, B., 1996. Evaluation of
metal retention in a wetland receiving acid mine drainage. Abstracts, 30th
International Geological Congress, Beijing, (oral presentation).
8. Jordan, G and Szucs, A. 1996. The role and future of geology in modern
environmental research and decision support. Abstracts, 30th Int‟l. Geol.
Congr., Beijing, (oral presentation).
9. Szucs A., 1995. An integrated systems approach to biogeochemical proc-
esses: evaluation of the control mechanisms of trace metal speciation in a
wetland stream environment affected by acid mine drainage. Licenciate
Thesis, Uppsala University.
Acknowledgements
I would like to thank Ulf Qvarfort for giving continuous support during the
years of my thesis work at the Uppsala University. He provided research
facilities and a good working environment for thesis work and fruitful dis-
cussions. His outstanding ability of solving emerging problems immediately
made my research work much efficient. I would like to express my gratitude
to Keith Bennett, head of Earth Sciences Department for his patience and
kind understanding during my PhD work at his department.
My sincere thanks to my friends at the Uppsala University. Bertram
Schott, Manhal Sirat, Barna Zsigmond, and many others. Without the per-
sonal support of friends outside of the university I would not have been able
to complete my thesis. Special thanks to Anikó and Edward Ledwaba and
the Heszler family for providing true friendship during my stay in Sweden.
This thesis would not exist without the support of the Geological Institute
of Hungary (MAFI). First of all, I would like to say thank you to Gábor
Gaál, former MAFI director, and Károly Brezsnyánszky, Director of MAFI
for their continuous interest in my work and their understanding support. I
thank István Horváth, György Tóth and Emőke Jocháné Edelényi of the
Hydrogeology Division for providing facilities and giving friendly support
for the preparation of this thesis.
The three months of stay as a visiting scientist at the International Insti-
tute for Geo-Information Science and Earth Observation (ITC) in Enschede,
the Netherlands was important for the success of my thesis. Professor An-
drea Fabbri, Head of Geological Survey Division, provided a welcoming
environment and important professional support. I thank the Vekerdy family
for being so friendly to me during my stay in the Netherlands.
The period of my PhD work has brought the greatest success and happi-
ness in my life: the birth of my three sons, Marcell, Milan and Nicholas.
Thanks for being with me.
Research has been carried out with the help of scholarships from the
Swedish Institute, the Soros Foundation and the Hungarian National Science
Fund ( OTKA T022133). Their assistance is gratefully acknowledged.
Contents
1 Introduction .............................................................................................. 11
2 Aim of the study ....................................................................................... 13
3 Study Area ................................................................................................ 14
4 Methods ..................................................................................................... 16 4.1 Geochemical landscape analysis for AMD contamination risk
assessment ............................................................................................... 16 4.2 Sampling and analytical procedures ............................................... 19 4.3 Data analysis and modelling ............................................................ 21
5 Sampling design based on geochemical landscape modelling .............. 22
6. Results ...................................................................................................... 25 6.1 Geochemical abundances ................................................................. 25
6.1.1 Stream ........................................................................................ 25 6.1.2 Mire ............................................................................................ 27
6.2 Geochemical gradients ..................................................................... 30 6.2.1 Stream ........................................................................................ 30 6.2.2 Mire ............................................................................................ 30
6.3 Geochemical processes ..................................................................... 32 6.3.1 Stream ........................................................................................ 33 6.3.2 Mire ............................................................................................ 35
7 Geochemical landscape analysis for AMD impact assessment ............ 37 7.1 Stream environment ......................................................................... 37 7.2 Mire environment ............................................................................. 40
8 Conclusions ............................................................................................... 43
Summary in Swedish .................................................................................. 45
References .................................................................................................... 47
11
1 Introduction
Since most of the elements used by the society come from mineral extrac-
tion, mining of mineral resources provide essential raw material for sustain-
able development. Mining however leads to severe impact on the environ-
ment, including contamination by toxic metals. A recent Europe-wide survey
identified wide-spread pollution problems caused by mining, abandoned
mines in particular (COM, 2003). Younger et al. (2002) estimated that about
1,000 to 1,500 km of watercourses are polluted by metal mine discharges in
the EU. Through mineral extraction and subsequent mineral processing huge
amount of wastes are produced. Metals and metal compounds in wastes tend
to become chemically more available, which can result in the generation of
acid mine drainage (AMD). Due to great volumes and slow chemical proc-
esses, mine waste can release toxic compound for a very long time on the
scale of centuries and thousands of years (BAT, 2003). Thus, remediation of
mine sites, including abandoned mines, has to consider long-term solutions
and remediation technologies have to be sustainable for a long time (Sind-
ing, 1999). As an alternative, passive in situ remediation methods for AMD
treatment using wetlands that provide means of cheap and sustainable reme-
diation has been proposed (PIRAMID, 2003). Around the mine site, soils
and surface water in the receiving environment are often contaminated with
harmful elements or compounds of AMD. These contaminated sites act as
secondary sources for pollution, especially for historic sites (Jordan and
D‟Alessandro, 2004).
AMD flowing from abandoned ore mines is a major environmental pollu-
tion problem in historic mining areas in central Sweden. If left untreated,
AMD can have adverse effects on natural ecosystems and human health.
Abundant natural wetlands may offer inexpensive treatment alternatives to
chemical neutralisation and precipitation processes in this region. However,
the effectiveness of wetland treatment depends on many chemical and bio-
logical factors, and the long-term capacity of wetlands to retain metals has to
be assessed by the modelling of metal-sediment interactions and by evaluat-
ing the stability of the wetland system as a whole. In this study the attenua-
tion of six metals Pb, Ni, Cu, Zn, Fe and Mn originating from abandoned
mines and waste rock dumps is investigated in the receiving stream (Paper
I) and mire (Paper II) sediments at Slättberg in central Sweden, where acid
mine leachate has been discharging for over 70 years into the wetland stream
12
and mire. Stream and mire sediments, as potential geochemical barriers for
AMD metals, are analysed using a sequential chemical extraction procedure
in order to investigate the control mechanisms of metal retention in these
sediments.
Among the various sampling media, stream sediments have been widely
used to investigate element distributions in surface environments (e. g., Car-
penter et al., 1975; Davenport, 1990; Mantei et al., 1996; Salminen et al,
2003). Stream sediments are a dynamic sampling medium that can represent
the geochemistry of nearby landscapes in terms of both the natural geo-
chemical heterogeneity, and the pollution caused by mining in the study area
(Plumlee and Logsdon, 1999). The abundant clays, Al, Mn, and Fe oxides,
and humic material in stream sediments collectively provide a large reactive
surface area for the effective control of dissolved metal content of aquatic
systems, therefore these sediments act as a geochemical barrier for metals.
The organic-rich reducing mire sediments studied in this research provide
abundant reactive surface for AMD toxic metal retardation by organic com-
plexation and metal sulphide formation. The chemistry of the contaminated
groundwater discharging into the studied stream and mire is well docu-
mented (Herbert, 1994) thus the Slättberg study area offers a good opportu-
nity for the geochemical modelling of AMD metal fate in the receiving wet-
land environment.
Assessment of long-term stability and contamination risk of AMD metals
requires contaminant fate modelling along the source-pathway-receptor
chain (van Leuwen and Hermens 1996; CARACAS 1999). In other words,
AMD contamination risk assessment requires not only the modelling of geo-
chemical processes but the analysis of the receiving landscapes (SENES
2000). In a review of the problem, Jordan (2004) points out that the complex
problem of AMD risk assessment requires methods that can address both (1)
geochemical contaminant fate modelling and (2) landscape analysis. In order
to link geochemical contaminant fate modelling and landscape analysis for
the risk assessment of AMD along the source-pathway-receptor chain, a
simple geochemical landscape analysis tool is developed using geochemical
modelling and landscape ecology spatial analysis methods (Paper III). Spa-
tial analysis of geochemical processes uses geochemical landscape analysis
implemented with GIS technology (Paper IV). Methods of geochemical
landscape analysis and spatial modelling are demonstrated through a case
study in the Slättberg study area in central Sweden.
13
2 Aim of the study
- The main objective of the thesis is to determine the relative partitioning
of six metals, Pb, Ni, Cu, Zn, Fe and Mn, into five sedimentary fractions
likely to be affected by various environmental conditions (Paper I and
II).
- The second objective is to evaluate the stability of various metal forms
within the stream and mire sediments by correlating the spatial distribu-
tion of chemical parameters and to determine the relation between these
parameters and samples using statistical techniques (Paper I and II).
- A third objective is to evaluate the stability of the wetland system, in an
effort to determine if wetland environment can act as permanent metal
sinks (Paper II).
- The final objective is to link geochemical contaminant fate modelling
and landscape analysis for the contamination risk assessment of AMD
(Paper III and IV).
- Specific objectives of investigation are
(1) the geochemical modelling of metal behaviour in the studied oxi-
dising wetland stream sediments (Paper I), and
(2) the geochemical modelling of metal behaviour in the studied re-
ducing mire sediments (Paper II).
The hypothesis is that, based on the partitioning of metals determined by
a sequential extraction procedure, metal retention by stream and mire sedi-
ments can be assessed, and results of geochemical process modelling can be
linked to spatial landscape analysis for the identification of geochemical
barriers and for the efficient risk assessment of AMD contamination. The
organic-rich reducing mire and oxidising stream sediments along the path-
way of flowing AMD is hypothetised to serve as double reducing-oxidising
barrier for metals in the mine effluent.
In this thesis geochemical modelling uses univariate and multivariate sta-
tistical and data analysis methods. The emphasis is on geochemical model-
ling and on linking geochemical process modelling to spatial analysis.
14
3 Study Area
The studied wetland and stream in the Slättberg nickel mine area is located
in central Sweden (Fig. 1). Vegetation on hill slopes surrounding the wetland
and stream consists of pine and spruce forest with mosses and heather. The
studied wetland is an ombrotrophic bog slightly raised with swampy parts
along wetland streams and with a 3-5 m wide shallow marginal open water
zone. In the marginal waters iron-oxihydroxide precipitates (bog iron) are
observable. The wetland is in a groundwater discharge area (Herbert, 1994).
According to field observations the groundwater table is about 50 cm below
surface in the dry season and 20 cm when highest in the autumn, giving a 30
cm wide zone of fluctuating reducing-oxidising conditions. The thickness of
the wetland sediments is 2-2.5 m on average, which is characteristic of bogs
in central Sweden (Franzen, 1985). The vegetation in the bog is typically
hummocky bog vegetation characterised by Sphagnum species, primarily
Sphagnum fuscum, with sparse pines. The bog is fed and intersected by a
braiding river and a smaller stream, both crossing the mineralization and
accompanying mining area upgradient to the mire. Down gradient to the
mire the stream widens into a fen, a shallow stagnant water. The stream is
1.5 deep on average and it runs on thick (1-2 m) organic rich sediment bed.
Both in the excavated ditches and in the fen ochreous precipitates are locally
observable on the sediment surface.
The geology of the area is characterised by island-arc igneous rocks of the
Early Proterozoic Svecofennian orogeny (Gaal, 1990) (Fig. 1). In the central
valley, the bedrock is covered by Quaternary gravely-sandy glaciofluvial
deposits of 6 to 15 metres in thickness and the surrounding area is dominated
by bouldery-sandy moraine covered by podzol soils. The main geological
feature of the area is an approximately 1.6 km long and 6 m wide mineral-
ised amphibolite dyke in a WSW-ENE orientation, which is intruded into a
hornblende bearing granitic gneiss host rock (Tegengren, 1924) (Fig. 1). The
ore association is characterised by pyrrhotite (Fe1-x
S), pyrite (FeS2), pent-
landite ((Fe,Ni)9S
8) with bravoite ((Fe,Ni)S
2), millerite (NiS) and linneite
((Co,Ni)3S
4) and chalcopyrite (CuFeS
2) (Magnusson, 1953; Wickman,
1994). As a result of metasomatic processes the dyke has been strongly al-
tered and feldspar is replaced to a great extent by sericite, chlorite and epi-
dote.
15
Six abandoned waste rock dumps and mine shafts are situated along the
dyke upstream to the studied stream segment (Fig. 1). Wastes derived from
the excavated mine shafts and tunnels have been dumped directly adjacent to
the mines, covering a surface area of about 1300 m2 with a thickness be-
tween one to two meters each. Mining of copper, iron and nickel ores com-
menced in 1805 and ceased during the 1940s. However, the bulk of the
waste was produced in the course of several years from 1917. Metals are
leached from waste rock dumps into groundwater in high concentrations
(pH: <3, Fe: 1110 mg/L, Mn: 4.4 mg/L, Cu: 40 mg/L, Zn: 18.5 mg/L, Ni:
32.1 mg/L, Pb: 0.29 mg/L, SO4: 3750 mg/L; Herbert, 1994) and are dis-
charged into excavated ditches and the studied stream segment. These con-
centrations are considerably attenuated by the point where AMD discharges
into the mire (pH: 3.96, Fe: 0.03 mg/L, Mn 1.98 mg/L, Cu: 4.4 mg/L, Zn:
4.6 mg/L, Ni: 2.5 mg/L, Pb: <0.05 mg/L, SO4: 460 mg/L; Herbert, 1994).
Figure 1. Location of the study area. Inset: Solid box indicates the location of the
area in Sweden (60 48'N, 15 15'E). Shading shows physiographic zones. The study
area is situated in the southern boreal zone of Scandinavia (medium grey shading)
(figures show average annual precipitation). The study area falls into the region of
the Svecofennian island arc lithologies of Early Proterozoic age (dotted line). Large
map: Map is the result of GIS overlay operations of topographical, hydrological,
geological and quaternary geological layers. The studied mire and stream section are
marked with thick lines.
16
4 Methods
4.1 Geochemical landscape analysis for AMD
contamination risk assessment
Landscapes result from the interaction of the geospheres, biosphere and
socio-economic systems (Forman, 1995). The physico-chemical properties
of landscape components (rock, soil, ground and surface water, vegetation
and atmosphere) and their spatial relationship relative to matter transport
determine the geochemical character of the landscape. Landscape structure
(pattern) is described in terms of geochemical abundances and geochemical
gradients. Landscape functioning is described in terms of element migration
(chemical reactions), geochemical flows (transport processes) and landscape
geochemical barriers. Landscape dynamics is described by landscape classi-
fication and historical geochemistry. The basic geochemical system, elemen-
tary landscape, is defined as a landscape volume which is homogeneous with
respect to matter transport pattern and forms a part of the land surface in
which the interactions between rocks, surface- and ground water, organisms,
atmospheric constituents and solar energy are maintained constant at any
given time (Fortescue, 1980). In practice, landscape units are mapped which
are areas homogeneous with respect to at least one landscape component.
Soil, the product of these interactions, serves as the criterion for determining
the lateral boundaries of the elementary landscapes (Glazovskaya 1963).
Elementary landscapes extend vertically from the top of unweathered rock to
the uppermost level of vegetation cover. Geochemical input and output is
called positive and negative extra landscape flow (+/- ELF), respectively.
Matter flow within the landscape can form the main migration cycle (MMC)
if element migration is cyclic in the landscape, and the landscape geochemi-
cal flow (LGF) when element migration is characterised by flow through the
landscape (Kozlovskiy 1972). In order to simplify mapping of soil types and
soil variants, soil and land cover types are clustered into three groups on the
basis of the migrational environment of chemical elements (Polynov, 1956).
Where the water table is always below the surface, an eluvial landscape
characterised by downward salt movement in well-drained soils is formed. If
the water table and the daylight surface coincide a super-aqual landscape
(bog or marsh) appears, where the dominant pattern of salt movement is
upward and horizontal. A lake or stream environment, where the water table
is above the land surface, is called an aqual landscape characterised by circu-
17
lar salt movement. Classification of elementary landscapes have been ex-
tended with the consideration of their spatial relationships through surface
and subsurface matter flow in a way analogous to the soil catena concept of
soil science (Glazovskaya 1963; Gerrard 1992). Landscape classes can be
extended with the „trans-„ prefix emphasising that the given landscape is
characterised by throughflow. For example, on the basis of transport intensi-
ties in surface waters active stream segments are classified as trans-aqual
landscapes while lakes and stagnant waters are called aqual landscapes.
Similarly, eluvial landscapes are subdivided as eluvial, trans-eluvial and
eluvial-accumulative landscapes on hill tops where erosion and downward
movement of elements dominates, on hill slopes where down-gradient
throughflow dominates and in valleys which are characterised by deposition
(Glazovskaya, 1963). In this study the delineation and classification of ele-
mentary landscapes were implemented with GIS by the overlay of the hy-
drology layer and the slope break model derived from the digital elevation
model (DEM) (Fig. 2) (Papers I, II and IV). Then, relationships between
landscapes were analysed by matter transport models, such as run-off and
watershed models derived also from the DEM (Fig. 2).
Elementary landscapes that are geochemically related by material trans-
port (landscapes in a watershed, or areas at recharge and discharge regions of
the same groundwater system, for example) form a geochemical landscape.
Geochemical landscape analysis studies geochemically „dependent‟ land-
scapes where the geochemistry of the area such as stream segments is influ-
enced by another landscape such as an area with mineralisation up-slope. In
this study a run-off model derived from the DEM was used for calculation of
surface matter transport vectors (Fig. 2).
Elementary landscapes were used to define geochemical barriers. These
were classified based on matter transport directions at elementary landscape
boundaries, and the geochemical character of landscapes determined by their
acid neutralising capacity and redox conditions on either side of the barrier
(Perelman, 1986) (see Fig. 3). For example, in the eluvial landscapes, char-
acterised by podzol soils underlain by granitic glacial till, groundwater is
likely to be oxidising, and in the super-aqual mire landscape it is reducing,
while active stream segments are also oxidising environments in the study
area (Figs 1, 2 and 3). This information was added to the attributes of ele-
mentary landscape polygons in the GIS database (Fig. 2). Then, the elemen-
tary landscape map was overlaid with the run-off model using GIS technol-
ogy in order to define flow direction at landscape boundaries (Fig. 2). Ac-
cordingly, the oxidising active stream segment receives weakly acidic (pH 3-
6.5) reducing gley water from the neighbouring mire and an oxidising barrier
is likely to form in the receiving stream sediments with an expected element
association of Fe, Mn and Co, for example (an A6-type barrier; see Perelman
18
1986) (Fig. 3). Where the water flow direction is just the reverse, that is
weakly acidic oxidising water from glacial eluvial landscapes discharges into
the reducing gley mire environment a reducing gley barrier forms (C2-type;
Perelman 1986) (Fig. 3).
Figure 2. A. Grey shading shows elementary landscapes derived from the hydrology layer (Fig. 2C) and the slope-break model (Fig. 2B). Geochemical landscape flow is also shown derived from the watershed (Fig. 2D) and run-off (Fig. 2E) models. The map shows elementary landscapes and their relationships by matter flow lines. Solid box: area shown in Fig. 3A. B. and C. Slope-break model derived from DEM and hydrology layer, respectively. D. and E. Watershed and run-off models derived from DEM, respectively.
19
In addition, adsorption is a major process for metal retardation in organic-
rich peat sediments, hence complex reducing gley-adsorption barriers form.
The landscape geochemical base map obtained in this way shows areas ho-
mogeneous with respect to geochemical flow patterns, their relationship by
surface and subsurface transport processes within and between geochemical
landscapes and matter flow control elements, such as geochemical barriers
(Fig. 3). The landscape geochemical base map was used for the selection of
sampling sites in this study.
4.2 Sampling and analytical procedures
Twenty-three stream sediment samples were collected with a Plexiglas
sediment-corer of 5 cm diameter. All samples were collected during Spring.
Each sample represented a composite of the loose, upper 10-15 cm of the
sediment column. At locations where the sediment was more compact and
SO4
2-
reduction was evidenced by dark colour and H2S odour, the core was
dissected into 5 cm sections and the uppermost (0-5 cm) and the lowermost
sections (25-30 cm) were reserved for separate analysis. Fifteen peat cores 5
cm in diameter were also taken using a hand-held steel tube corer in the wet-
land. Altogether fifty 10 cm long samples were selected 30 cm apart from
each other in the cores for chemical analysis. The uppermost 10 cm sections
of the cores were not selected for analysis in order to eliminate the effect of
atmospheric metal contamination. All samples were extruded into polyethyl-
ene bags in the field and transported to the laboratory on ice. Water pH was
measured in the field with a Beckman 21 pH meter. Loss-on-ignition (LOI)
was determined on sediment samples dried at 105 C for 48 hours and subse-
quently ignited at 550 C for 6 hours.
Samples were stored at +4 C until laboratory analysis. All samples for
chemical analysis were sieved with a stainless steel sieve of 2 mm mesh size.
The samples were thoroughly homogenised and 6 g test portions of each
sample were weighed into centrifuge tubes. The selective chemical extrac-
tion procedure based on the method of Tessier et al. (1979) was applied as
described in Table 1.
All extractions were made with chemicals of analytical grade or better.
Between each extracting steps the separation of solution and particulate mat-
ter was obtained by centrifugation at 3000 rpm for 15 minutes. The super-
natant was pipetted into glass tubes and acidified to pH<2 with concentrated
HNO3. The residue was rinsed with a small amount of deionised water (10
ml) and centrifuged for another 15 minutes. This second supernatant was
discarded. Analysis of test solutions for Pb, Ni, Cu, Zn, Fe and Mn was con-
20
ducted by flame atomic absorption spectrophotometry (Varian 1275). Com-
plementary to chemical extractions, XRD analysis was also performed on
selected sediment samples.
Table 1. Description of the applied sequential extraction method.
Pore water Sediment pore water was obtained by centrifugation of sediment
samples at 3000 rpm for 15 minutes.
Fraction 1 exchangeable metals and metals bound to carbonates The solid resi-
due of the pore water extraction was leached with 1 M ammonium-
acetate (NH4OAc) adjusted to pH 5 with acetic-acid (HOAc) at room
temperature for 6 hours under continuous agitation.
Fraction 2 metals bound to Fe and Mn hydroxides (reducible phase) The residue
from (1) was leached with 0.3 M sodium-dithionite (Na2S
2O
6) added
to 0.175 M sodium-citrate (NaC6H
5O
7) at pH adjusted to 4.8 with
HOAc. Continuous agitation was maintained for 6 hours at room
temperature.
Fraction 3 metals bound to organic matter and amorphous sulphides (oxidisable
phase) The residue from (2) was extracted with 0.02 M nitric-acid
(HNO3) mixed with 30% hydrogen-peroxide (H
2O
2) in 3:5 proportion
at pH 2 and heated gradually to 85°C with occasional agitation for 5
hours. After cooling 3.2 M NH4OAc was added in 20% HNO
3 and
double volume deionised water, and then kept under continuous agi-
tation for 30 minutes in order to prevent adsorption of extracted met-
als on oxidised sediments.
Fraction 4 residual metals The residue from (3) was digested with hot aqua regia
(HCl:HNO3 3:1) at boiling temperature for 1 hour, then the samples
were left to cool overnight.
Partitioning of metals between the various fractions should be considered as
operationally defined by the extraction procedure rather than a reflection of
scavenging by a specific mineral phase (Kheboian and Bauer, 1987; Rauret
et al., 1989; Nirel and Morel, 1990). Also, regarding the selectivity of the
extractants (Tessier et al., 1979; Bodek et al., 1988; Karlsson et al., 1988)
much care should be taken in the interpretation of the results. Low selectivity
has been reported for NH4OAc in fraction 1 (Gibbs, 1977) and in this study
no effort was made to extract the carbonate-bound fraction separately. No
attempt was made to separate metals adsorbed by different forms of organic
matter (Karlsson et al., 1988). Uncertainty in the metal content of the oxidis-
able phase (fraction 4) arises from the fact that some humic and fulvic acids
are extremely stable and the oxidation of all forms of organic matter may not
be complete. It is noted also that only 80-90% of residual metals are ex-
tracted by hot aqua regia (Alloway, 1990).
21
4.3 Data analysis and modelling
In this study, geochemical processes were studied as stochastic processes
and statistical tools were used for data analysis. Univariate exploratory data
analysis (EDA) techniques (Tukey, 1977; Hoaglin et al., 1983) followed by
multivariate statistical analyses, such as cluster and Q-mode principal com-
ponent analysis, were used to investigate element distribution patterns, geo-
chemical abundances and gradients in the sediments and to identify sample
populations. Correlation analysis and canonical correlations were applied to
investigate significant relationships among the variables measured. All cor-
relations were visually controlled and robustness of regression analysis was
assured by interactive outlier rejection. Throughout geochemical modelling
the use of robust and non-parametric techniques were preferred to classical
statistical methods because geochemical datasets are often characterised by
small sample sizes, multi-modal populations, outliers and non-normality
(Kurzl, 1988).
22
5 Sampling design based on geochemical landscape modelling
Sampling had a double-fold objective. In the mire, the objective was to re-
construct the spatial extent of AMD contamination plume and analyse the
controlling processes of metal transportation and retention. In the stream, the
goal was to investigate trace metal attenuation in stream sediments. Samples
representative of processes at the stream-mire oxidising barrier were col-
lected regarding the heterogeneity of this environment. For the design of
sample locations the local geochemical flow patterns were analysed by the
classification of elementary landscapes. The permanently water-saturated
wetland was classified as a reducing gley super-aqual landscape, and the
neighbouring areas as eluvial-accumulative landscapes (Glazovskaya, 1963;
Perelman, 1972) (Fig. 2). The active stream channel was classified as a
trans-aqual landscape and the stagnant fen water as an aqual landscape.
Using the DEM constructed by triangulated irregular network interpola-
tion the watershed, slope- and slope-break, and the run-off models were de-
rived, as shown in Figure 2. By the overlay of the classified elementary land-
scape map geochemical relationships between adjacent landscapes could be
analysed and geochemical barriers were classified as shown in Figure 3.
Landscape geochemical barriers were classified as complex reducing gley-
adsorption barriers where slightly acidic groundwater is discharging into the
water-logged mire sediments (Perelman, 1986). The wetland stream sedi-
ment was classified as a complex oxidising-adsorption barrier (A6-G6) ac-
cording to Perelman‟s classification scheme (Perelman 1986). This forms
when weakly acidic (pH 3-6.5) reducing gley water encounters relatively
oxidising geochemical conditions, and is characterised by the retention of
trace metals by adsorption on, or co-precipitation with, oxides and/or by
organic matter chelation. However, considering the vertical redox profile
likely to build up in the stream sediment column, the landscape geochemical
model of trace metal retention in the upper surface sediments can be refined,
as shown in Figure 3 (Paper I).
Based on the objectives of sampling and prior landscape geochemical model,
sediment-sampling locations were identified in the active stream channel
downstream to the waste rock dumps, in the effluent ditches and in the fen.
23
Figure 3. Sampling design with landscape geochemical methods and GIS process-ing. A. Landscape geochemical base map showing elementary landscapes of the study area, surface matter flow directions and classified landscape geochemical barriers. Locations of mine shafts and waste rock dumps are shown by the “■” sym-bol, stream sample locations by the “▲” symbol and locations of peat sediment cores in the mire by “●”. Drainage ditches in the mire are indicated by dotted lines. B. Landscape geochemical cross section indicating matter transport directions and patterns within and between landscapes (MMC: main migration cycle; LGF: land-scape geochemical flow; ELF: extra landscape flow). C. Conceptual model of land-scape geochemical barriers in the aqual landscape. Locations of the landscape sec-tion are shown in A. Sections 1 and 2 are located in the active stream channel and section 3 in the stagnant mire surface water (fen).
24
By the superimposition of potential metal contamination sources, such as ore
mines and waste rock dumps, the path of AMD from sources to end at com-
plex reducing hydrogen sulphide-adsorption barriers was traced (Fig. 3).
Although the water flow model is based on surface transport directions, rela-
tive homogeneity of the hydrological properties of the aquifer over the study
area was assumed and the surface run-off directions could be regarded as
reasonable approximations of groundwater flow as well.
Horizontal sampling in the mire density was higher (5 m separation in the
groundwater flow direction) close to the edge of the mire in order to detect
abrupt changes in concentration gradients due to spatially separated proc-
esses. Samples further along the flow direction were placed 10 m apart. All
samples were located 15 m apart from each other parallel to the edge of the
mire (Fig. 3). Vertical sampling density was designed to characterise geo-
chemical conditions in the peat profiles above, at and below the groundwater
fluctuation zone representing oxidising, fluctuating oxidising-reducing and
permanent reducing conditions (Fig. 3) (Paper II).
25
6. Results
6.1 Geochemical abundances
6.1.1 Stream
Robust summary statistics of metal concentrations in each fraction in the
stream sediments are listed in Table 2. According to the enrichment factors,
calculated as ratios of metal concentration in each fraction to the sums of
fractions 1 to 4 in the sediment, metals are mostly abundant in the oxide-
bound fraction (Fe 70-80%, Ni 50-60%, Pb ~40%, Zn ~30%), except for Cu
and Mn. This indicates the importance of the reducible fraction in the sedi-
ment studied. The higher proportion of Mn in fraction 1 (between 50-60%)
and the variability of Fe in the oxide-bound phase suggest that redox condi-
tions are heterogeneous and Mn is probably in its reduced form.
Organic matter, beside oxides, is a major sink for trace metals in the wet-
land stream, which is reflected by the high relative concentrations of metals
in the oxidisable phase, particularly for Cu (50-60%) and Pb (30%);
whereas, the residual fraction accounts in all cases for less than 20% of the
total metal content.
Stem-and-leaf displays, frequency histograms and box-plots of the data
sets showed polymodal behaviour with positively skewed populations and
were characterised by the presence of less than 20% outlying values, as de-
fined by the inner-fence criteria (Tukey, 1977), as shown in Figure 4A. Dis-
tribution shapes were similar for Fe and Mn in all fractions. For both ele-
ments, high frequency of values occurred at the lower limit of the range,
representing sediment samples from the stagnant surface water, and the dis-
tributions approximated the shape of a lognormal distribution. In order to
stabilise variability, the original variables were re-expressed by log-
transformation for further analysis (Velleman and Hoaglin, 1981). Cluster
analysis (CA) and Q-mode principal component analysis (PCA) of the log-
transformed data helped to further explore data structures in the multidimen-
sional space. Using the average linkage method and the Euclidean-distance
as a measure of similarity in the cluster analysis, Fe and Mn samples sepa-
rated into two groups representing sediments in the active stream channel
and sediments in the stagnant water areas.
26
Table 2. 5-number summaries of the measured variables in the stream sediments.
Metal concentrations in the fractions are in ppm. Loss-on-ignition (LOI) is in % of
dry (105 ºC) sample. Sample size: 23. Refer to Table 1 for a description of the ex-
traction method used to leach each fraction.
5-number summaries Fraction 1 Fraction 2 Fraction 3 Fraction 4
Pore water LOI
Pb Minimum 0.03 2.66 0.73 1.73 0 8.61
Lower quartile 0.63 3.66 1.93 1.96 0 44.88
Median 1.47 5 3.3 2.16 0 74.5
Upper quartile 3.13 6.99 4.06 2.53 0 88.98
Maximum 3.73 13.32 7.99 3.93 0.07 97.1
Ni Minimum 0.43 2.96 0.23 0.83 0
Lower quartile 2.03 9.99 1.83 1 0.025
Median 3.83 13.65 3.56 2 0.045
Upper quartile 5.66 20.65 5.43 3.93 0.01
Maximum 17.55 38.63 92.91 6.63 0.58
Cu Minimum 0.1 0.67 1.33 0.63 0
Lower quartile 0.7 1.43 4.66 0.93 0.02
Median 1.17 5 11.32 1.53 0.045
Upper quartile 2.2 10.32 30.3 1.9 0.1
Maximum 219.78 93.24 266.4 30.3 0.45
Zn Minimum 3 4 8.66 5.33 0
Lower quartile 10.99 20.65 21.98 10.99 0.08
Median 23.98 43.29 32.97 17.98 0.11
Upper quartile 54.61 106.56 45.29 34.3 0.53
Maximum 5661 899.1 702.63 40.29 2.9
Fe Minimum 1.33 85.25 3.33 13.99 0.15
Lower quartile 4.93 166.83 43.29 39.63 0.26
Median 104.23 1132.2 131.54 92.24 0.52
Upper quartile 278.06 2763.9 322.68 263.07 1.22
Maximum 1798.2 12187.8 648.35 1964.7 14
Mn Minimum 1.37 0.37 0.17 0.1 0.01
Lower quartile 4.6 1.03 0.37 0.53 0.28
Median 18.65 5.59 1.07 1.9 0.47
Upper quartile 41.29 24.98 1.93 6.99 0.95
Maximum 356.31 76.92 4.36 22.98 1.55
27
The observation that Fe and Mn appear to be bimodally distributed sug-
gests that Fe and Mn can selectively describe the redox conditions of the
system. In the PCA analysis of Fe fractions, shown in Figure 4C, the same
groups were apparent in the plane of the first two principal components
while no similar groups could be separated for the other metals.
Figure 4. Univariate and multivariate exploratory data structure analysis of iron in stream sediments. A. Frequency histogram of log-transformed iron data in the re-ducible phase. Note the separation of populations. B. Box-and-whiskers plot of original iron data in the reducible phase. Outlier and far-outliers are indicated by the symbols “□” and “”, respectively. C. Principal components analysis of log-transformed iron data. Mire- and stream surface water sediment samples are shown as solid boxes and empty circles, respectively, and numbers from 1 to 4 indicate the four metal fractions (see Table 1).
6.1.2 Mire
According to the data in Table 3, most of the Cu (around 50% of the total
concentrations) is found in the oxidisable phase demonstrating its high affin-
ity for organic and/or sulphide complexation. Manganese is probably in its
reduced Mn(II) state which explains its higher proportion in the exchange-
able and water soluble phases (70-80% of total concentrations). Ni is also
abundant in the exchangeable-phase (around 30% of total) indicative of its
mobility in the peat environment. Iron is mostly found in the reducible phase
(70-80% of total) which is somewhat surprising in an acidic reducing gley
environment. This can be explained, however, by the improper selectivity of
28
the extraction procedure 3 (see below) and the existence of resistant crystal-
line iron oxides which is supported by the high concentrations of iron in the
residual phase. All the metals are greatly enriched in the exchangeable or
water soluble phase in samples close to the AMD discharge showing the
location and extent of contamination.
Table 3. 5-number summaries of the measured variables in the peat sediments. Metal concentrations in the fractions are in ppm. Loss-on-ignition (LOI) is in % of dry (105 ºC) sample. Sample size: 50. Refer to Table 1 for a description of the ex-traction method used to leach each fraction.
5-number summaries
Pb Ni Cu Zn Fe Mn
Water soluble Upper eight 0.04 0.9 1.82 3.14 0.86 0.54 Upper quartile 0.02 0.58 0.98 1.38 0.34 0.42 Median 0 0.17 0.38 0.66 0.25 0.1 Lower quartile 0 0.04 0.16 0.28 0.16 0.02 Lower eights 0 0.04 0.06 0.18 0.12 0.02 Fraction 1 Upper eight 2.59 3.61 36.4 62.65 3.4 2.77 Upper quartile 1.16 3.08 9.8 36.75 2.14 2.14 Median 0.335 2.295 0.37 11.375 1.35 1.33 Lower quartile 0.11 0.18 0.18 5.6 0.77 0.95 Lower eights 0.04 0.07 0.14 3.15 0.6 0.6 Fraction 2 Upper eight 4.55 8.4 26.5 37.1 130.9 0.77 Upper quartile 4.55 6.65 10.85 28 71.05 0.39 Median 3.85 4.2 1.385 17.675 45.5 0.32 Lower quartile 3.5 2.45 1.09 7.7 21.7 0.25 Lower eights 3.5 1.82 0.98 5.95 16.1 0.18 Fraction 3 Upper eight 3.05 1.26 110.95 40.95 13.58 0.21 Upper quartile 2.38 0.95 21.7 31.15 7.74 0.21 Median 1.89 0.65 1.805 20.475 5.235 0.18 Lower quartile 1.51 0.46 0.91 5.25 3.82 0.11 Lower eights 1.23 0.39 0.7 1.58 3.01 0.11 Fraction 4 Upper eight 1.52 0.58 7 14 15.6 0.32 Upper quartile 1.36 0.44 1.9 10.8 6.6 0.2 Median 1.12 0.34 0.58 6.7 3.99 0.09 Lower quartile 1.04 0.28 0.44 4 2.38 0.02 Lower eights 0.92 0.26 0.36 3.6 1.8 0.02 pH LOI Upper eight 4.03 98 Upper quartile 3.87 98 Median 3.74 97 Lower quartile 3.64 95 Lower eights 3.54 89
29
If the AMD input were the only major source of heavy metals in the stud-
ied mire then the ratios of median total concentrations to the groundwater
concentrations would be the enrichment (or accumulation) factors due to
retardation which are in this case in the order:
Fe(530)>Pb(320)>>Cu(32)>Zn(18)>Ni(4.8)>Mn(1.4). The enrichment of
iron is hardly feasible since it is mostly mobile in reducing peat sediments
(Henrot and Wieder, 1990). Although Pb has high affinity to organic matter,
it is difficult to explain how it can have a much higher enrichment factor
than the even more strongly-bound Cu (Alloway, 1990) which is more than
100 times more abundant in the discharging AMD. Clearly, Fe and Pb have
an other, probably common, source, too, which is also suggested by their
relatively high concentrations in the residual phase. Manganese seems to
migrate with groundwater almost conservatively in the peat sediments.
In order to study the univariate structure of variables and obtain informa-
tion about their populations, stem-and-leaf displays were constructed for
each variable and outliers and far-outliers were identified using the robust
inner-fence outer-fence criteria (Tukey, 1977). Separation of populations in
the geochemical data was enhanced by the use of Sinclair's graphical method
based on probability graphs (Sinclair, 1974). For copper, as shown in Figure
5, two widely separated populations are apparent: the one with smaller val-
ues contains sample locations further away from the AMD discharge loca-
tion („background population‟) and the other all consists of higher or outlier
values and cover samples close to the AMD discharge location („anomalous
population‟) (Figs. 5A, 5B and 5C). This suggests that there are basically
two spatially separated regions in the course of Cu migration in which dif-
ferent processes are responsible for Cu distribution. (Obviously, if one proc-
ess or group of processes in a laterally homogeneous matrix, such as peat
could be assumed, one population would result.) Water soluble and ex-
changeable phase variables were most skewed probably representing the
effect of contaminant plume while residual phase distributions were most
symmetric for all metals. Symmetric distributions in the residual and oxidis-
able phases for Pb, Mn and Fe indicate that these metals are uniformly dis-
tributed in the sampled mire region because they represent the geochemical
background dominating in the time of mire formation.
Separation of populations in the multidimensional variable space was
studied by CA and PCA methods using log-transformed data (Howarth and
Sinding-Larsen, 1983). Cluster analysis for Cu using Euclidean distance as a
measure of similarity and average linkage method also confirmed the separa-
tion of a background cluster (n=32) and an anomalous one (n=18) with a
sub-cluster for samples at 20 cm depths in the latter which readily merged
into the background cluster at higher similarity levels. This result implies
that these samples are geochemically closer to the background cluster, which
30
supports indirectly the hypothesis that in the background only organic matter
controls Cu retention similarly to the relatively oxidising anomalous samples
at shallow depth. In the anomalous population strong sub-clusters were ap-
parent showing the higher inhomogenity of this region. In the plane of the
first two principal components (93% of the total variance) the same groups
and sub-groups became apparent (Fig. 5D). If water soluble phases represent
the propagating plume, then it is obvious from the figure that separation of
samples is due to differences in the adsorption-retardation processes repre-
sented by the first principal component. Although the presence of sub-
populations was observed for the other metals too, distinct separation such as
for Cu was not so obvious.
6.2 Geochemical gradients
6.2.1 Stream
The spatial heterogeneity of the barrier studied could be analysed using star
symbol plots (Chambers et al., 1983) of the Fe fractions. Star symbol plots
for samples in the active stream channel were similar in shape indicating
similar geochemical conditions (Paper I). The relative concentrations were
higher in the stream which reflect the importance of Fe in this environment
indicated in the plots by the dominance of the oxide-bound, organic-bound
and residual fractions. Deviations from this pattern indicate heterogeneity of
the stream sediments geochemical barrier. Separation of samples represent-
ing sediments in the active stream channel and sediments in the stagnant
water areas are also apparent in map (Paper I).
6.2.2 Mire
In the mire, horizontal metal distributions were studied by the construction
of isoconcentration maps for each fraction of each metal in three horizons
representing peat sediments above, at and below groundwater table fluctua-
tion (10-20, 40-50 and 70-80 cm below mire surface, respectively). As
shown in Figure 5F, it was possible to reconstruct the shape of the plume by
the collected samples, verifying the appropriateness of the sampling design.
It can also be seen that the highest gradient for copper is within about 10 m
from the AMD discharge location. The typical 'plume-shape' pattern similar
to that shown in Figure 5F was found in all phases and depths for Cu and to
a lesser extent for Ni and Zn, but was virtually absent for the other metals
with exceptions of the exchangeable phase for Mn, Fe and Pb and oxidisable
phase for Pb. Uniformly for all metals at all depths, the water soluble phase
and to a lesser extent the exchangeable phase provided the best means to
31
Figure 5. Geochemical modelling of metal behaviour in the studied mire demonstrated
on Cu. „Background‟ and „anomalous‟ samples are shown as empty circles and solid
boxes, respectively, and numbers from 1 to 5 in the figure indicate the five metal frac-
tions. A-C. Univariate exploratory data structure analysis. A. Quantile plot for Cu in
the oxidisable phase. B. Box-and-whiskers plot for Cu in the oxidisable phase. C.
Frequency histogram for Cu in the oxidisable phase after log-transformation. D. Com-
bined Q- and R-mode PCA display for Cu. Samples in the 'anomalous' population at
depth 10-20 cm are marked with 'x'. E-G. Spatial analysis. F. Horizontal distributions
of total Cu (fraction 1+..+5) at different depth as shown by contour lines using kriging
32
study the spatial distribution of the contaminant plume by yielding distribu-
tion patterns similar to that shown in Figure 5F. For copper, the gradients
were far higher in the anomalous population at all depths and phases com-
pared to the background population showing that copper retention is more
efficient in the first 10 m of contaminant propagation. Also, gradients were
higher in the anomalous population for oxidisable phase-bound Cu below
groundwater table indicating that copper is retained more efficiently in this
phase.
Vertical profiles for copper, nickel and ash content in the anomalous
population show a characteristic peak at depth 40-50 cm probably as a result
of remobilisation and deposition of reduced forms at and below the ground-
water fluctuation zone (Pakarinen et al., 1980) (Fig. 5G). The vertical profile
of Cu in peatlands following ash content was observed in other wetlands, too
(Sillanpää, 1972). This peak is absent in the background population (Fig. 5E)
which implies that similar remobilisation processes are insignificant further
away from the AMD input location. Iron has its highest concentration at 10-
20 cm depth and drops with depth because iron is in its mobile reduced
Fe(II) form below the groundwater table and it is likely to move along con-
centration gradients or be uplifted by plants to the oxidising top sediments.
6.3 Geochemical processes
Associations among metals were studied by means of the Pearson‟s correla-
tion analysis of log-transformed data. Where it was geochemically justified,
correlations were subjected to partial correlation analysis in order to correct
for possible induced correlations caused by a common variable. All correla-
tions were checked for significance at the 0.05 significance level and were
graphically examined to avoid spurious correlations. Correlation coefficients
are given in brackets in the discussion below. Potential associations were
first identified by the analysis of Spearman correlation matrices and con-
struction of Draftsman's plots (Chambers, 1983). All the correlations (Pear-
son linear correlation coefficient) were graphically examined to avoid spuri-
ous correlations. Variables were log-transformed to stabilise their variance
and robustness was achieved by the use of interactive outlier rejection. No
more than 10% of the samples were excluded in any case but in the majority
of cases it was below 5%. Pairwise regressions discussed were all significant
at the 0.05 level.
(continued) interpolation. Dark arrow shows AMD discharge into the mire. E and G.
Typical vertical profiles of partial abundances of Cu in the peat sediments in the
„background‟ and „anomalous‟ populations, respectively.
33
6.3.1 Stream
Correlation analysis for metals in the pore water of the stream sediment re-
vealed strong association between Ni and Zn (0.93), and Cu and Zn (0.85).
These elements are associated with the mineralisation and hence this fraction
could most probably reflect contamination caused by mining. In fraction 1
the correlation coefficients between Fe-Mn (0.91) and in the surface sedi-
ments also for Zn-Ni (0.82) suggest common pools for these metals. No cor-
relation with pH is apparent in this fraction, but the strong correlation be-
tween Fe and Mn and their respective strong correlation with reducible Fe
(0.92 and 0.94) suggest that they are specifically adsorbed on Fe oxy-
hydroxides. The lack of any correlation of Cu and Pb, as well as their low
enrichment in this fraction, suggests that these metals are preferably more
strongly bound in the sediment. The strong correlations of Fe with Pb (0.81),
Cu (0.84), Mn (0.89) and a somewhat weaker correlation with Ni (0.59) in
fraction 2 are indicative of Fe oxy-hydroxide scavenging. Reducible Fe and
associated metal concentrations are invariably lower in the more reducing
mire surface water sediments than in the stream sediments. Although, clus-
ters of mire and stream sediment samples can be recognised, they fall on the
same regression line for Pb (see Fig. 6A) and Mn. In the cases of Cu (Fig.
6B) and Ni, however, mire surface water sediment samples are displaced
away from the stream population regression line and do not show correlation
with Fe. A similar pattern was observed for Zn as for Ni and Cu, but was
blurred by the higher scatter of data points, thus did not lead to significant
correlation.
A significant correlation was observed between Pb and Cu (0.72) in frac-
tion 3 in most samples (Fig. 6D). This result suggests that oxidisable Pb and
Cu are extracted from a common pool. Regarding their high affinity for
complexing with organic ligands (Alloway, 1990) these metals are probably
retained by organic matter chelation in this fraction which is in accordance
with the high stability constants of Cu and Pb organic complexes (Stumm
and Morgan, 1981; Bodek et al., 1988). Based on this, one would expect
strong correlation for Pb and Cu with LOI that, however, was not found in
our case. On the other hand, Zn had the second largest concentration in this
fraction and showed positive correlation with LOI (0.72). One explanation of
these results could be that in the investigated size fraction LOI mainly repre-
sents the chemically inert, unprocessed organic debris (Tessier et al., 1980).
Taking into account that Zn has the highest biological accumulation coeffi-
cient (BAC) among the metals studied (Ni(0.03), Cu(0.13), Zn(0.9),
Fe(0.012), Mn(0.4); Brooks, 1973; Perelman, 1975) it is plausible that Zn in
this fraction is extracted from the unprocessed organic debris and, therefore,
represents metal uptake by plants. This would also explain the lack of corre-
34
lation between Pb, Cu and LOI if the amount of decomposed organic matter
is assumed to be subtle compared to the unprocessed organic remains.
Figure 6. Correlation analysis of log-transformed data (A, B, D and E). Mire- and stream surface water sediment samples are indicated by solid boxes and empty cir-cles, respectively, where relevant. Outliers rejected from the regression are marked with crosses. C shows logKa vs. pH for the Fe hydrous oxide system where solid lines refer to the limits compiled by Tessier et al. (1985). Symbols refer to data from the present study. F shows the canonical correlation between metals in fractions 1 and 3.
In the residual fraction (fraction 4) Fe and Pb (0.83) (Fig. 6E) and Fe and Mn
(0.89) were strongly related to each other. Regarding these element associa-
tions and correlations of Fe and Mn with the ash content (0.97 and 0.94,
respectively) it is supposed that the residual fraction represents the natural
geochemical background. Mainly minerals, such as feldspar, quartz and
hornblende, as shown by the XRD analysis, and perhaps amorphous oxides
and clay minerals characterise the composition of the residual fraction.
Moreover, the lack of association of Ni, Cu and Zn in the residual fraction
might further support our earlier hypothesis that these metals originate from
contamination by ore mining in the geochemical landscapes studied in this
research.
35
6.3.2 Mire
In the mire, the high negative correlations between pH and Ni (-0.76), Mn
(-0.76) and Zn (-0.63) in the water soluble phase suggest that the mobility of
these metals is most dependent on pH in peat sediments which is in accor-
dance with their higher relative concentrations in this (and exchangeable)
phase compared to the other phases. The strong association of Ni, Cu, Zn
and Mn as expressed by their mutually high correlations (between 0.72 and
0.90) in this phase shows that their main common source is the AMD dis-
charge. Direct relationship between exchangeable metals and pH was not
evidenced by correlation analysis. Negative correlations between pH and
oxidisable Pb (-0.75), Cu (-0.56) and Ni (-0.54) were found only for samples
identical to the background population of copper. The correlation can be
explained with the pH dependent 'hydrolysis + adsorption' model (Stumm,
1987) represented in a general form (M: metal):
(H2O)
5M(OH
2)
z+
+ SOH = SOM(H2O)
5
(z-1)
+ H3O
+
(1)
More interesting is the absence of similar correlations in the anomalous
region. This shows that processes regulating metal concentrations in the
oxidisable phase here do not influence or depend on pH directly. Partial cor-
relation on water soluble and oxidisable fractions and pH caused the elimina-
tion of the related correlations which means that these variables are mutually
interrelated in the studied peat system. Correlation between ash content and
metals in the anomalous samples in oxidisable phase was found only for Cu
(0.87) which means that ash content represents primarily oxidisable phase
Cu in this area. The pH dependence of the water soluble phase and the oxi-
disable phase metals and the absence of correlation with exchangeable met-
als suggest that metals are either in the pore solution or form strong inner-
sphere complexes with organic ligands. That is the exchangeable outer-
sphere positions are relatively unimportant in the adsorption process at the
given pH. This may imply that metal adsorption to organic substances is
kinetically fast. In the operationally defined residual phase (fraction 4) sig-
nificant correlation were found between Ni and Cu (0.63), Ni and Fe (0.64)
and Pb and Fe (0.54). The strong negative correlation (-0.71) between pH
and ash content in the background shows that metals in pH dependent phase
are most responsible for the ash content in this region. Considering the re-
sults discussed above, this implies that the ash content for these samples
represents organically bound metals mainly. The absence of similar relation-
ships in the anomalous sample group shows, in accordance with the finding
in fraction 3, that high ash content and metal retention is mainly caused by a
process related to the oxidisable phase in association with Cu. High canoni-
cal correlation was found only for the oxidisable phase (r=0.91) confirming
our hypothesis of previous modelling that retention of metals from solution
36
is primarily due to organic adsorption. The canonical correlation coefficient
is far higher than any individual correlation justifying the use of canonical
variables. It is interesting that samples from both the background and
anomalous region are scattered along the same line indicating that complexa-
tion by organic matter is the main process even in the latter population.
37
7 Geochemical landscape analysis for AMD impact assessment
7.1 Stream environment
Enrichment factors of metals, as well as results of the correlation analysis
confirm the importance of the reducible and the oxidisable phases in control-
ling metal speciation (Fig. 7). Since Mn does not show correlation with any
of the metals except for Fe, it is most probably found occluded in Fe oxy-
hydroxides and does not form a separate phase. Correlations indicate that Pb
and Mn are controlled by co-precipitation in Fe oxy-hydroxides both in the
stream and mire water sediments. Cu and Ni distribution, on the other hand,
can be related to Fe oxy-hydroxides only in the stream sediment samples. As
these metals are primary constituents of the ore mineralization and are also
characterised by associations in the pore water phase, in this fraction they are
regarded as recent contamination originating from ore mine waste rocks.
Therefore, the organic rich wetland stream sediment can be regarded as an
oxidising-adsorption barrier with respect to these metals, as it was hypothe-
sised in our prior model.
Taking into account these constraints and also that rigorous equilibrium
conditions required for thermodynamic calculations of metal adsorption
could not be assumed we applied the adsorption equations proposed by Ben-
jamin and Leckie (1981a,b) to further investigate the sediment samples. Ad-
sorption density ( ) for iron oxy-hydroxides and apparent equilibrium con-
stants (Ka) were calculated and compared to data on trace metal adsorption
on iron oxy-hydroxides compiled by Tessier et al. (1985). The calculations
for Cu, shown in Figure 6C, suggest also that Cu in the reducible fraction in
the stream sediment could only be controlled by adsorption on Fe oxy-
hydroxides. This is coherent with the results of the correlation analysis. Cal-
culations provided similar results for Zn but not for Ni.
Regarding organic matter in the sediments, its chemical activity can be
significantly different within the investigated size fraction due to the pres-
ence of both the decomposed, chemically reactive humic substances and the
chemically inert, unprocessed organic debris. However, fraction 3 could also
represent amorphous sulphide compounds. The formation of metal sulphides
38
is dependent on the availability of reducible sulphur and reactive metals, low
redox potential, pH influencing the solubility of sulphides, and microbial
activity of sulphate reducers for which decomposable organic matter and
circum-neutral pH are required (Rabenhorst and James, 1992).
Figure 7. Global cycle of iron and matter flow control processes. A. Qualitative description of global iron cycle. B. Control mechanisms of iron in landscape systems with emphasis on the fate of iron in AMD, indicated by bold arrow in A. (Compiled after Bigham et al., 1992 and Schwertmann and Fitzpatrick, 1992).
39
It has been found that sulphate reduction is usually restricted to microsites in
sediments where microbes can easily create living conditions favourable for
their activity (Pakarinen et al., 1980; Shotyk, 1988; Rabenhorst and James,
1992). In a tidal marsh, for example, it was found that sedimentary oxides
and sulphate reduction with subsequent formation of pyrite coexisted in the
same macro-environment (Rabenhorst and Haering, 1989). Thus the exis-
tence of oxidised forms extracted in fraction 2 does not exclude the possibil-
ity of sulphate reduction. In the sediments studied, metals and sulphate could
be available from the discharging AMD and pH (pH is around 6 on average
in the stream water) and decomposable organic matter in the sediments
would not limit metal sulphide formation. The characteristic H2S smell and
black colour around organic debris could be observed in some of the sedi-
ment samples in the field, these evidence that sulphate reduction does occur
locally but probably does not dominate organic matter decomposition in the
sediment.
Recent metal contamination caused by mining would appear in the labile,
surface-bound positions of the sediments, and its spatial distribution could be
studied by element associations in the pore water and exchangeable phases.
In the pore water fraction samples representing the stagnant surface water
sediments fell on the same regression line as those of the stream sediment
samples, metal concentrations being systematically higher in the mire pore
water fraction. This indicates that metal input reaches the stream from the
neighbouring landscapes and the spatial distribution pattern of metals in the
sediments does not result from a single upstream source. Potential pools for
the retention of metals in fraction 1 could be outer-sphere positions on min-
erals, such as clays, as well as organic matter surfaces and structural posi-
tions in carbonates. It is known that most of the outer-sphere positions in the
sediments are occupied by alkaline and alkaline earth metals, H and Al,
which exist in concentrations greater than trace metals in the stream water
(Stumm and Morgan, 1981).
Canonical correlation analysis was used to analyse the interaction of sev-
eral mutually related factors to model more realistically adsorption processes
in the sediments studied. High canonical correlation (0.91) was found be-
tween metals in fraction 1 and fraction 3 (Fig. 6F). The canonical correlation
coefficient is higher than any individual correlations justifying the use of the
canonical variables. Assuming that metals in fraction 3 are adsorbed on or-
ganic matter, this indicates that probably organic matter controls metals in
the exchangeable phase.
40
7.2 Mire environment
Results show that precipitation of insoluble Cu sulphides is a major process
close to the location of contaminant discharge while adsorption by organic
matter alone may control metal distribution further away along the migration
path. There are basically three limiting conditions for sulphide formation in
peat sediments: (1) reduction of SO4, for which, in turn, sulphate in suffi-
cient quantities and appropriate biogeochemical conditions are needed; (2)
the quantity and availability of reactive metal forms and (3) the pH depend-
ent solubility of metal sulphides (Boyle, 1977; Brown, 1985; Rudd et al.,
1986; Bodek et al., 1988). In our case it seems that the quantitative require-
ments are satisfied because there is abundant metal and sulphate in the dis-
charging AMD. Reduction of sulphate is conditioned on the presence and
activity of sulphate-reducing bacteria, availability of decomposable organic
matter and low redox potential. While sulphate-reducing bacteria are neces-
sary catalysts for sulphate reduction to occur in peat sediments and require
specific environmental conditions for optimal growth, they can be active in
these acidic peat soils as evidenced by the strong H2S odour of samples be-
low the groundwater table.
However, the high content of Cu in the oxidisable phase, the presence of
two distinct populations in variables of Cu also influencing relationships of
the other metals, the distinction of samples at shallow depth in the anoma-
lous region from samples below the groundwater table for Cu, the pH inde-
pendence of oxidisable Cu in the anomalous population as opposed to its pH
dependence in the background population, all provide indirect evidence that
the most insoluble copper sulphides, probably Cu2S may precipitate closer to
the discharge location. The high ash content of samples in the anomalous
population is in accordance with Lett and Fletcher's (1979) findings that
where microbial SO4 reduction takes place peats are rich in ash. Further sup-
porting evidence is that the H2S gas formation was observable only for cores
in the background region and was absent close to the AMD input.
As Cu is efficiently retained by organic adsorption and sulphide precipita-
tion, the migrating groundwater becomes depleted with respect to copper
within 10-15 meters in its course in the mire (Fig. 5). Beyond this distance
there is little copper sulphide formation, and organic complexation controls
primarily metal retardation. In this region SO4 is also reduced at microenvi-
ronments but the generated H2S remains in gas form in the absence of Cu, as
evidenced by the field observation of H2S only in samples in this area. Here
metals extracted in the oxidisable fraction are bound to organic matter and
give the ash contents which explains the significant correlations between pH
and these variables for samples from this region.
41
Figure 8. Control mechanisms in a mire ecosystem and its relation to the global Cu cycle. Signs „+‟ and „-„ shows positive and negative correlations, respectively. Dashed lines in this part indicate energy flow (R: respiration). The two diagrams are related by their similar organisation as well. Compiled from Cameron et al. (1989), Moor (1989), Butcher et al. (1992) and Schlesinger (1991).
It seems from the analytical results that Fe and Pb were already present in
higher quantities in the peat sediments prior to the beginning of AMD dis-
charge. As shown by the analysis of the reducible and residual phases, they
originate from the background and were transported and deposited in the
42
sediment phase by background surface drainage, probably as Fe/Mn or other
partly crystallised hydrous oxides and in clay minerals. The relatively high
concentrations of Pb in these phases compared to Cu and Ni is consistent
with the fact that Pb is several times more abundant in felsic rocks than in
basic rocks. The high concentrations of these metals in fraction 2 (reducible
phase) may well be the result of the poor selectivity of the extraction proce-
dure but, as we saw the importance of micro-environments in the reducing
gley super-aqual mire landscape, it is not unreasonable to assume the pres-
ence of some oxidised Fe in these sediments, as it was reported to coexist
with SO4 reduction in tidal marshes (Rabenhorst and Haering, 1979).
In order to assess the long-term retention of metals in natural wetlands the
stability of the whole peat sediment system also has to be studied. Although
the naturally high adsorption capacity of wetlands is attractive from a techni-
cal viewpoint for AMD treatment, they are selective towards metals usually
found together in mine waste waters as it was shown by the case study. Most
prominently, mires are extremely vulnerable to dewatering introduced either
by human activity or naturally, such as changes in climate or local drainage
systems (Fig. 8). Drying is not balanced by natural feedback mechanisms
and leads to the irreversible destruction of the mire-peat sediment system
(Moor, 1989; Cameron et al., 1989). Since their metal retention is based
almost exclusively on their reducing waterlogged environments, their long-
term metal retention potential and stability depends only on a single, easily
changeable variable. Upon drying the sediments are eroded and organic mat-
ter and sulphides are fast oxidised and released to the hydrological system.
43
8 Conclusions
- A geochemical landscape analysis method using geochemical modelling,
GIS technology and landscape ecology has been developed for the risk
assessment of heavy metals in the studied sediments (Papers III).
- In accordance with the needs of environmental risk assessment of AMD,
a sequential geochemical mapping procedure has been developed for the
spatial analysis of AMD in stream and wetland sediments in the study
area (Paper IV).
- The analysis of geochemical barriers performed by GIS-based geo-
chemical landscape modelling, in addition to a detailed geochemical
modelling, can contribute to the analysis of environmental geochemical
problems by providing means for the systematic analysis of the relation-
ships between related landscapes (Papers III and IV).
- Application of a sequential chemical extraction procedure could enable
the determination of different metal sinks in surface water and peat
sediments impacted by AMD (Papers I and II).
- Results of data analysis show that heterogeneity of the geochemical bar-
rier in the stream sediment is controlled by redox gradients (Paper I). It
is suggested by the analysis that this landscape geochemical barrier can
be sufficiently characterised by the distribution of Fe fractions, but more
data is needed for the verification of the refined hypothetical models
concerning the occurrence and extent of sulphate reduction in the sedi-
ments. Despite the high overall amount of organic matter in the stream
sediments, Fe oxy-hydroxides are abundant and play major role in the
retention of trace metals by adsorption on their surfaces.
- Mn is probably specifically adsorbed on Fe oxy-hydroxides and, beside
Zn, is least retained in the stream sediment (Paper I). Pb, Cu and Ni are
44
found in considerable quantity in the reducible fraction and are sug-
gested to occur occluded in Fe oxy-hydroxides. On the other hand, or-
ganic matter provides important adsorption sites for Cu and Pb and con-
trols exchangeable metals, too. Ni, Cu and Zn are suggested to represent
the mineralization based on enrichment factor calculation and correlation
analysis in the pore water and the oxide-bound fractions. More specific
investigation of the reducible fraction is required to differentiate be-
tween the natural geochemical background and contamination by min-
ing.
- Where AMD discharges into the studied mire a hydrogen sulphide-
adsorption barrier forms in the peat sediments (Paper II). It was also re-
vealed that under the acidic conditions in the bog only the most stable
copper sulphides may precipitate and control metal adsorption in a rela-
tively short distance of AMD plume propagation (about 10 m). Where
migrating groundwater becomes depleted with respect to Cu, H2S re-
mains in gaseous form and metal retardation is mostly controlled by or-
ganic complexation.
- Results suggest that both the general geochemical characters of the mire
and the metal composition of AMD are important when assessing reten-
tion capacity of wetlands (Paper II). This means that each AMD risk as-
sessment requires individual investigation.
- The evaluation of wetlands for long-term AMD treatment also has to
consider that mires are very sensitive to changes in hydrological condi-
tions and drying of the sediments leads to erosion and hence the release
of adsorbed metals to the environment (Paper II).
45
Summary in Swedish
Surt gruvvatten (AMD) innehållande toxiska element vilka ofta år en huvudor-
sak till mark- och vattenpåverkan inom många gruvområden i Europa. Dessa
kontaminerade områden verkar som sekundära källor för förorening, särskilt
vid platser där gruvbrytningen förekommit under lång tid. Riskbedömning av
försurat gruvavfall och relaterade förorenade landområden kräver ett holistiskt
tillvägagångssätt som gör det möjligt att studera föroreningens omfattning,
emissioner och påverkan på landskapet. Riskbedömningen innefattar följande
steg: (1) en beskrivning av föroreningarnas farlighet, (2) mängd/respons (toxi-
citet) analys, (3) ämnenas transport, (4) exponeringsbedömning, (5) risk karak-
tärisering, och (6) riskbehandling. Föroreningsmodellen i mark och vatten
inkluderar därför studier av steg 1, 3 och 4 längs med källan – väg (transport)
– mottagare (exponering för människor eller ekosystem). Målet med denna
undersökning var att utnyttja en geokemiskt kontamineringsmodell kopplad
till en landskapsanalys för riskbedömning med AMD. Modellering utförs via
källa-väg-receptor kedjan genom utnyttjandet av ett enkelt geokemiskt verktyg
för landskapsanalys som sedan kan användas för att följa planeringen av
markutnyttjandet inom gruvområden. Det specifika målet med den här studien
är att presentera en applikation av geokemisk landskapsanalys med AMD
riskbedömning via en fallstudie i mellersta Sverige. Studien behandlar de
AMD metaller som kan härleds till nedlagda gruvor och deponier och dess
uppträdande i vattendrag och en våtmark vid Slättberg i mellersta Sverige.
Inom detta område, där har försurat gruvvatten påverkat omkringliggande
vattendrag och myrar under ca 70 år.
Inom ramen för undersökningen insamlades tjugotre borrprover får en bäck
som rinner genom området. Varje prov representerade de övre 15-20 cm av
sedimenten i bäcken. Femton borrprov från en närliggande myr har också
samlats in. Totalt har femtio 10 cm långa prover valts belägna 30 cm ifrån
varandra i myren för kemisk analys. Alla prover var insamlade på våren. En
sekventiell extraktionsprocess har använts för att bestämma koncentrationer
av Pb, Ni, Cu, Zn, Fe och Mn som lösligt, utbytbart, reducerbart och oxide-
rande delar. Dessutom bestämdes den organiska andelen (LOI). XRD analys
har också utförts på de valda proverna i syfte att bestämma mineralsamman-
sättningen på sedimenten. Statistiska verktyg har använts för data analys och
modellering. (EDA) tekniker följd av multivariat analys och Q-mode princi-
46
piella komponent analyser, har använts för att bl.a. undersöka fördelnings-
mönster.
Den använda geokemiska modellen och landskapsanalysen för riskbedöm-
ningen av AMD längs med källa-väg-mottagare kedjan har länkats till ett
enkelt geokemiskt landskapsanalysverktyg med användandet av en land-
skapsekologiska rymdanalys.I studie av de geokemiska processer använde
också GIS verktyg. Beskrivningen och klassificerandet av det geokemiska
landskapet har genomfört med GIS med användandet av ett hydrogeologiskt
skikt och ett lutningsskikt (DEM). Den avrinningsmodell som erhölls från
DEM användes sedan för beräkning av ämnestransporten. Denna klassifice-
rades baserat på ämnestransporten och typen på det geokemiska landskapet
bestäms av dess syraneutraliserande kapacitet och redox förhållandet. I un-
dersökningen beskrivs också olika mekanismer som reglerar fördelning och
utflödet av metaller på lång och kort sikt inom ett gruvområde.
Resultaten visar sammanfattningsvis att den använda enkla geokemiska
lanskapsanalysmodellen kan användas effektivt för beskrivning och riskbe-
dömningen av gruvavfall inom ett komplext våtmarksområde. Resultaten
visar också att den generella geokemiska karaktären av både myren och me-
tallsammansättningen av AMD är viktig för uppskattning av kvarhållande
egenskaper för våtmarker.
47
References
Alloway B. J. (ed.), 1990. Heavy Metals in Soils. Blackie, London. BAT, 2003. Draft Reference Document on Best Available Techniques for Manage-
ment of Tailings and Waste-Rock in Mining Activities, 2003. European IPPC Bureau, Joint Research Centre of the European Commission, Seville.
Benjamin M. M. and Leckie J. O., 1981a. Conceptual model for metal-ligand-
surface interactions during adsorption. Environ. Sci. Technol. 15, 1050-1057. Benjamin M. M. and Leckie J. O., 1981b. Competitive adsorption of Cd, Cu, Zn and
Pb on amorphous iron oxyhydroxide. J. Colloid Interface Sci. 79, 209-221. Bigham J. M., Schwertmann U. and Carlson L., 1992. In: H. C. W. Skinner and W.
Fitzpatrick (eds), Biomineralization of Iron and Manganese: Modern and An-cient Environments. pp. 219-232. Catena Supplement 21, Catena Verlag, Crem-lingen-Destedt.
Bodek I., Lyman W. J., Reehl W. F. and Rosenblatt D. H. (eds.), 1988. Environ-
mental Inorganic Chemistry. Pergamon Press Inc., New York. Boyle R. W., 1977. Cupriferous bogs in the Sackville area, New Brunswick, Can-
ada, Journal of Geochemical Exploration 8, 495-527. Brooks R. R., 1973. Biogeochemical parameters and their significance for mineral
exploration. J. Applied Ecology 10, 825-836. Brown K. A., 1985. Sulphur distribution and metabolism in waterlogged peat, Soil
Biology and Biochemistry 17, 39-45. Butcher S. S., Charlson R. J., Orians G. H. and Wolfe G. V. (eds), 1992. Global
Biogeochemical Cycles. San Diego, Academic Press Ltd. Cameron C. C., Esterle J. S. and Palmer C. A.., 1989. The geology, botany and
chemistry of selected peat-forming environments from temperate and tropical latitudes, International Journal of Coal Geology 12, 105-156.
CARACAS (Concerted Action on Risk Assessment for Contaminated Sites in the
European Union), 1999. Risk Assessment for Contaminated Sites in Europe. Vol. 1, Scientific Basis. LQM Press, Nottingham.
Carpenter R. H., Pope T. A. and Smith R. L., 1975. Fe-Mn oxide coatings in stream
sediment geochemical surveys. J. Geochem. Explor. 4, 349-363.
48
Chambers, J. M., Cleveland W. S., Kleiner B. and Tukey P. A., 1983. Graphical Methods for Data Analysis. Wadsworth International Group, Belmont, Calif., Duxbury Press, Boston, MA.
COM(2003) 319 final, 2003/0107 (COD). Proposal for a Directive of the European
Parliament and of the Council on the Management of Waste from the Extractive Industries. Commission of the European Communities, Brussels.
Davenport P. H., 1990. A comparison of regional geochemical data from lakes and
streams in northern Labrador; implications for mixed-media geochemical map-ping. J. Geochem. Explor. 39, 1-13.
Franzen, L. G., 1985. Peat in Sweden - an important energy resource or just a paran-
thesis?. In: Proceedings of the Peat and Environment '85 International Peat So-ciety Symposium, The Swedish National Peat Commitee, Stockholm.
Forman, R. T. T., 1995. Land Mosaics. The ecology of landscapes and regions.
Cambridge University Press, New York. Fortescue J. A. C., 1980. Environmental geochemistry: a holistic approach Springer-
Verlag, Inc., New York. Gaál G. (1990) Tectonic styles of early proterozoic ore deposition in the Fennoscan-
dian Shield. Precambr. Res. 46, 83-114. Gerrard, J., 1992. Soil geomorphology. London: Chapman & Hall. Gibbs R. J., 1977. Transport phases of transition metals in the Amazon and Yukon
rivers. Geol. Soc. Am. Bull. 88, 829-843. Glazovskaya M. A., 1963. On geochemical principles of the classification of natural
landscapes. Int. Geol. Rev. 5, 1403-1430. Henrot J. andWieder R. K., 1990. Processes of iron and mangeneses retention in
laboratory peat microcosms subjected to acid mine drainage, J. Environ. Qual. 19, 312-320.
Herbert R., 1994. Metal transport in ground water contaminated by acid mine drain-
age. Nordic Hydrology 25, 193-212. Hoaglin D. C., Mosteller F. and Tukey J. W., 1983. Undertanding Robust and Ex-
ploratory Data Analysis. John Wiley and Sons Inc., New York. Howarth R. J. and Sinding-Larsen R., 1983. Multivariate analysis. In: R. J. Howarth
(editor), Statistics and Data Analysis in Geochemical Prospecting. Handbook of Exploration Geochemistry. Vol. 2, Elsevier, Amsterdam.
Jordan, G. and D‟Alessandro, M. (eds), 2004. Mining, Mining Waste and Related
Environmental Issues: Problems and Solutions in the Central and Eastern Euro-pean Candidate Countries. Joint Research Centre of the European Commission, Ispra. LB-NA-20868-EN-C, 208 p.
49
Jordan, G., 2004. Mining and Mining Waste: Problems and Solutions in the Central and Eastern European Candidate Countries. In: G. Jordan and M. D‟Alessandro (eds.), Mining, Mining Waste and Related Environmental Issues: Problems and Solutions in the Central and Eastern European Candidate Countries. Joint Re-search Centre, European Commission, Ispra. LB-NA-20868-EN-C, 208 p.
Kozlovskiy, F. I., 1972. Structural function and migrational landscape geochemical
processes. Pochvovedeniye 4: 122-138 (in Russian). Lett R. E. W. and Fletcher W. K., 1979. The secondary dispersion of transitional
metals through a copper-rich hillslope bog in the Cascade Mountains, British Columbia, Proc. 7th Int. Geochem. Explor. Symp., Denver, Colo.
Karlsson S., Allard B. and Håkansson K., 1988. Chemical characterization of
stream-bed sediments receiving high loadings of acid mine effluents. Chem. Geol. 67, 1-15.
Kheboian C. and Bauer C. F., 1987. Accuracy of selective extraction procedures for
metal speciation in model aquatic sediments. Anal. Chem. 59, 1417-1423. Kurzl H., 1988. Exploratory data analysis: recent advances for the interpretation of
geochemical data. J. Geochem. Explor. 30, 309-322. Magnusson N. H., 1953. Malm geology. Jernkontoret, Stockholm. Mantei E. J., Gutierrez M. and Zhou Y., 1996. Use of normalized metal concentra-
tions in the Mn oxides/hydrous oxides extraction phase of stream sediments to enhance the difference between a landfill emission plume and background. Appl. Geochem. 11, 803-810.
Moor P. D., 1989. The ecology of peat-forming processes: a review. International
Journal of Coal Geology, 12, 89-103. Nirel M. V. and Morel F. M. M., 1990. Pitfalls of sequential extractions. Water
Resources 24, 1055-1056. Pakarinen P., Tolonen K. and Soveri J., 1980. Distribution of trace metals and sulfur
in the surface peat of Finnish raised bogs. Proc. 6th Int. Peat Congr., Duluth, U.S.A.
Perelman, A. I., 1975. Landscape Geochemistry. Vysshaya skhola, Moscow (in
Russian). Perelman A. I., 1986. Geochemical barriers: theory and practical applications. Appl.
Geochem. 1, 669-680. PIRAMID, 2003. P. L. Younger, L. Batty, C. Ayora, J. Carrera, L. Lovgren, J.
Loredo, M. Sauter, C. Wolkersdorfer, M. Veseliè, M. LeBlanc, B. Oehlander and J. Landin, 2003. Passive In Situ Remediation of Acidic Mine/Industrial Drainage (PIRAMID), Final Report. http://www.piramid.org.
50
Plumlee, G. S. and Logsdon M. J. (eds), 1999. The environmental geochemistry of mineral deposits. Part A: Processes, techniques, and heath issues. Society of Economic Geologists, Michigan.
Polynov, B. B., 1956. Geochemical Landscapes. Geografgiz., Moscow (in Russian). Rabenhorst M. C. and Haering K. C., 1989. Soil micromorphology of a Chesapeake
Bay tidal marsh: implications for sulfur accumulation. Soil Sci. 147, 339-347. Rabenhorst M. C. and James B. R., 1992. Iron sulfidization in tidal marsh soils. In:
H. C. W. Skinner and W. Fitzpatrick(eds), Biomineralization of Iron and Man-ganese: Modern and Ancient Environments. pp. 203-217. Catena Supplement 21, Catena Verlag, Cremlingen-Destedt.
Rauret G., Rubio R. and López-Sánches J. F., 1989. Optimization of Tessier`s pro-
cedure for metal solid speciation in river sediments. Int. J. Anal. Chem. 36, 69-83.
Rudd J. W. M., Kelly C. A. and Furutani A., 1986. The role of sulfate reduction in
long term accumulation of organic and inorganic sulfur in lake sediments. Lim-nology and Oceanography 31, 1281-1291.
Salminen, R., M. J. Batista, M. Bidovec, A. Demetriades, B. De Vivo, W. De Vos,
M. Duris, A. Gilucis, V. Gregorauskiene, J. Halamic, P. Heitzmann, G. Jordan, G. Klaver, P. Klein, J. Lis, J. Locutura, K. Marsina, A. Mazreku, P.J. O‟Connor, S. A. Olsson, R. -T. Ottesen, V. Petersell, J. A. Plant, S. Reeder, I. Salpeteur, H. Sandström, U. Siewers, A. Steenfelt, T. Tarvainen, 2003. New European wide database and geochemical atlas. Proceedings, 4th European Congress on Reginal Geoscientific Cartography and Information Systems, 2003, Bologna, It-aly, Vol. 2, pp. 655-656. Regione Emilia-Romagna, Servizio Geologoco, Italy.
Schlesinger W. H. (1991) Biogeochemistry, An Analysis of Global Change. Aca-
demic Press, San Diego. SENES, 2000. State-of-the-Art of Risk Assessment Application to Acid Rock
Drainage (ARD). A study prepared for INAP, No. 32714. Schwertmann U. and Fitzpatrick R. W., 1992. Iron minerals in surface environ-
ments. In: H. C. W. Skinner and W. Fitzpatrick(eds), Biomineralization of Iron and Manganese: Modern and Ancient Environments. pp. 7-30. Catena Supple-ment 21, Catena Verlag, Cremlingen-Destedt.
Shotyk W., 1988. Review of the inorganic geochemistry of peats and peatland wa-
ters. Earth Sci. Rev. 25, 95-176. Sinding K., 1999. Environmental impact assessment and management in the mining
industry. Natural Resources Forum, 23, 57-63. Sillanpää M., 1972. Distribution of trace elements in peat profiles, Proc. 4th Int. Peat
Congr., Otaniemi.
51
Sinclair A. J., 1974. Selection of threshold values in geochemical data using prob-ability graphs, Journal of Geochemical Exploration 3, 129-149.
Stumm W., 1987. (Ed) Aquatic Surface Chemistry. John Wiley and Sons, Inc., New
York.. Stumm W. and Morgan J. J., 1981. Aquatic Chemistry. (2nd ed.) John Wiley and
Sons, New York. Tegengren F., 1924. Sveriges ädlare malmer och bergverk. Sveriges Geologiska
Undersökning, Series Ca No. 17, (in Swedish). Tessier A., Campbell P. G. C. and Bisson M., 1979. Sequential extraction procedure
for the speciation of particulate trace metals. Anal. Chem. 51, 844-851. Tessier A., Campbell P. G. C. and Bisson M., 1980. Trace metal speciation in the
Yamaska and Saint Francois Rivers (Quebec). Can. J. Earth Sci. 17, 90-105. Tessier A., Rapin F. and Carignan R., 1985. Trace metals in oxic lake sediments:
possible adsorption onto iron oxyhydroxides. Geochim. Cosmochim. Acta 49, 183-194.
Tukey J. W., 1977. Exploratory Data Analysis. Addison-Wesley, Reading, MA. Younger P. L., Banwart S. A. and Hedin R. S., 2002. Mine Water; Hydrology, Pol-
lution, Remediation. Kluwer Academic Publishers, Dodrecht. van Leuwen, C. J. and Hermens J. L. M. (eds), 1996. Risk Assessment of Chemi-
cals. An Introduction. Kluwer Academic Press, Dodrecht. Velleman P. F. and Hoaglin D. C., 1981. Applications, Basics and Computing of
Exploratory Data Analysis. Duxbury Press, Boston. Wickman F. E., 1994. The Siljan Ring impact structure: possible connections with
minor ores in its neighbourhood. GFF 116, 145-146.
Acta Universitatis UpsaliensisDigital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology 200
Editor: The Dean of the Faculty of Science and Technology
A doctoral dissertation from the Faculty of Science andTechnology, Uppsala University, is usually a summary of anumber of papers. A few copies of the complete dissertationare kept at major Swedish research libraries, while thesummary alone is distributed internationally through theseries Digital Comprehensive Summaries of UppsalaDissertations from the Faculty of Science and Technology.(Prior to January, 2005, the series was published under thetitle “Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Science and Technology”.)
Distribution: publications.uu.seurn:nbn:se:uu:diva-6992
ACTA
UNIVERSITATIS
UPSALIENSIS
UPPSALA
2006
Recommended