Environ Geol (2006) 49: 765–779

DOI 10.1007/s00254-005-0127-7 ORIGINAL ARTICLE

G. R. Hancock
E. Turley
Evaluation of proposed waste rock dump
designs using the SIBERIA erosion model

Abstract Computer-based landscape RIA erosion model over a 1000-year

Received: 16 June 2005
Accepted: 4 October 2005 evolution models offer the ability to simulation period demonstrates
Published online: 6 December 2005 evaluate landscape stability over the waste rock dump designs using a
Ó Springer-Verlag 2005 short (annual), medium (decades to conventional stepped design of
hundreds of years) and long-term backsloping benches and caps with
(thousands of years). Modeling has angle of repose slopes provide the
advantages in that design ideas can lowest average erosion rates and
be tested, different surface material depths of incision than do other de-
properties can be evaluated and risk signs. Caution should be applied in
analysis carried out. Landscape interpreting these results as the
evolution models allow landscape SIBERIA erosion model is sensitive
surface change through time. These to parameter input and in this case
models also offer the advantage that was calibrated and run using a gen-
the landscape can be evaluated eric set of parameters that are not
visually as it develops through time, site specific. Nevertheless, the results
which is not possible with other provide a guide as to the strengths
G. R. Hancock (&) types of models. Landscape evolu- and weaknesses of different rehabil-
School of Environmental and Life Sciences, tion models can be used for not only itation designs and demonstrate the
Geology Building, The University soil loss assessment (i.e. tonnes/ insights that modeling studies can
of Newcastle, 2308 Callaghan, Australia provide.
E-mail: gggh@alinga.newcastle.edu.au hectare/year), but also to evaluate
Tel.: +61-2-49215090 the method of soil loss (i.e. rill or
Fax: +61-2-49216925 interrill erosion). This study exam- Keywords Mine rehabilitation Æ
ines a range of waste rock dump SIBERIA Æ Erosion modeling Æ
E. Turley
Manager Environment NQ Division, designs for the Minera Alumbrera Waste rock dump Æ
Xstrata Copper, Mt Isa Mines, 4825 Ltd. copper mine, Argentina. An Open-cast mining Æ
Mt Isa, QLD, Australia erosion assessment using the SIBE- Argentina

Introduction To better understand a catchment system for post-

mining rehabilitation purposes it is advisable that long-
Open-cast mining is a very efficient and cost-effective term pre-mining studies be conducted to determine
method for the extraction of mineral resources. How- background environmental data. For soil erosion and
ever, with modern technology this mining method has slope stability assessment this requires field plot studies
the ability to disturb large areas of the landscape sur- investigating different landscape construction material
face. While, both mining companies and communities over a range of slope lengths and angles and over as long
recognise the economic benefits of mining, disturbance a time period as possible. The measurement of stream
to the environment should be minimised. Consequently, sediment concentrations can also provide an integrated
cost-effective and environmentally sustainable rehabili- measure of soil erosion over a catchment (Evans 2000).
tation design is crucial for the long-term stability of an Nevertheless, long-term field studies are often not
engineered post-mining landform. practical due to time and cost constraints of many
766

mining operations. The next best option is to use The climate in the region is considered arid with a
numerical modeling to assess and evaluate erosion risk. mean annual rainfall of 160 mm per year. The nature of
Computer-based landscape evolution models (Coult- rainfall is normally in the form of intense storm events.
hard 2001) offer the ability to evaluate landscape sta- The evaporation rates greatly exceed rainfall (1385 mm/
bility over the short (annual), medium (decades to yr). Some light snow events are also experienced, but
hundreds of years) and long-term (thousands of years) these generally occur infrequently (1 or 2 times per year
(Hancock 2004). Modeling has advantages in that design for 2 days).
ideas can be tested, different surface material properties The current waste rock dump design consists of two
evaluated and risk analysis carried out. Landscape major dump areas. The northern dump area has a final
evolution models allow the landscape surface to change height of approximately 100 m and the southern dump
through time, in contrast to other models. Evolution has an approximate final height of 180 m. The current
models also offer the advantage that the landscape can design of these dumps is a benched design, with maxi-
be evaluated visually as it develops through time, which mum lifts of 50 m and berm widths of 50 m. Waste rock
is not possible with other models. Landscape evolution is placed by end tipping.
models can be used for not only soil loss assessment (i.e. In terms of waste rock dump stability, it has long
tonnes/hectare/year), but also to evaluate the method of been considered by the mine management that a ben-
soil loss (i.e. rill or interrill erosion). ched design is the most practical. However, recent
This paper discusses an erosion assessment of reha- studies of long-term erosion of the benched design of
bilitation design options for the Xstrata Alumbrera waste rock suggest that the landform will eventually
copper mine using the SIBERIA erosion model (Will- form an angle of repose outer slope (Hancock and
goose and others 1991a–d). The design options exam- others 2003). In a benched waste rock dump design there
ined in this study are considered by the mining company is increased possibility of ponding and concentrating
to be the most practical, economic and environmentally surface water flow over the crest of the berm or the
sound options available to the mine at the present time. formation of ‘‘pipes’’ through the berm which can result
Underlying the closure strategy for the Alumbrera in significant gully erosion, thus removing the effective-
operation is a desire to engineer a closure program that ness of any cover of the outer slopes. Due to the lack of
allows for a ‘‘walk away’’ at the end of operations. To environmentally benign run of mine rock material, the
achieve a ‘‘walk away’’ result for the closure of the waste current cover thickness of outer slopes is approximately
rock dumps, it is expected that a significant research 20 m, consequently any erosion can expose suboptimal
effort will be required as they are considered one of the material. Therefore, it is important that rate and depth
greatest long-term liabilities, along with the tailings of erosion be known so that this can be taken into ac-
dam, because of the high potential to cause significant count for any rehabilitation design.
environment impacts.
In this study the SIBERIA erosion model will be used
to evaluate a range of waste rock dump design options Rehabilitation designs
over a 1000 year modelled (model output at 10, 50, 100
and 1000 years) period as this time period is considered A series of rehabilitation designs, are discussed in this
best practice for mines in sensitive areas (Evans 2000). paper. These designs are considered the most practical
Output from the modeling are presented visually, and and cost-effective rehabilitation structures and have
also as a maximum depth of erosion and average erosion been created to evaluate a range of design options at the
over the hillslope. Positions of maximum erosion will be mine. In all a total of six waste rock dump designs are
located and highlighted. evaluated and are described below (Fig. 1 to 3).
Figures 1a – d are conventional designs with varying
bench widths and lift heights. Figure 1(a) displays the
Study site proposed 50 m lift and 50 m bench design for the site.
Figure 1(b) displays a rehabilitation design with 25 m
Minera Alumbrera is the operator of the Alumbrera lifts at angle of repose and 30 m wide benches.
open-cut copper/gold mine in northwest Argentina. Ore Figure 1(c) displays a 50 m lift at angle of repose and
is treated through a flotation circuit at a rate of 30 m bench design hillslope. Figure 1(d) is a single linear
approximately 37 million tonnes p.a. Flotation opera- slope at angle of repose.
tions commenced in late 1997. A total of 400 million Figures 1(e) and 1(f) are single concave slopes with
tonnes of waste are expected to be generated over the different amounts of concavity. In previous mining
mine life. There are approximately 10 years of mining rehabilitation studies, Hancock and others (2003) have
remaining and 12 years of milling. The concentrator will shown that concave slopes show potential as rehabili-
be supplied with stockpiled ore for the latter part of the tation design. The design of concave slopes is based on
operation. The mine is currently in year six of operation. the understanding that catchments with uniform
 767

Fig. 1 Hillslope profiles for the 400 400

waste rock dump design. (a) is a b
the proposed 50 m lift and 50 m
bench design for the site. (b) 350 350
displays a rehabilitation design
with 25 m lifts at angle of

metres
metres
repose and 30 m wide benches 300 300
while (c) displays a 50 m lift at
angle of repose and 30 m bench
design hillslope. (d) is a single 250 250
linear slope at angle of repose.
(e) and (f) are single concave
slopes with different amounts of 200 200
0 100 200 300 400 500 0 100 200 300 400 500
concavity. See text for more
metres metres
details
400 400
c d
350 350

metres
metres

300 300

250 250

200 200
0 100 200 300 400 500 0 100 200 300 400 500
metres metres

400 400
e f
350 350
metres
metres

300 300

250 250

200 200
0 100 200 300 400 500 0 100 200 300 400 500
metres metres

underlying geology, and with fluvial erosion dominant, Eq. 1 predicts that the concave slope of fluvially
generally have a convex upper hillslope profile and a dominated landscapes typically is steeper in the higher
concave profile moving down the slope (Kirkby 1971). reaches of the catchment, with a decreasing slope as
The concavity in the lower section of the slope profile catchment area increases. This profile produces a rapid
can be described by the area-slope relationship, which drop in elevation in the upper reaches of the catchment,
relates the area draining through a point to the slope at where catchment area and discharge is small. As catch-
that point and thus effectively quantifies the local ment area increases, slope decreases; therefore reducing
topographic gradient as a function of drainage area. the velocity of discharge and, consequently, erosive po-
This concavity is given by a relationship of the form: tential. In contrast, linear slopes allow a buildup of
runoff down the slope with no reduction in velocity. The
A/ S ¼ constant; ð1Þ equation in the area-slope relationship accordingly
where A is the contributing area to the point of interest, provides a parameter for the construction of hillslope
S (metre/metre) is the slope at the point of interest and a profiles.
is the concavity index of the hillslope for natural In post-mining landscapes, the majority of erosion
catchments, with values ranging between 0.2 and 0.7 occurs by gullying caused by fluvial erosion (Hancock
(Hack 1957; Flint 1974; Tarboton and others 1989; and others 2000). Consequently, in this paper the
Willgoose 1994). authors are only concerned with fluvial erosion and the
768

Fig. 2 Rehabilitation hillslope profile design for the Alumbrera mine using 50 m high lifts at angle of repose with 50 m benches with 2 m
humps added to the cap and benches. All dimensions are metres

fluvial component of the area-slope relationship. Ero- their long-term effectiveness. For both the humpy design
sion caused by rainsplash is considered to be insignifi- and those waste rock dumps without the surface eleva-
cant when compared to fluvial erosion on post-mining tions, all have a small amount of surface roughness
landscapes. added (±1 m) to each elevation to represent surface
In this study the area-slope relationship was deter- roughness created during construction of the landform.
mined from undisturbed catchments within the mine At all breaks in slope a 1.5 m tall and 4 m wide bund
lease boundary. Using this data (a=0.24), and a given a wall has been constructed. This represents the wall that
maximum lift length of 150 m, a constant value can be is constructed as a safety measure (a safety bund) at all
calculated and the concave hillslope profile constructed breaks in slope. This also acts as a sediment trap and
(Fig. 1(e) and 1(f)). Figure 1(f) has a concavity matching may play an important role in controlling water and
that of the surrounding natural hillslopes while Fig- sediment movement.
ure 1(e) has a concavity double that of a natural hills- For all simulations the hillslope designs were modeled
lope. While it is recognised that concave hillslopes are on a 2 m grid as this will allow erosion/gullying of the
difficult to construct, modeling of these designs will cap to be quantified and visually assessed for incision.
provide some insight into the benefits and pitfalls of such Accordingly, the maximum size of the each rehabilita-
structures. tion design is 500 m by 500 m. Figures 2 and 3 dem-
On all waste rock dumps, the flat caps and benches onstrate the designs for profiles 1(a) and 1(d) on the 2 m
backslope away from the batter at 1–2%. This directs all by 2 m grid that were used for the simulations. All other
runoff away from the steep slopes. It is also proposed designs were constructed in the same way for input into
that on all flat caps and benches that benign waste rock the SIBERIA erosion model.
will be end dumped, creating a very rough ‘humpy’ In addition, an angle of repose design (with the same
surface with humps ranging in height from 2 to 3 m. profile as Fig. 1(d)) that is constructed in three lifts and
This rough surface creates ponding and acts as a has three different ranges of particle size distribution
sediment trap. To test this humpy design, a 2–3 m tall down the slope was also evaluated (Fig. 4). This will
random elevation has been added to all flat caps and simulate the effect of particle size segregation down the
benches to simulate paddock dumping. A second series slope. This is discussed further in the ‘‘Calibration of the
of hillslopes were created without the humps to evaluate SIBERIA model’’ section.
 769

Fig. 3 Rehabilitation hillslope profile design for the Alumbrera mine using a single hillslope at angle of repose with 2 m humps added to
the cap and benches. All dimensions are metres

The SIBERIA erosion model subjected to fluvial erosion and mass transport
processes. SIBERIA links widely accepted hydrology
SIBERIA is a physically-based predictive model that and erosion models under the action of runoff and
can simulate the geomorphic evolution of landforms erosion over variable-time scales. The SIBERIA erosion

Increased
Less segregation segregation of
of particle size particle size over
along a 50m >100m bench
bench height

a Current Benched Design b Benched/Angle of Repose Design c Angle of Repose Design

Fig. 4 Conceptual dump configurations for benched and angle of lift pushed out to the limit of the bottom lift. This will result in a
repose slopes. (a) displays a conventional benched design mix of both fine and coarse material down the hillslope. (c) is an
constructed in 3 lifts and conceptually how particle size distribution angle of repose waste rock dump constructed in a single lift.
is likely to vary down each slope from fine to coarse. The angle of Conceptually particle size is likely to increase down the hillslope
repose slope in (b) is constructed in two lifts with the second (top)
770

model (Willgoose and others 1991a-d) has recently been Livingston 1995), the main difference being that SIBE-
tested and evaluated for erosion assessment of post- RIA allows the landform to evolve in response to erosion.
mining landforms (Boggs and others 2000, 2001; Evans Consequently, SIBERIA output describes how the
and Willgoose 2000; Evans and others 2000; Hancock catchment is expected to look, on average, at any given
and others 2000, 2002). SIBERIA is an important tool in time. Long-term landscape evolution is the balance be-
the understanding of the interactions between geomor- tween fluvial processes that incise the landscape and
phology, erosion and hydrologic process because of its diffusive processes which round or smooth the landscape
ability to explore the sensitivity of a system to changes in (Willgoose 1994). The sophistication of SIBERIA lies in
physical conditions without many of the difficulties of its use of digital terrain models for the determination of
identification and generalisation associated with the drainage areas and geomorphology as well as its ability
heterogeneity encountered in field studies. to efficiently adjust the landform with time in response
The sediment transport equation of SIBERIA con- to the erosion that occurs on it. The testing and vali-
sists of two terms: dation of SIBERIA is ongoing (Hancock 2003; Will-
goose and others 2003).
qs ¼ qsf þ qsd ; ð2Þ
where qs is the sediment transport rate per unit width
(m3/s/m width), qsf is the fluvial or rill sediment trans- Calibration of the SIBERIA model
port term, and qsd is the diffusive or interrill transport
term (both m3/s/m width). Before SIBERIA can be used to simulate soil erosion the
The fluvial sediment transport term (qsf) is based on sediment transport equation (Eq. 3) and area-discharge
the Einstein-Brown model which is: relationship (Eq. 5) require calibration. The fluvial sed-
iment transport equation (Eq. 3) in SIBERIA is pa-
qsf ¼ b1 Qm1 Sn1 ; ð3Þ rameterised using input from field sediment transport
3 and hydrology data. This parameterisation process is
where Q is the discharge per unit width (m /s/m width),
described in detail by Evans and others (2000) and
S the slope (m/m) in the steepest downslope direction
Hancock and others (2000) and can use field data col-
and b1, m1, and n1 are constant non-dimensional
lected from rainfall/runoff plots or laboratory flume
parameters of the model with m1 and n1 determining the
and/or rainfall simulator data (Evans and others 1998;
type of erosion (Kirkby, 1971). The rate constant b1
Hancock and others 2002; Hancock 2004). Alterna-
controls the rate of erosion and can be scaled to match
tively, a data base of SIBERIA erosion model parame-
the observed erosion rate.
ters has been developed from existing rainfall simulator
The diffusive transport term is expressed as:
and flume studies that relates soil particle size distribu-
qsd ¼ DS; ð4Þ tion to erosion rates. This latter method is less reliable
than the former.
where D (m3/s/m width) is diffusivity and S is slope. The At this study site, as rainfall is often low and infre-
diffusive term can model creep, rainsplash and land- quent, it was not possible to derive calibration data from
sliding. natural rainfall events quickly. The erosion data of
SIBERIA does not directly model runoff (Q, m3 - for Sheridan and others (2000) provides a database for spoil
the area draining through a point) but uses a sub-grid textural properties as well as its erodibility. It has long
effective parameterisation based on empirical observa- been recognised that soil properties influence erosion
tions, and justified by theoretical analysis, which con- rates. While there are many factors, such as organic
ceptually relates discharge to area (A) (m2/m width) matter content, electrical conductivity, mineralogy and
draining through a point as: rock content that contribute to soil erodibility, one of
Q ¼ b 3 A m3 ; ð5Þ the strongest influences on erodibility is soil particle size
distribution. Consequently, the textural properties or
where b3 is the runoff rate constant and m3 is the particle size distribution of soil can be used to provide an
exponent of area, both of which require calibration for indication of erodibility and can be used to calibrate the
the particular field site. SIBERIA erosion model.
For long-term elevation changes it is convenient to In this study, calibration data was derived from
model the average effect of these processes over time. rainfall simulator and flume studies for a range of soil
Accordingly, individual events are not normally modeled, and spoil materials from the data of Sheridan and others
but rather the average effect of many aggregated events (2000). To calibrate Eq. 3 the sediment concentration
over time are modeled. The physics of runoff and erosion data was used in a multiple regression between sediment
that are included in the model are broadly similar to the concentration, slope and discharge. This provides a
well understood Water Erosion Prediction Project range of values for b1, m1, and n1. From this a database
(WEPP) model (Laflen and others. 1991; Flanagan and of input parameters for SIBERIA have been determined.
 771

Using particle size data determined by O’Kane Results

Consultants (2004) erosion parameters of m1= 1.5,
n1=2.1, b3=1, m3=1 and b1=0.00075 were determined Results are presented for simulations using the
for the Alumbrera rainfall and waste rock material high average annual rainfall of 400 mm/year and
(Table 1). The values of m1 and n1 determine the type of 160 mm/year representing the average annual rainfall
erosion. Values of m1=1 and n1=1 produce sheet wash for the Alumbrera site. Results are presented as average
erosion while m1=2 and n1=2 produce incision or rates and depths of erosion (Tables 2 to 5) and also
gullying (Kirkby 1971). Therefore, the parameters visually (Figs. 5 to 7), displaying how and where erosion
derived for this site indicate that erosion will be domi- occurs on the different waste rock dump designs over the
nated by incisive type processes. To provide a range of 1000-year simulation. Two simulations were carried out
possible outcomes, a high set of parameters were for each design, one with 2 m humps on the surface of
calculated and also used in the SIBERIA model to the flat cap and benches and one without humps. The
represent an average annual rainfall of 400 mm/year. results of the SIBERIA simulation for each dump design
Further simulations were also run using m1= 2.0, are discussed below.
n1=2.1, b3=1, m3=1 and b1=0.00075 to evaluate the
effect of enhanced gullying on the waste rock dump
designs (Table 1). Designs using multiple benches
It is well understood that rock particles segregate
from small to large when dumped down a hillslope. This Results of the simulations demonstrate over the 1000-
segregation effectively changes infiltration rates down a year modeled period show (Tables 2 and 3) the
slope, from a low value at the top of the hillslope to a majority of erosion occurs down the angle of repose
high infiltration at the bottom of the slope, where the slopes, with little erosion (other than redistribution) on
particles are large. SIBERIA has the ability to spatially the flat cap and benches (Fig. 5). Erosion down the
vary runoff over a hillslope. This allows different rainfall angle of repose slopes occurs mostly by gullying. Gul-
and/or infiltration rates to be input into the model as a lying is most evident for the high rainfall simulations.
result of segregation. In this study, the effect of spatially At no time over the 1000 year simulation is the bund
variable particle size distribution on runoff, and resul- wall (safety wall) at the edge of the flat cap and benches
tant erosion rates, were evaluated using SIBERIA on the breached. Considerable deposition is evident at the foot
angle of repose hillslope (Fig. 1d, 3 and 4). For this of all slopes.
waste rock dump design (Fig. 3), the hillslope has been The results demonstrate that maximum average ero-
constructed in three 50 m lifts to produce a single angle sion and maximum depth of erosion occur for the 50 m
of repose slope 150 m tall, producing three sets of par- bench and 50 m lift design (design 1a, average depth of
ticle size distributions down each 50 m lift (Fig. 4). erosion = 1.85 m, maximum depth of erosion =
Particle size distribution of mine waste rock is related 2.188 m), with the lowest erosion occurring for the 25 m
to the geology itself, blast energy used to break up the lift with 30 m benches (design 1b, average depth of
intact rock, as well as mechanical handling of the erosion = 1.50 m, maximum depth of erosion =
material. All these factors can and do vary on a minesite. 1.605 m). Nevertheless, the erosional differences be-
Consequently, there is little data in the literature, despite tween the various bench and lift designs are not large.
it being well recognised. For this study, there is no data The design using 25 m lifts and 30 m benches (design 1b)
for the Alumbrera mine in regard to particle size dis- has the lowest erosion of all designs due to the shorter
tribution down a slope. To assess the impact of different slope lengths.
particle size distributions down a hillslope, and spatially Simulations using the average annual rainfall data
variable infiltration rates, the b3 value (Eq. 5) was varied demonstrate a considerably lower rate of erosion com-
at constant rate from 1 to 0 for each of the three lifts. pared to the high rainfall data (Table 3). Similar to the
This effectively varied infiltration and consequent runoff high rainfall simulations, the designs using a 50 m bench
down the angle of repose slope. with 50 m lifts (design 1a) had the highest average an-

Table 1 Parameters used in the

SIBERIA erosion model (see High Rainfall Low Rainfall/Low Low Rainfall/High incision
text for an explanation of these (400 mm/year) incision (160 mm/year) (160 mm/year)
parameters)
m1 1.5 1.5 2.0
n1 2.1 2.1 2.1
b3 1 1 1
m3 1 1 1
b1 0.00225 0.00075 0.00075
772

Table 2 Results from the SIBERIA simulations for the waste rock dump designs displayed in Fig. 1 with humps on the caps and benches
using erosion model parameters for an average annual rainfall of 400 mm per year

Design Depth of erosion (m) 10 years 100 years 1000 years

Humps No humps Humps No humps Humps No humps

a Average 0.005 0.004 0.037 0.031 0.246 0.185

Minimum 0 0 0 0 0 0
Maximum 0.275 0.28 0.802 0.807 2.17 2.188
b Average 0.003 0.003 0.029 0.024 0.198 0.150
Minimum 0 0 0 0 0 0
Maximum 0.246 0.070 1.12 0.450 1.58 1.605
c Average 0.005 0.004 0.036 0.030 0.241 0.180
Minimum 0 0 0 0 0 0
Maximum 0.199 0.199 0.715 0.715 1.92 1.949
d Average 0.012 0.011 0.056 0.048 0.413 0.332
Minimum 0 0 0 0 0 0
Maximum 0.49 0.490 1.86 1.855 8.09 8.095
e Average 0.011 0.010 0.044 0.040 0.262 0.228
Minimum 0 0 0 0 0 0
Maximum 0.574 0.550 1.10 1.012 3.64 3.750
f Average 0.012 0.011 0.055 0.051 0.372 0.345
Minimum 0 0 0 0 0 0
Maximum 0.632 0.656 1.58 1.700 5.36 6.010

Table 3 Results from the SIBERIA simulations for the waste rock dump designs displayed in Fig. 1 with humps on the caps and benches
using erosion model parameters for an average annual rainfall of 160 mm per year

Design Depth of erosion (m) 10 years 100 years 1000 years

Humps No humps Humps No humps Humps No humps

a Average 0.002 0.002 0.016 0.014 0.094 0.074

Minimum 0 0 0 0 0 0
Maximum 0.114 0.114 0.633 0.634 1.210 1.226
b Average 0.001 0.001 0.011 0.009 0.078 0.061
Minimum 0 0 0 0 0 0
Maximum 0.101 0.025 0.548 0.220 1.330 0.937
c Average 0.002 0.002 0.016 0.014 0.094 0.072
Minimum 0 0 0 0 0 0
Maximum 0.079 0.079 0.415 0.415 1.320 1.34
d Average 0.005 0.005 0.030 0.027 0.150 0.121
Minimum 0 0 0 0 0 0
Maximum 0.255 0.254 1.12 1.12 3.282 3.28
e Average 0.007 0.007 0.0251 0.023 0.109 0.096
Minimum 0 0 0 0 0 0
Maximum 0.544 0.508 0.836 0.756 2.380 2.25
f Average 0.008 0.007 0.029 0.023 0.152 0.096
Minimum 0 0 0 0 0 0
Maximum 0.543 0.507 0.944 0.014 3.261 2.251

nual erosion of 0.074 m, with maximum depth of ero- Design using a single angle of repose slope
sion of 1.34 m occurring for the design using 50 m lifts
with 30 m benches (design c, Fig. 1). Results for design 1d with 2 m humps on the cap and
The simulations without the humps on the cap and benches demonstrate an average erosion rate of 0.413 m
benches show a slightly lower average erosion rate, but and 0.150 m, and a maximum depth of erosion of
similar maximum depths of erosion as the simulations 8.09 m and 3.28 m for the high and low rainfall simu-
using humps. The slightly lower average erosion rate is a lations, respectively, over the 1000-year modeled period
result of less erosion occurring on the cap and benches (Tables 2 and 3). In this design, the majority of erosion
as there is less soil redistribution due to the reduced occurs down the long angle of repose slope, with little
surface roughness. erosion (other than redistribution) on the flat cap. Ero-
 773

Table 4 Results from the SIBERIA simulations using different and 2.38 m for the high and low rainfall simulations,
erosion parameters for the waste rock dump design using 50 m lifts respectively, over the 1000-year modeled period, while
and 50 m benches (Figs. 1a and 2)
the shallow concavity slope produces an average ero-
Simulation Depth of 10 years 100 years 1000 years sion rate of 0.372 m and 0.152 m and a maximum
erosion (m) depth of erosion of 5.36 m and 3.26 m for the high
and low rainfall simulations, respectively, over the
m1=1.5 Average 0.002 0.016 0.094 1000-year modeled period (Tables 2 and 3). In this
Minimum 0 0 0
Maximum 0.114 0.633 1.210 design, the majority of erosion occurs down the long
m1=2.0 Average 0.005 0.038 0.244 concave slope with little erosion (other than redistri-
Minimum 0 0 0 bution) on the flat cap. Erosion down the concave
Maximum 0.409 1.065 5.707 slope occurs mostly by gullying (Fig. 7). Gullying is
most evident for the high rainfall simulations. At no
time over the 1000-year simulation is the bund wall
sion down the angle of repose slopes occurs mostly by (safety wall) at the edge of the flat cap and benches
gullying (Fig. 6). Gullying is most evident for the high breached. Considerable deposition is evident at the
rainfall simulations. At no time over the 1000-year foot of the slope. The shallow and deep concave slopes
simulation is the bund wall (safety wall) at the edge of have the second and third highest erosion rates of all
the flat cap and benches breached. Considerable depo- the designs.
sition is evident at the foot of the slope. This design has Further simulations without the humps on the cap
the highest erosion rate and the highest depth of incision show a slightly lower average erosion rate, but similar
of all the designs. maximum depths of erosion as the simulations using
Further simulations for design 1d without the humps humps. The slightly lower average erosion rate is a result
on the cap show a slightly lower average erosion rate, of less erosion occurring on the cap and benches as there
but similar maximum depths of erosion as the simula- is less soil redistribution due to the reduced surface
tions using humps. The slightly lower average erosion roughness.
rate is a result of less erosion occurring on the cap and The results demonstrate that concave designs with
benches as there is less soil redistribution due to the deeper concavity perform better in terms of erosion than
reduced surface roughness. those with shallow concavity over the same average
slope length and height. This is a result of the higher
concavity slopes losing elevation quicker in the higher
Design using a single concave slope reaches of the hillslope where there is less runoff and
with deep concavity having a less steep slope in the lower reaches of the
catchment where runoff is greatest.
Two simulations were carried out for the design using a
single concave slope with a shallow and deep concavity
(Figs. 1e and f). These were a design with 2 m humps on Design using a 50 m lift and 50 m benches
the surface of the flat cap and benches and one without and using a m1 =2.0
humps. It is recognized that this design is difficult to
construct, but the simulations provide an indication of As discussed in ‘‘Calibration of the SIBERIA model’’
the advantages and disadvantages of such design. section, the calibration process used in this study uses a
Results with 2 m humps on the cap and benches generic data set so it is appropriate that an evaluation of
demonstrate an average erosion rate of 0.262 m and different parameters be conducted. In this study a higher
0.109 m and a maximum depth of erosion of 3.64 m value of m1=2.0 (all other simulations used m1=1.5)
was run to assess the impact of enhanced gullying (as a
result of the higher value of m1) on the 50 m lift and
Table 5 Results from the SIBERIA simulations for the waste rock
dump designs with humps on the cap and single angle of repose 50 m bench design (Figs. 1a and 2). This simulation
slope constructed in three lifts used the waste rock dump design that had 2 m humps of
the cap and benches and high rainfall.
Simulation Depth of 10 years 100 years 1000 years Results of the simulation demonstrate an average
erosion (m)
erosion rate of 0.244 m and maximum depth of 5.71 m
Constant runoff Average 0.005 0.030 0.150 over the 1000 year simulation (Table 4). This compares
Minimum 0 0 0 to an average erosion value of 0.094 m, a maximum
Maximum 0.255 1.12 3.282 depth of erosion of 1.21 m using m1=1.5 (as discussed
Spatially varied Average 0.002 0.019 0.082 in the ‘‘Design using multiple benches’’ section) over the
Minimum 0 0 0
Maximum 0.125 0.690 2.029 1000-year modeled period. This represents a three-to-
four fold increase in erosion over the modelled period
774

Fig. 5 SIBERIA simulation of a rehabilitation hillslope design for the Alumbrera mine using 50 high lifts at angle of repose with 50 m
benches using high rainfall of 400 mm/year (top) and low rainfall of 160 mm/year (bottom) after 1000-years of erosion. All dimensions are
metres

and demonstrates the sensitivity of the SIBERIA model gullying with little erosion (other than redistribution) on
to different parameters. It also highlights the need for the flat cap and benches. Despite the enhanced rate and
the determination of correct erosion model parameters depth of erosion at no time over the 1000-year simula-
for long-term simulations. tion is the bund wall (safety wall) at the edge of the flat
As for the simulations using m1=1.5, the majority of cap and benches breached. Considerable deposition is
erosion occurs down the angle of repose slopes by evident at the foot of all slopes.
 775

Fig. 6 SIBERIA simulation of a rehabilitation hillslope design for the Alumbrera mine using a single hillslope at an angle of repose using
high rainfall of 600 mm/year (top) and low rainfall of 160 mm/year (bottom) after 1000 years of erosion. All dimensions are metres

Design using a single angle of repose slope angle of repose slope (Fig. 1d) and spatially variable
and with spatially varied runoff runoff down the angle of repose slope (see the ‘‘Cali-
bration of the SIBERIA model’’ section). For this
It is well understood that rock particles segregate from waste rock dump design (Fig. 4), the hillslope has been
small to large when dumped down a hillslope. A constructed in three 50 m lifts to produce a single angle
simulation was carried out for the design using a single of repose slope 150 m tall, this producing three sets of
776

Fig. 7 SIBERIA simulation of a rehabilitation hillslope design for the Alumbrera mine using a single concave hillslope with deep
concavity using high rainfall of 400 mm/year (top) and low rainfall of 160 mm/year (bottom) after 1000 years of erosion. All dimensions
are metres

particle size distributions down each 50 m lift. These (Table 5). This compares to an average erosion rate of
simulations used 2 m humps on the surface of the flat 0.150 m and a maximum depth of erosion of 3.282 m,
cap. with no segregation taken into account. This represents
Results demonstrate an average erosion rate of a reduction of erosion of approximately half when
0.082 m and a maximum depth of erosion of 2.029 m compared to the same slope with no segregation.
 777

As in all other designs, the majority of erosion occurs tained on the benches. While a small amount of erosion
down the long angle of repose slope with little erosion occurred on the benches, the majority of erosion oc-
(other than redistribution) on the flat cap with erosion curred on the angle of repose slopes. The results dem-
down the angle of repose slopes occurring mostly by onstrate that the longer the angle of repose slope, the
gullying. At no time over the 1000-year simulation is the greater the erosion. Deposition can be observed at the
bund wall (safety wall) at the edge of the flat cap and foot of all slopes.
benches breached. Considerable deposition is evident at In terms of depth of incision the waste rock dumps
the foot of the slope. with design b (Fig. 1) provides the lowest average ero-
Overall, this simulation greatly reduced erosion when sion rate and depth of incision of 1.58 m and 1.33 m for
compared to the simulation which did not take into the high and low rainfall simulations respectively. The
account particle sorting down the slope. Nevertheless, as highest depth of incision, with a maximum depth of
there is no field data with which to justify the particle 8.09 m and 3.28 m for the high and low rainfall simu-
sorting on the waste rock dumps and therefore the lations over the 1000-year modeled period, occurs on the
runoff calculations used in the model the results found in angle of repose slope (Fig. 1) (design d). Nevertheless,
this study should be viewed as indicative only. This is an designs b and c also have low erosion rates and maxi-
area for further research. mum depths of incision. Overall, the results demonstrate
that constructed hillslopes with benches provide the
lowest erosion rates. The design using 30 m benches with
Discussion 25 m lifts (Fig. 1b) provides the lowest overall erosion
rate, but the difference is small compared to the other
Computer modeling provides a design tool with which to benched designs.
assess the benefits and limitations of different waste rock Nevertheless, caution should be applied in inter-
dump designs. It also provides a method to evaluate the preting these results. The SIBERIA erosion model was
strengths and weaknesses of available environmental calibrated and run using a generic set of parameters
knowledge and provides a framework for designing fu- that are not site specific. For the calibration, a particle
ture research programs with which to fill knowledge size distribution from fresh waste rock was used. No
gaps. This aims to ensure that environmental rehabili- account was taken of the chemical properties of the
tation is conducted with the best technology using the spoil and its effect on erodibility as the data was not
best information. available. Also, the effect of weathering and armouring
The SIBERIA model used here to assess the various of fresh spoil together with the establishment of vege-
waste rock dump designs not only allows average erosion tation may retard or enhance erosion over the long-
rates to be assessed, but also how and where erosion oc- term and it is likely that erosion rates will change
curs. It also allows erosion to be visualized, something through time as weathering and armouring occurs
that is not possible with WEPP (Flanagan and Livingston which may either increase or enhance erosion through
1995) or USLE (Wischmeier and Smith 1978). This is time. The importance of these factors for long-term site
particularly important for mines such as Alumbrera where stability is unknown at the present time. Long-term
potentially acid generating material is encapsulated by a rainfall data is required to correctly input rainfall and
thick cover of benign material. The modeling results resultant runoff into the model. Accordingly, to ensure
demonstrate that any of the designs are sufficient to pre- the validity of these modeled results site specific cali-
vent exposure of potentially acid generating material if a bration data is required. This requires that field sites be
conservative cover thickness of 20 m of benign waste rock established for long-term monitoring of runoff and
is used for periods of at least 1000 years. sediment transport on rehabilitated spoil over arrange
The results demonstrate that waste rock dump de- of slopes and slope lengths.
signs using angle of repose slopes with benches provide Also the hillslopes investigated in this study are
the most erosionally stable design over the modelled constructed to precise designs. There has been no ac-
period. The waste rock dumps with a single slope at count of construction error that may result in ponding
angle of repose and the single concave slopes all have or redirection of flows and their influence on erosion.
higher erosion rates than the benched designs. Small These factors require further investigation to better
amounts of incision can be observed on the angle of guarantee the reliability of the modeling. This is of
repose slopes with benches while considerable incision particular importance as if ponding occurs due to poor
exists on the single angle of repose and concave slopes. construction and overtops a contour bank, severe gul-
There appears to be little difference between simulations lying may result.
using the humpy design on the flat cap and benches While the simulation investigating the effect of par-
compared to those with no humps. ticle segregation down an angle of repose slope shows
For all designs, the backsloping benches prevented great promise as a potential rehabilitation design, fur-
the bunds being breached and all sediment was con- ther information is required on the effect of particle
778

segregation in infiltration and runoff rates. There ap- Conclusion

pears to be a dearth of information in the literature
regarding particle size distribution down constructed Erosion models can be used as rehabilitation design tools
angle of repose slopes and this study’s result should be for post-mining landscape design. In this study the
viewed as indicative only. This is an area for consider- authors have used the SIBERIA landscape evolution and
able further research. erosion model to evaluate a series of potential waste rock
There is no data on background or natural soil ero- dump designs. The modeling used a generic set of erosion
sion rates in the area. While the waste rock dump de- parameters as well as a set of potential rehabilitation
signs with benches are relatively simple and easy to design options that are cost-effective and practical to
construct and provide the lowest erosion rates, it is not construct. The use of this modeling approach shows that
known how these erosion rates compare to those of the this type of evaluation can be an effective design tool as
surrounding undisturbed environment. It is important not only average erosion rates, type and depth of erosion
that background erosion rates be determined and a but position of erosion can be evaluated. This is of par-
waste rock dump design be used that has an erosion rate ticular importance if the waste rock is covering or
similar to that of the surrounding undisturbed area as encapsulating tailings or potentially acid generating
enhanced erosion rates and resultant poor water quality material. The results of this study demonstrate that a
may impact on off-site biota for a considerable time waste rock dump design that has short slope lengths
post-mining. produces the lowest erosion rates while those designs with
The slope designs used in this report also assume that the longest slope lengths had the highest erosion rates. The
the mine equipment can build waste rock dumps to the model shows that not only can erosion rates be determined
specifications examined. This assumes that the caps and but maximum depths of incision and where this incision
benches are correctly graded. This report does not take occurs can be evaluated. The methodology outlined in this
into account undulations and cracking of the top sur- study offers a framework for the evaluation of rehabili-
faces resulting from settling. tation designs for other post-mining landscapes. Never-
Finally, while there are many unknowns in the cali- theless to ensure the validity of the results, site specific
bration of this model, the results have provided a risk calibration is necessary as well as field validation of the
assessment using the limited environmental data avail- modelling results over a number of years.
able. The work demonstrated here provides a framework
for determining the effectiveness of different waste rock Acknowledgements The very helpful comments of the anonymous
dump designs. reviewers are very much appreciated.

References

Boggs G, Evans K, Devonport C, Moliere Evans KG, Willgoose GR, Saynor MJ, Flint JJ (1974) Stream gradient as a func-
D, Saynor M (2000) Assessing catch- House T (1998) Effect of vegetation and tion of order, magnitude and discharge.
ment-wide, mining related impacts on surface amelioration on simulated Water Resources Research 10(5):969–
sediment movement in the Swift Creek landform evolution of the post-mining 973
Catchment, Northern Territory, Aus- landscape at ERA Ranger Mine, Hack JT (1957) Studies of longitudinal
tralia, using GIS and landform evolu- Northern Territory, Supervising Scien- stream profiles in Virginia and Mary-
tion modeling techniques. Journal of tist Report 134, Australian Government land. United States Geological Survey
Environmental Management Publishing Service, Canberra Professional Paper 292(B):45–97
59:321–334 Evans KG, Willgoose GR, Saynor MJ, Hancock GR (2004) The use of landscape
Boggs G, Devonport C, Evans K, Puig P Riley SJ (2000) Post-mining landform evolution models in mining rehabilita-
(2001) Rapid assessment of erosion risk evolution modeling. I. Derivation of tion design. Environmental Geology
in a small catchment in the wet/dry sediment transport model and rainfall- 46:561–573
tropics of Australia using GIS. Land runoff model parameters. Earth Surface Hancock GR (2003) The effect of catch-
Degradation and Development Processes and Landforms 25(7):743–763 ment aspect ratio on geomorphological
12(5):417–434 Evans KG, Willgoose GR (2000) Post- descriptors. In: Prediction in Geomor-
Coulthard TJ (2001) Landscape evolution mining landform evolution modeling: 2. phology, Wilcock P, Iverson R (eds)
models: a software review. Hydrological Effects of vegetation and surface rip- Geophysical Monograph Series Volume
Processes 15:165–173 ping. Earth Surface Processes and 135, American Geophysical Union,
Evans KG (2000) Methods for assessing Landforms 25(8):803–823 Washington, DOI: 10.1029/135GM015
mine site rehabilitation design for ero- Flanagan DC, Livingston SJ (1995) Water Hancock GR, Loch R, Willgoose GR
sion impact. Australian Journal of Soil Erosion Prediction Project (WEPP) (2003) The design of post-mining land-
Research 38:231–247 Version 95.7 User Summary. In: Flan- scapes using geomorphic guidelines.
agan DC, Livingston SJ (eds) ‘WEPP Earth Surface Processes and Landforms
user summary’, NSERL Report No 11, 28:1097–1110
July 1995
 779

Hancock GR, Willgoose GR, Evans KG O’Kane Consultants (2004) Report on Willgoose GR, Bras R, Rodriguez-Iturbe I
(2002) Testing of the SIBERIA land- particle size distribution of waste rock (1991a) A physically based coupled
scape evolution model using the Tin prepared for Minera Alumbrera Ltd network growth and hillslope evolution
Camp Creek, Northern Territory, Aus- Sheridan GJ, So HB, Loch RJ, Pocknee C, model: 1 Theory. Water Resources Re-
tralia, field catchment. Earth Surface Walker CM (2000) Use of laboratory search 27(7):1671–1684
Processes and Landforms 27:125–143 scale rill and interrill erodibility mea- Willgoose GR, Bras RL, Rodriguez-Iturbe
Hancock GR, Willgoose GR, Evans KG, surements for the prediction of hills- I (1991b) A physically based coupled
Moliere DR, Saynor MJ (2000) Med- lope–scale erosion on rehabilitated coal network growth and hillslope evolution
ium term erosion simulation of a mines soils and overburden. Australian model: 2 Applications. Water Resources
abandoned mine site using the Journal of Soil Research 38(2):285–298 Research 27(7):1685–1696
SIBERIA landscape evolution model. Tarboton DG, Bras RL, Rodriguez-Iturbe Willgoose GR, Bras RL, Rodriguez-Iturbe
Australian Journal of Soil Research I (1989) Scaling and elevation in river I (1991c) A physical explanation of an
38:249–263 Networks. Water Resources Research observed link area-slope relationship.
Kirkby MS (1971) Hillslope process-re- 25(9):2037–2052 Water Resources Research 27(7):1697–
sponse models based on the continuity Willgoose GR, Hancock GR, Kuczera G 1702
equation, Slopes Form and Process. (2003) Variability of catchment statis- Willgoose GR, Bras RL, Rodriguez-Iturbe
Institute of British Geographers Serial tics. In: Prediction in Geomorphology, I (1991d) Results from a new model of
Publication No.3, p15–30 January 1971 Wilcock P, Iverson R (eds), Geophysi- river basin evolution. Earth Surface
Laflen JM, Elliot WJ, Simanton J R, cal Monograph Series Volume 135, Processes and Landforms 16:237–254
Holzhey CS, Kohl KD (1991) WEPP American Geophysical Union, Wash- Wischmeier WH, Smith DD (1978) Pre-
soil erodibility experiments for range- ington, DOI:10.1029/135GM014 dicting rainfall erosion losses-A guide to
land and cropland soils. Journal of Soil Willgoose GR (1994) A physical explana- Conservation Planning. US Depart-
and Water Conservation pp 39–34 tion for an observed area-slope-eleva- ment of Agriculture, Agriculture
tion relationship for catchments with Handbook No. 537.2
declining relief. Water Resources Re-
search 30(2):151–159