You are on page 1of 22

Applied Geochemistry 57 (2015) 267–288

Contents lists available at ScienceDirect

Applied Geochemistry
journal homepage: www.elsevier.com/locate/apgeochem

Review

Modeling and management of pit lake water chemistry 1: Theory


D.N. Castendyk a,⇑, L.E. Eary b, L.S. Balistrieri c
a
Department of Earth & Atmospheric Sciences, State University of New York, Oneonta, Oneonta, NY, United States
b
InTerraLogic, Fort Collins, CO, United States
c
U.S. Geological Survey, Seattle, WA, United States

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

Article history: Pit lakes are permanent hydrologic/landscape features that can result from open pit mining for metals,
Available online 20 September 2014 coal, uranium, diamonds, oil sands, and aggregates. Risks associated with pit lakes include local and
regional impacts to water quality and related impacts to aquatic and terrestrial ecosystems. Stakeholders
rely on predictive models of water chemistry to prepare for and manage these risks. This paper is the first
of a two part series on the modeling and management of pit lakes. Herein, we review approaches that
have been used to quantify wall-rock runoff geochemistry, wall-rock leachate geochemistry, pit lake
water balance, pit lake limnology (i.e. extent of vertical mixing), and pit lake water quality, and conclude
with guidance on the application of models within the mine life cycle. The purpose of this paper is to bet-
ter prepare stakeholders, including future modelers, mine managers, consultants, permitting agencies,
land management agencies, regulators, research scientists, academics, and other interested parties, for
the challenges of predicting and managing future pit lakes in un-mined areas.
Ó 2014 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
2. Literature review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
3. Predicting pit wall runoff geochemistry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
3.1. Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
3.2. Observed effects of wall-rock processes on pit lake water quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
3.3. Approaches for modeling wall-rock geochemical processes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
3.3.1. Estimating wall-rock runoff from laboratory and field tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
3.3.2. Estimating wall-rock runoff from mineral dissolution kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274
3.3.3. Estimating wall-rock leaching from oxidation modeling and laboratory leaching tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274
3.3.4. Estimating wall-rock leaching from batch tests with representative water compositions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
4. Predicting the pit lake water balance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276
4.1. Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276
4.2. Direct rainfall and evaporation volumes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276
4.3. Wall-rock runoff volume. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
4.4. Groundwater volume . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
4.5. Water balance design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280
5. Predicting lake mixing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280
5.1. Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280
5.2. Stratification and mixing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280
5.3. Hydrodynamic predictions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
6. Predicting lake geochemistry. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
6.1. Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
6.2. Conceptual models of pit lake geochemistry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
6.3. Model limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284

⇑ Corresponding author. Tel.: +1 607 436 3064; fax: +1 607 436 3547.
E-mail address: Devin.Castendyk@oneonta.edu (D.N. Castendyk).

http://dx.doi.org/10.1016/j.apgeochem.2014.09.004
0883-2927/Ó 2014 Elsevier Ltd. All rights reserved.
268 D.N. Castendyk et al. / Applied Geochemistry 57 (2015) 267–288

7. Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285

1. Introduction (see impacts caused by the 1986 limnic eruption of Lake Nyos, a
volcanic crater lake in Cameroon; Halbwachs et al., 2004).
Society’s future demand for metals will be met in large part by In addition, stakeholders need to consider post-closure risks
open pit mining techniques where considerable volumes of ore and associated with global climate change. Predicted warmer air tem-
waste rock are excavated from one or more surface excavations peratures will increase evaporation rates in the Southwest United
and metallurgical processes are used to recover the desired States and elsewhere above present levels (Eary, 1998; Hoerling
metal(s) (Eary and Castendyk, 2012). Often open pit mines become and Eischeid, 2007; Grimaldi, 2009). Increased evaporation will
deep enough to intersect the local water table and thereafter increase efflorescence formation affecting runoff chemistry, and
require groundwater pumping to maintain a dry mining surface. modify pit lake water balances and geochemical predictions. For
Once mining concludes, pumping wells are turned off, which arid regions, related increases in water scarcity may also affect
enables the water table to rebound and allows water to flood the the future value of water resources and the public perception of
pit over time. Depending upon pit volume and annual rates of terminal pit lakes which perpetually lose groundwater to the
groundwater discharge and surface water diversion into the pit, atmosphere. Conversely, predicted increases in the intensity of
it can take anywhere from 1 year to 500 years to attain a steady- rainfall events will directly impact wall rock runoff and leaching
state lake level. This flooding results in the formation of permanent predictions, and may lead to greater exposure of unreacted mate-
anthropogenic water basins known as ‘‘pit lakes.’’ rials and more frequent pit wall failures than presently predicted.
In the absence of surface water diversions, most of the water Predictive computer models of pit lake water quality allow
added to pit lakes will come from rain water and snowmelt wash- stakeholders to assess post-closure risks associated with pit lakes
ing over the pit walls in addition to groundwater seeping through during the planning phase of mining when this knowledge can
fractures in the pit walls and floor. This water reacts with certain have the most influence on mine design. Because pit lakes will
minerals that are exposed along the pit surface or along the become permanent landscape features, risks need to be assessed
groundwater flow path producing acidic, neutral, or basic drainage for a long time period after mine closure. The durations of pub-
with elevated concentrations of dissolved solids. The resulting lished water quality predictions range from 8 to 230 years
solution is called ‘‘mine impacted water’’ (MIW) (McLemore, (Werner, 2009; Kempton et al., 1997). Presently, there is no ‘‘stan-
2008). Pit lakes collect and store large volumes of MIW and provide dard pit lake prediction time,’’ and the length used most likely
unique settings where additional geochemical and biogeochemical depends upon mining company policy, regulatory policy, and mine
reactions can occur. Additions of MIW and subsequent reactions site characteristics. As with any predictive models (i.e. weather
can potentially degrade the water quality of the pit lakes plus forecasts or economic forecasts), the accuracy of pit lake predic-
down-gradient groundwater and surface water resources, and tions will decrease with each time-step due to errors caused by
represent long-range, post-closure, environmental risks for the uncertainties and assumptions used in designing the model.
stakeholders (i.e. mining companies, environmental consultants, For this reason, short-term predictions may have greater validity
environmental regulators, land management agencies, non- than long-range predictions, and models should be updated
government organizations, and average citizens). throughout the mine life as new information becomes available.
Ideally, mining companies and their consultants will assess the Additional benefits of pit lake prediction models include: 1. The
risks associated with future pit lakes during the exploration and designation of ‘‘action levels’’ based on the dissolved concentration
planning phases of mining, and regularly update their risk models of ‘‘contaminants of concern’’ that specify when post-closure mit-
during mine development, closure, and post-closure phases. In igation measures are necessary (Lee, 1999); 2. Cost/benefit analysis
addition, companies will optimally have extensive consultation of various mitigation options (Castendyk and Webster-Brown,
with, and oversight from, other stakeholders during this process. 2007b); 3. Selection of sampling locations, sampling parameters,
Lee (1999) defined ‘‘risk’’ in the context of mine waste manage- and sampling frequency used in sampling and monitoring pro-
ment as the probability that a given hazard will occur multiplied grams; 4. Investigation of ‘‘what if’’ scenarios for post-closure man-
by the magnitude of the consequence of the hazard. The most agement decisions that affect the physical limnology and water
‘‘severe’’ hazards potentially associated with pit lakes include: 1. quality of the lake (Balistrieri et al., 2006; Castendyk and
Human fatalities resulting from drowning in a pit lake (e.g. aban- Webster-Brown, 2007b); and 5. Consideration of post-closure,
doned coalfields, Virginia; OSMRE, 2007); 2. Degradation of human ‘‘beneficial uses’’ for the pit lake in order to achieve sustainable
drinking water resources due to surface and/or groundwater dis- development objectives (McCullough et al., 2009).
charge from a pit lake (e.g. Harvard Lake, California; Savage et al., One of the best ways to validate the accuracy of a pit lake pre-
2000); 3. Chronic and/or acute health injuries to terrestrial organ- diction is to compare predicted concentrations to observed, post-
isms utilizing the surface water of a pit lake (e.g. Berkeley Lake, mining, water quality data collected through a lake monitoring
Montana; Hagler Bailly Consulting, 1996). In particular, migratory program, as demonstrated by Werner (2009) and Oldham et al.
birds can be attracted to pit lakes, and should be considered a spe- (2009). Monitoring programs also allow stakeholders to evaluate
cific class of terrestrial organisms with unique risks; 4. Chronic the evolution of lake chemistry over time, to initiate corrective
and/or acute impacts to aquatic ecosystems that receive surface measures to avoid poor water quality (if necessary), and to verify
water and/or groundwater originating from a pit lake (see US the achievement of closure objectives for company, community,
EPA rational for establishing a ‘‘critical level’’ in Berkeley Lake, and regulatory purposes. Monitoring programs also facilitate the
Montana discussed in Part 2 by Castendyk, Balistrieri, Gammons, improvement of pit lake predictions by identifying additional pro-
and Tucci, this volume); and 5. Acute impacts to aquatic ecosys- cesses that need to be included in the conceptual model and guid-
tems and terrestrial organisms caused by the rapid turnover and ing the calibration of model parameters. Gammons (2009)
degassing of a meromictic pit lake, known as a ‘‘limnic eruption’’ provided detailed guidance on the design of monitoring programs.
D.N. Castendyk et al. / Applied Geochemistry 57 (2015) 267–288 269

This paper is the first of a two part series on the modeling and pit lake research from the broader field of acid-mine drainage
management of pit lakes. Part 1 presented herein, reviews research.
approaches used to develop conceptual models of wall-rock runoff Early research into pit lakes was largely motivated by the closure
geochemistry, lake water balance, lake mixing, and lake water of open pit mines and the need to comply with regulatory water
quality, and concludes with guidance on the application of models quality guidelines. Between 1990 and 1999, numerous studies
within the mine life cycle. It is not the intent to recommend one reported limnologic and geochemical observations from existing
approach as being better than others, but to describe the pit lakes (Table 1), geochemical predictions of future pit lakes
approaches used and to discuss their various merits according to (Table 2), and remediation strategies for existing pit lakes (Table 3).
the descriptions provided by the developers of the approaches in From a modeling standpoint, some of the most influential studies
the literature. To demonstrate the application of certain identified general geochemical trends in existing pit lakes using
approaches, Part 2 provides case studies of a hydrodynamic predic- data from multiple mine sites (Price et al., 1995; Miller et al.,
tion for the existing Dexter Lake in Nevada, a geochemical predic- 1996; Davis and Eary, 1997). Notably, Shevenell et al. (1999) dem-
tion for a future pit lake at the Martha Mine in New Zealand, and onstrated the influence of mine geology on pit lake water chemistry
the interpretation of recent monitoring data from the Berkeley by comparing the geochemistry of 16 hard rock pit lakes in Nevada
Lake in Montana (Castendyk, Balistrieri, Gammons, and Tucci, this based on ore deposit type. This work prompted similar studies in
volume). The purpose of these papers is to better prepare future the Iberian Mining District (Sánchez España et al., 2008) and the
modelers, mine managers, consultants, permitting agencies, land Central German Lignite Mining District (Schultze et al., 2010), and
management agencies, regulators, research scientists, academics, inspired the recent development of an online, global pit lake data-
and other interested parties for the challenges of predicting and base containing data from 72 pit lakes at the time of writing
managing future pit lakes in un-mined areas. (Johnson and Castendyk, 2012; http://pitlakesdatabase.org). Com-
bining data from 24 hard rock pit lakes in the United States and
2. Literature review 66 coal pit lakes in Germany, Eary (1999) identified minerals likely
to precipitate from pit lake water as a function of pH and dissolved
Pit lake research is a relatively young field of study spanning ion concentrations. Pit lake modelers have frequently used this
just two-and-a-half decades. Davis and Ashenberg (1989) provided paper during the design of geochemical predictions to specify min-
one of the first widely-published studies on pit lake geochemistry. erals that are likely to precipitate from solution and thereby influ-
Their work on the Berkeley Lake, Montana showed pit lakes as ence the concentrations of contaminants of concern.
layered geochemical systems influenced by climate, limnologic The perception of the ‘‘best case’’ scenario for pit lakes began to
mixing events, and water–rock reactions, and thereby differentiated change following the publication of the Minerals, Mining and

Table 1
Select publications on pit lake observations.

Category Name Location Reference


Physical limnology Gunnar Saskatchewan, Canada Tones (1982)
Multiple Nevada, USA Atkins et al. (1997)
Multiple Saskatchewan, Canada Doyle and Runnells (1997)
Brenda British Columbia, Canada Stevens and Lawrence (1998)
Brenda British Columbia, Canada Hamblin et al. (1999)
Sleeper Nevada, USA Atkin and Schrand (2000)
Island Copper British Columbia, Canada Fisher and Lawrence (2000)
Berkeley Montana, USA Jonas (2000)
Enterprise NW Territory, Australia Boland and Padovan (2002)
Yerington Nevada, USA Jewell and Castendyk (2002)
Goitsche Central Germany Boehrer et al. (2003)
Lake 111 Lusatian, Germany Karakas et al. (2003)
Summer Camp Nevada, USA Parshley and Bowell (2003)
East Sullivan Quebec, Canada Tassé (2003)
Island Copper British Columbia, Canada Stevens and Fisher (2005)
Geochemistry Berkeley Montana, USA Davis and Ashenberg (1989)
Multiple Nevada, USA Price et al. (1995)
Multiple USA Miller et al. (1996)
Multiple USA Davis and Eary (1997)
Multiple Germany Klapper and Schultze (1997)
Berkeley Montana, USA Robins et al. (1997)
Multiple Nevada, USA Shevenell et al. (1999)
Harvard California, USA Savage et al. (2000, 2009)
Multiple Nevada, USA Shevenell (2000)
Multiple Nevada, USA Shevenell and Connors (2000)
Multiple Global Bowell (2002)
Berkeley Montana, USA Madison et al. (2003)
Summer Camp Nevada, USA Bowell and Parshley (2005)
Berkeley Montana, USA Pellicori et al. (2005)
Multiple Germany Schultze et al. (2010)
Combined physical limnology and geochemistry Spenceville California, USA Levy et al. (1997)
Multiple Utah, USA Castendyk and Jewell (2002)
Udden Northern Sweden Ramstedt et al. (2003)
Elizabeth Vermont, USA Seal et al. (2006)
Berkeley Montana, USA Gammons and Duaime (2006)
Multiple Spain Sánchez España et al. (2008)
270 D.N. Castendyk et al. / Applied Geochemistry 57 (2015) 267–288

Sustainable Development Report (MMSD, 2002). Instead of devel- emphasized pit lakes found in soft-rock, lignite mines found in
oping lakes that only comply with regulatory water quality guide- Lusatia, Germany. In 2009, Gammons et al. (2009) produced a guid-
lines, the MMSD Report challenged mining companies to develop ance document for the Canadian government called Creating Lakes
pit lakes into post-mining resources in order to achieve industry- from Open Pit Mines: Processes and Considerations, with Emphasis
wide sustainable development objectives. Several pit lakes now on Northern Environments. This 106-page document provides a con-
exist that provide beneficial uses such as public recreation areas cise and comprehensive review of physical, chemical, and biological
and aquaculture industries (McCullough et al., 2009). Between processes occurring in pit lakes, the characteristics needed for pit
2000 and 2012, the U.S. Environmental Protection Agency hosted lakes to become self-sustaining aquatic ecosystems, options for
the first two conferences exclusively devoted to pit lake research post-closure pit lake use, and case studies of the closure and treat-
(McCready, 2001; McCready and Ober, 2006), the journal South- ment of North American pit lakes located above 40 degrees North.
west Hydrology published a special issue on the hydrology of pit Later that year, the Acid Drainage Technology Initiative, Metal Min-
lakes (Woodhouse, 2002), and several international conferences ing Sector (ADTI–MMS) in association with the Society for Mining,
hosted topical sessions exclusively on pit lake research (i.e. 2003 Metallurgy and Exploration (SME), published a guidance workbook
Annual Meeting of the Geological Society of America in Seattle, Wash- titled Mine Pit Lakes; Characteristics, Predictive Modeling, and
ington; the 2006 7th International Conference on Acid Rock Drainage Sustainable Development (Castendyk and Eary, 2009). This compre-
(ICARD) in St. Louis, Missouri; the 2009 8th ICARD in Skellefteå, hensive, 300-page document was an international collaboration
Sweden; and the 2012 9th ICARD in Ottawa, Canada). among 26 authors from four countries that addressed global pit
Several pit lake guidebooks have been produced over the past lakes in both arid and humid climates resulting from metal and coal
15 years. Researchers in Germany published the first pit lake mining. Workbook sections focused on regulatory issues, character-
guidebook in 1998, called Acidic Mining Lakes: Acid Mine Drainage, istics and classifications, conceptual models, sampling and
Limnology, and Reclamation (Geller et al., 1998). This book monitoring, predictive modeling, remediation, and post-mining

Table 2
Select publications on pit lake prediction studies.

Category Name Location Reference


Water–rock reactions Multiple Nevada, USA Connors et al. (2000)
General theory Morin and Hutt (2006)
Martha New Zealand Castendyk et al. (2005)
Climate General theory Eary (1998)
General theory Flite and Eidson (2003)
Hydrology General theory Havis and Worthington (1997)
General theory Marinelli and Niccoli (2000)
Getchell Nevada, USA Shevenell and Pasternak (2000)
Niederlausitz Germany Werner et al. (2001)
General theory Kuma et al. (2002)
Physical limnology Multiple Nevada, USA Lyons et al. (1994)
Merseburg-Ost Central Germany Böhrer et al. (1998)
General theory Stevens et al. (2002)
Dexter Nevada, USA Balistrieri et al. (2006)
Martha New Zealand Castendyk and Webster-Brown (2007a)
Geochemistry Models Bird et al. (1994)
General theory Pillard et al. (1996)
Twin Creeks Nevada, USA Kempton et al. (1997)
General theory Davis and Fennemore (1998)
General theory Lewis (1999)
Getchell Nevada, USA Tempel et al. (2000)
Martha New Zealand Castendyk and Webster-Brown (2007b)
Kepwari WA Australia Oldham et al. (2009)
Harvard California Savage et al. (2009)
Baerwalde Germany Werner (2009)

Table 3
Select publications on pit lake remediation and sustainability.

Category Pit lake name Location Reference


Remediation Multiple Germany Stottmeister et al. (1999
Summer Camp Nevada, USA Castro et al. (1999)
General theory Castro and Moore (2000)
Multiple British Columbia, Canada Martin et al. (2003)
Anchor Hill South Dakota, USA Park et al. (2006)
South Tennessee, USA Wyatt et al. (2006)
Multiple Canada Kalin and Wheeler (2009)
Island Copper British Columbia, Canada Pelletier et al. (2009)
General theory Weilinga (2009)
Sustainability General Theory Global MMSD (2002)
Multiple WA Australia Evans et al. (2003)
Multiple Global McCullough et al. (2009)
D.N. Castendyk et al. / Applied Geochemistry 57 (2015) 267–288 271

uses and considerations. The document concluded with a chapter the most sulfidic ore is often located in the lower portion of the
on the state-of-the-art of pit lake water quality prediction, data pit, and oxide ore and non-mineralized overburden are often
gaps and research needs, recommendations for the mining indus- located nearer the top of the pit. Likewise, leaching processes are
try, and guidelines for developing and presenting the results of also most influential during the early stages of lake filling when
pit lake predictive models. In 2011, the Australian Centre for the volume of water is small and the potential reactive area of
Geomechanics published a compendium titled Mine Pit Lakes: wall-rock is relatively large. The influence of leaching decreases
Closure and Management, which provided advice for mine planners, with time due to the relatively low solubility of oxygen in lake
regulators, environmental managers and communities to design water, and the development of oxidation rimes which isolate
sustainable mine pit lakes that provide regional benefit to stake- reactive materials from oxidants. However, future wall rock
holders (McCullough, 2011). Recognizing the potential impacts of failures occurring above and below the lake surface can expose
36 pit lakes that will result from oil sands mining, the Cumulative fresh reactive material, allowing for runoff and leaching processes
Environmental Management Association (CEMA) of Fort McMurray, to begin anew.
Alberta, Canada, conducted a 2-year review of oil sands pit lakes While conceptual models typically seem to show these wall-
and published the End Pit Lakes Guidance Document (CEMA, 2012). rock processes as somewhat simple functions of the wall-rock sur-
Most recently, Geller et al. (2012) provided a significant update to face area and various chemical release functions, they are in fact
their 1998 guidebook titled Acidic Pit Lakes – Legacies of Surface one of the more difficult parts to simulate with reliability in predic-
Mining of Coal and Metal Ores. tive models of pit lake chemistry. The walls of pit lakes are not
smooth surfaces but more typically are covered with rock debris
and are heavily fractured due to the blasting during mining.
3. Predicting pit wall runoff geochemistry Fig. 2 shows typical pit walls with high amounts of rubble left over
after mining has moved on to other parts of the pit. Fig. 3 shows pit
3.1. Overview walls with high amounts of fracturing and friable material created
from weathering and acid rock drainage during mining and after
Most geochemical modeling studies of mine pit lakes begin mining ended prior to filling with water. Rain falling on high walls
with the development of a conceptual model that describes the may take a rapid path down to the pit lake. However, it is clear
hydrological and geochemical processes affecting the water and from the images in Figs. 2 and 3 that much of the rain that falls
chemical mass balances of the pit lake. A typical conceptual model on the other pit walls will not take a rapid path to the water sur-
is shown in Fig. 1. These conceptual models commonly include the face but will encounter significant amounts of rubble of variable
processes of wall-rock runoff (also referred to as ‘‘washoff’’) and grain size and in fracture zones in the pit walls. These pathways
wall-rock leaching (also referred to as ‘‘wall-rock flushing’’). can be expected to increase the potential for leaching and solute
Wall-rock runoff is a surface hydrology process generated by rain- release due to slowed water movement and an increase in the
storms and snowmelt. Ephemeral washoff dissolves efflorescent rock-to-water ratio. The heterogeneity of exposed rock can be
salts and oxidation products that may be present on exposed rock under-represented when using laboratory tests on a limited num-
surfaces, and transports dissolved constituents to the lake. Wall- ber of samples. In addition, many pits have remnant piles of waste
rock leaching is both a groundwater and lake water process occur- rock deposited in inactive parts of the pit providing additional
ring where water is in direct contact with wall rocks over a longer material with potential for chemical release to the pit lake.
time scale. For groundwater, chemical leaching occurs in both the The process of wall-rock leaching is commonly considered to
unsaturated and saturated zones along pore spaces, fractures, and result from the flushing of the wall-rocks by groundwater entering
historic mine tunnels that can intersect the pit wall at length scales the pit lake and also by wave action of the lake itself at the water–
ranging from millimeters, to tens of meters, to kilometers, respec- wall-rock interface. The lake level may fluctuate up and down,
tively. For lake water, leaching occurs at the submerged water– depending on the water balance between inflow and outflow rates.
rock interface. Transportation rates are generally much slower in These fluctuations also provide a potential mechanism for rinsing
environments where leaching occurs compared to steep, exposed weathering products from the wall-rocks and bringing them into
surfaces where runoff occurs, therefore leaching processes involve the pit lake.
significantly longer reaction times. This section provides a detailed review of different methods
The influence of wall-rock runoff generally decreases over time that have been used in modeling studies to represent the processes
as the pit progressively fills with water and wall-rocks become of wall-rock runoff and wall-rock leaching. These methods define
submerged. This is particularly true for sulfidic ore bodies where the geochemical characteristics of wall-rock runoff and wall-rock
leachate which are key input variables in pit lake prediction
models.

Fig. 1. Typical conceptual model of pit lake hydrogeochemical processes for a


closed basin (i.e. terminal pit lake) without groundwater or surface water discharge.
The water table is indicated by a blue line. (For interpretation of the references to
color in this figure legend, the reader is referred to the web version of this article.) Fig. 2. Typical rubble-fill benches of an active hardrock open pit.
272 D.N. Castendyk et al. / Applied Geochemistry 57 (2015) 267–288

Fig. 3. Images of a weathered pit wall rock above a dry bench that have produced ARD over time.

3.2. Observed effects of wall-rock processes on pit lake water quality maintain near-neutral pH. However, a subsequent waste dump
failure sent 40,000 tons of sulfide waste into the pit lake caused
The processes of wall-rock runoff and leaching are not purely a drop in pH and increased sulfate and metals. More lime was
hypothetical but have been observed to affect water quality in added to control the pH. The chemistry of the pit lake eventually
pit lakes. The typical pattern is: 1. Initial rapid concentration stabilized and met all primary Nevada drinking water standards.
increase due to the release of soluble constituents from weathered The excellent study and account of events at the Sleeper pit lake
wall-rocks and waste rock left in the pit bottom; 2. Dilution by by Dowling et al. (2004) underscored the importance of developing
groundwater entering the pit which typically has lower metal con- an understanding of the potential for wall-rock leaching and runoff
centrations; and 3. In terminal pit lakes in arid climates, gradual to have an adverse effect on water quality in sulfidic pits and to
increase in metal concentrations over time due to evapoconcentra- plan accordingly to mitigate those effects.
tion (i.e. removal of fresh water from evaporation resulting in the
concentration of remaining lake water). For example, Bowell and 3.3. Approaches for modeling wall-rock geochemical processes
Parshley (2005) reported that the presence of reactive secondary
minerals in wall rocks of the Summer Camp pit lake provided a A number of different approaches have been developed to rep-
source of easily leached arsenic, metals, acidity, and sulfate that resent wall-rock processes in predictive modeling studies for pit
were released upon exposure to seasonal rain events. Barker lakes. Often, the distinction between wall-rock runoff and wall-
et al. (2004) concluded that washoff of soluble sulfate minerals rock leaching is not clear in the modeling approaches. An effort
(gypsum and ferricopiapite) from pit walls affected acidity and sul- is made here to categorize the approaches for purposes of discus-
fate levels in an alluvial gold-mine pit lake. Perhaps an extreme sion into the following generalized areas:
example of wall-rock effects was described by Triantafyllidis and
Skarpelis (2006), who reported that an assortment of secondary  Wall-rock runoff (above the pit lake surface).
sulfate and arsenate minerals in wall-rocks and on the bottom of – Laboratory and field tests.
an acidic pit lake (pH 2.9–3.1) had a major effect on water chemis- – Mineral dissolution kinetics.
try. Savage et al. (2009) reported that efflorescences of sulfate salts  Wall-rock leaching (below the pit lake surface).
comprised of copiapite along with hexahydrite and jarosite accu- – Oxidation modeling and laboratory leaching tests.
mulated in weathering sulfide zones in wall-rocks of the Harvard – Batch testing with representative water compositions.
pit lake at the Jamestown Mine, California. Arsenic concentrations
fluctuated seasonally with the highest concentrations tending to The processes of wall-rock runoff and wall-rock leaching are
occur in association with seasonal rain, indicating that washoff of conceptualized in Fig. 4. Wall-rock runoff is defined as the release
the efflorescence from wall-rocks is a major source of arsenic and of solutes from the pit walls due to incident rain and flow down the
sulfate to the pit lake. Pyrite and other sulfides oxidize in the pres- pit walls into the pit lake. Wall-rock runoff decreases over time as
ence of humid air (Jerz and Rimstidt, 2004) such that a surface the pit lake fills, submerging increasing proportions of the wall-
layer of oxidation products may continually build up and be rinsed rock until hydrologic steady state is reached. Wall-rock leaching
into a pit lake during rainfall. is defined as the process of solute release at the wall-rock–water
The likelihood for wall-rock processes to have a deleterious interface due to groundwater inflow into the pit lake and rinsing
effect on water quality has been recognized during pit lake model- of the wall rock due to wave action. Wall rock leaching may also
ing studies as part of closure planning (Meyer et al., 1997; Warren include mineral dissolution and sulfide oxidation below the water
et al., 1997; Dowling et al., 2004). Work on the Sleeper pit lake in surface especially in acidic pit lakes with appreciable ferric ion
Nevada by Dowling et al. (2004) led to a mitigation plan to cover concentrations.
remnant waste rock in the pit with oxide waste and alluvial mate-
rial to isolate sulfidic material and provide alkalinity. In addition, 3.3.1. Estimating wall-rock runoff from laboratory and field tests
alkaline groundwater was pumped into the pit to rapidly submerge The most common approach to representing wall-rock runoff
sulfidic wall rocks, thereby minimizing the potential for oxidation chemistry in pit lake predictive models involves extrapolating or
and release of acidity from wall-rock leaching reactions. The chem- ‘‘scaling’’ up results from laboratory kinetic tests to field condi-
istry of the pit lake evolved as expected with an initially slightly tions. The laboratory test procedure most commonly used is the
acidic water (pH < 4) due to washoff and release of acidic salts from humidity cell test (ASTM, D5744-07e1), although one-time rinse
wall rocks as it rapidly filled. Lime was added during filling to tests, such as the meteoric water mobility procedure (MWMP)
D.N. Castendyk et al. / Applied Geochemistry 57 (2015) 267–288 273

converting the laboratory test results to be a function of surface


area. The specific surface area (m2/kg) of spherical mineral grains
(Amineral) can be estimated from (White, 1995):
0 1
 2
m @ 6k A
Amineral ¼ h i Gd ½m ð1Þ
kg q m3 kg
Oxidized
Wallrock Runoff: Walls
Runoff down pit walls Rubble
where q is the mineral density (kg/m3), Gd is the grain diameter (m),
and fractured zones
and k is a surface roughness factor (dimensionless) defined as the
ratio of measured or true surface area to the equivalent geometric
Waste surface area.
Wallrock Leaching: Flushing Given a median particle size distribution, Eq. (1) may be used to
by groundwater and rinsing
by pit water of fractured,
calculate the surface area of the sample used in the leaching tests
oxidized wallrock such that release rates are expressed as a function of surface area
(fExpA) rather than sample mass:
Fig. 4. Conceptualization of wall-rock runoff and wall-rock leaching for pit lakes. h i
Sub-horizontal arrows reflect the orientation of the hydraulic gradient and the   f mg
mg ExpM kg yr
direction of groundwater flow, whereas vertical dashed arrows indicate the f ExpA ¼ h i ð2Þ
direction the water table will rise over time. m2 yr A m2
mineral kg

Eq. (2) may be used to calculate the rate of chemical mass additions
(NDEP, 2010) and synthetic precipitation leaching procedure of individual solutes (Mi) given an estimate of the surface area con-
(SPLP) (ASTM D6234-98(2007); also EPA Method 1312) have also tributing wall-rock runoff (Awallrock) to the pit lake:
been used. In general, humidity cell tests and MWMP tests are  
mg
expected to provide more useful information on leaching because Mi ¼ f ExpA 2
 Awallrock ½m2  ð3Þ
m yr
they rely on a relatively low water-to-rock ratio (1:1) compared
to SPLP tests (20:1). In cases where weathered rock samples are While these equations are relatively simple, the problem for
tested, the first flush from the humidity cell tests may be the most predictive models is in estimating values for the important param-
representative solution composition because it should be at least eters of mineral surface roughness (k) in Eq. (1) and the wall-rock
partially indicative of the release of secondary salts that may have surface area (Awallrock) in Eq. (3). One approach to calculate mineral
built up over time in the wall-rock samples used in the tests. Alter- surface roughness involves the following: 1. Measure the specific
natively, the steady state rates of solute release from humidity cells surface area of the mineral (m2) using the BET gas adsorption
may be used, depending on climatic conditions. For example, a method (Brunauer et al., 1938), where surface area is proportional
tropical climate with regular and heavy rainfall may preclude the to the volume of gas adsorbed by the sample; 2. Measure the aver-
buildup of oxidation products; hence a steady-state release rate age radius (r) of mineral grains using optical microscopy; 3. Calcu-
may be appropriate. In contrast, the first flush data from humidity late the geometric surface area (m2) using r and the equation for
cell tests may be more appropriate for a climate with long periods the area of a sphere (i.e. A = 4pr2); and 4. Divide the specific surface
between rainfall or wet season-dry season cycles. area by the geometric area to calculate mineral surface roughness.
The validity of using laboratory testing data on wall-rock sam- According to the compilation of White (1995), values of surface
ples to represent a long-term weathering process of wall-rock run- roughness range from 2 to 10 for unweathered silicates, from 50
off is dependent on the level of oxidation of the tested rock to 200 for sand-sized soils developed from granitic bedrock, and
samples, which is a difficult parameter to estimate. Test results up to 1000 for very weathered soils. The range of 10–200 is prob-
on fresh, unoxidized core samples would underestimate leaching ably reasonable for most mine pits that are comprised of relatively
rates (with the possible exception of calcite-bearing rocks, where fresh rock exposures. Surface roughness factors are generally
calcite leaches first). Tests on oxidized material can be expected independent of grain size for a given sample type and degree of
to be more representative of the long-term leaching rate but such weathering (White, 1995).
material may not be available for feasibility and environmental Alternatively, estimations of the wall-rock surface areas
permitting studies conducted prior to mine construction. (Awallrock) can be made from detailed drawings of the pit walls
For mines already in operation, the option to use samples of and benches generated using computer aided design (CAD)
actual wall-rock material in laboratory tests may be available. software that show the areas of the important rock or alteration
The use of material from the operating mine would increase the types. Surface area estimates may also be made from surveys of
likelihood that the material has a representative oxidation and pit geometry integrated into geographic information systems
weathering state more representative of the long-term than sam- (GIS) software. Both approaches tend to oversimplify small-scale
ples of unexposed core. In addition, mine-wall stations may be variations in wall-rock roughness which in turn underestimates
used to isolate a portion of wall rock and collect leachate samples surface area. Recent advances in remote sensing technology, called
from a measured surface area following the procedures pioneered ground-based LiDAR, can measure wide swaths of pit walls with
by Morin and Hutt (2006). centimeter-scale resolution by illuminating wall-rocks with a laser
The data from laboratory kinetic tests provide a measure of the and analyzing the reflected light, and can minimize this error.
mass of soluble constituents released per mass of rock (fExpM) per However, the walls of open pits have been subjected to blasting
unit of time [(mg/kg)/yr]. Laboratory kinetic tests are generally and depressurization leading to abundant fractures and joints.
conducted on samples ground to a small size, such as less than Fractures from blasting in granite are reported to have depths that
6.4 mm (0.25 in) for humidity cell tests. The difficulty in applying range up to 1.1 m (Siskind and Fumanti, 1974), although the actual
these data to make predictions of pit lake chemistry is in develop- range at any given pit will be dependent on the rock type and wall
ing a reasonable method to scale up from the laboratory system to stability. Morin and Hutt (2006) reported ratios of actual exposed
the field system. Kinetic processes are primarily a function of wall-rock surface area to geometric surface area measurements
surface area rather than mass. Thus, the first step in scale up is range from 27:1 to 161:1 for three case studies of pits.
274 D.N. Castendyk et al. / Applied Geochemistry 57 (2015) 267–288

An additional consideration in the determination of wall-rock for each rock type. Tempel et al. (2000) used the approach based on
surface area is that open pits often contain some amount of waste Eq. (4) along with water balance calculations to model the chemis-
rock and possibly tailings that may be deposited in various non- try of an existing pit lake. The term freactivity is a scaling factor used
active parts of the pit to save on hauling costs at the end of mining. to extrapolate laboratory data to field conditions. It is empirical in
Also, erosion and collapse of benches may yield large amounts of nature and can only be reasonably estimated with confidence by
variably sized material in the pit walls (Fig. 2). These materials will comparison of model simulation results to measured data through
provide an extra source of leachable solutes that will be released to a calibration process. Tempel et al. (2000) determined a value of
the pit lake. The same approach outlined above for the pit walls 1000 for the reactivity factor (freactivity) through a calibration pro-
may be applied to these waste materials to estimate solute releases cess. The best guidance that is available at the current time is
but with modification of the surface area factors and consideration through consideration of the range of potential values; the value
of oxidation processes that will largely cease once waste materials of 1000 for the reactivity factor is within the approximate range
are submerged. that could be obtained from the combination of surface roughness
Laboratory leaching tests provide data on solute releases for an factors (10–200) and ratios of exposed wall rock to geometric wall
aggregate rock sample representing a particular rock or alteration rock (27:1–161:1), discussed above.
type in the pit walls as a function of the lake surface elevation. A potential problem with both of these mineral dissolution
The elevation function is important in accounting for the submer- approaches is that the potential process of washoff of sulfide oxi-
gence of at least the lower portions of the pit walls as the pit lake dation products accumulating between infrequent rain events in
fills over time thereby decreasing the chemical load from wall-rock arid locations may not be specifically represented in calculations.
runoff. Thus, Eqs. (1)–(3) will generally need to be applied to each For mines with relatively non-reactive or very low amounts of sul-
rock or alteration type present in the pit walls given their respec- fides, washoff of oxidation products may not be important. For
tive wall-rock surface areas and potential time dependence in the other mines with exposed sulfide minerals, oxidation and washoff
event that they become submerged. may need to be explicitly included in modeling formulations.

3.3.3. Estimating wall-rock leaching from oxidation modeling and


3.3.2. Estimating wall-rock runoff from mineral dissolution kinetics
laboratory leaching tests
A second approach for representing wall-rock runoff chemistry
Many of the first pit lake predictive models used the results
involves calculation of solute release rates based on empirical rate
from kinetics tests as direct representations of the chemistry of
expressions for mineral dissolution. This approach fundamentally
leachates produced from the flushing of wall rocks with groundwa-
requires determination of the area of specific reacting minerals in
ter filling the pit (Kirk et al., 1996; Pillard et al., 1996). However, it
the pit wall rocks. Tempel et al. (2000) developed the following
was recognized there should be a finite amount of solute release
equation for this approach (see also Bird and Mahoney, 1994 and
from groundwater flushing, and over time, the inflowing water
Castendyk and Webster-Brown, 2007b):
from the groundwater system should approach the groundwater
   
mol mol chemical composition. The approach used to determine the finite
Mi ¼ Awallrock ½m2   P mineral%  Di 2  3:15E amount of solute release commonly has been based on the
yr m sec
sec Davis–Ritchie equations (Davis and Ritchie, 1986a; Davis et al.,
þ7  f reactivity ð4Þ 1986b) that describe pyrite oxidation as a function of the rate of
yr
oxygen diffusion into the wall-rock (Davis et al., 2006;
where Mi is the rate of molar additions of individual solutes per Fennemore et al., 1998; Kempton et al., 1997). The Davis–Ritchie
year, Pmineral% is the weight percentage of the mineral in the wall approach represents the rock as a homogeneous mass into which
rock, Di is the published dissolution rate for the individual mineral oxygen may diffuse and travel through air and water-filled pore
making up the wall rocks, and freactivity is a dimensionless reactivity spaces. Fennemore et al. (1998) revised the Davis–Ritchie approach
factor. to account for moisture content and fracture density. These model-
Rate expressions for dissolution of a wide range of minerals are ing approaches yield a thickness for the oxidized rim of the wall
available in the compilation of Palandri and Kharaka (2004). The rock as a function of time where the thickness is a function of
more difficult parameters in Eq. (4) to determine are again the wall the rock physical properties (porosity, fracture density), pyrite con-
rock surface area (for the same reasons as described above), the tent (plus any other oxidizing sulfides), and time. The time period
weight percentage of different minerals comprising the wall rocks for oxidation is the time of exposure during mining plus the time
(Pmineral%), and the reactivity factor (freactivity). The percentages of after mining ends until submergence. Generally, it is assumed that
the different minerals in the wall rocks can be determined from oxidation reactions are negligible after the wall rock becomes
mapping mineral assemblages in wall rocks and detailed mineral- submerged by the rising pit lake thereby greatly reducing the
ogical analyses of representative samples from each assemblage as availability of oxygen (Davis et al., 2006).
described by Castendyk et al. (2005). Their approach quantified The concepts of oxidation modeling and solute release are
mineral concentrations by: 1. Using X-ray diffraction (XRD) for shown in idealized form in Fig. 5. The sulfide minerals in exposed
mineral identification; 2. Using X-ray fluorescence (XRF) and Leco wall rock are oxidized as oxygen diffuses into the rock matrix along
furnace (required for sulfur and carbon species) to quantify ele- grain boundaries and cracks. The oxygen is consumed by oxidation
mental masses; and 3. Assigning elemental mass to known miner- thereby limiting the thickness of the oxidized zone to a thin rim
als which defined the weight percent of each mineral in each that gradually thickens over time as the sulfides are dissolved.
mineral assemblage. This study was more directly applied to The Davis–Ritchie type of equations and subsequent improve-
assessing mineral leaching in the submerged portion of the pit ments by Fennemore et al. (1998) provide a means to calculate
lake, however, the same approach could be applied to determine the thickness of the oxidation zone with time. The products of oxi-
minerals reacting with wall rock runoff. By using a reactivity factor dation are assumed to accumulate as soluble secondary solids
of unity (freactivity = 1), this study performed a sensitivity analysis within the oxidized rim during the period of exposure during min-
using order-of-magnitude variations in wall rock surface area ing and prior to mine closure and submergence. After mine closure
(Awallrock) to account for the error in surface area measurements. and cessation of dewatering systems, groundwater flows into the
Alternatively, mineral percentages may be estimated from pit flushing the accumulated oxidation products into the emerging
maps of rock type areas and available data on mineral abundances pit lake. Eventually as the oxidation products are rinsed out and
D.N. Castendyk et al. / Applied Geochemistry 57 (2015) 267–288 275

  hmgi
Oxidized Rim Non-Oxidized mg 1
f CR ¼ Concentration  V rinse ½L 
kg L L Msample ½kg
O2 1
 ð6Þ
/  V sample ½L
O2
Multiplication of the chemical release function (fCR) by the mass
of oxidized rock estimated from oxidation modeling produces the
rate of mass input of solutes to the pit lake as a function of the pore
volumes of groundwater passing through the wall rock en route to
the pit lake. The size of the pore volumes is typically approximated
9 10

from estimates of blast depths and porosity of the blasted zone.


Blast-zone depths are generally less than 1.1 m (Siskind and
Oxidaon Thickness (m)

Fumanti, 1974) and porosities may range from 5% to 45%, depend-


8

ing on rock type and blast pattern. Pore volumes are approxima-
7

No fractures
tions due to the fact that most pit lake water quality predictions
6

are made prior to mining, and during mining, safety concerns


5

preclude working directly on pit walls to make close inspections.


4

The chemical release function for solutes produced from oxida-


With fractures
tion should theoretically decrease over time as the oxidation prod-
3

ucts are washed out, at which time concentrations should


2

approach the composition of the influent groundwater. This trend


0 1

may be followed if samples were already oxidized prior to testing.


0 10 20 30 40 50 60 70 80 90 Under these circumstances, Schafer et al. (2006) pointed out that if
Years representative oxidized samples are used in kinetic tests, the
concentration after any number of pore volumes (CVpore) can be
Fig. 5. Idealized depiction of leached rim of a wall rock with calculations of
described by a simple mixing calculation between the end mem-
oxidation thickness adapted from Fennemore et al. (1998).
bers consisting of the initial flush concentration and influent
groundwater concentration:

wall rocks submerged, the influent water composition should C Vpore ¼ C Vpore¼0 for V pore 6 1 ð7Þ
approximate the groundwater composition.
With the thickness of the oxidized wall rock estimated by oxi- M Vpore¼0  MGW
C Vpore ¼ C GW þ for V pore > 1 ð8Þ
dation modeling, the next step is to determine rates of chemical V pore
release resulting from oxidation. The typical procedure for deter-
where CVpore=0 is the initial flush concentration obtained from the
mining chemical release functions is to use data from laboratory
first pore volume rinsed through an oxidized sample, MVpore=0 is
kinetic tests (e.g., humidity cell tests or column leaching tests).
the mass within the initial flush calculated by CVpore=0  Vpore, CGW
Kinetic tests yield data on the concentrations of solutes as a func-
is the groundwater concentration, and MGW is the mass in an equiv-
tion of the time obtained from a weekly cycle of three days of dry
alent volume of groundwater calculated by CGW  Vpore. Using the
air circulation, three days of humid air circulation, and one day of
initial flush may not well represent rocks that have a long lag period
rinsing with water. Ideally, this cycle is continued until steady
prior to consumption of neutralization capacity and initiation of
state concentrations are obtained, which may take up to a year
acidity. Under these circumstances, using the leaching rates from
(i.e. 52-week cycle) or more. Often 20-week or 40-week cycles
the period of acid generation will likely yield a more reasonable
are used, however, experience shows that the longer the cycle
representation of potential solute loads created from sulfide oxida-
time, the better. The concentration data from the kinetic tests are
tion and subsequent metal leaching.
normalized to concentrations per pore volume (Vpore) of solution
In practice, there are some potential problems in obtaining data
rinsed through the samples during testing where one pore volume
that are clearly indicative of an eventual end to flushing of oxida-
is the amount of water filling the entirety of the pore space of the
tion products. First, the endpoint in Eq. (8) may take a very long
granular rock sample used in the test, and porosity (/) is defined as
time to establish in kinetic tests. Second, if the pit lake modeling
the pore volume divided by the total sample volume of rock used in
effort is being conducted prior to mining, then the only available
the kinetic test (Vsample):
samples for testing may be core samples that have not been sub-
V pore ½L jected to oxidation. Third, the release of nitrogen compounds from
/¼ ð5Þ blast residue is very difficult to estimate without samples from
V sample ½L
active mining operations. Under these circumstances, it may be
Normalizing to pore volumes allows the data to be related to reasonable to make the assumption that the fCR asymptotically
the rate of inflow of groundwater through the oxidized wall-rock approaches groundwater composition over some set number of
rim. For example, a test conducted on a 1.0 kg mass of sample pore volumes. Typical numbers range from 3 to 20 pore volumes,
(Msample) will usually be leached with an equivalent solution depending on the reactivity of the rock and length of time required
weight of 1.0 kg, or 1.0 L water, called the weekly rinse volume for pit lake filling.
(Vrinse). [For example, if the sample has with a bulk density (qb)
of 1.0 kg/L and a porosity (/) of 40%, the sample will have a total 3.3.4. Estimating wall-rock leaching from batch tests with
volume (Vtotal) of 1.0 L and a pore volume (Vpore) of 0.4 L. By divid- representative water compositions
ing the weekly rinse volume (1.0 L) by one pore volume, we calcu- The methods described above rely on a combination of labora-
lated that each weekly cycle of oxidation and rinsing is equal to 2.5 tory test interpretation and mathematical models. Both of these
pore volumes (Vpore).] The chemical release function (fCR) for a aspects require estimating a number of parameters that are extrap-
given dissolved parameter in units of mg/kg per pore volume olated to field conditions. This approach results in a range of
(Vpore) (mg/kg/L) is obtained from: uncertainties that are very difficult to quantify. Another approach
276 D.N. Castendyk et al. / Applied Geochemistry 57 (2015) 267–288

described in the literature involves creating analogue pit lake and has the advantage of yielding directly measureable informa-
water compositions by mixing together solutions with composi- tion about secondary mineral precipitation processes and their
tions meant to represent each of the major chemical loads to a effects on metal and metalloid concentrations.
pit lake (Davis, 2003; Schafer et al., 2006). The solution mixture For any of the above approaches for representing the solute load
is allowed to equilibrate for a period of time after which the chem- generated from water–rock reactions in pit walls, it is important to
ical composition is analyzed along with any mineral precipitates consider the geoenvironmental characteristics of the deposit
that may have formed as a result of mixing. Colloquially, such exploited in the open pit when evaluating their effects on water
bench-top column experiments (i.e. microcosm) and field-based chemistry through numerical modeling. The pit lake database
tank experiments (i.e., mesocosm) experiments have been called (Johnson and Castendyk, 2012) provides data from existing pit
the ‘‘pit lake in a bucket’’ approach. lakes that can be used to make comparisons. Eary and Castendyk
Davis (2003) described this type of procedure as an analogue pit (2012) provide summaries of water chemistry trends for different
lake test. Davis (2003) developed a water balance for a pit lake to types of ore deposits that can also provide some means for making
determine the proportions of different inflows, including rainfall, comparisons.
stream water additions, wall-rock runoff, and groundwater. These
proportions were determined for the early period of filling and the
4. Predicting the pit lake water balance
late stage of filling to take into account the increase in proportion
of rainfall that was predicted to occur as the pit lake increased in
4.1. Overview
size over time. A portion of the water was allowed to evaporate
as well to represent the real system over time. Precipitates were
The pit lake water balance specifies the volumes of water added
collected and analyzed for mineralogy and metal content. Finally,
to and removed from the pit lake by various sources and outputs
the solution mixing and resulting mineral precipitation were sim-
over time. Sources include pit wall runoff, groundwater inflow,
ulated with the PHREEQC geochemical model (Parkhurst and
direct precipitation, and possibly surface water via the diversion
Appelo, 1999) for comparison to the experimental results. Good
of stream water into the pit. Pit wall runoff is typically the single
agreement was observed between measured and simulated solu-
largest contributor of MIW to the lake, however, groundwater
tion compositions. The advantages to this approach include: 1.
can also be significant in some cases (Werner et al., 2001). Lake
Experiments are relatively simple in concept and easy to conduct;
outputs include evaporation, groundwater outflow, and possibly
and 2. Solutions can be mixed in different proportions to examine
surface water discharge. Fig. 6A shows a conceptual model of a
the effects of potential mitigation measures or addition of other
‘‘flow-through pit lake’’ that discharges groundwater and/or sur-
materials to the pit lake such as tailings.
face water to the surrounding environment, whereas Fig. 6B shows
Schafer et al. (2006) also developed an improved approach to
a conceptual model of a ‘‘terminal pit lake’’ that only loses water
predicting pit lake water quality through batch mixing tests. They
via evaporation. Key questions asked by stakeholders include:
conducted large-scale column tests to obtain the chemical compo-
sitions of leachates created by the movement of groundwater
 How long will it take for the pit lake to fill to a constant, steady-
through weathered rock representative of pit wall rocks and mate-
state level?
rial under consideration as a partial backfill for the pit. The solu-
 How much water will discharge from the lake after lake filling?
tions from the column leaching tests were mixed with other pit
 What is the water quality of lake discharge?
lake inflows according to proportions obtained from hydrologic
 What aquifers and surface water bodies will ultimately receive
modeling of the pit lake water balance. The other inflows included
pit lake discharge?
precipitation, runoff from exposed high walls as represented by
 How will pit lake discharge affect the short and long-term water
solutions from unsaturated column leaching tests, groundwater
quality, aquatic ecology, and resource potential of surface water
flow-through wall rocks and backfill material as represented by
and groundwater resources receiving pit lake water?
first flush solutions from saturated column leaching tests con-
ducted with deep groundwater and weathered rock samples. The
This section describes the use of a pit lake water balance to
solutions were allowed to equilibrate and evaporate. Mineral pre-
answer the first two questions which inform subsequent pit-
cipitates formed from solution mixing were analyzed by XRD and
lake-water-quality prediction models. Both ‘‘analytical’’ and
scanning electron microscopy (SEM).
‘‘numerical’’ approaches are referenced below. In this context, ana-
The PHREEQC geochemical model (Parkhurst and Appelo, 1999)
lytical problems are those which can be solved using algebraic
was used by Schafer et al. (2006) to simulate the equilibration and
equations in a spreadsheet program (e.g. Microsoft Excel), whereas
mineral precipitation processes. The PHREEQC model was found to
numerical problems involve simultaneously solving a matrix of
provide a reasonably accurate simulation of major ions, pH, and
equations by iteration using a computer program with a numerical
calcite precipitation but simulations of trace elements (e.g., anti-
solver package (e.g. MODFLOW).
mony, arsenic, nickel, zinc) were of variable accuracy (Schafer
et al., 2006). To improve the predictions of metal concentrations,
Schafer et al. (2006) adjusted equilibrium constants and adsorption 4.2. Direct rainfall and evaporation volumes
constants in the PHREEQC thermodynamic database through an
iterative calibration process. The adjustments were based on the Local climate dictates the volume of water added to a lake via
mineralogical analyses of precipitates from the batch tests that direct rainfall. In the United States, regional monthly and annual
showed the association of the metals with specific solid phases rainfall data can be obtained through the National Weather Ser-
in forms not included in the PHREEQC database. For instance, the vice. For new mine sites, it is advisable to install a weather station
analyses showed that arsenic was associated with calcium carbon- on-site to record local variations in rainfall as well as air tempera-
ate phases (calcite and aragonite) rather than ferric hydroxide. The ture, wind speed, and wind direction because local microclimates
calibrated model provided an accurate prediction of the metal con- can significantly modify local conditions from regional trends. Typ-
centrations in the batch tests providing assurance that the model ically, the average annual rainfall received over a 30-year period is
was capable of yielding a reliable prediction of the field system a useful starting point for pit lake predictions. To account for
over time. The approach developed by Schafer et al. (2006) annual variations in rainfall, the modeler may use a cumulative
provides an alternative to purely numerical modeling approaches frequency plot showing annual rainfall versus probability for all
D.N. Castendyk et al. / Applied Geochemistry 57 (2015) 267–288 277

Runoff
High Precipitation Low Evaporation

Water Table Overflow

Water Table

Groundwater IN Groundwater OUT

Groundwater IN/OUT

Fig. 6A. Simplified water balance model for a flow-through pit lake in a humid climate (modified from Castendyk, 2009a). Both groundwater outflow and surface water
overflow are shown, however, it is unlike that both will be significant in a flow-through pit lake.

Runoff

High Evaporation Low Precipitation

Water Table Water Table

Groundwater IN Groundwater IN

Groundwater IN

Fig. 6B. Simplified water balance model for a terminal pit lake in an arid climate (modified from Castendyk, 2009a).

available data. From this plot, the modeler can also determine a pit lake does not exist. Moreover, the walls of the pit are likely to
minimum rainfall value (i.e. 10% of all years receive a rainfall 6x) reduce the wind velocity such that wind speeds measured near
and a maximum rainfall value (i.e. 10% of all years receive a rainfall the rim of the pit are likely to differ from conditions that will exist
of Px), which can also be considered in subsequent models. Dunne over the future lake surface (Stevens et al., 2002). As such, the
and Leopold (1978) provided instruction on the development of modeler may need to assume an evaporation rate from regional
cumulative frequency plots. maps which will introduce errors to the prediction, rather than
The quality of rainwater local rainwater is also an important measuring a site specific value. Additional errors will be introduced
factor to characterize, as rainwater pH and composition change if climate changes significantly over the life of the pit lake. For
from region to region (Drever, 1997). Direct collection of water example, Hoerling and Eischeid (2007) demonstrated that present
samples from local rain events, coupled with published rainwater evaporation and evapotransporation rates (both of which are tem-
data can be used to define this variable. perature dependent) in the western United States are predicted to
Annual evaporation rates are more difficult to estimate than increase over time as a function of increased surface air tempera-
rainfall rates. Common methods include: 1. Direct measurement ture. Modelers can account for this uncertainty by conducting a
of evaporation pan or lake evaporation; 2. Analytical calculations sensitivity analysis on evaporation rates.
based on observed solar radiation, air temperature, relative humid- Some generalizations regarding pit lake behavior may be sur-
ity, and wind speed; and 3. Numerical models. Not all weather sta- mised once values for rainfall and evaporation are defined. Humid
tions have a 30-year record of daily pan evaporation, however, the climates exist in areas where annual rainfall exceeds annual evapo-
National Weather Service provides the input data for the analytical ration, such as pit lakes in the eastern United States (Seal et al., 2006;
calculations and numerical models, plus regional evaporation Wyatt et al., 2006; Mase et al., 2008) and in central and northern Eur-
maps (http://www.cpc.ncep.noaa.gov/soilmst/e.shtml). Wind has ope (Ramstedt et al., 2003; Schultze et al., 2010). During lake filling,
a strong influence on evaporation rates because it removes the sat- these lakes tend to receive a greater proportion of wall-rock runoff
urated air lying just above the lake surface, which promotes further that can strongly influence lake water chemistry relative to other
evaporation. Dingman (2002) provided several methods to calcu- input sources. To minimize these reactions, the Island Copper Pit
late the free-water evaporation rate based on regional meteorolog- Lake on the west coast of Canada was artificially flooded with sea
ical or climatic conditions. The lake evaporation is determined by water over several days (Pelletier et al., 2009). The addition of sur-
adjusting the free-water evaporation rate to account for advection face water to a pit during mine closure may only be an option in
and heat-storage effects in a lake. Obviously, it is difficult to humid climates where the diversion of surface water from local
directly apply these methods in advance of mining because the rivers will not have excessive impacts on river ecosystems or
278 D.N. Castendyk et al. / Applied Geochemistry 57 (2015) 267–288

downstream water users. Hypothetically, once steady state condi- of exposed rock, the steepness of pit walls, and the compaction
tions are achieved in a humid climate pit lake, excess direct rainfall of loose sediments along hall roads by machinery. As a result, a
should lead to the dilution of dissolved concentrations over time. high percentage of rainwater volume landing inside the basin will
However, Castendyk and Webster-Brown (2007b) predicted that become surface water in the form of wall-rock runoff (Fig. 4).
MIW from unmitigated wall-rock runoff at a future pit lake in New One of the simplest analytical methods to calculate the volume
Zealand may gradually increase dissolved concentrations and acid- of runoff added to a pit lake during a rain event is the Rational
ity over time (see Part 2; Castendyk, Balistrieri, Gammons and Tucci, Method (Fetter, 2001):
this volume). Humid climate lakes are also likely to exhibit flow-    
m3 m
through conditions once steady-state conditions are achieved Q runoff ¼ c  Irain  Acatchment ½m2  ð9Þ
(Fig. 6A). In comparison to a terminal pit lake with an equivalent vol- day day
ume, a flow-through pit lake will have a shorter residence time for The average rainfall rate (Irain) is multiplied by the area of the
lake water. Consequently, lake water in direct contact with wall pit catchment (Acatchment; this area excludes the area of the devel-
rocks will have less time for leaching reactions, and the magnitude oping lake surface), and a dimensionless runoff coefficient (c) to
of maximum concentrations will be limited by mass loss via surface give the volume of runoff over time (Qrunoff). The runoff coefficient
discharge and by dilution from direct precipitation. is an empirically-determined constant that defines the fraction of
Conversely, arid climate pit lakes exist where annual evapora- rainwater that becomes runoff, and the fraction of water that infil-
tion exceeds rainfall, such as the western United States (Davis trates the land surface to become vadose water and ultimately
and Eary, 1997; Shevenell et al., 1999), Western Australia groundwater [see Fetter (2001) for a table of representative c val-
(Oldham et al., 2009; McCullough et al., 2009), and southwest ues which are ranked according to properties of the land surface].
Europe (Sánchez España et al., 2008). Most pit lake studies to date The Rational Method is designed for catchments with areas less
have been conducted in arid regions. On an annual basis, ground- than 100 hectares, and only applies to storms that are longer in
water provides the largest portion of input to the developing lake duration than the length of time required for water to flow from
with occasional pulses of direct rainfall and wall-rock runoff from the farthest catchment divide to the lake shoreline (Fetter, 2001).
storms and snowmelt. The annual volume of water lost by evapo- With these assumptions in mind, modelers can calculate an initial
ration increases as the lake fills and the lake surface area increases. approximation of the volume of runoff added annually to the lake
Ultimately, the rate of groundwater input roughly equals the rate by multiplying the annual rainfall rate by the catchment area and
of evaporation at which point the lake has reached its steady-state an appropriate mine runoff coefficient. The Rational Method
water level. The lake filling process is slow in arid climates and should be used as a first order approximation of runoff volumes.
lakes in arid regions may take hundreds of years to achieve Dingman (2002) provided additional analytical equations for
steady-state (Kempton et al., 1997). If evaporation is the only rainfall–runoff relationships. Several numerical models are also
process that removes water from the lake, then the lake becomes available, such as PRMS offered by the U.S. Geological Survey
a terminal point for local groundwater flow and is called a terminal (http://water.usgs.gov/cgi-bin/man_wrdapp?prms).
pit lake (Fig. 6B). This does not imply that the local water table is The fraction of rainwater that does not become wall-rock runoff
flat, but rather that a local groundwater divide surrounds the pit will either take a shallow flow path and discharge to the pit as
lake. Fetter (2001) and Castendyk (2009a) provide flownets of ter- vadose zone seepage (Fig. 4), or it will infiltrate to the water table
minal lakes which illustrate this hydrogeology. and become part of local groundwater (see below). In a simple
All pit lakes are initially terminal pit lakes and only become model where all rainwater flows into the pit without contacting
flow-through lakes if the elevation of lake water exceeds the sur- groundwater, the volume of seepage is equal to the daily volume
rounding water table or overtops a surface water divide during of water infiltrating through the land surface (Qinfiltration). This infil-
steady-state conditions. Pit lakes in arid climates may remain ter- tration term is derived from the Rational Method as follows:
minal during steady-state conditions, although mines excavated in    3  3
mountainous terrains high above the surrounding topography may m m m
Irain  Acatchment ½m2  ¼ Q runoff þ Q infiltration ð10Þ
discharge groundwater to deep aquifers and thus exhibit flow- day day day
through conditions. Local groundwater quality strongly influences Finally, an areal recharge rate (W; m/day) is defined as the daily
the water quality of arid climate pit lakes. As noted above, these linear addition of infiltration across the entire pit catchment area:
lakes have a higher residence time compared to flow through lakes h i
of equal volume. The loss of water via evaporation increases the   Q m3
m infiltration day
concentrations of dissolved solids over time until the water W ¼ ð11Þ
day Acatchment ½m2 
achieves equilibrium with specific minerals; a process known as
evapoconcentration. Eary (1998, 1999) discussed the effects of A numerical vadose-zone flow model may be required to handle
evapoconcentration on pit lakes in arid climates, and defines min- scenarios where a fraction of infiltration recharges soil moisture, a
erals that are likely to precipitate from solution and thereby limit fraction recharges the water table, and a fraction discharges to the
the maximum dissolved concentrations of some elements. pit wall as seepage.

4.3. Wall-rock runoff volume 4.4. Groundwater volume

Open pit mines are anthropogenic topographic depressions that Permanent pit lakes will only develop at open pit mines that
can become closed surface water basins after closure. The highest extend below the pre-mining water table. In general, the pre-min-
elevations surrounding the pit become temporary (i.e. flow- ing water table will be relatively close to the surface for mines sit-
through lakes) or permanent (i.e. terminal lakes) surface water uated in humid climates or topographic low points, whereas it will
divides. As such, most of the rain (or snow) landing inside the pit be relatively deep for mines situated in arid climates or topo-
divide that does not evaporate (or sublimate) will flow to lake as graphic highs. As previously discussed, all pit lakes begin as termi-
either wall-rock runoff or shallow groundwater seepage. The dis- nal discharge points for local groundwater and fill to a steady-state
turbed surfaces created during mining, including the pit walls, lake level where lake inputs balance lake outputs. In some cases,
bench tops, pit floor, and roads (Figs. 2 and 3), will typically have ephemeral pit lakes may develop above the water table in response
a low capacity for rainwater infiltration due to the low porosity to extreme rain events or snow melt, especially if pits are underlain
D.N. Castendyk et al. / Applied Geochemistry 57 (2015) 267–288 279

by low-permeability bedrock or if a layer of low-permeability sed- lateral extent of the proposed open pit, and conduct field tests
iment accumulates at the bottom of the pit. Such lakes eventually to define aquifer characteristics. These tests involve drilling a
drain through a combination of evaporation and infiltration. pumping well and an observation well at the proposed mine site,
The following site-specific information may be needed to fully and monitoring the depth of the water table in the observation
assess the groundwater hydrology of a pit lake: well during a 24-h pump test. With this information, the Theis
Equation provides a simple analytical tool to calculate the depth
 Pre-mining depth of the water table below the land surface. and shape of the cone of depression required to dewater the open
 Depth and geometry of the pit. pit at any given time (Bair and Lahm, 2006):
 Hydrogeologic parameters of each lithologic unit inter-
Q1 Q Q
sected by the pit, including; hydraulic conductivity, effec- stotal ¼ Wðu1 Þ þ 2 Wðu2 Þ þ    þ n Wðun Þ ð12Þ
4p T 4pT 4pT
tive porosity, thickness, transmissivity, and storativity.
 Potentiometric surface(s) and hydraulic gradients surround- where stotal is the composite drawdown (m) of the water table
ing the open pit before, during, and after pumping. below the mine produced by n pumping wells, Qn (m3/day) is the
 The groundwater recharge rate to each lithologic unit inter- pumping rate of each well, T (m2/day) is the transmissivity of the
secting the pit. aquifer, u is a dimensionless constant and W(u) is the dimensionless
 The locations of groundwater divides and impervious ‘‘Theis well function.’’ The well function is based on the properties
boundaries that surround the pit both horizontally and of the aquifer, the distance between the pumping wells and the cen-
vertically. ter of the pit, and the pumping time, and is commonly determined
 Pumping rates and pumping times, plus pump locations and using the ‘‘Theis curve-matching technique’’ which fits observed
screen intervals. drawdown data to the ‘‘Theis type curve.’’ For more information
 Hydrologic characteristics of major structural features, such on the well function, see Bair and Lahm (2006) or Fetter (2001). It
as faults, and historic workings (i.e. vertical and horizontal is important to make sure that the assumptions used to evaluate
tunnels) intersected by the pit; and drawdown data are consistent with the characteristics of the mine
 Hydrologic characteristics of blast-generated fractures in site and the wells. Key assumptions required for the Theis Method
wall-rock, including the fracture-density, depth, area, and are that the pumping well is fully penetrating, and the aquifer is
aperture. homogeneous and isotropic. Several other methods exist which
may be more applicable to a given mine site or well configuration.
One of the most important and time-intensive aspects of The reader is referred to Fetter (2001) for a review of other
groundwater modeling is the hydrogeologic characterization of methods.
wall-rock. Ore deposits develop in complex hydrogeologic settings The pumping wells are turned off at mine closure allowing the
that involve a range of lithologies and geologic structures. Faults, water table to rebound and the open pit to fill with water. At this
fractures, and joints create secondary porosity and preferential point, the open pit essentially acts like a very large diameter well
flow paths in preexisting rocks, whereas the precipitation of new that fills with water from the pit floor and walls. The groundwater
minerals in pore spaces reduces both primary and secondary discharge area (Adischarge; m2) is equal to the sum of the vertical and
porosity. Superimposed upon these natural characteristics is the horizontal areas through which groundwater discharges, and is ini-
anthropogenic porosity caused by mining, including underground tially limited to the pit floor. A first-order approximation of the
workings and shafts, and blast-generated fractures. As such, mod- daily groundwater discharge rate during lake filling can be taken
elers should expect groundwater flow to be heterogeneous and from the daily pumping rate prior to mine closure. For a more
anisotropic around open pit mines. To define flow characteristics, accurate value, Marinelli and Niccoli (2000) provided a series of
modelers need to consider variations in wall-rock lithology, major analytical equations for calculating changes to groundwater input
faults, joints, and fractures, as well as blast-generated fractures in to a filling pit lake:
the wall rock. If the open pit exists in a previously mined area, his-    3  3
m3 m m
torical mine tunnels and shafts that intersect the pit wall can be Q gwtotal ¼ Q wall þ Q floor ð13Þ
day day day
high-porosity, preferential flow paths for groundwater and may
have considerable lateral and vertical extent beyond the open pit.    
m3 m
It is difficult to know the exact location and extent of historical Q wall ¼W  p  ðr 20  r 2p Þ ð14Þ
works, and groundwater flow in this terrain may exhibit similar day day
complexity as karst groundwater environments. A careful study  
of historical mine maps and geologic maps combined with field m3 K h2
Q floor ¼ 4  rp   ðh0  dÞ ð15Þ
based observations, aquifer pump tests, and lab permeability tests day m2
may be required to account for the heterogeneity and anisotropy of sffiffiffiffiffiffiffiffi
site parameters. Niccoli (2009a) provided further discussion of K h2
hydrogeologic characterization. m2 ¼ ð16Þ
K v2
Ultimately, the modeler will need to develop a conceptual
model of groundwater flow that accounts for the major groundwa- where Qgwtotal is the total groundwater inflow to the pit per day,
ter flow features and simplifies the minor variations. Oversimplifi- Qwall is the groundwater entering through the walls of the open
cation will lead to errors in groundwater predictions, whereas too pit, and Qfloor is the groundwater entering through the floor of the
much detail increases the time, effort, and numerical complexity of pit. For groundwater entering through the pit walls (Eq. (14)), W
the model (Anderson and Woessner, 1992). Typically, errors in a (m/day) is the aerial recharge defined in Eq. (11), ro is the radius
prediction are caused by errors in the conceptual model, which (m) of influence which the pit has on groundwater (i.e. half of the
makes the development of a conceptual model one of the most aerial width of the cone of depression), and rp is the geometric radius
important and time consuming aspects of modeling. (m) of the inside of the pit. For groundwater entering through the pit
During the planning phase of mining, companies are interested floor (Eqs. (15) and (16)), Kh2 is the horizontal hydraulic conductivity
in the volume of water that must be pumped in order to lower the (m/day) of the pit wall, Kv2 is the vertical hydraulic conductivity (m/
water table below the active mine surface. To address this day) in the pit wall, m2 is a dimensionless term equal to the square
question, the company needs to define the required depth and root of the ratio of horizontal to vertical hydraulic conductivity, ho is
280 D.N. Castendyk et al. / Applied Geochemistry 57 (2015) 267–288

the initial hydraulic head (m) before dewatering, and d is the depth typically controlled by regional surface water regulators and
of water (m) in the pit lake. Both ho and d are defined with respect to may fluctuate from year to year depending upon the availability
height above the same groundwater datum. Before applying this of excess surface water. Most importantly, several terms are
approach, the modeler should review the work by Marinelli and dependent upon the geometry of the mine site which often
Niccoli (2000) to ensure that the mine site in question meets the cri- changes over the mine life cycle between initial mine planning,
terion specified by the authors. mine development, and mine closure. Because the pit lake water
Three-dimensional, numerical groundwater models such as balance is dependent upon these variables, each change to the
MODFLOW are commonly used to estimate the volume of flow into mine plan will affect the pit lake water balance to some
and out of a pit lake during mine closure. Niccoli (2009b) noted degree as discussed by Castendyk and Webster-Brown (2010). To
that numerical models of a pit lake may not necessarily generate minimize this error, stakeholders should update the pit lake water
more accurate groundwater prediction than analytical methods, balance throughout mine development each time major changes to
however, numerical models can handle more complexity and het- the mine plan are approved and implemented.
erogeneity than analytical models. Ultimately, all groundwater
predictions should be calibrated using two or more observations
5. Predicting lake mixing
of pit lake water levels after mine closure, and long-range, post-
closure audits should be conducted to ensure that predicted levels
5.1. Overview
remain within specified confidence intervals (Anderson and
Woessner, 1992).
The physical limnology of a pit lake is defined by processes that
physically separate the water column into two or more horizontal
4.5. Water balance design
layers and processes that vertically mix these layers. Periods of
stratification can produce chemically different water layers
Water balances are used to determine the time required to fill a
whereas periods of mixing can homogenize portions of the water
pit to a steady-state water level, the volume of water expected to
column. For these reasons, pit lake modelers need to determine
discharge from flow-through pit lakes under steady state-condi-
the frequency and depth of seasonal mixing periods, called ‘‘turn-
tions, and the volumes of specific lake inputs and outputs used
over,’’ in order to specify the volumes of mixing layers in geochem-
in both hydrodynamic and geochemical prediction models. Typi-
ical predictions and to identify if a permanently-isolated bottom
cally water balances are calculated analytically on an annual basis
layer will develop.
using a spreadsheet programs to tabulate the inputs and outputs of
Turnover events influence water chemistry by mixing dissolved
sources and sinks. Spreadsheet programs also provide an indepen-
oxygen from the lake surface throughout the water column and by
dent check on the accuracy of numerical groundwater and hydro-
transporting deep, carbon dioxide-rich water to the lake surface
dynamic models. Discrepancies between analytical and numerical
where it may exsolve raising epilimnion pH. Suboxic or anoxic
predictions of lake filling times and lake overflow volumes indicate
hypolimnion may become oxic as a result of lake mixing. This sec-
that inconsistencies exist within the model design.
tion summarizes the procedures used to define a conceptual model
To construct a spreadsheet water balance, the modeler must
of physical limnology. Numerical models of physical limnology are
know the following data:
called ‘‘hydrodynamic models,’’ and Part 2 of this series describes
the development of a hydrodynamic model for the Dexter Lake,
 Inputs
Nevada (Castendyk, Balistrieri, Gammons, and Tucci, this volume).
– Annual rainfall volume landing on the lake surface over time.
– Annual volume of surface water (i.e. streams located outside
the pit catchment area) diverted into the pit over time (if 5.2. Stratification and mixing
applicable).
– Annual groundwater input volume over time. Two simple conceptual models for pit lake limnology describe
– Annual runoff volume over time. holomictic and meromictic pit lakes (Fig. 7). Holomictic pit lakes
 Outputs have two layers, called the epilimnion (shallow layer) and the
– Annual evaporation volume from the lake surface over time. hypolimnion (deep layer), respectively, which completely mix at
– Annual groundwater outflow volume from the lake over least once a year (Fig. 7A). Meromictic pit lakes have at least three
time. layers: epilimnion and hypolimnion layers that mix at least once a
– Annual surface water volume leaving the lake over time (if year and a basal, monimolimnion layer that remains permanently
applicable). isolated from the surface environment (Fig. 7B). Although these
 Depth and geometry relationships of the proposed mine site terms characterize the most common limnologic behavior
– Pit volume. observed in pit lakes, there are a range of other possibilities includ-
– Ultimate steady-state lake depth ( pre-mining water table ing shallow pit lakes that are perpetually mixed by wind and never
elevation). develop stratification, pit lakes that develop four or more density-
– Lake volume and aerial surface area as a function of lake separated layers, lakes that begin as holomictic and become
depth (i.e. hypsograph). meromictic over time, and lakes that are initially meromictic that
– Pit wall catchment area as a function of lake depth. undergo complete top-to-bottom mixing in response to extreme
– Groundwater inflow/outflow rates as a function of lake weather events, landslides, or lake management activities (see
depth. Berkeley Lake in Part 2; Castendyk, Balistrieri, Gammons, and
– Groundwater inflow/outflow areas as a function of lake Tucci, this volume).
depth. The fundamental variables that control physical limnology are
upward stabilizing forces generated by the buoyancy of layers,
Often these variables are subject to change over time which and downward mixing forces generated by wind moving over the
introduces errors to the predicted water balance. Climate- lake surface. Imberger and Patterson (1990) described the seasonal
dependent terms may change significantly if climate changes over changes in the density structure and buoyancy within the water
the life of the mine or the long term life of the pit lake. Surface column of a temperate zone lake. In general, the temperature
water available for diversion into the pit during lake filling is profile during summer and winter conditions produces low density
D.N. Castendyk et al. / Applied Geochemistry 57 (2015) 267–288 281

Several mechanisms have been proposed that explain the devel-


(A) Holomictic pit lake
opment of meromictic pit lakes: 1. The geometry of the lake com-
bined with pit-wall wind sheltering impede whole-lake mixing; 2.
Oxygen-saturation
Epilimnion
Downward-settling mineral precipitates re-dissolve under reduc-
ing conditions near the lake bottom and increases the salinity
Thermocline
and density of the basal layer, which can result in a monimolim-
nion; and 3. The addition of both high-salinity groundwater and
Low-oxygen low-salinity surface water to a filling pit lake results in a density
conditions
stratified lake that resists wind mixing. The flooding and
Hypolimnion
Hypolimnion induced-stratification of the Island Copper Pit Lake demonstrated
the significance of the salinity of lake inputs in the development
of meromictic pit lakes, and thereby supports the last mechanism
(Pelletier et al., 2009; Schultze and Boehrer, 2009).
Groundwater in,
In some cases, such as the Berkeley Lake, Montana (see Part 2;
Groundwater out
Castendyk, Balistrieri, Gammons, and Tucci, this volume), mero-
mictic lakes may periodically exhibit whole-lake mixing or gradu-
ally evolve into a holomictic lake. Schultze and Boehrer (2009)
(B) Meromictic pit lake
hypothesized that unforeseen, whole-lake turnover events in mer-
omictic lakes could be caused by extreme weather events or major
Oxygen-saturation
Epilimnion wall-rock failures. Whole-lake mixing in meromictic lakes could
result in acute environmental impacts to terrestrial and aquatic
Thermocline
ecosystems owing to the transport of monimolimnion water to
Low-oxygen Hypolimnion
the lake surface and the rapid release of stored gases such as,
H2S, CH4, and CO2. However, in the case of the Island Copper Pit
Lake, meromictic conditions have benefited lake mitigation efforts
Chemocline
by isolating metal-rich water from the surface environment
Anoxic (Pelletier et al., 2009). For these reasons, it is important for stake-
conditions
Monimolimnion holders to know if future pit lakes will become meromictic and if
stratification will endure extreme weather events and periodic
landslides.

5.3. Hydrodynamic predictions


Fig. 7. Conceptual models of holomictic (A) and meromictic (B) pit lakes with
limnologic terminology (modified from Castendyk, 2009a).
Three tools have been used to predict the physical limnology of
epilimnion water overlying higher density hypolimnion water. As a future pit lakes: observations from local natural lakes and reser-
result, the buoyancy of the epilimnion is strong enough to resist voirs, analytical models based on future lake geometry, and
wind driven mixing and the lake is seasonally stratified. Surface numerical models. Because local lakes exist in a similar climate
cooling during the fall and surface warming during the spring min- as the future pit lake, data from local lakes can provide a sense
imize the density difference between layers. Under these condi- of whether local winds tend to be strong enough to perpetually
tions, successive storm events mix epilimnion water with deeper or seasonally mix water columns and can also define a range of
and deeper portions of the hypolimnion until the lake is likely epilimnion depths during summer stratification, months
homogenized. when turnover occurs, and months when ice cover exist (if any).
Early pit lake researchers noted that pit lakes have a unique Although site specific conditions at a pit mine can lead to signifi-
geometry compared to natural lakes, and questioned whether the cant differences between natural lakes and pit lakes, nevertheless
shape of pit lakes could limit whole-lake mixing (Lyons et al., these data provide a sense of how a future pit lake in a similar cli-
1994; Doyle and Runnells, 1997). By comparing lake surface area mate may behave and aid the evaluation of numerical prediction
(Alake; m2) to maximum depth (zmax; m), sometimes called the results. The identification of multiple meromictic lakes can indi-
‘‘aspect ratio,’’ researchers concluded that pit lakes tend to be much cate local sources of saline groundwater as discussed by Schultze
deeper than natural lakes of similar surface area. Hutchinson (1957) et al. (2010) in a statistical analysis of over 50 pit lakes in the Cen-
provided a useful term for comparing these dimensions called ‘‘rel- tral German Lignite District.
ative depth’’ (zrelative) which is expressed as a percentage: Several authors have suggested that geometry-based compari-
sons alone can provide indications of whether future pit lakes will
pffiffiffiffi become holomictic or meromictic (Lyons et al., 1994; Doyle and
50  zmax  p
zrelative ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð17Þ Runnells, 1997; Castro and Moore, 2000). Owing to the simplicity
Asurface of the calculations, relative depth has been widely adopted by
According to Wetzel (2001), natural lakes tend to have the mining industry and pit lake modelers alike, and relative
zrelative 6 2% whereas existing pit lakes have zrelative ranging from depths are frequently reported in pit lake case studies. However,
1% to 45% (Castendyk, 2009b). In addition, pit lakes tend to have extensive observations of aspect ratios, relative depths, and the
steep pit walls that are thought to shelter the lake surface from physical limnology of existing lakes have yet to identify a consis-
the maximum wind speeds observed on the rim of the pit. Pit walls tent relationship between geometry and physical limnology that
also shade the surface of the lake from incoming solar radiation applies to a wide range of lakes. Joehnk (2001) plotted the maxi-
which reduces epilimnion temperature and lowers the density mum depth and surface area of 135 meromictic and 684 holomictic
and buoyancy of the layer. Hypothetically, the combination of an lakes (mostly natural) and concluded that meromictic lakes could
elevated relative depth with wind sheltering should make pit lakes not be differentiated from holomictic lakes based on geometry
more likely to develop meromictic conditions than natural lakes alone (for further discussion, see Schultze and Boehrer, 2009).
and reservoirs. Castendyk and Webster-Brown (2007a) compared the relative
282 D.N. Castendyk et al. / Applied Geochemistry 57 (2015) 267–288

depths and physical limnology of 22 pit lakes and could not define of pit lake geochemical models: 1. Define model purpose (see
a relationship between these parameters. Gammons and Duaime above); 2. Develop a conceptual model of the most significant lake
(2006) noted that the Berkeley Lake does not exhibit the limnologic processes; 3. Define the representative water chemistry for each
behavior that would be expected using relative depth comparisons. lake input [see Predicting Wall-rock Runoff Geochemistry; see also
These studies suggested that geometry-based comparisons have a Castendyk and Webster-Brown (2007b) for treatment of
low probability for accurately predicting future pit lake limnology. groundwater, surface water, and rainwater inputs]; 4. Design and
Moreover, geometry-based comparisons also provide no informa- run the model (see Part 2 for an example from Martha Mine,
tion on the likely volumes of lake layers that will circulate during New Zealand; Castendyk, Balistrieri, Gammons, and Tucci, this vol-
turnover events; information that is needed for subsequent geo- ume); 5. Validate model results using direct observations of lake
chemical modeling. water chemistry, laboratory experiments, observations from
Numerical hydrodynamic predictions provide detailed predic- existing pit lakes in similar ore bodies (note: the INAP Pit Lakes
tions of future pit lake limnology, however, these models take a Database was designed for this type of post-modeling comparison;
long time to develop and require large amounts of input data. see Johnson and Castendyk, 2012), or other methods (see Eary and
Model inputs include the future lake geometry as specified by Castendyk, 2009); 6. Expand and adjust the conceptual model as
the mine plan, predicted climate data for the duration of the lim- needed to improve the accuracy of the pit lake prediction. Repeat
nology prediction, predicted groundwater inputs and outputs Steps 3–5 until a satisfactory prediction is achieved. The initial
defined by the hydrogeologic model, and the predicted pit lake conceptual model and prediction should be updated over time as
water balance. As previously stated, changes to the mine plan new information about the pit lake is obtained during mining;
can affect hydrodynamic predictions, and so regular updates to and 7. Compare predicted values to water quality guidelines. If
the hydrodynamic model may be required over the mine life. Pop- necessary, use a sensitivity analysis to explore potential mitigation
ular hydrodynamic programs include CE-QUAL-W2 (Wells and options (see Castendyk and Webster-Brown, 2007a,b).
Berger, 2010) and DYRESM (Imberger and Patterson, 1990), which Modelers have used analytical and numerical approaches to
predict the temperature, salinity, and density profiles in lakes over generate annual predictions of water chemistry based on concep-
time. These programs have been used to model several existing pit tual models. Lewis (1999) used a spreadsheet program to calculate
lakes (Atkins et al., 1997; Fisher and Lawrence, 2000; Balistrieri the mass balance of epilimnion and hypolimnion layers based on
et al., 2006) and to develop a predictive model of a future pit lake the concentrations and volumes of annual lake inputs and outputs
(Castendyk and Webster-Brown, 2007a). Part 2 of this study pro- plus the timing and depth of mixing events. Such models are sim-
vides a detailed example of the development of a numerical hydro- ple to construct but neglect geochemical reactions. Bird et al.
dynamic model for the existing Dexter Lake in Nevada after the (1994), Kempton et al. (1997) and Tempel et al. (2000) discussed
work by Balistrieri et al. (2006). the use of popular geochemical speciation programs, such as PHRE-
EQC (Parkhurst and Appelo, 1999) and MINTEQA2 (Allison et al.,
1991), to predict the water chemistry of a unit volume (e.g. 1 L)
6. Predicting lake geochemistry that results from mixing proportions of representative lake inputs
specified by the annual pit lake water balance (e.g. 80% groundwa-
6.1. Overview ter, 10% surface water, 5% wall-rock runoff, and 5% rainwater). Part
2 of this series provides an example of the design of geochemical
Existing pit lakes display a wide range of water quality ranging prediction for a future pit lake at the Martha Mine, New Zealand
from circum-neutral systems with low total dissolved solids (TDS), using PHREEQC (Castendyk, Balistrieri, Gammons, and Tucci, this
acidic systems with high TDS, circum-neutral systems with high volume). The most recent approach used by modelers has been
TDS, and basic systems with high TDS. All pit lakes will eventually to design hybrid models that directly link groundwater, limnology,
develop an aquatic ecosystem based on lake water quality, nutrient and geochemistry predictions into a single model (Oldham et al.,
concentrations, and introduction of organisms. Some existing pit 2009; Werner, 2009). These models are advantageous because
lakes contain sustainable fish populations whereas other lakes con- changes to one component, such as predicted groundwater dis-
tain only algal communities or, in the case of low pH systems like charge, will be automatically integrated into the overall prediction.
Berkeley Lake, extremophile communities. The growth, population The accuracy of pit lake predictions can be difficult for stake-
size, sedimentation, and decomposition of these organisms will holders to evaluate. Comparing geochemical pit lake predictions
influence pit lake water quality in addition to physical and chem- to observed data from the same pit lake is the only direct way to
ical processes. quantify the accuracy of a given prediction, although several indi-
The following section focuses on the development of a pit lake rect methods exist. Unfortunately, it is likely to be decades
water chemistry prediction based on physical and chemical pro- between the original pit lake prediction and the formation of a
cesses alone. The results provide a useful indication of future water pit lake, and few such studies have been conducted and published.
quality and can be added to a hydrodynamic model, such as CE- Two notable exceptions show good correlation between predicted
QUAL-W2 (Wells and Berger, 2010) or CAEDYM (Hipsey et al., and observed water chemistry especially with respect to major
2006), to explore the effects of biology on water quality. The reader ions (Oldham et al., 2009; Werner, 2009). The principal disadvan-
is directed to the technical manuals for these programs for further tage is the time and effort required to construct such models, their
discussion on modeling biologic processes. A first-pass approxima- complexity, and the possible need for existing lake observations in
tion of potential lake productivity can be derived from the nutrient order to appropriately define model parameters and processes. As
load added to the lake from wall-rock runoff and groundwater. noted by Drever (2011), two of the biggest problems with modern
Many mining companies use nitrogen-based explosives during geochemical pit lake predictions are the challenge for model
pit excavation, the residuals of which can add a pulse of nitro- reviewers to independently repeat and check calculations, and
gen-rich water to the pit lake during lake filling and thereby stim- the lack of understanding of the uncertainty associated with
ulate initial algal growth. predictions. Nevertheless, these predictions represent the best
Geochemical predictions of pit lake water chemistry build upon estimate stakeholders have for what will happen in the future.
the results of the water balance and mixing models. Anderson and This section reviews some of the key geochemical processes
Woessner (1992) provided a flow chart for the development of occurring within pit lake water (Fig. 8). Model design is
groundwater models which we have adapted for the development demonstrated using a case study of the Martha Mine, New Zealand
D.N. Castendyk et al. / Applied Geochemistry 57 (2015) 267–288 283

Mass Rain Mass


Input Evaporation
water Input
from from
Runoff Runoff

Gas Exchange

Dissolution/ High O2,


Oxidation Low CO2
Speciation/
Annual Mixing Redox

High CO2,
Low O2

Ion Adsorption

Mass Input Mass Input


Precipitation from
from
Groundwater Groundwater

Settling

Sediment

Fig. 8. Conceptual model of geochemical reactions occurring within a holomictic pit lake (modified from Castendyk, 2009a).

presented in Part 2 of this series (Castendyk, Balistrieri, Gammons, applications specifically designed to model iron hydroxide
and Tucci, this volume). For guidance on other components of adsorption. Adsorption onto aluminum hydroxides (see Karamalidis
model development, the reader is directed to the references shown and Dzombak, 2010), manganese hydroxides (see Visual MINTEQ
in Table 2. by Gustafsson, 2010), clay minerals, organic matter, and other
surfaces will also influence dissolved concentrations.]; 5. The
dissolution or oxidation of submerged wall-rock minerals (see
6.2. Conceptual models of pit lake geochemistry
Section 3); and 6. Sedimentation of downward settling solids
coupled with the removal of metals and metalloids from the water
Geochemical processes can be divided into external processes
column (an implicit assumption in geochemical programs).
that occur before water enters the lake, and internal processes that
One process that may be oversimplified in present models is the
occur inside the lake. The most significant external processes
partitioning of gases between the lake surface and the atmosphere.
involve the generation of mine impacted water (MIW) along flow
Models typically assume that equilibrium between epilimnion
paths leading to the lake as discussed in Section 3. Mine drainage
water and atmospheric gases occurs on an annual basis following
can also contain elevated levels of nutrients due to nitrogen-based
lake turnover (see Part 2). This process increases the dissolved oxy-
blasting materials used during pit excavation that can affect the
gen concentration of the epilimnion and chemically differentiates
microbial productivity of the initial lake. Both MIW and nutrients
the epilimnion from the hypolimnion. At the same time, equilib-
should be considered when defining the representative wall-rock
rium will remove dissolved CO2 and alkalinity from epilimnion
runoff chemistry used in the model.
water which raises pH according to the following reactions:
Existing geochemical predictions focus on the following physi-
cal and chemical processes occurring within the lake (see Fig. 8): Hþ þ HCO3 ¼ CO2ðaqÞ þ H2 O ð18Þ
1. Vertical mixing of lake water during turnover events; 2.
Exchange of dissolved oxygen and carbon dioxide between the
CO2ðaqÞ ¼ CO2ðgÞ ð19Þ
atmosphere and the lake surface; 3. Precipitation of minerals from
lake water under saturated conditions [see Eary (1999) for a list of Consequently, geochemical predictions may show epilimnion
minerals likely to precipitate under specific pH conditions]; 4. water to have a higher pH than hypolimnion water which may or
Adsorption of trace metals and/or metalloids onto fine-grained may not reflect reality. Literature on the treatment of MIW as well
mineral surfaces and organic matter coupled with the downward as the treatment of municipal drinking water provides many
settling of these particles [see Tempel et al. (2000) for a discussion examples where stagnant surface water does not necessarily
on modeling arsenic concentrations in a pit lake as a result of sur- release CO2(g) unless it is physically agitated. Eary (1999) showed
face adsorption. Iron hydroxides are typically used as the adsorp- that existing pit lakes with pH less than 8.25 have partial pressures
tion surface in pit lake predictions because of their common of CO2 (PCO2) that exceed the atmospheric PCO2, and that observed
occurrence in MIW and because existing software include PCO2 generally increases as pH decreases. Thus, the assumption
284 D.N. Castendyk et al. / Applied Geochemistry 57 (2015) 267–288

Gather site data: the job of the modeler to minimize these errors. Ultimately, the
Pre-mining Water
Quality Survey
Geology, Hydrology, accuracy of a geochemical prediction will be evaluated by stake-
Climate
holders based on the difference between predicted water quality
and observed water quality over time. Not unlike economic fore-
Develop Mine
Establish Post-Mining Model Limnology and casts or weather forecasts, pit lake water chemistry predictions
Geometry &
Objecves Water Quality
Water Balance will never have 100% accuracy. It is important for the modeler to
convey the associated uncertainty to all stakeholders at the onset
of predictive modeling so that all parties can agree on the purpose
Compare
of the model and develop realistic expectations for model results.
As such, stakeholders should expect predictions to provide a likely
Predicon Does Meet Predicon Does Not Explore Migaon range of concentrations for principal contaminants of concern over
Objecves Meet Objecves Opons time, which can be used for risk assessment and mitigation plan-
ning. Stakeholders should not expect predictive models to provide
Submit Mine Permit the exact concentration of a given dissolved species during a spe-
Applicaon Postpone Mining cific year following mine closure.

Fig. 9. Flow chart illustrating the ideal use of geochemical predictions during the 7. Conclusions
planning phase of open pit mine development. This flow chart ends with the
initiation of mining activity and continues on Fig. 10. This paper is Part 1 of a two-part series on contemporary
approaches to pit lake modeling and management. Part 1 reviews
the theory behind pit lake research with emphasis on the methods
commonly used to characterize wall-rock runoff geochemistry,
Mine Closure and predict future lake hydrology, predict future physical limnology,
Lake Filling and predict pit lake water chemistry. The principal findings of Part
1 are summarized below:
Lake and Ecosystem Post-Mining
Compare
Monitoring Objecves
 Wall-rock geochemistry models estimate the composition of pit
wall runoff that will be added to the lake after mine closure.
Most of these models rely on a combination of extrapolation
Observaons Do Not Observaons Do
Meet Objecves Meet Objecves of laboratory kinetic data and theoretical modeling of
sulfide oxidation by oxygen in pit wall rocks. These methods
Implement
Develop Post-Mining
require estimation of a number of parameters that have
Remediaon
Lake Use non-quantifiable levels of uncertainty, but have the advantage
of relying primarily on data obtained from typical baseline
Model Remediaon
Opons
File for Regulatory hydrogeochemical characterizations and numerical models;
Release
hence, the predictions can be made prior to mine startup. The
alternative approach is to conduct experiments in which ana-
Fig. 10. Following from Fig. 9, this flow chart begins at the conclusion of mining
logue pit lake compositions are created by mixing solutions rep-
activity and illustrates the ideal use of pit lake geochemical predictions during the
closure phase of open pit mining. resenting the major chemical loadings to the pit lake, such as
wall-rock runoff and wall-rock leaching. The analogue experi-
mental approach offers the advantages of providing measure-
that epilimnion pit lake water will achieve equilibrium with atmo- able solution compositions and secondary mineral precipitates
spheric gases may cause discrepancies between predicted and resulting from solution mixing, but has the disadvantage that
observed surface water pH. weathered rock samples and groundwater samples must be
Several additional processes occur in pit lakes that are generally available for use in the experiments; a requirement that may
not included in current predictive models due to their complexity: be difficult to meet for modeling studies conducted prior to
1. Sediment diagenesis and recirculation of pore-water ions into mine startup.
the water column (Wendt-Potthoff, 2006); 2. Aerobic microbial  Hydrologic models indicate the volumes of water that will be
productivity (especially iron-oxidizing bacteria), incorporation of added to a pit lake during lake filling from wall-rock runoff,
dissolved metals and metalloids into organic matter, and adsorp- groundwater, rainfall, and in some cases, surface water. Mine
tion of dissolved metals and metalloids onto organic surfaces planners use these predictions to estimate the depth, volume,
(Kalin and Wheeler, 2009; Pelletier et al., 2009); 3. Anaerobic and surface area of the final pit lake, plus the duration of time
microbial productivity (especially iron-reducing bacteria) under between mine closure and the establishment of a steady-state
reducing conditions found in monimolimnion water (Weilinga, lake level. A key environmental question is whether a future
2009); and 4. Re-dissolution of downward-settling solids into pit lake will be a terminal hydrologic sink or a flow-through sys-
monimolimnion water. These processes represent data gaps in cur- tem. Terminal pit lakes are more likely to form on flat topogra-
rent predictions and opportunities for future research. phy in arid climates and for shallow lakes with large surface
areas. These lakes have the advantage of keeping MIW on the
6.3. Model limitations mine site, but are likely to show increasing concentrations of
dissolved solids and trace metals over time owing to evapocon-
Modelers are required to make assumptions in the development centration. Flow-through lakes are more common on sloping
of these initial models, and it may be impossible to validate all topography in humid climates and for deep lakes with small
model assumptions. Consequently, errors generated by initial surface areas. The total dissolved solids concentrations in
hydrological and limnological models will be carried over into flow-through lakes cannot achieve the same concentrations as
the geochemical model. These errors are unavoidable, and it is TDS in terminal pit lakes owing to the constant flushing of lake
D.N. Castendyk et al. / Applied Geochemistry 57 (2015) 267–288 285

water over time. However, once the pit lake achieves compared to existing lakes in similar ore deposits, or the model
steady-state conditions, MIW collected and stored within a approach could itself be ‘‘tested’’ on an existing pit lake with pub-
flow-through pit lake will discharge to streams and/or aquifers lically available data. More field observations will help to reduce
surrounding the mine and will migrate away from the mine site. the degree of uncertainty within, and quantify the significance of,
Both analytical and numerical approaches have been used to some of the parameters discussed herein. Also, more publications
determine hydrologic models for pit lakes. of pit lake predictions and data will help to expand the collective
 Limnological models show the frequency and maximum depth understanding of pit lake prediction between scientists while
of vertical mixing events within the lakes. These events homog- simultaneously raising stakeholder confidence in modeled results.
enize the geochemistry of upper lake layers, transport dissolved Finally, modelers should continue to convey a high level of trans-
CO2 from depth to the lake surface where it may exsolve, renew parency in their work so that model reviewers and stakeholders
dissolved oxygen levels at the wall-rock-water interface, and are able to ascertain what is known with a high degree of confi-
influence the biological structure of the lake. Most importantly, dence from what is suspected with a lesser degree of confidence
the physical limnology dictates whether a lake will fully mix on (Drever, 2011). To this end, Part 2 of this study provides three case
an annual basis or whether meromictic conditions will develop. studies on the development of a physical limnology prediction, the
The evolution of a permanently isolated bottom layer can have a development of a geochemistry prediction, and the interpretation
significant impact on the closure plan and risk assessment for a of results from a long-term lake monitoring program (Castendyk,
given pit lake. In some cases, it may be possible to utilize the Balistrieri, Gammons, and Tucci, this volume).
monimolimnion to sequester trace metals from the water
column. In other cases, the accumulation of H2S and CO2 in References
monimolimnion water may present an additional risk for mine
planners to consider should the lake undergo complete turn- Allison, J.D., Brown, D.S., Novo-Gradac, K.J., 1991. MINTEQA2/PRODEFA2, a
over. Existing analytical models of physical limnology (i.e. inter- geochemical assessment model for environmental systems. Version 3.0 User’s
Manual: U.S. Environmental Protection Agency, Report 600/3-91/021, Athens,
pretations of relative depth) are inconsistent with the observed Georgia, 93 pp.
behavior of pit lakes. Numerical hydrodynamic models may Anderson, M.P., Woessner, W.W., 1992. Applied Groundwater Modeling; Simulation
provide more accurate predictions. of Flow and Advective Transport. Academic Press, San Diego, California, 381 p.
Atkin, S.A., Schrand, W.D., 2000. Limnology of the Sleeper Pit Lake, Humboldt
 Geochemical predictions integrate the results of these compo- County, Nevada. 5th International Conference on Acid Rock Drainage,
nents to generate a long-range forecast of pit lake water quality. proceedings. Society for Mineral Exploration, Littleton, Colorado, pp. 347–357.
Geochemical-speciation and forward-modeling software pack- Atkins, D., Kempton, J.H., Martin, T., Maley, P., 1997. Limnologic conditions in three
existing Nevada pit lakes: observations and modeling using CE-QUAL-W2. 4th
ages are well suited for modeling pit lake geochemistry because
International Conference on Acid Rock Drainage Proceedings, Mine
they allow the modeler to systematically explore multiple geo- Environment Neutral Drainage Program, Ottawa, Ontario, pp. 697–713.
chemical processes observed within existing pit lakes. However, Bair, E.S., Lahm, T.D., 2006. Practical Problems in Groundwater Hydrology. Pearson,
all predictions are limited to some extent. For example, most Upper Saddle River, New Jersey, Chapter 3.
Balistrieri, L.S., Tempel, R.N., Stillings, L.L., Shevenell, L.A., 2006. Modeling spatial
published geochemical predictions do not include the impacts and temporal variations in temperature and salinity during stratification and
of biological processes on lake water chemistry. Modelers can overturn in Dexter Pit Lake, Tuscarora, Nevada. Appl. Geochem. 21, 1184–1203.
improve the accuracy of predictions by comparing predicted Barker, S.L.S., Kim, J.P., Craw, D., Frew, R.D., Hunter, K.A., 2004. Processes affecting
the chemical composition of Blue Lake, an alluvial gold-mine pit lake in New
results to direct water chemistry observations for the pit lake Zealand. Mar. Freshw. Res. 55, 201–211.
being studied, observations from existing pit lakes in similar Bird, D.A., Lyons, W.B., Miller, G.C., 1994. An assessment of hydrogeochemical
ore bodies, lab experiments that attempt to synthesize lake computer codes applied to modeling post-mining pit water geochemistry. In:
Tailings and Mine Waste ‘94, Proceedings: Balkema. Rotterdam, Holland, pp.
water chemistry, and other methods. Using these approaches, 31–40.
the modeler may be able to quantify the uncertainty in predic- Bird, D.A., Mahoney, J.J., 1994. Estimating post-mining pit lake geochemistry
tions and to convey this information to mine planners and reg- utilizing geochemical and numerical modeling. Reprint of paper presented at
1994 Annual Meeting of Society for Mining, Metallurgy, and Exploration,
ulators. It is recommended that mine planners and regulators Albuquerque, New Mexico, 5 p.
use the results of pit lake geochemical predictions to indicate Boehrer, B., Schultze, M., Liefold, S., Behlau, G., Rahn, K., Frimel, S., Kiwel, U., Kuehn,
the likely range of future water chemistry values rather than B., Brookland, I., Büttner, O., 2003. Stratification of mining lake Goitsche during
flooding with river water. Tailings and Mine Waster ‘03, Proceedings, Balkema,
viewing predictions as a definitive statement of the exact water
Lisse, Holland, pp. 223-231.
chemistry after a given period of time. Böhrer, B., Heidenreich, H., Schimmele, M., Schultze, M., 1998. Numerical prognosis
for salinity profiles of future lakes in the opencast mine Merseburg-Ost. Int. J.
Geochemical predictions are most useful to stakeholders when Salt Lake Res. 7, 235–260.
Boland, K.T., Padovan, A.V., 2002. Seasonal stratification and mixing in a recently
they are generated during the planning phase of the mine life cycle flooded mining void in tropical Australia. Lakes Reservoirs: Res. Manage. 7,
(Fig. 9). The information received from predictions can be incorpo- 125–131.
rated into the development of the risk management plan and the Bowell, R.J., 2002. The hydrogeochemical dynamics of mine pit lakes. In: Younger,
P.L., Robins, N.S. (Eds.), Mine Water Hydrogeology and Geochemistry:
closure plan for the mine site, and the estimated cost of these plans Geological Society. Special Publications 198, London, pp. 159–185.
should be factored into the net earnings projected for a new mine. Bowell, R.J., Parshley, J.V., 2005. Control of pit-lake water chemistry by secondary
Updated models are also useful during the closure and post-closure minerals, Summer Camp pit, Getchell mine, Nevada. Chem. Geol. 215, 373–385.
Brunauer, S., Emmett, P.H., Teller, E., 1938. Adsorption of gases in multimolecular
monitoring of pit lake water chemistry by allowing mine managers layers. J. Am. Chem. Soc. 60, 309. http://dx.doi.org/10.1021/ja01269a023.
to consider a variety of mitigation, remediation, and lake manage- Castendyk, D.N., 2009a. Conceptual models of pit lakes. In: Castendyk, D.N., Eary,
ment options (Fig. 10). Like all predictive models, pit lake geo- L.E. (Eds.), Mine Pit Lakes: Characteristics, Predictive Modeling, and
Sustainability. Society for Mining, Metallurgy, and Exploration Inc, Littleton,
chemical predictions will never be 100% accurate. Through Colorado, pp. 61–76.
careful characterization, conceptualization, design, and validation, Castendyk, D.N., 2009b. Predictive modeling of the physical limnology of future pit
the difference between predicted and observed concentrations can lakes. In: Castendyk, D.N., Eary, L.E. (Eds.), Mine Pit Lakes: Characteristics,
Predictive Modeling, and Sustainability. Society for Mining, Metallurgy, and
be minimized, and a useful prediction can be produced.
Exploration Inc, Littleton, Colorado, pp. 101–114.
Looking forward, the overall accuracy of pit lake modeling can Castendyk, D.N., Eary, L.E., 2009. Mine Pit Lakes: Characteristics, Predictive
be improved by making more observations of physical data (see Modeling, and Sustainability. Society for Mining, Metallurgy, and Exploration
also Drever, 2011). Where possible, predictions should be scruti- Inc, Littleton, Colorado, 304 p.
Castendyk, D.N., Jewell, P., 2002. Turnover in pit lakes: I. Observations of three pit
nized and updated using direct comparisons to observed pit lake lakes in Utah, USA. In: Tailings and Mine Waste ‘02: Balkema, Lisse, Holland, pp.
chemistry. Where this is not possible, the model results could be 181–188.
286 D.N. Castendyk et al. / Applied Geochemistry 57 (2015) 267–288

Castendyk, D.N., Mauk, J.L., Webster, J.G., 2005. A mineral quantification method for Gammons, C.H., 2009. Sampling and monitoring of pit lakes. In: Castendyk, D.N.,
wall rocks at open pit mines, and application to the Martha Au-Ag mine, Waihi, Eary, L.E. (Eds.), Mine Pit Lakes: Characteristics, Predictive Modeling, and
New Zealand. Appl. Geochem. 20, 135–156. Sustainability. Society for Mining, Metallurgy, and Exploration Inc, Littleton,
Castendyk, D.N., Webster-Brown, J., 2007a. Sensitivity analysis in pit lake Colorado, pp. 77–90.
prediction, Martha Mine, New Zealand 1: Relationship between turnover and Gammons, C.H., Duaime, T.E., 2006. Long term changes in the limnology and
input water density. Chem. Geol. 244, 42–55. geochemistry of the Berkeley Pit Lake, Butte, Montana. Mine Water Environ. 25,
Castendyk, D.N., Webster-Brown, J.G., 2007b. Sensitivity analyses in pit lake 76–85.
prediction, Martha mine, New Zealand 2: Geochemistry, water-rock reactions, Gammons, C.H., Harris, L.N., Castro, J.M., Cott, P.A., Hanna, B.W., 2009. Creating lakes
and surface adsorption. Chem. Geol. 244, 56–73. from open pit mines: processes and considerations, with emphasis on northern
Castendyk, D.N., Webster-Brown, J.G., 2010. Effects of mine expansion on environments. Can. Tech. Rep. Fish. Aquat. Sci. 2826, 106 p.
geochemical predictions of pit lake water quality: an example from Martha Geller, W., Klapper, H., Solomons, W., 1998. Acidic Mining Lakes: Acid Mine
Mine, Waihi, New Zealand. New Zeal. J. Geol. Geophys. 53, 143–151. Drainage, Limnology, and Reclamation. Springer, Germany, 435 p.
CEMA, 2012. End Pit Lakes Guidance Document. Cumulative Environmental Geller, W., Schultze, M., Kleinmann, B., Wolkersdorfer, C., 2012. Acidic Pit Lakes –
Management Association, Fort McMurray, Alberta, Canada, 436 p. <http:// Legacies of Surface Mining of Coal and Metal Ores. Springer, Germany, 504 p.
cemaonline.ca/index.php/news-a-events/cema-press-releases/89-cema-news/ Grimaldi, R., 2009. Climatologic characteristics. In: Castendyk, D.N., Eary, L.E. (Eds.),
press-releases/press-release-articles/196-press-release-cema-delivers-oilsands- Mine Pit Lakes: Characteristics, Predictive Modeling, and Sustainability. Society
mine-end-pit-lake-guidance-document-october-4-2012> (accessed 11.05.14). for Mining, Metallurgy, and Exploration Inc, Littleton, Colorado, pp. 19–31.
Connors, K., Shevenell, L., Lyons, W.B., 2000. Pit water-pit wall interactions in Gustafsson, J.P., 2010. Visual MINTEQ ver. 3.0: KTH. Department of Land and Water
Nevada precious metal mines: experimental investigation of four Resources Engineering, Sweden. <www2.lwr.kth.se/English/OurSoftware/
representative geologic systems. In: Cluer, J.K., Price, J.G., Struhsacker, E.M., vminteq/index.html> (accessed 15.04.11).
Hardyman, R.F., Morris, C.L. (Eds.), Geology and Ore Deposits 2000: The Great Hagler Bailly Consulting Inc, 1996. Supplemental injury assessment report. Clark
Basin and Beyond, Proceedings. Geological Society of Nevada, Reno, Nevada, pp. Fork River NPL Sites NRDA; Lethal Injuries to Snow Geese, Berkeley pit, Butte,
825–848. MT. Montana Natural Resource Damage Assessment Litigation Program, Helena,
Castro, J.M., Moore, J.N., 2000. Pit lakes: their characteristics and the potential for 34 p. + appendices.
their remediation. Environ. Geol. 39, 1254–1260. Halbwachs, M., Sabroux, J.C., Grangeon, J., Kayser, G., Tochon-Danguy, J.C., Felix, A.,
Castro, J.M., Wielinga, B.W., Gannon, J.E., Moore, J.N., 1999. Stimulation of sulfate- Beard, J.C., Vilevielle, A., Vitter, C., Richon, P., Wüest, A., Hell, J., 2004. Degassing
reducing bacteria in lake water from a former open-pit mine through addition the ‘‘Killer Lakes’’ Nyos and Monoun, Cameroon. EOS 85, 281–284.
of organic waste. Water Environ. Res. 71, 218–223. Hamblin, P.F., Stevens, C.L., Larwrence, G.A., 1999. Simulation of vertical transport in
Davis, A., 2003. A screening-level laboratory method to estimate pit lake chemistry. a mining pit lake. J. Hydraul. Eng. 125, 1029–1038.
Mine Water Environ. 22, 194–205. Havis, R.N., Worthington, S.J., 1997. A simple model for the management of mine
Davis, A., Ashenberg, D., 1989. The aqueous geochemistry of the Berkeley Pit, Butte, pit-lake filling and dewatering. In: Tailings and Mine Waste ‘97, Proceedings.
Montana, U.S.A. Appl. Geochem. 4, 23–36. Balkema, Rotterdam, Holland, pp. 478–549.
Davis, A., Bellehumeur, T., Hunter, P., Hanna, B., Fennemore, G.G., Moomaw, C., Hipsey, M.R., Romero, J.R., Antenucci, J.P., Hamilton, D., 2006. Computational
Schoen, S., 2006. The nexus between groundwater modeling, pit lake Aquatic Ecosystem Dynamics Model: CAEDYM v2; v2.3 Science Manual: Centre
chemogenesis and ecological risk from arsenic in the Getchell Main Pit, for Water Research, University of Western Australia, Perth, 102 p.
Nevada, U.S.A. Chem. Geol. 228, 175–196. <www.cwr.uwa.edu.au> (accessed 14.04.11).
Davis, A., Eary, L.E., 1997. Pit Lake water quality in the western United States: an Hoerling, M.P., Eischeid, J., 2007. Past peak water in the Southwest. Southwest
analysis of chemogenetic trends. Min. Eng. 49, 98–102. Hydrol. 6, 18–19.
Davis, A., Fennemore, G.G., 1998. A method to predict evolving post-closure pit lake Hutchinson, G.E., 1957. A Treatise on Limnology: Part I – Geography, Physics, and
chemistry. 15th Annual Meeting of the American Society for Surface Mining and Chemistry. John Wiley & Sons, New York, 1015 pp.
Reclamation: American Society for Surface Mining and Reclamation, Lexington, Imberger, J., Patterson, J.C., 1990. Physical limnology. In: Hutchinson, J.W., Wu, T.Y.
Kentucky, pp. 315–321. (Eds.), Advances in Applied Mechanics, v. 27. Academic Press, New York, pp.
Davis, G.B., Ritchie, A.I.M., 1986a. A model of oxidation in pyritic mine wastes: part 303–475.
1: equations and approximate solution. Appl. Math. Model. 10, 314–322. Jerz, J.K., Rimstidt, J.D., 2004. Pyrite oxidation in moist air. Geochim. Cosmochim.
Davis, G.B., Doherty, G., Ritchie, A.I.M., 1986b. A model of oxidation in pyritic mine Acta 68, 701–714.
wastes: part 2: Comparison of numerical and approximate solutions. Appl. Jewell, P., Castendyk, D., 2002. Turnover in pit lakes: II. Water column stability and
Math. Model. 10, 323–329. anoxia. Tailings and Mine Waste ‘02, Proceedings, Balkema, Lisse, Holland, pp.
Dingman, S.L., 2002. Physical Hydrology, second ed. Waveland Press Inc., Long 189–194.
Grove, Illinois, 646 p. Joehnk, K., 2001. 1-D Hydrodynamische Modelle in der Limnophysik: Turbulenz,
Dowling, J., Atkin, S., Beale, G., Alexander, G., 2004. Development of the Sleeper pit Meroxixis, Sauerstoff: Limnophysics Reports, v. 1. Institute of Limnophysics,
lake. Mine Water Environ. 23, 2–11. Konstanz, Germany.
Doyle, G.A., Runnells, D.D., 1997. Physical limnology of existing mine pit lakes. Min. Johnson, E., Castendyk, D., 2012. The INAP Pit Lakes Database: a novel tool for the
Eng. 49, 76–80. evaluation of predicted pit lake water quality. In: Price, W.A., Hogan, C.,
Drever, J.I., 2011. Prediction is hard – particularly about the future. Elements 7, Tremblay, G. (Eds.), 9th International Conference on Acid Rock Drainage,
363. Proceedings, Ottawa, Canada, May 20–26, 2012. Technical Paper 0037, pp. 1–12.
Drever, J.I., 1997. The Geochemistry of Natural Waters: Surface and Groundwater Jonas, J., 2000. Current seasonal limnology of the Berkeley Pit Lake. 5th
Environments, third ed. Prentice Hall, Englewood Cliffs, New Jersey, 436 p. International Conference on Acid Rock Drainage, Proceedings, Society for
Dunne, T.D., Leopold, L.B., 1978. Water in Environmental Planning. Freeman and Mineral Exploration, Littleton, Colorado, pp. 359–366.
Company, New York, pp. 35–74 . Kalin, M., Wheeler, W.N., 2009. Biogeochemical remediation of pit lakes. In:
Eary, L.E., 1998. Predicting the effects of evapoconcentration on water quality in Castendyk, D.N., Eary, L.E. (Eds.), Mine Pit Lakes: Characteristics, Predictive
mine pit lakes. J. Geochem. Explor. 64, 223–236. Modeling, and Sustainability. Society for Mining, Metallurgy, and Exploration
Eary, L.E., 1999. Geochemical and equilibrium trends in mine pit lakes. Appl. Inc, Littleton, Colorado, pp. 203–214.
Geochem. 14, 963–987. Karakas, G., Brookland, I., Boehrer, B., 2003. Physical characteristics of acidic Mining
Eary, L.E., Castendyk, D.N., 2012. Hardrock metal mine pit lakes: occurrence and Lake 111. Aquat. Sci. 65, 297–307.
geochemical characteristics. In: Geller, W., Schultze, M., Kleinmann, B., Karamalidis, A.K., Dzombak, D.A., 2010. Surface Complexation Modeling: Gibbsite.
Wolkersdorfer, C. (Eds.), Acidic Pit Lakes – Legacies of Surface Mining of Coal Wiley, New York, 294 p.
and Metal Ores. Springer, Germany, pp. 75–106. Kempton, J.H., Locke, W., Atkins, D., Nicholson, A.D., Bennett, M., Bliss, L., 1997.
Eary, L.E., Castendyk, D.N., 2009. The state of the art of pit lake predictions. In: Probabilistic prediction of water quality in the Twin Creeks mine pit lake,
Castendyk, D.N., Eary, L.E. (Eds.), Mine Pit Lakes: Characteristics, Predictive Golconda, Nevada, USA. 4th International Conference on Acid Rock Drainage,
Modeling, and Sustainability. Society for Mining, Metallurgy, and Exploration Vancouver, B.C., Canada, May 31–June 6, 1997, Proceedings, Mine Environment
Inc, Littleton, Colorado, pp. 275–289. Neutral Drainage Program, Ottawa, Ontario, pp. 1729–1744.
Evans, L., Rola-Rubzen, F., Ashton, P., 2003. Beneficial end uses for open cut mine Kirk, L.B., Schafer, W., Volberding, J., Kranz, S., 1996. Mine lake geochemical
sites: Planning for optimal outcomes. 2003 Annual Meeting of the Minerals prediction for the SPJV McDonald project. Planning, Rehabilitation, and
Council of Australia, Proceedings, Kingston, Australia. Treatment of Disturbed Lands, Billings Symposium, Proceedings, pp. 393–403.
Fennemore, G.G., Neller, W.C., Davis, A., 1998. Modeling pyrite oxidation in arid Klapper, H., Schultze, M., 1997. Sulfur acidic mining lakes in Germany-Ways of
environments. Environ. Sci. Technol. 32, 2680–2687. controlling geogenic acidification. 4th International Conference on Acid Rock
Fetter, C.W., 2001. Applied Hydrogeology, fourth ed. Prentice Hall, Upper Saddle Drainage, proceedings, Mine Environment Neutral Drainage Program, Ottawa,
River, New Jersey, 598 p. Ontario, pp. 891–904.
Fisher, T.S.R., Lawrence, G.A., 2000. Observations at the upper halocline of the Island Kuma, J.S., Younger, P.L., Bowell, R.J., 2002. Hydrogeological framework for assessing
Copper pit lake. In: Lawrence, G.A., Pieters, R., Yonemistsu, N. (Eds.), 5th the possible environmental impacts of large-scale gold mines. In: Younger, P.L.,
International Symposium on Stratified Flows, Proceedings. Vancouver, British Robins, N.S. (Eds.), Mine Water Hydrogeology and Geochemistry: Geological
Columbia, Canada, pp. 413–418. Society, Special Publications 198, London, pp. 121–136.
Flite, O.P., Eidson, G.W., 2003, Impact of wind: Planning for sustainable pit lakes. Lee, M., 1999. Risk Assessment Framework for the Management of Sulfidic Mine
Abstracts with Programs, 2003 Geological Society of America Annual Meeting, Wastes. Australian Centre for Mining Environmental Research (ACMER),
v. 35, p. 268. Kenmore, Queensland, Australia, 16 p.
D.N. Castendyk et al. / Applied Geochemistry 57 (2015) 267–288 287

Levy, D.B., Custis, K.H., Casey, W.H., Rock, P.A., 1997. The aqueous geochemistry of Pelletier, C.A., Wen, M.E., Poling, G.W., 2009. Flooding pit lakes with surface water.
the abandoned Spenceville copper pit, Nevada County: California. J. Environ. In: Castendyk, D.N., Eary, L.E. (Eds.), Mine Pit Lakes: Characteristics, Predictive
Qual. 26, 233–243. Modeling, and Sustainability. Society for Mining, Metallurgy, and Exploration
Lewis, R.L., 1999. Predicting the steady-state water quality of pit lakes. Min. Eng. 51, Inc, Littleton, Colorado, pp. 187–202.
54–58. Pellicori, D.A., Gammons, C.H., Poulson, S.R., 2005. Geochemistry and stable isotope
Lyons, W.B., Doyle, G.A., Petersen, R.C., Swanson, E.E., 1994. The limnology of future composition of the Berkeley pit lake and surrounding mine waters, Butte,
pit lakes in Nevada: the importance of shape. In: Tailings and Mine Waste ‘94: Montana. Appl. Geochem. 20, 2116–2137.
Balkema. Rotterdam, Holland, pp. 245–248. Pillard, D.A., Runnells, D.D., Doyle, T.A., Young, J., 1996. Post-mining pit lakes:
Madison, J.P., Gammons, C.H., Poulson, S.R., Jonas, J.P., 2003. Oxidation of pyrite by predicting lake chemistry and assessing ecological risks. In: 3rd International
ferric iron in the acidic Berkeley pit lake, Montana, U.S.A. In: Farrell, T., Taylor, G. Conference on Tailings and Mine Waste ‘96, proceedings: Balkema, Rotterdam,
(Eds.), 6th International Conference on Acid Rock Drainage, Proceedings. pp. 469–478.
Australasian Institute of Mining and Metallurgy, Carlton, Australia, pp. 1073– Price, J.G., Shevenell, L., Henry, C.D., Rigby, J.G., Christensen, L.G., Lechler, P.J.,
1078. Desilets, M.O., Fields, R., Driesner, D., Durbin, B., Lombardo, W., 1995. Water
Marinelli, F., Niccoli, W.L., 2000. Simple analytical equations for estimating ground Quality at Inactive and Abandoned Mines in Nevada: Nevada Bureau of Mines
water inflow to a mine pit. Ground Water 38, 311–314. and Geology, Open-File Report 95-4, Reno, Nevada, 71 p.
Martin, A.J., Crusius, J., McNee, J.J., Whittle, P., Pieters, R., Pedersen, T.F., 2003. Field- Ramstedt, M., Carlsson, E., Lövgren, L., 2003. Aqueous geochemistry in the Udden pit
scale assessment of bioremediation strategies for two pit lakes using lake, northern Sweden. Appl. Geochem. 18, 97–108.
limnocorrals. In: Farrell, T., Taylor, G. (Eds.), 6th International Conference on Robins, R.G., Berg, R.B., Dysinger, D.K., Duaime, T.E., Metesh, J.J., Diebold, F.E.,
Acid Rock Drainage, Proceedings. Australasian Institute of Mining and Twidwell, L.G., Mitman, G.G., Chatham, W.H., Huang, H.H., Young, C.A., 1997.
Metallurgy, Carlton, Australia, pp. 529–539. Chemical, physical and biological interaction at the Berkeley Pit, Butte,
Mase, D.F., Greenemeier, R.J., Castendyk, D.N., 2008. Investigation of hydroxide Montana. In: Tailings and Mine Waste ‘97: Balkema. Rotterdam, Holland, pp.
mineral precipitation and the fate of trace metals in an acidic pit lake. 529–541.
Anthracite District, Pennsylvania. Abstracts with Programs, 2008 Northeast Sánchez España, J., López Pamo, E., Santofimia Pastor, E., Diez Ercilla, M., 2008. The
Section Meeting of the Geological Society of America, vol. 40, no. 2, p. 74. acidic mine pit lakes of the Iberian Pyrite Belt: an approach to their physical
McCready, J., 2001. Mining Impacted Pit Lakes 2000 Workshop, Proceedings. U.S. limnology and hydrogeochemistry. Appl. Geochem. 23, 1260–1287.
Environmental Protection Agency, Washington, DC, EPA/625/C-00/004. Savage, K.S., Ashley, R.A., Bird, D.K., 2009. Geochemical evolution of a high arsenic,
McCready, J., Ober, D., 2006. Interactive Pit Lakes 2004 Conference: U.S. alkaline pit-lake in the Mother Lode Gold District, California. Econ. Geol. 104,
Environmental Protection Agency, Washington, DC, EPA/625/C-06/002. 1171–1211.
McCullough, C.D. (Ed.), 2011. Mine Pit Lakes: Closure and Management. Australian Savage, K.S., Bird, D.K., Ashley, R.P., 2000. Legacy of the California Gold Rush:
Centre for Geomechanics, Nedlands, Western Australia, 182 p. environmental geochemistry of arsenic in the Southern Mother Lode Gold
McCullough, C., Hunt, D., Evans, L., 2009. Sustainable development of open pit District. Int. Geol. Rev. 42, 385–415.
mines: creating beneficial end uses for pit lakes. In: Castendyk, D.N., Eary, L.E. Schafer, W.M., Logsdon, M., Zhan, G., Espell, R., 2006. Post-Betze pit lake water
(Eds.), Mine Pit Lakes: Characteristics, Predictive Modeling, and Sustainability. quality prediction, Nevada. In: Barnhisel, R.I. (Ed.), 7th International Conference
Society for Mining, Metallurgy, and Exploration Inc, Littleton, Colorado, pp. on Acid Rock Drainage, Proceedings. American Society of Mining and
249–267. Reclamation, Lexington, Kentucky, pp. 1863–1886.
McLemore, V.T. (Ed.), 2008. Basics of Metal Mining Influenced Water. Society for Schultze, M., Boehrer, B., 2009. Induced meromixis. In: Castendyk, D.N., Eary, L.E.
Mining, Metallurgy, and Exploration Inc., Littleton, Colorado, 103 p. (Eds.), Mine Pit Lakes: Characteristics, Predictive Modeling, and Sustainability.
Meyer, W.A., Williamson, M.A., Warren, G.C., Brown, A., 1997. Suppression of sulfide Society for Mining, Metallurgy, and Exploration Inc, Littleton, Colorado, pp.
mineral oxidation in mine pit walls Part II. Geochemical modeling. Fourth 239–246.
International Conference on Tailings and Mine Waste ‘97, Fort Collins, Colorado, Schultze, M., Pokrandt, K., Hille, W., 2010. Pit lakes of the Central German lignite
Jan 13–17, 1997, Proceedings, Balkema, Rotterdam, pp. 433–441. mining district: creation, morphometry and water quality aspects. Limnologica
Miller, G.C., Lyons, W.B., Davis, A., 1996. Understanding the water quality of pit 40, 148–155.
lakes. Environ. Sci. Technol. 30, 118A–123A. Seal, R.R., Balistrieri, L.S., Piatak, N.M., Garrity, C.P., Hammarstrom, J.M., Hathaway,
MMSD, 2002. Breaking New Ground: Final Report of the Mining, Minerals and E.M., 2006. Processes controlling geochemical variations in the South Pit lake,
Sustainable Development Project. International Institute for Environment and Elizabeth mine Superfund site, Vermont, USA. In: Barnhisel, R.I. (Ed.), 7th
Development, and World Business Council for Sustainable Development, International Conference on Acid Rock Drainage, Proceedings. American Society
Earthscan Publications Ltd., London, 418 p. of Mining and Reclamation, Lexington, Kentucky.
Morin, K.A., Hutt, N.M., 2006. The MEND minewall technique: overview and details. Shevenell, L., 2000. Water quality in pit lakes in disseminated gold deposits
13th Annual BC MEND Workshop, Open Pits and Underground Mine Workings, compared to natural, terminal lakes in Nevada. Environ. Geol. 39,
Nov. 29-30, 2006, Presentation, Vancouver, BC. 807–815.
NDEP, 2010. Meteoric Water Mobility Procedure (MWMP) Standardized Column Shevenell, L., Connors, K.A., 2000. Observed and experimental pit lake water quality
Percolation Test Procedure. Nevada Division of Environmental Protection. in low sulfidation (quartz-adularia) gold deposits, Nevada. In: Cluer, J.K., Price,
<ndep.nv.gov/bmrr/mobilty1.pdf> (accessed 07.10.10). J.G., Struhsacker, E.M., Hardyman, R.F., Morris, C.L. (Eds.), Geology and Ore
Niccoli, W.L., 2009a. Hydrologic characteristics and classification of pit lakes. In: Deposits 2000: The Great Basin and Beyond: Geological Society of Nevada
Castendyk, D.N., Eary, L.E. (Eds.), Mine Pit Lakes: Characteristics, Predictive Symposium, Proceedings. Reno, Nevada, pp. 855–867.
Modeling, and Sustainability. Society for Mining, Metallurgy, and Exploration Shevenell, L.A., Connors, K.A., Henry, C.D., 1999. Controls on pit lake water quality at
Inc, Littleton, Colorado, pp. 33–43. sixteen open-pit mines in Nevada. Appl. Geochem. 14, 669–687.
Niccoli, W.L., 2009b. Predicting groundwater inputs to pit lakes. In: Castendyk, D.N., Shevenell, L.A., Pasternak, K.I., 2000. Modeled versus observed filling rates of the
Eary, L.E. (Eds.), Mine Pit Lakes: Characteristics, Predictive Modeling, and historical Getchell pit lakes, Nevada. In: Cluer, J.K., Price, J.G., Struhsacker, E.M.,
Sustainability. Society for Mining, Metallurgy, and Exploration Inc, Littleton, Hardyman, R.F., Morris, C.L. (Eds.), Geology and Ore Deposits 2000: The Great
Colorado, pp. 91–99. Basin and Beyond, Proceedings. Geological Society of Nevada, Reno, Nevada, pp.
Oldham, C.E., Salmon, S.U., Hipsey, M.R., Ivey, G.N., 2009. Modeling pit lake water 869–871.
quality: coupling of lake stratification dynamics, lake ecology, aqueous Siskind, D.E., Fumanti, R.R., 1974. Blast-produced Fractures in Lithonia Granite. U.S.
geochemistry, and sediment diagenesis. In: Castendyk, D.N., Eary, L.E. (Eds.), Bureau of Mines, Report of Investigations 7901, 38 p.
Mine Pit Lakes: Characteristics, Predictive Modeling, and Sustainability. Society Stevens, C.L., Castendyk, D., Fisher, T.S.R., 2002. The open-cast mine Pit Lake:
for Mining, Metallurgy, and Exploration Inc, Littleton, Colorado, pp. 127–136. environmental fluid mechanics and long-term prediction. New Zeal. Min. 31,
OSMRE, 2007. Annual evaluation summary report for the regulatory and abandoned 25–29.
mine land programs administered by the Commonwealth of Virginia for Stevens, C.L., Lawrence, G.A., 1998. Stability and meromixis in a water-filled mine
evaluation year 2007. United States Department of Interior, Office of Surface pit. Limnol. Oceanogr. 43, 946–954.
Mining Reclamation and Enforcement (OSMRE), Washington, D.C., 26 p. Stevens, C.L., Fisher, T.S.R., 2005. Turbulent layering beneath the pycnocline in a
Palandri, J.L., Kharaka, Y.K., 2004. A Compilation of Rate Parameters of Water- strongly stratified pit lake. Limnol. Oceanogr. 50, 197–206.
Mineral Interaction Kinetics for Application to Geochemical Modeling: Open Stottmeister, U., Glässer, W., Klapper, H., Weißbrodt, E., Eccarius, B., Kennedy, C.,
File Report 2004–1068. U.S. Geological Survey, Menlo Park, California. Schultze, M., Wendt-Potthoff, K., Frömmichen, R., Schreck, P., Strauch, G., 1999.
Park, B.T., Wangerud, K.W., Fundingsland, S.D., Adzic, M.E., Lewis, N.M., 2006. In situ Strategies for remediation of former opencast mining areas in eastern Germany.
chemical and biological treatment leading to successful water discharge from In: Azcue, J.M. (Ed.), Environmental Impacts of Mining Activities. Springer,
Anchor Hill pit lake, Gilt Edge Mine Superfund Site, South Dakota, USA. In: Berlin, pp. 63–296.
Barnhisel, R.I. (Ed.), 7th International Conference on Acid Rock Drainage: Tassé, N., 2003. Limnological processes in the development of acid pit lakes. In:
American Society of Mining and Reclamation, Lexington, Kentucky, pp. 1065– Farrell, T., Taylor, G. (Eds.), 6th International Conference on Acid Rock Drainage,
1069. Proceedings. Australasian Institute of Mining and Metallurgy, Carlton, Australia,
Parkhurst, D.L., Appelo, C.A.J., 1999. User’s Guide to PHREEQC (Version 2)-A pp. 559–570.
Computer Program for Speciation, Batch-reaction, One-dimensional Transport, Tempel, R.N., Shevenell, L.A., Lechler, P., Price, J., 2000. Geochemical modeling
and Inverse Geochemical Calculations: Water-Resources Investigations Report approach to predicting arsenic concentrations in a mine pit lake. Appl.
99–4259. U.S. Geological Survey, Denver, Colorado. Geochem. 15, 475–492.
Parshley, J.V., Bowell, R.J., 2003. The limnology of Summer Camp pit lake: a case Triantafyllidis, S., Skarpelis, N., 2006. Mineral formation in an acid pit lake from a
study. Mine Water Environ. 22, 170–186. high-sulfidation ore deposit: Kirki, NE Greece. J. Geochem. Explor. 88, 68–71.
288 D.N. Castendyk et al. / Applied Geochemistry 57 (2015) 267–288

Tones, P.I., 1982. Limnological and Fisheries Investigation of the Flooded Open Pit at Werner, F., 2009. Integration of prediction models. In: Castendyk, D.N., Eary, L.E.
the Gunnar Uranium Mine. Saskatchewan Research Council Publication No. C- (Eds.), Mine Pit Lakes: Characteristics, Predictive Modeling, and Sustainability.
805-10-E-82, Saskatoon, Canada. Society for Mining, Metallurgy, and Exploration IncLittleton, Littleton, Colorado,
Warren, G.C., Brown, A., Meyer, W.A., Williamson, M.A., 1997. Suppression of sulfide pp. 159–165.
mineral oxidation in mine pit walls Part I: Hydrologic modeling. 4th Werner, F., Bilek, F., Luckner, L., 2001. Impact of regional groundwater flow on the
International Conference on Tailings and Mine Waste ‘97, Fort Collins, water quality of an old post-mining lake. Ecol. Eng. 17, 133–142.
Colorado, Jan 13–17, 1997, Proceedings, Balkema, Rotterdam, pp. 425–432. Wetzel, R.G., 2001. Limnology: Lake and River Ecosystems, third ed. Academic
Weilinga, B., 2009. In situ bioremediation of pit lakes. In: Castendyk, D.N., Eary, L.E. Press, San Diego, California, 1006 pp.
(Eds.), Mine Pit Lakes: Characteristics, Predictive Modeling, and Sustainability. White, A.F., 1995. Chemical weathering rates of silicate minerals in soils. In: White,
Society for Mining, Metallurgy, and Exploration Inc, Littleton, Colorado, pp. A.F., Brantley, S.L. (Eds.), Chemical Weathering Rates of Silicate Minerals.
215–223. Mineralogical Society of America, Reviews in Mineralogy, vol. 31, pp. 407–461.
Wells, S., Berger, C., 2010. CE-QUAL-W2, Version 3.6. Department of Civil and Woodhouse, B., 2002. Hydrology of pit lakes. Southwest Hydrol. 1, 40 p.
Environmental Engineering, Portland State University. <www.cee.pdx.edu/w2/> Wyatt, G., Miller, F., Chermak, J., 2006. Innovative water treatment plant utilizing
(accessed 10.02.11). the South mine pit at the Copper Basin mining site in Tennessee, USA. In:
Wendt-Potthoff, K., 2006. In situ remediation of an acidic pit lake: stimulation of Barnhisel, R.I. (Ed.), 7th International Conference on Acid Rock Drainage,
microbial sulphate reduction in enclosure experiments. In: McCready, J., Ober, Proceedings. American Society of Mining and Reclamation, Lexington,
D. (Eds.), Interactive Pit Lakes 2004 Conference: U.S. Environmental Protection Kentucky, pp. 2529–2539.
Agency, Washington, DC, EPA/625/C-06/002.

You might also like