ENVIRONMENTAL RESEARCH REPORT
HYDRAULIC AND WATER QUALITY MODELING OF SILVER BOW CREEK-UPPER
CLARK RIVER. A SIIPERFUND SITE IN MONTANA
Kendall P. Brown and Zia Hosseinipour
ABSTRACT
Water quality modeling and
related exposure assessments at
a Superfund site, Silver Bow
Creek-Clark Fork River in
Montana, demonstrate the CEAM
modeling capability to predict
the fate of mining waste
pollutants in the environment.
A linked assessment system—
consisting of hydrology and
erosion, river hydraulics,
surface water quality, metal
speciation, non-point source
and groundwater mixing and
transport models—has been
applied to the site to show the
applicability of such modeling
schemes and the complexities
involved in this application.
Some of the models had to be
modified to match the physics
of! this project. Graphs of the
water quality parameters show
good fit between the measured
and predicted concentrations at
some stations while substantial
deviations are observed at other
stations along the course of the
stream.
EPA's Center for Exposure
Assessment Modeling (CEAM) was
established in July 1987 to meet the
scientific and technical exposure
assessment needs of EPA's Program and
Regional Offices and Superfund
Technology Support Center for Exposure
and Ecorisk Assessment. The Center
is also the focal point for a variety
of general Agency support activities
related to the scientifically
defensible application of state-of-
the-art exposure assessment technology
for environmental 'risk-baaed
decisions. CEAM provides analysts and
decision-makers operating under
various legislative mandates with
relevant exposure assessment
technology, training and consultation,
technical assistance, and
demonstration of new or innovative
applications. This research brief
describes one such demonstration
project - analysis of metals
contamination of the upper Clark Fork
River, Montana.
INTRODUCTION
This project was initiated
to analyze the hydrology and
water quality of the Silver Bow
Creek-Clark Fork River in
response to years of mining
activity in the surrounding
areas. Heavy metal
concentrations in the surface
and subsurface waters of the
Clark Fork River basin have
diminished aquatic life in many
of the region's streams. The
principal sources of these
metals are the waste byproducts
of copper mining in the Silver
Bow Creek watershed above and
around the town of Butte,
Montana. The Superfund site
includes Silver Bow Creek with
the Warm Springs Ponds, and the
old mine and dump sites at
Butte. Remedial Investigation
-------
has been conducted in the Silver
Bow Creek region and a set of
.rojmrts that comprise the Silver
Bow Creek Remedial Investigation
Final Report (RIFR) have been
prepared (Tuesday et. al. ,
1987) .
Description of the Study Area
Silver Bow Creek flows as
a small stream from Butte at
the Metro Storm Drain, to the
Warm Springs Ponds 27 miles
downstream. Both acid mine
drainage and contaminated
groundwater seepages enter
Silver Bow Creek within the
Butte town limits before
reaching the Colorado Tailings .
Below the town boundary, „ the
creek continues to pass large
streamside tailings deposits
and flood-deposited tailings
banks. Some small tributaries,
including Missoula Gulch,
Brown's Gulch, and German Gulch,
flow into Silver Bow Creek above
the Warm Springs Ponds. The
stream flows through Pond 3
(100 hectares of open water)
into Pond 2 (32 hectares) and
into Pond 1 (8 hectares).
Sediment deposits rise above the
surface of these shallow ponds.
Below Pond 2 the creek reaches
the confluence with the Mill-
Willow Bypass, which drains the
waste sites surrounding the
Anaconda smelter, the flows
from Mill and Willow Creek, and
seepage from the Warm Spring
Ponds. Below this confluence,
the creek combines with the flow
from Warm Springs Creek and
becomes the Upper Clark Fork
River. The creeks that enter the
river below the Warm Springs
Ponds include Modesty Creek,
Lost Creek, Dempsey Creek and
Racetrack Creek, and a
substantial increase in flow
occurs. (Figure 1)
More than 100 years of
continuous mining operations
and related activities have
changed the area's natural
environment greatly. Waste
rocks, ore process tailings,
acid mine drainage and smelting
wastes are the primary sources
of heavy metal loadings to the
Silver Bow Creek and Clark Fork
River via surface runoff and
ground water flow. Settling
ponds have limited capacity and
during high flows (discharges
of greater than 700 cfs) the
flow is by-passed without any
treatment. Further, geotechnical
studies have revealed that a
flow of 4000 cfs can result in
the failure of diversion and
control structures such that a
large amount of contaminated
sediments are released into the
Clark Fork River. Hydrological
investigations estimate the 100-
year flood to be 3600 cfs (Ch2M-
Hill 1988). The Remedial
Investigation Final Report
indicates that the seepage from
under the Warm Springs Ponds
can act as a major source of
ground water pollution. Toxic
elements from tailing deposits
are arsenic, cadmium, copper,
lead, iron, and zinc. The
severity of the contamination
problem was such that, in 1983,
EPA declared the area along the
course of the Silver Bow Creek
and Clark Fork River from Butte
to the Milltown Dam as a high
priority Superfund site. As this
brief introduction indicates,
the site hydrology, hydrogeology
and geochemistry are very
complex due to the variety and
magnitude of contaminant sources
and the multitude of pathways
to the surface water and ground
water resources. Therefore, a
detailed modeling scheme was
-------
-
Upper Clark Fork Basin of
Silver Bow Creek Superfund Site
MINING SITES
BLACKTAIL CREEK
YANKEE DOODLE POND^BUTTE
•Q,
BERKELEY PIT Y (7
CLARK TAILINGS
COLORADO^ TAILINGS
CLARK FORK
WARM SPRINGS PONDS
SILVER BOW CREEK
WARM SPRINGS CR EK
BROWNS GULCH
GERMAN GULCH
RACETRACK CREEK
WILLOW CREEK
ANACONDA
SMELTING SITES
-------
employed to delineate the
pathways and the fate of
pollutants.
History of Metals in the River
Since 1866, when large
scale mining and smelting of
copper began, the valley and
the stream have been used as
dumping areas for wastes. Wastes
in Butte include tailings from
the flotation process that
separates copper from the ore
and rock that was removed as
backfill and overburden from
either pit mines or underground
mines or was discarded as being
too low-grade to be put through
a separator.
Pollution problems began
early along the Silver Bow
Creek. The first industrial
operations sluiced the wastes
directly into the stream. Later,
the mine operators constructed
settling ponds and streamside
tailings piles as part of an
attempt to preserve water
quality. The wastes in the
stream moved down the river,
especially during floods which
caused erosion and transport of
sediment, and were widely
distributed over the flood plain
and the stream bed.
Between 1918 and 1920, Warm
Springs Ponds 1 and 2 were
constructed on Silver Bow Creek
above the confluence of Silver
Bow Creek with its two principal
tributaries, Warm Springs Creek
and Mill-Willow Creek. These
ponds were originally designed
to settle the metals carried by
Silver Bow Creek and prevent
contamination further
downstream.
As the two ponds lost
capacity due to sediment
accumulation, treatment
efficiency declined and
particles remained suspended in
the effluent. To remedy the
problem a larger pond was
constructed above the first two
ponds between 1953 and 1959.
Pond 3 was improved between 1959
and 1969 to increase the
capacity for metal removal.
Lime was added to the ponds
to precipitate and flocculate
the metals. This method of
settling the metal colloids and
particles from Silver Bow Creek
is successful in reducing the
metal content of the Clark Fork
River during periods of normal
flow. During high flows,
however, the ponds are bypassed
and the flow is directed to the
upper Clark Fork River without
treatment.
Between 1933 and 1937 the
stream itself was channelized
to prevent further erosion of
the tailings from the banks.
The first alteration to the
stream course of the Silver Bow
Creek was a channelization of
the flow between smelting slag
blocks placed along the stream
channel. This was done to
prevent the downstream transport
of newly deposited mine and slag
tailings on the old banks during
periods of bank overflow due to
flooding.
Biological Impairment
Adjacent to Silver Bow
Creek and the Clark Fork River
are flood plains and low banks
that have been covered with
waste sediments. In some areas,
the sediments have released
sufficient metal to either limit
plant growth to metal-tolerant
species or eliminate plant
-------
coverage entirely. These areas
are called "slickens," a term
that applies to all of the areas
that are either dead or have
visible biological impairment.
The past use of water from the
Clark Fork River for irrigation
has led to the contamination of
grazing lands.
The biological habitat of
the river and creek has been
damaged as well. Silver Bow
Creek does not support trout,
whereas the Upper Clark Fork
River supports a brown trout
population that suffers from
both chronic and acute
toxicity. The river always
contains metal contaminants/
such as iron, aluminum, zinc,
copper, arsenic, lead, and
cadmium.
Mass fish kills occur
during floods, and the acute
toxic ity of the stream due to
elevated metals concentrations
is thought to be the cause.
During flood flows (which can
occur during late winter snow
melts and during early and late
summer heavy rain storms), flow
in Silver Bow Creek can bypass
the settling ponds and enter
untreated into the Upper Clark
Fork River. Such events are
known to have caused mass fish
kills in the Clark Fork River.
Objectives of the Study
The modeling effort at the
Center for Exposure Assessment
Modeling (CEAM) will focus on
the prediction of the
frequencies of exposure of fish
to toxic metals at different
concentration levels in the
stream by using the metal
spociation and water quality
models combined with historical
data on the site. Investigations
to date have focused on the
typical flood events. The main
objective is to complete a
description of metals exposures
and anticipated effects on the
entire river during historical
periods of flooding. The
mechanisms affecting the native
fish, including the exposure
time(s) and concentrations that
produce mortality, will be
modelled by US EPA, ERL at
Duluth. '
Models and Methods
The general conceptual
model describes sources of
various metals (species) in
waste dumps and on the river
banks.
Chemical partitioning
between water and soil during
transport and transport of heavy
metals into the stream were
analyzed in the following
manner. The chemistry of the
tailings deposits was used to
determine the form of the
metals; the flow behavior of
rain on the banks as well as
overland flow determined the
principal transport mechanisms.
The rates of metal transport
depend on the rate of advection
predicted by PRZM (Carsel et
al. 1984), GCTRAN , and NPSOUT
(Brown 1989) and the solubility
of the oxidized metal at sulf ide
particle surfaces predicted by
MINTEQA2 (Brown et al. 1987).
Surface water transport is
simulated by WASP4. The
hydraulic parameters for WASP4
(Ambrose et al. 1987) are
provided by a river hydrodynamic
and sediment transport model
RIVERMOD (Hosseinipour, 1988).
The toxicity evaluations will
be performed by a Fish Acute
-------
Toxicity model developed at the
ERL-Duluth (Figure 2).
Geology and Geochemistry
The Super fund site has been
divided into geographic subsites
for the purpose of exposure
assessment modeling, with each
subsite having its own geology,
hydrology, and geochemistry. A
substantial body of scientific
literature already exists on the
chemical processes that create
acid mine drainage from mine
wastes. What is known about the
mine and process wastes is
summarized below.
The sidestream tailings
deposits are sandlike fine
particles of metal sulfides.
The primary sulf ide minerals in
the waste rock, overburden, and
processed ore include iron
pyrite (FeS2, iron sulfide),
chalcopyrite (FeCuS2, iron-copper
sulfide), realgar (AsS, arsenic
sulfide), chalcocite (Cu2S,
copper sulfide), and galena
(PbS, lead sulfide). These
minerals are geologically
embedded in monzonite porphyry
(primarily feldspar).
The ores are blasted, then
mined, crushed, and leached
during the initial processing
steps. Smelting is an oxidative
roasting process that leaves
slag as process waste.
Mechanical reduction in size of
the ore particles increases the
availability of surface area
for oxidation of the sulfide
minerals.
The source of the copper
in assumed to be a sulfide-
bearing waste tailings particle
that is corroded by exposure to
water and oxygen. The limits on
metal transport and solubility
will determine the magnitude of
the tailing's role as sources
of contaminants to the river,
and under certain circumstances
MINTEQA2 can be used to indicate
the partitioning of metal
between soil and water. MINTEQA2
is applied with the assumption
that oxidized heavy metal is
always present (metal
availability is not rate-
limited). Eh is set by the
balance between oxygen'diffusion
and consumption, and decreases
with depth. pH is set by
advection and diffusion of H*
away from the particle. pH is
increased by limitations on
diffusive and advective
transport of hydrogen ions and
this increases the diffusive
driving forces for oxidized
metals. pH increases with
depth: with less oxygen, less
H+ is created because of the
lower oxidation rate.
Oxygen Transport
Eh is controlled by the
availability of oxygen. The
sulfides react with oxygen and
water to form metal ions and
sulfuric acid. The process of
oxidation is dependent on a
supply of reactants (water,
sulfides, oxygen) and on water
as a transport medium for
reactants and products. The
process of oxidation must be
limited by transport of product
or reactants; otherwise rock
sulfides would not exist in
native form.
Oxygen is transported from
the surface of the tailings
where oxygen in the water phase
is in equilibrium with the
atmosphere. Oxygen diffuses into
the tailing deposit on the
-------
FIGURF. 2
DIAGRAM OF THE MODELING SYSTEM USED
IN THE CLARK FORK RIVER EXPOSURE ASSESSMENT
MINTEQA2
(MINTEQA2: CHEMICAL EQUILIBRIA FOR AQUEOUS SOLUTIONS)
I
I
v
PRZM
(PRZM: UNSATURATED ZONE, EROSION, AND RUNOFF TRANSPORT)
I
I
t
V
GCTRAN >NPSOUT
(NPSOUT: SATURATED ZONE MIXING AND EFFLUENT MODEL)
(GCTRAN: LARGE COLLOID TRANSPORT MODEL)
v
RIVERMOD >WASP4/TOXI4
(RIVERMOD: RIVER HYDRAULICS ROUTING)
(TOXI4: SURFACE WATER CONTAMINATION)
I
v
FISH ACUTE MORTALITY MODEL
(CURRENTLY UNDER DEVELOPMENT BY DULUTH ERL, US-EPA)
-------
stream bank and is transported
by diffusion and advection into
each unsaturated soil stratum
where sulfides are present.
Transport of oxygen to the metal
sulfide particle core depends
on the rate of diffusion through
metal oxide and metal sulfate
layers on a particle surface.
The rate of metal oxidation and
sulfate formation is assumed to
be equivalent to the oxygen
transport rate through the soil
pores (that is, the oxygen
transport rate is assumed to be
the rate-limiting step).
The rate of oxygen
transport into the oxidized
layer and through the sulfide
particle fissures to unreacted
metal sulfides is greater than
or equal to the transport away
of some oxidation products (FeO
or CuO). This was concluded from
the observed buildup of metal
oxides on particles in the
unsaturated zone oxidized soil
stratum.
Oxidation Product Transport
As the sulfides are
oxidized and byproducts (such
an hho sulfates, the oxides,
H') build up, they must be
carried away in order to prevent
a halt to oxidation. As
corrosion of the sulfide
proceeds, the pH of the particle
surface drops and the copper at
the particle surface and within
particle fractures and crevices
becomes more soluble. Oxidation
products are transported away
by diffusion out of the particle
fissures into the soil pore
water and subsequent advection.
If this did not occur, oxidation
products (H*, S042" ) would build
up and kinetically hinder
oxidation. Some oxidation
products (H+, S042" ) will be
removed faster than others
(CuO,FeO), and so a buildup of
some products (metal sulfates,
oxides, and carbonates) occurs
at the oxidation front between
solid oxidized metal and metal
sulfides. The buildup rate of
this process is determined by
the kinetics of oxidation and
by the reaction-limiting
mechanisms for transport of
products and react ants'. (Figure
3)
Copper diffuses into the
groundwater and enters other
transport pathways. The
oxidizing particles are the
major source for copper entering
subsequent transport pathways
such as runoff, erosion, and
leaching. Loss advection occurs
during dry periods, so that the
H* ion is trapped and increases
the solubility of metal.
Consequently, during dry
periods, there is an increase
in the source term for mobile
and soluble metal outside the
particles. During wet periods,
pH at the particle surface is
lowered sufficiently to reduce
transport outside the particle
and reduce the source term.
This can serve as an explanation
why the magnitude of the source
term varies seasonally.
In the groundwater away
from the sulfide particle, pH
drops, oxygen concentration
increases, and the copper
precipitates to form a metal
oxide or carbonate colloid in
suspension in the unsaturated
zone. The groundwater around
the sulfide particles has a
lower hydrogen ion concentration
than the water film around the
sulfide particles, and the
-------
Oxidation of a Sulfide
Metal Oxide
Precipitate
VD
lowpH
H2O
-------
higher pH reduces copper
solubility relative to the water
film. The higher pH will result
in precipitates being formed in
the groundwater, and these
precipitates represent the bulk
of the copper transported by
leaching and runoff. An example
of this will be shown later: pH
is 4.5 in the unsaturated zone
of the oxidized soil layer, but
may be as low as 2.5 in the
water film around the oxidizing
sulfide particle. Copper will
be carried in colloidal form
into the rain runoff and into
saturated groundwater.
Dependence of Geochemistry on
Soil
In the absence of a model
that will define the geochemical
conditions for a vertical soil
core (1-D), the first step in
the analysis of the waste site
is to develop a general modeling
framework. The soil core can be
described as a mixture of
tailings and alluvium 250 cm
deep, with a water table that
can vary from a depth of 0 cm
adjacent to the stream bank to
a depth of 250 cm at the edge
of the tailings furthest away
from the surface water (Figure
Ench soil stratum below
tli-"? surface of the stream-side
tailings deposit is
geochemically defined by the
oxygen consumption and oxygen
concentration in that stratum.
The soil core model corresponds
to both measurements of depth
vs . geochemistry in the Remedial
Investigation Final Report
(RIFR) and othor literature on
mine tailings geochemistry
(Tetra Tech, 1986). Oxygen
diffusion to lower strata is
limited by consumption in upper
soil strata, particularly for
tailings that are compacted and
fine-grained, and have a small
void fraction. In such
situations, the limited amount
of diffusing oxygen can be
totally consumed in the upper
soil strata. (Figure 5)
The conditions in the soil
and tailings mixture are
represented by two geochemical
parameters: Eh (electrochemical
potential) and pH (hydrogen ion
abundance). In some cases the
tailings have been mixed by
alluvial or mechanical processes
with the common associated
bedrock of the Butte valley.
This native bedrock has large
portions of calcium carbonate
that can neutralize acid
produced by oxidation and
corrosion of the sulfide
particles.
The oxidized stratum in
the unsaturated zone is the top
layer, below which are, in
sequence, the intermediate
mixing layer, the mixing layer,
and the' reduced stratum. The
top 100 cm of soil-tailings
layer has excess oxygen
available, and the Eh
consequently is at an oxidizing
potential of approximately
0.45V. Because of the rapid rate
of hydrogen ion production, the
pH in this layer is about 4.5.
The mixing layers have an
electrochemical potential of
0.45V, and pHs of 5 and 7. The
intermediate mixing layer is
between 100 and 150 cm deep,
with Eh=0.45V, and pH=5. The
mixing layer is between 150 and
200 cm and has an Eh=0.45V and
pH=7. In these layers, the
10
-------
FIGURE 4
Geochemical Profile
of the Unsaturated
Tailings and Alluvium
Variable
Water Table Depth
Units of Kd:
Moles/kilogram solid
Moles/liter water.
A
A
Soil Surface
100 cm
100 cm
Oxidizing zone pH = 3 Eh = .45V Kd=785
Cuprous ferrlte
A
50 cm
150 cm
Y
Intermediate pH a 5 Eh = .45V Kd = 7.86E4
Cuprous ferrlte
A
50cm
200 cm
Mixing zone pH = 7 Eh = .45V Kd • 7.64E6
Cuprous ferrite
A
50cm Reducing zone pH»7 Eh = .2V Kd = 4.3E15
Chalcopyrlte
250 cm _ _
Bedrock and/or Original Alluvium
Above diagram shows the geochemical
properties of the mixed tailings and alluvium
when they are above the saturated water level.
-------
Profile of the River, Water Table, and Tailing Deposit
Reducing Zone
Saturated Zone
-------
hydrogen ions mix with the
reducing compounds in the soil
core, but because some oxidation
is still occurring, Eh remains
at 0.45V. Oxidation can also
occur by copper sulfate
oxidizing iron sulfide to form
copper sulfide and iron sulfate.
Even in the presence of hydrogen
ions from this layer and the
oxidation layer above, the pH
is higher, because of lower
reaction rates and the
neutralizing effect of the
carbonates and the unoxidized
sulfides.
In this idealized
description of native soil and
tailings, the bottom stratum is
assumed to be a reduced layer,
where there is a negligible
concentration of oxygen. This
layer is represented as having
a reducing Eh of -0.2V.
The reduced soil layer has
the least oxygen available and
considerable reserves of
reducing sulfide, so that both
Eh and pH are in reducing ranges
(Eh=-.2V, pH=7). Metal sulfates
find ^ulfuric acid reaching this
layer are precipitated and
neutralized.
The saturated zone has
little oxygen because of oxygen
consumption by reaction in the
upper layers and because
diffusive transport through
water is slow. Oxidative
conversion of sulfides in the
saturation zone, therefore, is
assumed to be small relative to
the source term for copper in
the oxidized stratum. The
saturated zone is rauch better
mixed than the unsaturated zone
because of the continuous water
phase.
Geochemistry Controls Leachate
Metal is mobilized by the
hydrogen ions for transport into
the unsaturated zone away from
the sulfide particle surface to
the groundwater and can then be
carried to the surface water by
runoff, erosion, or leaching.
Hydrogen ions are assumed
to be neutralized in the reduced
layer by the metal sulfides and
calcium carbonate present in the
alluvium, and the pH is assumed
to remain at about 7. If the
saturated layer surface is
adjacent to the reduced layer
of soil and tailings, the model
we have proposed will predict
less copper in the leachate than
if the saturated layer was
adjacent to the oxidation layer
(the top 100 cm of the soil
core) or the mixing layer (the
middle 100-200 cm of the soil
core).
A lower oxygen
concentration causes a lower
oxidation potential in the lower
soil and tailings strata that
receive leachate from the upper
layers of the deposit, including
sulfate ions, hydrogen ions,
calcium, iron, and copper. The
sulfuric acid may either be
neutralized by calcium carbonate
(calcareous bedrock and
alluvium) or reduced by iron
pyrite and other unoxidized
sulfides.
To characterize the
leachate that enters the water
table, it is necessary to
determine solubilities in the
bottom unsaturated zone soil
and tailings layer and the metal
concentrations in the leachate
in this layer. The geochemical
13
-------
conditions in the bottom
UMH a I, u rated zone layer will
(l°i °i mine solubility, and the
depth of the soil and tailings
layer will dictate the
approximate values of the
principal parameters (Eh, pH) .
Depending on the depth of
the water teble and the width
of the tailings deposits,
leachate will either pass
primarily through the shallow
oxidized surface layers, or will
penetrate more deeply into the
mixing and reducing zones. The
water table level affects the
quality and properties of the
leachate entering the water
table and helps determine, the
form and the concentrations of
metals that pass into the
aquifer and into the stream.
The groundwater table can
intercept any of the geochemical
strata in the unsaturated zone,
depending on the distance from
the stream and the slope of the
stream bank. At the edge of the
stream, the water table and the
soil/tailing deposit surface
are the same. Away from the
stream edge, the water table
has a shallower gradient than
the deposit surface slope. The
wator table gets deeper and the
water table intercepts lower
geochemical strata as the
distance from the stream
increases. A lower slope for a
stream bank has a more shallow
water table. (See Figure 4)
The framework of a water
table that intercepts the
leachate from deeper unsaturated
zone layers as the distance from
the stream increases provides
a means of evaluating the
leachate loading from each of
the layers. To do this, the
intersecting area of each layer
that is bounded by the water
table needs to be evaluated.
Each intersecting area
represents an interface between
the saturated and unsaturated
zone that carries a leachate
across the boundary. In order
to characterize the leachate at
this boundary, it is necessary
to determine solubilities in the
unsaturated layer and the metal
concentrations in the leachate.
(See Figure 5)
Principal Models
The needs of this project
dictate that a metal speciation
model be employed to predict
metal solubility and that
surface and subsurface flow and
transport models be used to
predict the rate of contaminant
transport from the tailings
deposit. The metal speciation
model used is MINTEQA2.
Metal Speciation Model—MINTEOA2
With MINTEQA2 the
thermodynamic activities of all
possible compounds in aqueous
solution are solved for by the
iterative reduction of the
combined mass balance and mass
action equations. The model uses
a thermodynamic database of
formation constants and reaction
stoichiometries. The mass
balance equations are
established by the initial
amount of each component in the
initial mixture. Mass action
equations are established by
the formation constants for each
complex. The simultaneous
equations are determined by the
14
-------
complexes for which data exist
in the thermodynamic database
for a set of specified chemical
components. The combined set of
equations is solved using the
Newton-Raphson iteration method
(Brown et al. 1987)
In addition to solution
chemistry, MINTEQA2 predicts
the formation of precipitates;
i t. aJ.no predicts adsorption of
metals and the formation of
metal-organic complexes when
the sorption and formation
constants are included in the
database.
The Pesticiaa Root Zone Model-
-PRZM
The water balance on the
surface and through the
unsaturated zone includes
precipitation, infiltration,
runoff, and evapotranspiration.
The origin of flow in the
unsaturated zone and on the
tailing deposit surfaces may be
from one or more of the
following sources: overland
flooding, infiltration,
underflow, and runoff from the
slopes above the stream bank.
Rainfall is the major source of
contribution to all of the above
four flows. Rainfall can either
flow over the surface as runoff
and cause erosion, or infiltrate
and flow in the subsurface
through the open pores and reach
surface water flows as interflow
before reaching groundwater
table. It can cause overbank
flooding during a flash flood
or sudden snow melt, and
ultimately it may percolate
through the entire unsaturated
zone and reach the aquifer to
become deep groundwater storage
and/or the underflow (baseflow)
component of streams.
PRZM was developed to model
the transport of pesticides in
and below root zone in
agricultural fields. Flow of
water across the tailings
deposit surface, and through the
deposit is predicted by PRZH.
The PRZM model uses the SCS
Curve Number method to partition
the precipitation between
runoff, infiltration, and
evaporation (Carsel' et al.
1984).
The predicted contaminant
transport modes in PRZM include
runoff (dissolved metal carried
in overland flow), erosion
(metal in solid phase carried
by suspension in overland flow),
and leaching (gravity-driven
unsaturated zone transport of
soluble metal). Two modes of
transport not presently modeled
in PRZM are capillary transport
of contaminants to the surface
from the subsurface and overland
flood transport of metals.
Capillary transport occurs when
the soluble metal is carried to
the surface of the soil by
upward flow when
evapotranpiration is occuring
rapidly. The soluble metal
collects at the soil crust, and
appears as a salt with white or
green crystals of zinc and
copper sulfates and carbonates.
Overland flood transport of
metals is distinct from
transport by overland flow
erosion, and is similar to
sediment transport in stream
and rivers. (Figure 6)
15
-------
FIGURE 6
METAL LOADINGS
PATHWAYS
(CAPILLARY
\ TRANSPORT/
--5Z.
EROSION
PRZM evaluates the magnitude of metals loading
to the river for each of the three pathways.
16
-------
PRZM includes a parameter
describing the relative amount
of contaminant in solid phase
versus aqueous phase. This
parameter is called the
partition coefficient, and is
predicted using a measured
solid phase concentration of
metal along with a soluble
fraction of metal predicted with
MINTEQA2.
The Interface Programs between
PRZM and WASP4
Calibration of PRZM was
performed using data on surface
water runoff from tailings sites
along Silver Bow Creek and data
on sediment erosion in Butte at
the headwaters of Silver Bow
Creek (Brown, 1989). Using the
contaminant release time series
from PRZM program runs, daily
releases of metals and eroded
sediments can be determined for
each site using NPSOUT and
GCTRAN. The daily release of
copper into a surface-water
transport model WASP4 is
predicted using the geography
of the subsites. Output from the
NPSOUT program is transferred
to WASP4 as a non point source
loadings file.
These daily loadings into
each portion of the river take
into account measured slopes,
areas, and aquifer types. For
each contaminated subsite, the
surface area in hectares is
needed. Geographical data is
sparse for the off-stream sites.
The width of stream-side
tailings were measured for the
entire river course, and the
average width of the stream-side
tailings was used to calculate
the area of the streamside
tailings subsites. The areas of
these subsites are shown in
Table 1.
The NPSOUT Program
NPSOUT is a program written
for this project to predict the
transport of contaminant between
the plant root zone and the
surface water by simulating
mechanisms for contaminant
transport through the near-
stream saturated zone (Brown et.
al. 1989). The NPSOUT program
obtains and uses output data
from both PRZM and GCTRAN. PRZM
output data are used to predict
the movement of small particles
(less than 100 nanometers
diameter). The GCTRAN program
simulates large particle
generation and transport in the
unsaturated and saturated zone.
Predictions of metals
transport in groundwater will
depend on whether the metals
are in particulate or soluble
form. If the metals are in
particulate form, then the
phenomena that affect particle
mobility will become important
for predicting contaminant
transport. Typical pH and Eh of
the saturated zone (pH about
6.5, Eh about 0.2V) indicate
that the metal will be in a
colloidal precipitate form. The
transport of this precipitate
should not be retarded by the
adsorption of colloid onto the
soil; the precipitates should
have a very low adsorption rate,
because of electrostatic
stabilization.
The program NPSOUT provides
metals loadings from both the
sidestream deposits and other
sources; it adjusts the metals
17
-------
Location
Butte surface
Table I
Waste Site Areas
Area (hectares)
Butte subsurface
drainage
Anaconda
streamside
and alluvium
1049
not applicable
1500
648
Type of wastes
waste rock
acid mine
smelter waste
mixed tailings
18
-------
source magnitudes using various
parameters. The principal
parameters that can be modified
in NPSOUT and PRZM input files
to calibrate the predicted
surface water concentrations to
known data are the metals
concentrations at the surface
of the soil and tailings
mixture, Kd, and the recharge
flow to streamside aquifer
volume ratio Q/V. Rates of
transport of metal through
groundwater also will depend
upon the hydraulic gradients and
the hydraulic conductivity of
the soil through which the
metals pass on the way to the
river.
The principal modes of
jiKM-.nls transport in the
saturated and unsaturated zone
are transport as colloidal
particles and as solutes.
Colloids are a product of
precipitation from oversaturated
solutions. It is assumed that
the smaller colloidal particles
are transported by advection.
One simplifying assumption is
that larger precipitates formed
in the unsaturated zone are
immobilized and only soluble
compounds and small colloids
are carried from the unsaturated
zone to the saturated zone; this
assumption leads to one
transport model for the larger
colloidal particles and another
for the smaller colloidal
particles.
GCTRAN output data are used
to model the transport of larger
particles (between 100 and
10,000 nanometers in diameter) .
The particle size division is
arbitrary as there is no general
theory for predicting particle
size and mobility in unsaturated
media during oxidation processes
at this time.
One purpose of the NPSOUT
program is to provide a mixing
model for the transport of
solutes and the smaller classes
of colloids (100 nanometers to
1 nanometer) assuming a particle
velocity identical to the water
flow velocity in both the
saturated and unsaturated zones.
The rate of movement of water
and contaminants in the solute
and small colloid forms through
the porous media (saturated and
unsaturated) is assumed to be
the same. The overall velocity
for contaminant in the
unsaturated zone may be much
lower than in water because of
the confined movement of large
particles.
Small particle and solute
transport in the groundwater is
described with a Continuous
Stirred Tank Reactor mixing
model. Smaller particles and
solutes are treated as first
entering a well-mixed partial
volume, V, of the aquifer and
then entering the stream with
a fixed flow rate, Q. The
smaller' particles have an
average residence time in the
aquifer that depends on the
ratio Q/V. Calculation of Q is
based on the width of the
tailings, hydraulic gradient,
porosity, and hydraulic
conductivity. Calculation of V
is based on the width of the
tailings, porosity, and the
depth of the aquifer. Porous
media flow is governed by
Darcy's Law. Flow gradient and
aquifer permeability are
measured or estimated. Average
groundwater velocity can be
calculated using the above
parameters. Local velocities
can be obtained by the solution
19
-------
of groundwater flow equations
subject to the prevailing
boundary conditions. The
velocity and aquifer thickness
is used to estimate the total
flow through the aquifer. For
contaminated aquifers that are
at a distance from the stream,
a time of travel parameter for
travel from the mixing volume
in the aquifer to the streambank
where recharge is released by
the aquifer also can be included
in NPSOUT.
The contaminant flux into
the saturated zone was
determined using the leachate
fluxes predicted by PRZM. The
fluxes entering the groundwater
are treated as entering a
partial aquifer of fixed depth
and a width equal to the average
stream-side tailings deposit
width (i.e. 38m). The flow out
of the aquifer and into the
surface water is based upon
calibrated groundwater
velocities between 1.6 and 16.0
m/day. The parameter used within
the program NPSOUT to represent
the rate of decline in the
contaminant concentration in the
aquifer is the ratio Q/V, where
Q is recharge flow per day from
a partial aquifer volume and V
is the maximum partial volume
of the aquifer that is well-
mixed. Two trial values of Q/V
are 0.4 for the streamside
tailings, and 0.04 for the main
tailings deposits in Anaconda
and Rutte (based on a lower
hydraulic conductivity for the
off-stream sites). The
difference in the parameters
also could be attributed to the
waste deposit widths, which
could be much larger for the
off-stream sites in Anaconda and
Butte. A value of 0.04 leads to
a time scale of 120 days for
complete emptying of the aquifer
of a single 1-day contaminant
pulse, and 0.4 to a time scale
of 10 days for the same
scenario.
The GCTRAN Model
The Groundwater Colloid
Transport program GCTRAN was
written for this project for
use in conjunction with PRZM to
account for the special
transport characteristics of
particles that range from 0.1
to 10.0 micrometers in size
(large-size colloids).
GCTRAN simulates how the
mobility of larger colloids will
change as the water table
fluctuates and as percolating
water during floods and heavy
rainstorms removes significant
portions of the available
colloids generated by oxidation
processes in the unsaturated
zones.
The transport of large
colloids as modeled by GCTRAN
assumes that these particles
are moved with a velocity in
the saturated zone that exceeds
the average pore water velocity
and that these particles have
a velocity of zero in the
unsaturated zone.
The excess velocity of
larger colloids in the saturated
zone is based on the assumption
that excluded volume effects
increase large particle
velocities above pore water flow
velocities. Excluded volume
refers to the portions of the
pore water where the average
velocity field is small and the
diameter of the pores is small.
The larger a particle is, the
less chance that a particle will
20
-------
pass through the smaller pores
during its flow through
connected passageways in the
open spaces of the soil. The
large particle only passes
through the pores that have a
minimum required diameter and
that have higher pore water
velocities. Thus the average
flow field that the large
particle is suspended in exceeds
the average flow field velocity
of all tha soil pores. This
increase cf the particle
velocity to ebove the average
velocity of pore water can be
mathematically predicted given
a known distribution of pore
sizes and assumptions about the
pore networks. The differences
between average velocities of
particles with different sizes
in porous matrices is the basis
for two commonly used
chromatographic separation
techniques—size exclusion
chromatography and gel
permeation chromatography. Size
exclusion chromatography has
been applied to both polymers
in solution and to particles,
and gel permeation
chromatography is a technique
applied to the characterization
of polymer mixtures.
NPSOUT treats the larger
particles as if they enter the
surface water at the same time
as runoff and eroded sediment,
once they enter the aquifer.
The GCTRAN program uses a
number of parameters to control
the way in which large colloids
leach into the aquifer. The
first parameter is Kd, which
predicts a copper metal
concentration that corresponds
to pH and Eh within a boundary
layer of still water surrounding
each oxidizing sulfide particle.
The parameter PTRANS defines the
magnitude of the transport of
metal ions to the water outside
the boundary layer. In the
unsaturated zone outside the
boundary layer, the metal ions
are assumed to precipitate and
then accumulate. When the water
table rises, colloids in the
risen water are advected by the
saturated water flow to the
surface water. The parameter
controlling water table
fluctuation is RIVRIS, a
constant that relates cumulative
daily rainfall to change in the
water table level WATTAB. Daily
outflow and a decrement in the
water table is assumed to be
constant, so that the stream
recharge outflow results in a
decrement to the water table
level RECHDEL, with values
ranging from 0.01 to 0.2 cm/day.
Using the above hypotheses,
transport for large particles
and changes in the water table
level are predicted for the soil
and tailings core.
Removal of Metals in the Warm
Spring Ponds in NPSQUT
The? NPSOUT program performs
the additional function of
removing copper in the stream
at the end of Segment 6 of the
surface water model (that is,
at the exit to Warm Springs Pond
3). The measured removal
efficiency (mass removed/initial
contaminant mass) of total
copper (suspended and dissolved)
between the inlet to Warms
Springs Pond 3 and the outlet
of Warm Springs Pond 2 is 80%
on average. This removal
efficiency does not have any
apparent dependence on flow rate
through the ponds. The apparent
independence of removal rate
efficiency from flow rate leads
21
-------
to the hypothesis that residence
time for the metal in the
liming/precipitation ponds does
not affect removal efficiency,
and that removal efficiency is
constant. The approach taken in
NPSOUT to account for removal
of copper metal from the stream
in the Warm Springs Ponds was
to use NPSOUT to remove 80% of
all copper loadings from stream
loadings to surface water
segments 1 through 6.
The quantity and
characteristics of available
sediments affect transport,
removal, storage, and release
of heavy metals. Adsorption
parameters and sediment/eroded
soil particle size distributions
are not currently available. The
nonpoint source sediment loads
are included in PRZM output;
sediment data are provided to
the WASP4 model as a nonpoint
source from the erosion of the
stream banks. The model results
indicate that the sediment
transport is less than 5% of the
metals transported.
transport model (RIVERMOD) was
adopted. This model provided
the hydraulic parameters for the
water quality model WASP4/TOXI4.
In the development of RIVERMOD,
the Saint Venant equations are
used for the conservation of
mass and momentum. The sediment
transport module uses a sediment
yield equation as well as a
sediment continuity equation in
the sand size range. The
hydraulics and sediment
transport equations are solved
uncoupled, that is, first the
Saint Venant equations are
solved for hydraulic parameters
(Q,Y,V, etc.) and then these are
used for the computation of
sediment yield and sediment
transport from a given segment.
Details of this model are given
by Hosseinipour (1988, 1989).
In this application, the
hydraulic model was modified to
include time-variant lateral
inflows to better match the flow
characteristics of the stream.
The sediment transport module
was not activated in this
project.
The Water Quality and Transport
Modeling Package
The surface water quality
modeling package consists of
two separate models, RIVERMOD
and WASP4, linked by an
interface program.
The
River Hydraulics
and
Sediment
Transport
Model-
RIVERMOD
To simulate tho hydraulics
of combined Silver Bow
Creek/Clark Fork River and their
tributaries, a fully implicit
river hydrodynamic and sediment
The Water Quality Model-WASP4
The model readily available
for predicting surface water
metal concentrations was
WASP4/TOXI4. It was developed
for sediment and toxic material
transport and eutrophication
processes in surface water
systems, mainly estuaries and
wide river basins. The package
includes sub-models for
hydraulics, eutrophication and
toxic constituents. Details of
the model are given by Ambrose
et al. (1987). In this project,
the creek and river from Butte
to Deer Lodge was divided into
model segments for the use of
22
-------
WASP4.
The primary source of
surface water analytical data
is the Remedial Investigation
Final Report study (RIFR) on
the Silver Bow Creek (Tuesday,
1987); therefore, the main
stream segmentation was chosen
to coincide with the measurement
stations of that study as
outlined in Table 2. The surface
water modeling results are
compared to data from the RIFR
and from the Montana Water
Quality Bureau.
WASP4 was used to predict
the surface water quality for
the entire stream reach, with
11 segments representing
different reaches of the river.
Segments 1 to 5 represent Silver
Bow Creek. Segments 6 and 7
represent tho Warm Springs Ponds
or the Mill-Willow Bypass,
depending on the volume of the
flow. Segments 8 to 11 represent
the Upper Clark Fork River.
NPSOUT converts the PRZM
unit loading outputs into
specific loadings for each
segment. The time series from
NPSOUT are reformatted as WASP4
reference files. These time-
variable and space-variable
source terms help determine the
surface water contaminant
concentrations.
Application of the
simplified surface water
transport model WASP4 results
in a metal concentration time
series for each surface water
segment.
Results and Discussion
Surface Water Modeling Results
The surface water modeling
results of this study are
summarized in the following
manner. First, the copper
concentrations in Silver Bow
Creek, the Mill Willow Bypass,
and Warm Springs Creek are
compared with the predicted
metal loadings from the off-
stream subsites. The subsites
include the Butte mine drainages
and all other subsites for
Segment 1 and the Anaconda
smelter with the other subsites
for Segment 7. Each subsite has
a distribution coefficient Kd,
and a parameter Q/V.
Second, the in-stream
copper concentration predictions
for the Upper Clark Fork River
basin from this modeling
framework are compared to
measured surface water
concentrations. The comparison
is examined for conclusions
about the modeling system and
the assumptions used/ and for
observations about the approach.
Results of Tributary and Main
Dump Sites
The contaminants at the
end of Segment 1 are taken as
a measure of the Butte source
terms, such as the principal
Butte waste dump seepages and
drainages. The principal
drainages and seepages from the
Anaconda smelter area enter
Segment 7. The measurements of
loadings into segments 1 and 7
are used as reference data for
the calibration of the NPSOUT
model for contaminant sources
23
-------
Table 2.
WASP4 Segments Along the Clark Fork
River and Silver Bow Creek
WASP4 Segment RIFR Station and
Segment Length(km) Location at End of Segment
1 5.0 SS-07 Below Colorado Tailings
2 7.9 SS-10 Near Silver Bow
3 8.1 SS-14 Near Miles Crossing
4 8.0 SS-16 At Gregson Bridge
5 v 9.9 SS-17 At Stewart Street Bridge
6 5.6 Exit from Warm Springs Pond
#3
7 8.1 SS-29 At Perkins Lane Bridge
8 5.0 Between Measurement Stations
9 8.2 SS-30 Near Racetrack
10 8.4 SS-31 Below Dempsey Creek
Confluence
11 10.6 SS-32 At Deer Lodge
24
-------
in Butte and Anaconda. During
high flows, these off-stream
source terms may not describe
the entire metal flux observed
in the stream but they should
represent 50 to 100% of the
copper loadings during medium
and low flows.
Predictions of copper flux
from the tributary sites were
performed with NPSOUT, and the
results are compared with a
selection of 15 measurements
that date from 1984 through 1986
(Bahls 1987). This comparison
is made graphically by plotting
loading estimates from model
runs versus tributary loading
data (Brown 1989).
The first two sets of
conipArisons (Figures 7 through
.1.0) show how increasing Kd
reduces the predicted metal
fluxes and improves the
resemblance between model
predictions and the measured
data for Segment 1 and Segment
7.
Figures 10 and 11
demonstrate the results of a
change in the Q/V ratio (called
TQDIW in the model) for the
off-stream sites and an improved
resemblance to the calibration
data for Segment 7. This change
is equivalent to a large
increase in the rate of flow
from the aquifer.
_\
The type of groundwater
transport and mixing model used
strongly affects the predicted
loadings of metal into the
surface water. Changes in the
mixing model to include more
than one mixing volume might be
appropriate. The efficiency of
mixing in the model also might
be changed in order to improve
the model. Only the rate of flow
was varied in the model (Q/V
parameter) as applied to date.
The PRZM model results
reveal that the allocation of
metal is dependent on the
partition coefficient Kd.
Partition coefficient may vary
within a given subsite, but the
current approach does not
provide for that possibility.
The use of a single soil core
geochemistry (2-dimensional
isotropy) for each subsite is
the most significant assumption
in the model. Figures 7,9, and
12 compare surface water copper
concentrations with
concentration predictions for
several values of Kd.
In-Stream Predictions
The in-stream modeling
effort is based upon generating
stream hydraulics and loadings
for five different
meteorological and flow
scenarios. Two flood periods
were examined—a typical winter
flood due to runoff and a
typical spring flood due to
snowmelt. The seasonality of
the flow periods and contaminant
transport rates has already been
discussed. Figures 13 through
16 below show the model
predictions for the above
scenarios.
The normal flows for spring
and winter also were examined.
Finally, the low flow period
during late summer and the fall
were examined (see Figure 17).
Without having experimental data
for comparison, one can
nonetheless see the severity of
high copper concentration in the
stream during low flow periods.
25
-------
FIGURE 7
(0
•
(/)
c
f.l
Q.
200
150
100
50
Segment 1
Kd = 234, TQDIVV =0.04
a
a
a
a
1-1
CD
a
I I till
S « S 12
tO(O m 1O «f| ID
cDa)CDma3ai
^JiSrt^^
^ 3 S ^ § ^
U3 tp
Comparison Dates,, 1934-1986
Figure ^. Predicted vs.Measured Copper Loadings
26
-------
FIGURE 8
(O
tn
TD
td
o
Q.
Q.
O
U
400
300
200
100
Segment 1
Kd = 234, TQDIVV =0.04
a
1 1 T T
D
El
a
a
a
T 1 1 T T 1 T 1 1
Comparison Dates, 1984-1986
Figure 8. Measured vs. Predicted Copper Loadings
27
-------
FIGURE 9
8
o>
.v
m
c
'•D
(0
0)
CL
Q.
O
U
2DO
150
50
Segment 1
Kd = 328, TQDIVV =0.04
a
a
n n
1 — 1 LJ
1 1 1 1 1 1 1 1
Comparison Dates^ 1984-1986
Figure $. Measured vs. Predicted Copper Loadings
28
-------
FIGURE 10
(0
"O
(fl
D>
C
CO
O
QL
Q.
O
U
400
300
200
100
Segment 7
Kd = 328J TQDIVV =0.04
S »
a i
X
n
n
a
J8-
Si.
]—I T T I
2
Heasured
n
Predicted
8
&
a
-
JO
CD
CD to
a ?a
Comparison Dates^ 1984-1986
Pigureio. Measured vs. Predicted Copper Loadings
29
-------
FIGURE 11
r-\
>\
(0
rs
-v.
D>
^:
\_/
(/)
o>
b
(0
3
^3
Q.
a.
8
Segment 7
Kd = 328, TQDIVV =0.12
400
u>
§
to
1
i_»
§
n
a
T T T
X
Heasured
CD
Predicted
S
Comparison Dates, 1984-1986
Figure 10. Measured vs. Predicted Copper Loadings
30
-------
FIGURE 12
Segment 1
Kd = 656j TQDIVV =0.04
<5UU
3 ings c kg/ day}
,.» i-»
0 tfl
0 0
(O
O 50
0)
Q.
Q.
8
0
\|
X
a
sg^Sg^i s§§@§s°
i i i i i i i i i i i i > i i i
X
Heasured
J=J
Predicted
_mi/-,,,-1,,,cniDaiSaju>iJjeD(DfDa)lDCD
CM^^tM^^^oi'-^wj^^Jj^
Comparison Dates, 1984-1986
Figure 12. Measured vs. Predicted Copper Loadings
31
-------
inter Flood
Segment 1 Predicted, Measured
1400
-
j
g
-t— '
oo
o
CD
400
200 —
0
02/13/86
Predicted
Measured
02/23/86
03/15/86
Event Dates
-------
Winter Flood
Segment 11 Predicted, Measured
..
.-:
00
.c
v.
O
cn
1200
1000
800
600
400
200
0
i i i i i i i i i i i i i i i i I i i i i i i i i r
Predicted
Measured
02/13/86
02/23/86
03/05/86
03/15/86
Event Dates
-------
ui -+-1
*• co
.£
=5
a
pring Flood
Segment 1 Predicted, Measured
3000
2500 -—
2000
1500
Predicted
Measured
1000
05/14/86
05/24/86
06/03/86
Event Dates
06/13/86
-------
pring Flood
-JJ- V _J
Segment 11 Predicted, Measured
1000
p
(D
-b
CO
.c
O
en
800-
600-
400
200
x
0-fr
05/14/86
i i i i i i i i i i
05/24/86
i i [ i*"i
06/03/86
Predicted
x—
Measured
06/13/86
Event Dates
-------
CO
§
GO
CT>
Normal Fall Flow
Segment 7 Predicted, Measured
Predicted
10
0
07/14/86
07/24/86
08/03/86
i
08/23/86
08/13/86
Event Dates
-------
Conclusions
The variance between the
measured winter flood
concentrations in the Upper
Clark Fork River and the
prpdictions using the methods
and models depicted in this
report, show that, for the main
stem of the Clark Fork River,
copper in the river may have
unidentified sources currently
not included within the model.
For example, the stream
bed may hold a large reserve of
tailings deposited during floods
that bypassed the Warm Springs
Ponds or during floods before
the Warm Springs treatment ponds
were constructed. The outflows
from the Warm Springs Ponds are
accurately predicted.
Resuspension of sediments in
the ponds is accounted for with
the metal removal efficiencies.
During a flood, resuspension of
stream-bed contaminant may be
the primary contaminant source
if flows have sufficient power
to strip the armored cobble
bottom of the river and
resuspend sediments.
One suggestion and
conclusion for improving the
quality and predictive power of
the modeling system is to modify
the program GCTRAN, which
represents large colloid
transport in groundwater. If
improvements are made in
estimating the rate of
groundwater discharge to the
stream, infiltration, changes
in the water table, and the
effect of bank storage and
evapotranspiration, then a more
accurate simulation of the
sidestream contaminant sources
may improve the predictive
capability of this modeling
system.
For the winter flood, major
changes occur to the transport
patterns when the subsurface is
frozen. Freezing of the
subsurface reduces infiltration,
creates ice lenses, and affects
the chemical and microbiological
activity in the frost zone.
These phenomena are not included
in PRZM. Therefore, the runoff
in winter should include a much
larger portion of the snowfall,
rainfall, and snowmelt than is
currently predicted, and a
larger runoff could result in
a narrowing of the hydrograph
for the winter floods.
A narrow hydrograph may
lead to concentration versus
time peaks that are sharper than
the 1 day resolution permitted
by PRZM. Concentration may vary
significantly over a single day
for cases when the hydrograph
is narrow. An improvement in
model resolution for such cases
may be necessary.
The goal of improving
prediction of contaminant
transport raises the need for
a linkage between surface and
ground waters based on: (1)
better equations of flow in
groundwater and surface water
for this site; (2) improved
equations that define the rates
of infiltration from the surface
to the groundwater; (3)
equations that define kinetics
of particle oxidation and
changes in metal form, including
changes in contaminant
distribution and form over
different particle sizes,
changes in contaminant
distribution and form over
37
-------
different particle structures
(porosity, morphology), and
changes in contaminant metal
distribution and form over
different compositions of the
particles to which the metal
attaches (such as sulfides,
quartz, limestone, iron oxide,
mixed silicon and aluminum
oxides (clays)) and (4) expanded
equations that describe the
of fort of groundwater flow on
Llio movement of contaminants in
solution,in suspensions and in
immobile solids. Having an
accepted and reviewed linkage
would be a great advantage on
any other project where the
transport and flow in a
saturated zone or in a variably
porous and conductive subsurface
controls the contaminant
release.
Finally, we can conclude
that there may be substantial
contaminant transport events in
the river that have not been
looked for previously. This
conclusion is suggested by the
fall and late summer surface
water model results. The model
therefore, could be used to plan
river sampling for components
such as heavy metals, sulfates,
metal-bearing colloids, and
sediments.
REFERENCES
Ambrose, R. B., T. A. Wool, J.P.
Connolly, and R. W. Schanz.
1987. WASP4, A Hydrodynamic and
Water Quality Model-Model
Theory, User's Manual, and
Programmer's Guide. U.S.
Environmental Protection Agency,
Athens, GA. EPA/600/3-87/039.
Bahls, L. 1987. Water Quality
Data for the Clark Fork River
1985-1986, Montana State Water
Quality Bureau (unpublished
data).
Brown, D. S., R. E. Carl ton and
T. A. Wool 1987. MINTEQA1
Equilibrium Metal Speciation
Model: A Users Manual U.S.
Environmental Protection Agency,
Athens, GA, EPA/600/3-87/012..
Brown, K. P. 1989. Prediction
of Metal Speciation and
Transport Using Models of
Straamside Tailings Deposits
In: Proceedings of the Hazardous
Waste and Hazardous Materials
Conference, Hazardous Materials
Control Research Institute,
Silver Springs, MD.
Brown, K. P.. and Z. Hosseinipour
1989 Water Quality Modeling
and Transport Analysis of Heavy
Metal in the Clark Fork River
In: Symposium Proceedings on
Headwaters Hydrology. American
Water Resources Association,
Bethesda, MD. 708 pages.
Carsel, R. F., C. N. Smith, L.
A. Mulkey, J. D. Dean, P.
Jowise *• 1984. Users Manual for
the Pesticide Root Zone Model
(PRZM) U.S. Environmental
Protection Agency, Athens,
Georgia, EPA/600/3-84/109.
CH2M-Hill 1988 Silver Bow Creek
Flood Modeling Study (Draft),
CH2M-Hill, Boise, ID.
Hosseinipour, Z. 1989 Fluvial
Hydrodynamic and Sediment
Transport Model for the
Cheasapeake Bay Watershed. In:
Proceedings of the International
Conference on Cheannel Flow and
Catchment Runoff for Centennial
of the Manning's Formula and
Kuichling's Rational Formula,
38
-------
Charlottesville, Virginia
Hosseinipour, Z. "Development
of a Fluvial River Flow Routing
and Sediment Transport Model
for the Chesapeake Bay
Watershed," Report. U.S. EPA
CBLO, August 1988.
Tetra Tech. 1986. Anaconda
Smelter RI/FS. Tetra Tech,
Bellevue, Washington
Tuesday, D. S. Grotbo,and W. M.
Schafer 1987 Silver Bow Creek
Remedial Investigation Work Plan
and Draft Final Report
Multitech, Butte, MT.
39
------- |