LAKE ONTARIO TCDD MODELING REPORT
by
Douglas D. Endicott
William L. Richardson
Large Lakes Research Station
U.S. Environmental Protection Agency
9311 Groh Road
Grosse lie, Michigan 48138
Dominic M. Di Toro
Department of Environmental Engineering
Manhattan College
Bronx, New York 10471
-------
TABLE OF CONTENTS
List of Figures iv
List of Tables vii
1. Introduction 1
2. Results and Recommendations 3
Recommendations to Improve Lake Ontario Toxics Modeling . . 6
3. Model Description 8
Modeling Framework 8
Mass Balance Equations 11
Model Parameterization 15
Lake Ontario 15
Suspended Solids and Sediment 17
Partitioning 20
Volatilization and Photolysis 25
Lake Ontario meteorology 25
Volatilization 26
Photolysis 30
Loading . . . 38
4. Model Application 40
Steady State Model 40
Dynamic Model 46
Sensitivity Analysis 48
Monte Carlo Simulation of Uncertainty 59
Comparison to Data . 82
References 85
Appendices
A. Workplan 91
B. Expert Panel Contribution 97
C. Sample WASP4 Input and Output 98
iii
-------
LIST OF FIGURES
Number Page
1 Predicted Steady-State Response of Lake Ontario Sediment
to Constant TCDD Input 4
2 Predicted Steady-State Response of Lake Ontario Water
Column to Constant TCDD Input 5
3 Schematic of Lake Ontario TCDD Model 12
4 Estimated Monthly TCDD Volatilization and Photolysis
Loss Rates 36
5 Predicted Steady-State Response of Lake Ontario Sediment
to Constant TCDD Input 42
6 Predicted Steady-State Response of Lake Ontario Water
Column to Constant TCDD Input 43
7 Steady-State Fate and Transport of TCDD in Lake Ontario ... 44
8 Predicted Relationship Between Steady-State TCDD
Concentrations in Water Column and Sediment 45
9 Predicted Dynamic Response of Lake Ontario to TCDD Input
and Cutoff 47
10 Dynamic Input-Response Function for Lake Ontario TCDD
Input 49
11 Sensitivity Analysis: Suspended Solids Variation 51
12 Sensitivity Analysis: Active Sediment Layer Thickness .... 52
13 Sensitivity Analysis: Resuspension Flux Variation 53
14 Sensitivity Analysis: Sedimentation Flux Variation 54
15 Sensitivity Analysis: Suspended Solids fQC Variation .... 55
iv
-------
LIST OF FIGURES (CONTINUED)
Number Page
16 Sensitivity Analysis: Kpoc Variation . . 56
17 Sensitivity Analysis: Volatilization Rate Constant
Variation 57
18 Sensitivity Analysis: Photolysis Rate Constant Variation . . 58
19 Histograms of the Log Photolysis Rate Constant and the
Log Volatilization Rate 64
20 Log Probability Plots of the Volatilization Rate, the
Photolysis Rate, the Sedimentation Velocity, and the
Depth of the Active Sediment Layer 66
21 Scatter Plots of the Log of the Volatilization Rate, the
log Photolysis Rate, the Log Water Column Suspended Solids
Concentration, the Log Sediment Velocity, and the Log
Depth of the Active Sediment Layer 67
22 Log Probability Plots of the Total Water Column TCDD
Concentration, the Dissolved Concentration, the Total
Sediment TCDD Concentration, and the Sediment Sorbed TCDD
Concentration 68
23 Histograms of the Dissolved Water Column and Sediment
Sorbed TCDD Concentration for the Base Case Uncertainty ... 70
24 Scatter Plots of the Logs of the Total Water Column
Concentration, the Dissolved Water Column Concentration,
the Suspended Solid Sorbed Concentration, the Total Sediment
Concentration, and the Sediment Sorbed Concentration 71
25 Scatter Plots of the Concentrations and the Parameters
That Affect Removal of TCDD 72
26 Scatter Plots of the Concentrations and the Other
Uncertain Parameters 74
27 Histograms of the Dissolved Water Column and Sorbed
Sediment TCDD Concentration for the Base Case Uncertainty,
Modified to Include Lognormal Distribution of Photolysis
Rate Uncertainty 75
v
-------
LIST OF FIGURES (CONTINUED)
Number Page
28 Histograms of the Dissolved Water Column and Sorbed
Sediment TCDD Concentration for the Case of No Uncertainty
in Photolysis Rate 77
29 Scatter Plots of the Concentrations and the Parameters
Affecting TCDD Removal and Partitioning, With Zero
Photolysis Uncertainty 78
30 Histograms of the Dissolved Water Column and Sediment
Sorbed TCDD Concentration for No Volatilization Uncertainty,
and in Addition No Sediment Velocity Uncertainty, and in
Addition No KQC Uncertainty ... 79
31 Median, 2.5 and 97.5 Percent Confidence Limits for the
Dissolved Water Column and Sediment Sorbed TCDD
Concentrations 80
32 Lake Ontario TCDD Concentrations in Top 3 cm of Sediment
Cores 83
vi
-------
LIST OF TABLES
Number Page
1 Literature Values of L0G(K ) and L0G(K ) for TCDD 23
oc ow
2 Volatilization Rate Computations 31
3 Photolysis Rate Computations 34
4 Summary of Lake Ontario TCDD Model Parameters 39
5 Uncertainty Bounds for the Model Parameters 62
6 Results of the Uncertainty Analysis 81
vii
-------
SECTION 1
INTRODUCTION
In February, 1988, the U.S. Environmental Protection Agency's Large Lakes
Research Station initiated a modeling project at the request of Region II
Superfund, to support the Lake Ontario TCOD Biฉaccumulation Study.
The primary objective of the modeling project was to determine the
relationship between 2,3,7,8-tetrachlorodibenzo-p-dioxin (2,3,7,8-TCDD or
"TCDD") input to the Niagara River from Occidental Chemical Corporation's
Hyde Park hazardous waste site and resulting contaminant levels in the
surficial sediment of Lake Ontario. The possibility of other unquantified
sources of TCDD to Lake Ontario precluded any direct relation of Hyde Park
input to observed lake sediment contamination. However, by neglecting other
sources, the TCDD contamination of Lake Ontario attributable to Hyde Park
could be predicted. Hyde Park was modeled as an incremental source of TCDD
to Lake Ontario, an exercise consistent with the objectives of the
Bioaccumulation Study (U.S. Environment Protection Agency, 1987).
A mass balance model of the lake was developed, based upon models of
fallout radionuclides and PCB contamination in the Great Lakes. With the
assistance of a panel of modeling experts, the model was refined and
reparameterized for TCDD. Model simulations were then performed to determine
the response of Lake Ontario to an incremental input of TCDD. These included
1
-------
steady-state as well as dynamic simulations. Hodel sensitivity analysis was
conducted to demonstrate how the individual inputs affect the model
simulation. Uncertainty analysis was also performed upon the model to
determine the probabilistic behavior of model predictions given uncertainty
in the model inputs. Finally, modeling results were compared to available
Lake Ontario TCDD data.
2
-------
SECTION 2
RESULTS AND RECOMMENDATIONS
Model simulations were performed to estimate the response of Lake Ontario
sediment and water to TCDD input. The results for steady-state
sediment-sorbed TCDD and dissolved water concentrations are presented in
Figures 1 and 2. The confidence intervals in Figures 1 and 2 reflect the
uncertainty in model predictions due to uncertain input parameters. Model
predictions indicate that 50% of the steady-state TCDD concentrations are
reached within 10 years in the sediment and within less than one-half year in
the water column. Lake Ontario sediment TCDD concentrations are predicted to
approach steady-state in 40 years while the water column is predicted to
approach steady-state in 15 years.
According to the Stipulation for Requisite Remedial Technology at the
Hyde Park Landfill (U.S. Environmental Protection Agency, 1987) input of TCDD
from Hyde Park to the Niagara River and Lake Ontario is limited to 0.5
gm/year, the "Interim Aqueous Phase Liquid (APL) Flux Action Limit". The
predicted steady-state TCDD concentrations for an input of 0.5 gm/year to
Lake Ontario are 0.026 ng/kg (sediment sorbed concentration) and 9.5 x 10"^
pg/L (water column dissolved concentration). These model predictions may be
used to relate assessments of human and ecosystem risk due to TCDD
contamination of Lake Ontario to TCDD input from Hyde Park.
3
-------
Figure 1. Predicted Steady-State Response of Lake Ontario
Sediment to Constant TCDD Input
1000 1
Predicted Response
95 % Confidence Limits
100 ฆ:
a.
"O
.001
1
100
1000
1 0
10000
TCDD Input (grams/year)
4
-------
Figure 2. Predicted Steady-State Response of Lake Ontario
Water Column to Constant TCDD Input
Predicted Response
95% Confidence Limits
.0001 ฆ:
1
1
10
100
1000
10000
TCDD input (grams/year)
5
-------
Comparison of steady-state model predictions to the average, deposition
zone Lake Ontario surficial sediment TCDD concentration of 110 ng/kg provides
an estimated total TCOD loading to the lake of 2.1 kg/year. The predicted
steady-state water column dissolved TCDD concentration for this load is 0.4
pg/L.
RECOMMENDATIONS TO IMPROVE LAKE ONTARIO TOXICS MODELING
Through the process of developing and applying the Lake Ontario TCDD
model, every attempt has been made to make full use of available
information. However, the reliability and accuracy of the model are
constrained by the limitations of this infornation. In particular, one may
question whether much of the information used to characterize the
distribution, transport and fate of TCDD in the model is applicable to the
Lake Ontario system. As indicated by results of the Monte Carlo analysis of
uncertainty, model predictions depend upon TCDD loss mechanisms the
magnitudes of which are unique to Lake Ontario. Unless the distribution,
transport and fate of a contaminant are studies in the field system, the
possibility of mis-specifying or omitting processes of first-order importance
cannot be discounted. Improvement of the Lake Ontario TCDD model would
require such studies.
Application of this model as a predictive tool for toxics management is
precluded by a lack of information of historical and contemporary loadings of
TCDD to Lake Ontario. For example, the assumption of a steady-state system
cannot be tested without a characterization of loadings. Measurement of
Niagara River toxics loadings has proven to be difficult (Niagara River
Toxics Management Plan, 1988) and has yielded no information for TCDD. An
6
-------
alternative strategy of estimating loadings from sediment deposition records
is recommended. Such efforts have been successful for dioxins in Si skiwi t
Lake, Isle Royale (Czuczwa et al.., 1985) and southern Lake Huron (Czuczwa and
Hites, 1984). The deposition record of a sediment core from near the Niagara
River mouth (Station 208) has been shown to preserve a strong signal of
chlorinated hydrocarbon contamination from the Niagara River (Durham and
Oliver, 1983). A deep sediment core was collected at this location during
the Bioaccumulation Study. Analysis of TCDO and radionuclides from this core
may be used to estimate the historical load from the Niagara River to Lake
Ontario. With such an estimate, better predictions of the change in Lake
Ontario contaminant levels resulting from actions taken to control TCDD input
to the Niagara River could be made.
Consumption of TCDD-contaminated Lake Ontario fish has been identified as
the primary hazard to humans posed by leakage of TCDD from Hyde Park (U.S.
Environmental Protection Agency, 1987). The Lake Ontario TCDD model should
be extended to incorporate the aquatic food chain into the simulation, so
that the relationship between TCDD loading and fish contamination can be
predicted. The result would be a true exposure model, which could be linked
to risk assessments. The laboratory studies of lake trout TCDD
bioaccumulation (U.S. Environmental Protection Agency et al., 1989) should
provide information useful for calibrating a food chain model.
7
-------
SECTION 3
MODEL DESCRIPTION
MODELING FRAMEWORK
Models are constructed to simulate the behavior of complex systems. They
are necessarily simplifications of the real world, synthesizing information
of the processes controlling the system's behavior. If a model is accepted
as a reasonable representation, then it can be used to predict future system
behavior.
For this project, a model was constructed from a mass balance framework
and implemented by a computer program, WASP4 (Ambrose et al., 1987).
Essentially, the model is a set of equations derived from the principle of
mass conservation. The aquatic system is divided into homogeneous control
volumes, or segments. The mass of contaminants entering a segment is either
transported out of the segment, accumulates in the segment, or is transformed
by chemical reaction or biological degradation. In the case of a system
composed of several segments, transport of contaminant may occur between
segments as well as across the system boundaries. By accounting for the
accumulation of mass in the system segments over time, the model computes
concentrations of water quality parameters of interest.
Two predecessors of the WASP4 computer program were TOXIWASP (Ambrose et
al., 1983) and WASTOX (Connolly and Winfield, 1984). Both were used to
8
-------
implement fate and exposure models of toxics. They evolved from a general
water quality model, WASP (Di Toro et al., 1983). The descriptions of
transfer and transformation of pollutants in WASP4 are modifications of those
included in another program, EXAMS (Burns ei al., 1982).
The WASP mass balance formulation and computer programs have been used to
implement several water quality models. These include eutrophication and PC6
models of the Great Lakes including Lake Ontario (Thomann ฃt al., 1975;
Thomann et al., 1976; Thomann et aL> 1979; Di Toro and Connolly, 1980;
Thomann and Di Toro, 1983; Connolly et al-> 1987), eutrophication, heavy
metal, and PCB models of Saginaw Bay (Bierman and Dolan, 1981; Oolan and
Bierman, 1982; Richardson, 1982), eutrophication of Potomac Estuary (O'Connor
et al, 1983), and volatile organics in the Delaware Estuary (Ambrose, 1987).
Mass balance model computer programs similar to WASP have been applied to
model mirex in Lake Ontario (Halfon, 1983; Eadie et aL> 1983).
The Lake Ontario PCB model of Connolly et al*> was used as a prototype
for the model developed in this project. In a critical review conducted by
the International Joint Commission (IJC, 1988), this model was found to be
reasonably accurate, despite its simplicity. This model was calibrated using
"3Pu (a radioisotope from nuclear bomb testing fallout) data, for which
the history of loading and levels of contamination in the Great Lakes are
well known (Thomann and Di Toro, 1983). Because the persistent, hydrophobic
nature of 239Pu is similar to that of PCBs and TCDD, and the model fits the
radionuclide data well, there is a high degree of confidence in the
predictive ability of this model.
9
-------
The Lake Ontario TCDO model is based upon the general behavior of
hydrophobic contaminants in aquatic systems. Because of their
hydrophobicity, these contaminants will partition onto surfaces, such as
particles suspended in the water column. Karickhoff et al. (1979) determined
that partitioning of hydrophobic organic contaminants from water onto
particles (sorption) is normally controlled by the organic carbon content of
the particles.
The association of hydrophobic contaminants with particulate matter links
their transport and fate with that of particles as well as water. Water
column transport is governed by the customary advection-dispersion equation.
TCDD sorbed to solid particles is affected by settling in addition to
transport. Particles settle from the water column and accumulate in the
sediment where, over time, they are buried. Deep lake currents, due to
events such as spring runoff, severe storms, or lake overturn, resuspend
settled particles back into the water column. Association of hydrophobic
organic contaminants with particles act as a "capacitor" on water quality,
slowing the rate at which contaminant levels change. Partitioning also
influences fate, because dissolved and sorbed TCDD may be subject to
different transformation processes. Significant transformations of TCDD
include photolysis and volatilization. This model also considers contaminant
partitioning to a third phase, non-settling organic matter (NSOM), in
accordance with recent evidence from laboratory experiments (Carter and
Suffet, 1982; Voice and Weber, 1985; Gschwend and Wu, 1985; Eadie et al.,
1988) and field measurements (Baker et al., 1986). A schematic diagram of
10
-------
the overall WASP4 mass balance framework as applied in modeling TCDD in Lake
Ontario is presented in Figure 3.
WASP4 treats the aquatic system as an assemblage of segments. Two
segments were used to model Lake Ontario. A completely-mixed water column
was the first segment, and the underlying active sediment layer was the
second. This is the simplest configuration to include the interaction of
sediment with the water column, and is consistent with the general uniformity
of Lake Ontario sediment contamination data. A single inflow (Niagara River)
and outflow (St. Lawrence River) were included in the model.
MASS BALANCE EQUATIONS
The mass balance for the two-segment Lake Ontario model consists of four
coupled, linear differential equations. The first two equations describe the
mass balance of solids in the water column and sediment, and the second two
provide the mass balance for TCDD. WASP4 computes concentrations by
integrating the mass balance equations.
The mass balance equation for solids In the water column is:
Vj d M1 - Wm + vr A2 M2 - Q Mj - vs A2 M, (Equation 1)
dt
where Vj = volume of water column (L^)
Mj = suspended solids concentration (M/L^)
Wm = solids input to water column (M/T)
vr = resuspension velocity (L/T)
A2 - area of sediment-water interface (L^)
M2 = sediment solids concentration (M/L^)
II
-------
Figure 3. Schematic of Lake Ontario TCDD Model
TCDO Input
ro
Volatilisation
Photolysis
Sorbed
TCDD
Settling
Sorption
TCDD
Photolysis
Photolysis
Binding
Suspended Solids
Transport
Resuspenslon
TCDD
Water Column Transport
Diffusive
Exchange
Diffusive
Exchange
iiiTiiiiii^/.. -A xj ฃ
N Y, *ป
Spfpwa
Bound
Dissolved
TCDD
Sorbed
TCDD
- TCDD
-------
Q = flow rate through lake (L3/T)
vs = settling velocity (L/T)
The accumulation of suspended solids in the water column equals the sum of
solids inflow and resuspension of sediment, minus the outflow and settling of
suspended solids.
The mass balance equation for solids in the active sediment layer is:
v^ = sedimentation velocity (L/T)
The accumulation of solids in the active sediment layer equals the sum of
suspended solids settling, minus sediment resuspension and sedimentation
(incorporation of solids into deeper sediment layers).
For TCDD in the water column, the following mass balance equations
applies:
vr A2 M2 - vd Mg
(Equation 2)
where V2 ฆ volume of active sediment (L3)
ป! 111_ - W + Kf A2[(fd2 ~ fM) Cj/Oj -
-------
where
Cj = water column total TCDO concentration (M/L^)
W = TCDD input to water column (M/T)
Kf = diffusive exchange coefficient (L/T)
fdiป fbi, fsj ฆ dissolved, bound (to NSOM), and sorbed (to solid
particles) fractions of TCDD in water column
^d2' ^b2' ^s2 18 dissolved, bound, and sorbed fractions of TCDD
in active sediment
Cฃ = active sediment total TCDD concentration (M/L^)
r\2 = porosity of active sediment
k"p = first-order photolysis rate constant (T-*)
kv = first-order volatilization rate constant (L/T)
Aj = area of air-water interface (L^)
Water column accumulation of TCDD equals the sum of TCDD input, diffusion
from sediment pore water, and resuspension of sediment-sorbed TCDD, minus
outflow, photolysis, volatilization and settling of suspended solid-sorbed
Finally, the mass balance equation for TCDD in the active sediment layer
is:
TCDD.
vs A2 fsl C1 "
Kf A2^fd2 + fb2^ C2/n2 ' ^dl + fbl^ Cl^
A0 f , C
d 2 s2 u2
(Equation 4)
14
-------
Accumulation of TCDD in the active sediment equals the sum of settling of
suspended solids-sorbed TCDD, minus pore water diffusion to the water
column, and resuspension and sedimentation of sediment-sorbed TCDD.
MODEL PARAMETERIZATION
This section describes the development of parameters for the Lake Ontario
TCDD model, including sources of information, assumptions, and calculations.
An expert panel assisted in assembling and interpreting the information
necessary to estimate the transport and transformation of TCDD in Lake
Ontario. Appendix B, Expert Panel Contribution, describes the activities of
the expert panel.
Parameter values which are lake-wide, annual averages, were selected from
the literature and obtained directly from researchers. The objective of
parameterization was to identify expected or 'best estimate" values of model
parameters from the available information. In addition, ranges of parameter
values based upon data and/or a consensus of expert panel opinion were
selected to reflect uncertainty in these expected values. Parameter ranges
were used in model sensitivity and uncertainty analyses. Model parameters
have been grouped in the following manner:
Lake Ontario Water Column,
Suspended Solids and Sediment,
Partitioning,
Photolysis and Volatilization,
Loading,
and this section is organized in the same fashion.
Lake Ontario
Lake Ontario has been modeled as a completely-mixed water body. This
simple approach is appropriate for modeling long-term response to
15
-------
contamination, and is consistent with the limited amount of data available
for this modeling effort. This is also the simplification used in models of
fallout radionuclides (Thomann and Di Toro, 1983) and PCBs (Connolly fit al.,
1987). It has been observed that the lake waters are fairly well mixed by
the annual circulation (Simons et al., 1985). The eastward current along the
Lake's southern shore in winter, for example, transports ten times more water
than discharges to the St. Lawrence River.
The general distribution of hydrophobic contaminants in Lake Ontario
depositional sediments supports the completely-mixed assumption. No
significant differences in the distribution of TCDD in surficial sediments
among the three major deposition basins of Lake Ontario were observed in the
Bioaccumulation Study (U.S. Environmental Protection Agency fit al., 1989).
Jaffe and Hites (1986), using fluorinated aromatic compounds as tracers of
Hyde Park leakage, determined from analysis of sediment cores that
contaminants leaving the dump are rapidly and uniformly distributed
throughout Lake Ontario.
Similarly, fish analysis revealed no significant differences in TCDD body
burden for lake trout and brown trout collected from different zones of Lake
Ontario (U.S. Environmental Protection Agency et al., 1989). However, the
mobility and migration patterns of fish and prey makes this a weak test of
the completely-mixed assumption.
The physical parameters of Lake Ontario include volume and surface area
as well as flow. The lake volume is 1.68 x 10^ m^ and the surface area
is 1.95 x 1010 m2 (Aubert and Richards, 1981). The average lake depth is
86 m. Lake outflow, through the St. Lawrence River, has been regulated since
16
-------
1958. Over the 23 years following regulation the average flow has been 7100
m3/s (Quinn and Kelley, 1983) and this value was used. The hydraulic
residence time of the lake is 7.5 years.
Suspended Solids and Sediment
Transport of TCDD associated with solid particles is modeled by including
suspended and sedimented solids as part of the mass balance simulation.
Solids were modeled as a steady-state system in the water column and
underlying sediment layer in this application. The active, resuspendable
sediment is treated as a homogeneous layer, nixed by processes such as the
feeding activities of sediment biota (Lee and Swartz, 1980). The sources of
natural particles in the Great Lakes include shoreline erosion, resuspension
of surficial sediments, primary production, riverine and atmospheric
transport, and calcite precipitation (Eadie and Robbins, 1987). Model
parameterization includes the specifications of solids and sediment
characteristics, the fluxes of solids between the water column and sediment
segments, and exchange of TCDD by pore water diffusion.
Solids are characterized by average values of density, organic carbon
content, and concentration in the water column and sediment layers. The
density of solids was estimated to be 2.4 gm/cm3, a typical value for
fine-grained sediments. The average organic carbon fraction of suspended
solids, foc, was set at 12%, with a range of 10 - 15% (Rosa, 1985; Eadie,
1988), and the suspended solids concentration in Lake Ontario assigned an
average value of 1.2 mg/L, with a range of 1.0 - 1.5 mg/L (Rosa, 1985; Eadie,
1988). The average value of porosity in surficial sediments of Lake Ontario
was reported to be 0.887 (Connolly et al., 1987), from which a sediment
17
-------
solids concentration of 270 gm/L was calculated. The minimum and maximum
values of surficial sediment porosity (of four locations sampled) were 85%
and 95%, providing a range of sediment solids concentrations from 130 to 355
gm/L. The average organic carbon fraction of sediment solids is 3% based
upon 30 measurements of deposition basin sediment (U.S. Environmental
Protection Agency et aJL., 1989). The 95% confidence interval about the mean
of these data, 2.6 - 3.5%, provided the range of sediment organic carbon
fraction. The same average value of organic carbon fraction was reported for
deposition-zone sediment solids by Thomas et al. (1972). Connolly et al..
(1987) reported a lake-wide average value of 3.5%. The sediment layer
thickness was calculated to be 1.8 cm, based upon a 20 year residence time
for resuspendable sediment (Robbins, 1988). A 1-5 cm range was selected for
the active sediment layer thickness, which agrees with other Great Lakes
modeling studies (Thomann and Di Toro, 1983; Robbins, 1985) and sediment
surveys (Kemp and Harper, 1976).
Solids transport is parameterized in the model as fluxes of settling,
resuspension and sedimentation. Based upon an extensive survey of Lake
Ontario sediments, Kemp and Harper (1976) estimated a lake-wide sedimentation
flux of 240 gm/mVyear. The calibration of a Lake Ontario model to ^39Pu
data provided values of 224 and 197 gm/mfyyear for sedimentation and
resuspension fluxes, respectively (Thomann and Di Toro, 1983; Robbins,
1985). Based upon these data, a sedimentation flux of 240 gm/m*/year, and
a resuspension flux of 200 gm/mfyyear were selected as "best estimates".
These values correspond to sedimentation and resuspension velocities of 0.90
and 0.73 mm/year. Multiplying the resuspension flux by the organic carbon
18
-------
fraction (foc) of sediment solids (0.03) gives a model POC resuspension
flux of 6 gm/m2/year, which agrees closely with the 6.2 gm/m2/year
measured by Rosa (1985). To balance the steady-state sediment solids, a
settling flux of 440 gm/m2/year (and corresponding settling velocity of 1.0
m/day) was chosen. This value is equivalent to a POC settling flux of 53
gm/m2/year, similar to the flux of 64 gm/m2/year measured by Rosa (1985).
A solids loading of 1.4 x 107 kg/day is required to balance suspended
solids in the lake water column.
Due to the difficulty in accurately determining lake-wide solids fluxes,
a factor of five variation was assigned to these parameters. The bounds
of the solids fluxes were evenly spaced on a logarithmic scale below and
above the expected values. Calculated in this way, the ranges for
sedimentation and resuspension fluxes were 110-550 and 88-440 gm/m2/year,
respectively.
Diffusive exchange of TCDD between the sediment and the water column is
parameterized by a diffusive transport coefficient, Kf. A correlation for
Kf is calculated (Di Toro et al_.1981) as:
- 19 n MW"2/3 (Equation 5)
where n = sediment porosity
MW = molecular weight (gm/mole)
The value of Kf was calculated to be 0.36 cm/day. An order-of-magnitude
range, 0.11 - 1.1 cm/day was assumed.
19
-------
Partitioning
TCDD has been modeled as partitioning among three phases - dissolved,
sorbed to the organic fraction of solid particles, or bound to non-settling
organic matter (NSOM). Partitioning between phases is modeled as a linear,
reversible and rapidly-equilibrating process. This representation allows
compound phase distribution in the water column and sediment to be specified
by two coefficients, a sorption coefficient (Kp) and a binding coefficient
(Kjj). Define:
Kp = VCw (Equation 6)
where Cs = sorbed concentration (mass of TCDD/mass of particles)
Cw = dissolved concentration (mass of TCDD/volume of water)
and:
Kb = VCw (Equation 7)
where = bound concentration (mass of TCDD/mass of NSOM)
Kp and depend upon properties of both the compound and the phases into
which it partitions. For the sorption coefficient, these effects are
separated by the parameters Kp0C and foc:
Kp = foc ' Kpoc (Equation 8)
where foc = fraction organic carbon of particles
Kpoc = particulate organic carbon partition coefficient
20
-------
From the partition coefficients, the fractional distribution of compound
between the three phases (and thus the relationship between the total
concentration and the concentration in each phase) may be calculated:
fd ฆ 1+ K. B + K . M. (Equation 9)
where f,j = dissolved fraction of chemical
B = concentration of non-settling organic matter (mass of
NSOM/volume of water)
M$ = concentration of solids (mass of solids/volume of water)
fs - Kp * Ms * (Equation 10)
where fs = fraction of chemical sorbed to solids
fb = Kfa B fj (Equation 11)
where f^ = fraction of chemical bound to NSOM
Although this simple representation of partitioning may not be completely
accurate, it has been found to be useful for modeling the distribution of
hydrophobic organic compounds in aquatic systems. A possible enhancement to
the partitioning model is to make Kp a function of solids concentration.
Several investigators have observed an inverse relationship between Kp and
solids concentration, which Di Toro et al. (1985) have formulated as:
21
-------
where K'p = solids-dependent sorption coefficient
This "solids effect" is invoked only for partitioning in the water column.
The water column Kp adjustment was applied in all TCDD model simulations.
The contaminant-specific information necessary to parameterize
partitioning in the model are the values of KpOC and Values of
Kpoc may be obtained from laboratory measurements of the organic carbon
partition coefficient, Koc (Lodge and Cook, 1988), or from correlations
using solubility, S:
log(K0c) = 4.04 - 0.557 log(S) (Chiou et al.., 1979; Equation 13)
log(K0c) = 3.64 - 0.55 log(S) (Kenaga and Goring, 1980; Equation 14)
or Kow, the octanol-water partition coefficient:
lฐg(Knr) = lฐ9(Kftul) - 0.21 (Karickhoff et al., 1979; Equation 15)
UL UW
Using a TCDD solubility of 10.4 ng/L (Marple et al.., 1986a), the average
log(Koc) calculated by Equations 13 and 14 is 6.46. Table 1 lists values
of Kow and Koc from the literature. Five of the Kow measurements were
judged to be reliable values; they were applied to Equation 15. In
combination with two reliable Koc measurements from Table 1, eight values
of Koc for TCDD were obtained. The average log(Koc) was 6.5, and the 95%
confidence interval about the mean was 6.3-6.8. These resulting values were
22
-------
TABLE 1. LITERATURE VALUES OF LOG(Koc) AND LOG(Kow) FOR TCDD
SOURCE OF VALUE
LOG(Koc)
7.25 Lodge and Cook (1988)
6.6* Walters and Guiseppi-Elie (1988)
6.4* Marpleetal- (1987)
7.39 - 7.58 Jackson et al. (1985)
L0G(Kow)
6.64*
Marple et al. (1986b)
7.02*
Burkhard and Kuehl (1986)
8.93
Sarna et al. (1985)
6.15*
Kenaga (1980)
7.16*
Perkow et al. (1980)
6.84*
Johnson (1982)
5.50
Kaiser (1983)
*Va1ues used in determining Kpoc.
23
-------
used as the expected value and range of log(Kpoc) in the model. Kp was
calculated from Kp0C using Equations 8 and 12.
Experiments to evaluate the binding of a suite of hydrophobic organics to
non-settling organic matter in Great Lakes water indicate that the values of
Kb are typically two orders of magnitude less than Koc (Eadie et a]..,
1988). The fraction of chemical bound to NSOM (fjj) was found to be less
than 10% in these experiments. Two compounds of similar hydrophobicity to
TCDD, DDT and 2,2',4,4',5,5'-hexachlorobiphenyl (HCB), were found to have
1 og(Kjj) values of 4.3 - 4.4. Based upon this information, a best estimate
of 4.5 was selected for 1og(K^) of TCDD. The 95% confidence intervals
about the mean for log(Kjj) of DDT and HCB were slightly smaller than ฑ 0.2;
a range of 4.3 - 4.7 was selected for logfK^) of TCDD. Some values of
log(Kjj) for tetra-, penta- and hexachlorinated dioxins in solutions of
fulvic acid have also been reported (Webster et al., 1986), on the order of
o.9 - 5.4. Because experimental evidence suggests that the capacity of NSOM
to bind hydrophobic organic compounds (i.e., the value of Kf,) is dependent
upon the source of NSOM (Morehead et al.., 1986; Chiou et al.., 1987), the
values estimated for TCDD based upon Great Lakes water experiments were
judged to be most appropriate for this model.
Nonsettling organic matter may be operationally defined as the fraction
of total organic carbon which will pass a 0.45 ym filter. This includes
humic substances, lipids, carbohydrates, and other macromolecules, as well as
colloidal and micro-particulate organic matter. Concentrations of
nonsettling organic matter in the water columns of the Great Lakes range from
2 to 4 mg/L (Eadie, 1988; Bukata et al.., 1985), from which a best estimate of
24
-------
3 mg/L was selected. The concentration range of nonsettling organic matter
in Great Lakes sediment pore water is 5 to 20 mg/L (Eadie et al.., 1982), from
which a best estimate of 10 mg/L was selected.
The values of Kp0C and fysoM' together with the solids and sediment
parameters from the previous section, allow the calculation of the following
partitioning parameters for the TCDD model:
Parameter Water Column Sediment
K' 290,000
Kdm 95,000
Ku 32,000 32,000
fj 0.70 0
f" 0.24 1.0
fj 0.066 0
VOLATILIZATION AND PHOTOLYSIS
Volatilization and photolysis have been modeled as first-order processes
removing TCDD from the total water column. To parameterize these processes
in the model, the first-order rate constant for each must be determined.
Because no direct measurements of these loss rates are available,
physicochemical parameter estimates are used instead. This section describes
the calculation of rate constants for TCDD volatilization and photolysis.
Lake Ontario Meteorology
The calculations of volatilization and photolysis rates described below
require information characterizing the meteorology of the local environment.
For Lake Ontario, 40 years of daily over-lake meteorologic information has
been synthesized by NOAA's Great Lakes Environmental Research Laboratory.
While developed to estimate lake evaporation and heat budgets (Croley, 1989),
25
-------
this data is also directly useful for estimating hydrophobic contaminant
loss rates.
Volatilization
Volatilization is a mechanism of transport between water and air, driven
by non-equilibrium between the chemical fugacities of the two phases.
Depending upon the direction of the fugacity gradient, this process may
represent either a source or sink of contaminant to/from the water body.
Because this model treats Hyde Park as the only source of TCDD to Lake
Ontario, volatilization was considered as only a loss mechanism. It should
be recognized, however, that atmospheric sources contribute significantly to
the total input of hydrophobic contaminants, such as lead, polycyclic
aromatic hydrocarbons (PAH), PCBs and some dioxins to surface waters (Atlas
et al.., 1981; Bidleman et a].., 1981; Eisenreich et al., 1981; Astle et al..,
1987).
The conventional two-film theory (Liss and Slater, 1974) was used to
calculate the rate of TCDD volatilization. Resistance to volatilization
occurs in both the liquid and gas film at the water-air interface, with the
overall resistance given as:
R T
k = (4 + K Ha T1 = (4 + -Jf1 (Equation 16)
v ^ Kg nlc *i k g
where kv = volatilization rate constant
K-| = liquid film mass transfer coefficient
Kg = gas film mass transfer coefficient
R = gas law constant, 8.319 Pa*m3/(K-mole)
26
-------
Ta - air temperature ("K)
H-jc = Henry's law constant (Pa*m3/mole)
Henry's Law Constant for TCDD has been reported to be in the range of 1.63
10.34 Paซm3/mole at 25'C, with a median reported value of 3.34 (Shiu et
al.., 1988). These values were selected as the range and best estimate for
H-|c. According to Shiu et al.., an increase in water temperature of 16*C
doubles the expected value of the H^c for TCDDs; this temperature-
dependence was included in the calculation of TCDD volatilization rates.
The liquid film mass transfer coefficient, K-|, was calculated from the
correlation of O'Connor (1983) for large bodies of water. According to this
correlation:
1
1
K1 ^ D1 j2/3 k1/3
v
w
(M
r(uj < pw > u*
l
k ZfUJ
+ Pa Va } 1/2
pwvw '
(Equation 17)
where D-j = liquid diffusivity
k = von Karmen's constant, 0.4
Pa, Pw = density of air and water
va' vw = viscosity of air and water
U* = wind shear velocity
Vu*)
U* - u*/u
+ (-4)
*t
(Equation 18)
27
-------
where D-| = liquid diffusivity
k = von Karmen's constant, 0.4
Paป Pw - density of air and water
va, vw = viscosity of air and water
U* = wind shear velocity
1 1 ^1 e"
1 m \ ~ ~2 + ( ^ ) (Equation 18)
ov *' e a
where Ze = 0.15 cm
M - 2
un = 9
U*
(u- + d
r(U*) = rQ () e *c (Equation 19)
where r0 = 2.5
U*c = 6.2
K-j was computed by this formulation using over-lake wind speed and water
temperature derived from daily observations of 1948 - 1985 Lake Ontario
meteorology, obtained from NOAA (Croley, 1988).
The TCDD gas phase mass transfer coefficient, Kg, was calculated by
multiplying the water vapor mass transfer coefficient (K^w) by a ratio of
gas-phase molecular diffusivities of TCDD (Dg) and water (DWg)
(O'Connor, 1985):
28
-------
where Ze = 0.15 cm
Xj = 2
U*t - 9
(Equation 19)
where rQ = 2.5
U*c - 6.2
K-j was computed by this formulation using over-lake wind speed and water
temperature derived from daily observations of 1948 - 1985 Lake Ontario
meteorology, obtained from NOAA (Croley, 1988).
The TCDD gas phase mass transfer coefficient, Kg, was calculated by
multiplying the water vapor mass transfer coefficient (KAW) by a ratio of
gas-phase molecular diffusivities of TCDD (Dg) and water (DWg)
(O'Connor, 1985):
Values of the water vapor mass transfer coefficient were obtained from
computed Lake Ontario evaporative mass transfer coefficients, M
(Pa'mV'K'mol):
K = (D /Dw)0-6 KAw
g g g w
(Equation 20)
(Equation 21)
where Vw = wind speed (m/s)
Ta = air temperature (#K)
29
-------
Kaw and Kg computations were made from daily (1948-1985) lake-average
evaporation estimates, again obtained from NQAA (Croley, 1988).
Daily volatilization rate constants were calculated from K-j and K'g
using Equation 16, then averaged by month. The results, presented in Table
2, indicate that gas film resistance controls the rate of TCDD
volatilization. The annual average volatilization rate constant calculated
by this method was 0.15 m/day. Two other correlation methods, O'Connor's
(1985) and Mackay's (1983) result 1n somewhat higher annual average values of
the volatilization rate constant, 0.38 and 0.20 m/day, respectively. Monthly
values of kv from these two correlations are included in Table 2. A
best-estimate TCDD volatilization rate constant, 0.22 m/day, was obtained by
taking the geometric mean of the three annual average values of kv.
A range for the rate constant of TCDD volatilization was determined by
calculating annual average kv by the three correlation methods, using the
range of Henry's law constants. The resulting kv range is 0.074 - 0.47
m/day.
Photolysis
The rate of photolysis at the water's surface is a function of the
quantum yield (4>) and extinction coefficient (e^) of the contaminant, and the
light intensity (L^) across the contaminant's absorbance spectrum. The rate
constant for photolysis at the water's surface, kpS, is:
kps = I (Equation 22)
Extinction coefficient and light intensity vary with wavelength, hence the
subscripts. incorporates the effects of atmospheric light scattering and
water surface reflection upon the intensity of light entering the water.
30
-------
TABLE 2. VOLATILIZATION RATE COMPUTATIONS
Hlc Vw K-| K'g kw kv kv
MONTH Ta(#c) Tw^#c^ (Pa#m /mole) (nป/s) (m/day) (m/day) (m/day) (O'Connor) (Mackay)
January
-2.5
3.6
1.3
8.0
3.6
.22
.20
.40
.25
February
-2.6
1.7
1.2
7.4
2.9
.17
.16
.35
.21
March
0.35
1.5
1.2
6.7
2.4
.14
.13
.31
.18
April
4.8
2.9
1.3
5.6
1.7
.086
.081
.27
.15
May
8.9
5.7
1.5
4.7
1.2
.059
.056
.26
.13
June
14
12
1.9
4.4
1.3
.075
.071
.31
.15
July
19
18
2.5
4.5
1.8
.13
.12
.41
.19
August
21
21
2.8
5.0
2.5
.20
.18
.48
.23
September
17
18
2.5
5.6
3.0
.22
.21
.50
.26
October
11
12
1.9
6.0
2.9
.19
.18
.43
.23
November
5.5
7.3
1.6
6.8
3.1
.19
.18
.40
.23
December
0.30
4.9
1.4
7.7
3.6
.21
.20
.41
.25
Annual
Average
8.2
9.1
1.8
6.0
2.5
.16
.15
.38
.21
-------
The rate of photolysis over the depth of the water column depends upon
the surface rate as well as the attenuation of light in water. If
essentially all of the light is absorbed within the water column, as is the
case for Lake Ontario, the depth-integrated photolysis rate constant, kp,
is:
kp " "D" 1 "a (Equation 23)
A X
where = light attenuation coefficient
D = depth of the water column
The quantum yield of TCOD was measured as 0.0022 in a 50/50 solution of
water and acetonitrile (Dulin et al.., 1986). Extinction coefficients have
also been measured for TCDD (Podoll gt al., 1986), indicating strong light
absorbance in the UV-B wavelength range (300-340 nm). Tables of day-averaged
solar light intensity, under cloudless conditions, on the four mid-season
days were obtained from Stanford Research Institute (Mill, 1988). The
wavelength-dependent light attenuation coefficient in the open waters of Lake
Ontario has been measured for this project; at 315 nm, was measured as
1.0/meter (Mill, 1988). At this value, 90% of light attenuation and, hence,
photolysis takes place in the top meter of the lake. This value of is in
agreement with UV-B light attenuation measurements reported by Calbins (1975)
for Lake Erie. Using a Robertson-Berger meter with mean response
corresponding to 312 nm, Calbins measured as 1.1/meter. Values of kps
were calculated using Equations 22 and 23, for the four mid-season dates.
Daily values of kp were then obtained by interpolation.
32
-------
The depth-averaged rate constants were adjusted for the effects of
cloudiness upon light intensity, assuming a 50% reduction in UV-B intensity
for fully-overcast conditions (Zepp and Cline, 1977):
kp s (1 - 0.5 fcc) kp (Equation 24)
where k'p = cloud-corrected photolysis rate
fcc = fractional cloud cover
Daily observations of Lake Ontario cloud cover, collected from 1948 to 1985,
were obtained from NOAA's Great Lakes Environmental Research Laboratory
(Croley, 1988). Photolysis rate constants were also adjusted (Ambrose et
al., 1988) to account for difference in light intensity between the reference
latitude (40ฐN) and the latitude of Lake Ontario (42.7*N):
kJ = ฐ-95 kp (Equation 25)
where k"p = effective photolysis rate
Daily values of k"p were computed and then averaged by month. A tabulation
of monthly surface photolysis rate constants, depth-averaged photolysis rate
constants, cloudiness, and effective first-order photolysis rate constants is
presented in Table 3. The resulting "best estimate" annual average rate
constants for TCDD photolysis, as presented in Table 3, is 2.3 x 10"^
day"1.
Uncertainty in the rate of TCDD photolysis may be largely attributed to
the quantum yield. The value of used in the proceeding computations is
based upon a single measurement, in a solution dissimilar to Lake Ontario
33
-------
TABLE 3. PHOTOLYSIS RATE COMPUTATIONS
MONTH
-------
water. Consequently, 0.0022 may be a rather poor estimate of TCDD quantum
yield for this system. Furthermore, the use of the same photolysis rate to
model the loss of dissolved, bound, and sorbed TCDD (an assumption made for
the model) is questionable (Mill, 1988; Zepp, 1988; Zepp and Wolfe, 1987).
To reflect these uncertainties, a wide range of quantum yields were used to
calculate the range of TCDD photolysis rate. A lower value of zero (no
photolysis) and an upper value of 0.5 (observed upper limit for chlorinated
organics) were selected for The resulting range of the photolysis rate
constant is 0 - 0.52/day.
Mean monthly rates of volatilization and photolysis of TCDD in Lake
Ontario are plotted in Figure 4. The rate of TCDD volatilization is
calculated to reach maxima in September and January, while the photolysis
rate maximum occurs in July (corresponding to maximum solar intensity). The
coincidence of the photolysis rate maximum with the thermal stratification of
Lake Ontario, typically from June to October, may reduce the importance of
photolysis as a lake-wide TCDD loss mechanism. The completely-mixed model
assumes that TCDD throughout the water column is available for loss via
photolysis whereas, in a stratified Lake Ontario, only TCDD in the epilimnion
(estimated to be 20% of the lake volume) will be subject to photolysis.
Stratification may similarly reduce the volatilization loss, although to a
lesser extent because the volatilization rate is generally higher during the
unstratified seasons. Results of modeling Lake Ontario as a seasonally-
stratified water body indicate that the effect of stratification may be
neglected if annual average loss rates are used.
35
-------
Figure 4. Estimated Monthly TCDD Volatilization and Photolysis Loss Rates
0.004
0.003
0.002
0.001 -
Volatilization Loss Rate: k^ fjj/D
Photolysis Loss Rate: k "
0.000
1
2
3
7
4
5
6
9
8
10
12
1 1
Month
36
-------
A more significant shortcoming of depth-integrating photolytic loss over
the 86-meter water column is that all the loss is occurring very near the
surface of the lake (Zepp and Cline, 1977). 90% of the loss of TCDD by
photolysis is confined to the top 1% of the lake volume. For this loss to
occur, the flux of TCDD to the lake surface by mixing must be greater than
the "flux" of TCDD photolytic loss. The former may be approximated by K-j,
the liquid film transfer coefficient. The lesser of the two fluxes will
control the overall photolysis rate. Comparison of Tables 2 and 3 indicates
that K"p does, indeed, control the rate of TCDD photolysis. However, the
K"p range upper limit, 0.52/day, exceeds the annual average K-j, 0.029/day
(2.5 m/day * 86 m). The upper limit of photolysis rate constant was,
therefore, lowered to 0.029/day, to reflect the rate limitation of TCDD flux
to the lake surface.
The effect of ice upon TCDD photolysis and volatilization was also
evaluated, because ice cover should reduce these rates. Lake Ontario is
subject to the least ice cover of all the Great Lakes. In normal winters,
ice cover begins late in January with maximum coverage (15%) occurring in
mid-March (Phillips and McCulloch, 1972). Photolysis and volatilization rate
reductions were assumed to be proportionate to ice cover; this probably
overestimated the effect of ice. Assuming 15% ice cover for the months of
February and March, the estimated reduction of annual photolysis and
volatilization rates due to ice was found to be about 2 - 2.5%, and was,
therefore, neglected.
37
-------
LOADING
The objective of the modeling study was to estimate the relationship
between TCDD leakage from Hyde Park and TCDD contamination in Lake Ontario.
Because the model does not distinguish TCDD leaking from Hyde Park from other
sources to the lake, the model was used to relate incremental Hyde Park
leakage to incremental Lake Ontario contamination.
The Hyde Park landfill is located in Niagara Falls, New York,
approximately 2,000 feet from the gorge of the Niagara River, the channel
connecting Lake Erie with Lake Ontario. It is estimated that 630-1500 kg of
TCDD were buried at Hyde Park. The landfill has been leaking chemicals for
the past 35 years (U.S. Environmental Protection Agency, 1988). For modeling
purposes, Hyde Park leakage to the Niagara River and Lake Ontario has been
treated as a constant input. It has also been assumed that TCDD is
transported conservatively to Lake Ontario. Halfon (1986) has suggested
that, for relatively non-volatile and recalcitrant contaminants such as TCDD,
the Niagara River acts only as an area of transition. McLachlan and Mackay
(1987) reached a similar conclusion. Their Niagara River model predicted
that greater than 75% of TCDD in the river would be transported into Lake
Ontario. No significant solids deposition occurs in the river, and the short
travel time from Hyde Park to Lake Ontario precludes significant TCDD losses
by volatilization or photolysis. Therefore, modeling the transfer of TCDD
from its input to the Niagara River at Hyde Park to the river's discharge
into Lake Ontario was not considered necessary.
A summary of the Lake Ontario TCDD model parameters is presented in Table
4.
38
-------
TABLE 4. SUMMARY OF LAKE ONTARIO TCDD MODEL PARAMETERS
PARAMETER BEST ESTIMATE RANGE
Volume (m3)
Surface Area (nr)
Flow (m3/s)
1.68 x loJJ
1.95 x 1010
7100
(Based Upon 1958-1980 data)
Suspended Solids Concentration (mg/L)
Sediment Solids Concentration (gm/L)
Active Sediment Layer Thickness (cm)
Solids Settling Flux (go/m/yr)
Resuspension Flux (gm/nr/yr)
Sedimentation Flux (gm/rrr/yr)
Pore Water Exchange Coefficient (cm/day)
Solids Loading (kg/day)
1.2
270
1.8
440
200
240
.36 ,
1.4 x 107
1.0 - 1.5
130 - 360
1 - 5
Adjust to Balance Solids Flux
88 - 440
110 - 550
0.11 - 1.1
Adjust to Balance Solids Flux
Water Column NSOM Concentration (mg/L)
Pore Water NSOM Concentration (mg/L)
Suspended Solids foc
Sediment Solids foc
Log POC Partition Coefficient (L/kg)
Log Binding Coefficient (L/kg)
3
10
.12
.030
6.5
4.5
2 - 4
5 - 20
.1 - .15
0.026 - 0.035
6.3 - 6.8
4.3 - 4.8
Volatilization Rate Constant (m/day)
.23
.075 - .47
Photolysis Rate Constant (m/day)
.20
0 - 2.5
-------
SECTION 4
MODEL APPLICATION
The Lake Ontario TCOD model was used to perform several kinds of
simulations. The first were simulations of expected steady-state and dynamic
lake responses to contaminant input based upon best-estimate model
parameters. Because the "black box" nature of models inhibits an intuitive
understanding of the significance of model parameters, sensitivity analysis
was used to show how the variation of individual model parameters affected
the predictions. Uncertainty analysis was performed in order to examine the
probabilistic response of the model to simultaneous random variation of the
input parameters. Thus, the deterministic model was converted into a
probabilistic one, allowing a measure of the model's predictive uncertainty
to be made.
STEADY STATE MODEL
The model predicts contaminant concentrations in the lake sediment and
water column in response to contaminant input. Predicted concentrations
eventually reach constant, or steady-state, values when the input is
constant. Model simulations were made to predict the steady-state response
of Lake Ontario to a range of TCDD input, from 0.5 - 1000 gm/year. The
results, for sorbed sediment TCDD and dissolved water TCDD concentrations,
are presented in Figures 5 and 6. These figures show that the predicted
40
-------
response of the lake to TCDD input is a linear function. The solid lines in
Figures 5 and 6 ("Predicted Response") are predictions made with
best-estimate model input parameters. The 95% confidence intervals in
Figures 5 and 6 are based upon the results of model uncertainty analysis,
described below. There is a 95% probability that the model prediction of
steady-state TCDD concentrations will lie within the bounds of these
confidence limits.
Figure 7 provides a steady-state mass balance accounting for TCDD in Lake
Ontario. The arrows in the diagram represent the processes of TCDD fate and
transport in the lake. The percentages associated with each arrow indicate
the magnitude of TCDD mass flux, relative to an Input of 100%, for each model
process. Because UASP4 is linear in its equations, the percentage of the
mass flux attributable to each process is independent of the magnitude of
TCDD input. By summing the TCDD fluxes into and out of each of the three
"boxes" in the diagram (water column, sediment, and whole lake system), one
can verify that TCDD mass is conserved in the steady-state model predictions.
The relationship between water column and sediment TCDD concentrations
predicted at steady state is displayed in Figure 8.
According to the stipulation for Requisite Remedial Technology at the
Hyde Park Landfill (U.S. Environmental Protection Agency, 1987) input of TCDD
from Hyde Park to the Niagara River and Lake Ontario is limited to 0.5
gm/year, the "Interim Aqueous Phase Liquid (APL) Flux Action Limit". The
predicted steady-state TCDD concentrations corresponding to this input are
0.026 ng/kg (sorbed to sediment) and 9.5 x 10"ฎ pg/L (dissolved in water
column). The predicted 95% confidence intervals for these concentrations are
41
-------
Figure 5. Predicted Steady-State Response of Lake Ontario
Sediment to Constant TCDD Input
1000 -a
Predicted Response
95 % Confidence Limits
100 -
CL
o>
o
JO
.01
.001
1
1
10
100
1000
10000
TCDO Input (grams/year)
42
-------
Figure 6. Predicted Steady-State Response of Lake Ontario
Water Column to Constant TCDD Input
10
Predicted Response
- 95% Confidence Limits
1
a.
.1
o>
o.
.01
.001
.0001
0
1
1
10
100
1000
10000
TCDD Input (grams/year)
43
-------
Figure 7. Steady-State Fate and Transport of TCDD in Lake Ontario
Photolysis
38%
Volatilization
31%
Inflow
Outflow
Input
100%
6%
Sedimentation wI
Sedimented Solids
25%
Sediment
Settling w/ Suspended Solids
46%
Resuspension w/ Sedimented Solids
20%
Pore Water Diffusion
1% ป
Water Column
44
-------
Figure 8. Predicted Relationship Between Steady-State
TCDD Concentrations in Water Column and Sediment
TCDD Sorbed to Sediment (ng/kg = ppt)
45
-------
0.0030 - 0.056 ng/kg and 1.9 x 10"ฎ - 1.8 x 10"* pg/L. Because the
steady-state load-response functions are linear, these are also the
incremental concentrations attributable to a 0.5 gm/year input from Hyde
Park, regardless of the total TCOD input to Lake Ontario. The predicted
incremental concentrations of TCDD in Lake Ontario attributable to other
values of Hyde Park input can be obtained from Figures 5 and 6.
DYNAMIC MODEL
Dynamic model simulations were made to estimate the time rate-of-change
of Lake Ontario TCDD concentrations in response to changing input.
Prediction of the time scale for significant changes in water quality to
occur are useful for remedial action decision-making (International Joint
Commission, 1988).
To demonstrate the dynamic behavior of the Lake Ontario TCDD model, the
response to a hypothetical input function was simulated. A constant, 1 kg
TCDD/year was input to the lake until steady-state was reached; then, the
input was cut off (reduced to zero). The predicted sediment and water column
concentrations (Figure 9), display the lake's response to an increase
followed by a decrease in contaminant input. The concentration profiles are
exponential in shape, indicating rapid initial response to a change in input
followed by a more gradual approach to steady-state. Water column
concentrations respond much more rapidly to an input change than sediment
concentrations. 50% of the change in TCDD concentrations, due to either
input or cutoff, are reached in 10 years for sediment and in less than
on-half year for the water column. Steady-state, defined here as 95% of the
46
-------
Figure 9. Predicted Dynamic Response of Lake Ontario
to TCDD Input and Cutoff
1 kg/year input begins
0.20 -|ฃ
TCDD input cutoff
0.15
0.10-
0.05-
0.00
Dissolved in Water
Sorbed to Sediment
40 60
Elapse Time (years)
100
47
-------
total concentration change, is reached in sediment 40 years after the input
change and in 15 years for the water column.
Dynamic model prediction may also be viewed as an input-response
function, Figure 10. This function, shown for different times following
commencement of TCDD input to an initially-clean lake, forms a series of
parallel lines converging upon the steady-state input-response function. For
a lake responding to an input increase, concentrations will be lower than
predicted by the steady-state input-response function. Conversely,
concentrations will be higher for a lake responding to a decrease in
contaminant input.
A caveat should be applied to dynamic model simulations, however.
Because assumptions of complete mixing and annualized process rates are
incorporated in the model, only long-term predictions are expected to be
accurate. Short-term dynamic response to input change may be predominantly
controlled by spatial and seasonal gradients, predictable only with a higher
resolution model.
SENSITIVITY ANALYSIS
Sensitivity analysis was used to examine the influence of individual
model parameters upon model prediction. Sensitivity analysis also provides
some insight into the model's behavior. The sensitivity of the model to an
individual parameter was evaluated by varying that parameter value through
its range, while holding all other parameters constant at their expected
values, and repeatedly running the model. Expected values and ranges for the
Lake Ontario TCDD model parameters are shown in Table 4. The observed change
48
-------
Figure 10. Dynamic Input-Response Function
for Lake Ontario TCDD Input
100
Legend:
5 years following input
10years" "
20years" "
30years" "
a.
10
Steady State
1
1
.01
1
10
100
1000
TCDD Input (grams/year)
49
-------
in model results due to varying a single parameter is, then, a measure of the
"sensitivity" of the model to that parameter.
The following model parameters were included in the sensitivity analysis:
Suspended Solids Concentration
Active Sediment Layer Thickness
Resuspension Flux
Sedimentation Flux
Suspended Solids fQC
POC Partition Coefficient
Volatilization Rate Constant
Photolysis Rate Constant.
Results of the analysis are presented as a series of plots, Figures 11 - 18,
displaying predicted sediment TCDD concentrations in response to a 1 kg/year
input. Variation in suspended solids concentration (Figure 11) affects the
eventual sediment concentration at steady state, while sediment layer
thickness (Figure 12) and resuspension flux (Figure 13) affect only the
kinetics of sediment TCDD accumulation. Sedimentation flux variation (Figure
14) affects both the kinetics and the steady state. The remaining model
parameters, suspended solids foc (Figure 15), POC partition coefficient
(Figure 16), and volatilization (Figure 17) and photolysis (Figure 18) rate
constants, affect the eventual steady-state TCDD concentration.
Sensitivity analysis indicates that steady-state model predictions are
most strongly affected by variation in the photolysis rate constant, followed
by variation in Kp0C, volatilization rate constant, and sedimentation flux.
Sensitivity analysis cannot be readily used to examine model response to
simultaneous parameter variation, nor to relate model uncertainty to
uncertainty in parameter values. These issues are addressed in the next
section.
50
-------
Figure 11. Sensitivity Analysis: Suspended Solids Variation
1 kg/yr TCDD Input
Suspended Solids (mg/l):
10
1.2
1.5
0
40
80
120
160
200
Elapse Time (years)
51
-------
Figure 12. Sensitivity Analysis: Active Sediment Layer Thickness
50 -
oป
it
o>
c
40 -
c
E
ฆo
a
to
30
o
1 kg/yr TCDD Input
Active Sediment Thickness (cm):
ฆo
A
w
O
(0
1.0
1.8
O
a
o
H
5.0
0
40
80
120
200
160
Elapse Time (years)
52
-------
Figure 13. Sensitivity Analysis: Resuspension Flux Variation
1 kg/yr TCDD Input
Resuspension Flux (gm/m2/yr):
S3
200
440
160
200
Elapse Time (years)
53
-------
Figure 14. Sensitivity Analysis: Sedimentation Flux Variation
80
o>
.ae
0>
c
60 -
c
o
E
ฆo
e
<0
40 -
o
1 kg/yr TCDD Input
Sedimentation Hux (gm/m2/yr):
o
a
JQ
k
o
CO
a
a
o
H
240
550
0
40
80
120
160
200
a
Elapse Time (years)
54
-------
Figure 15. Sensitivity Analysis: Suspended Solids foe Variation
1 kg/yrTCDD Input
Suspended Solids foe:
01
0.12
0.15
->I
120
160
200
Elapse Time (years)
55
-------
Figure 16. Sensitivity Analysis: Kpoc Variation
80
60 -
40 -
1 kg/yr TCOD Input
log (Kpoc):
6.3
6.5
6.8
0
40
80
120
160
200
Elapse Time (years)
56
-------
Figure 17. Sensitivity Analysis: Volatilization Rate Constant Variation
o> 60-
c
c
E
TS
CO
40 -
o
1 kg/yr TCDD Input
Volatilization Rate Constant (m/day):
o
0)
0.075
0.23
0.47
Q
O
O
b-
0
40
80
120
160
200
Elapse Time (years)
57
-------
Figure 18. Sensitivity Analysis: Photolysis Rate Constant Variation
100
80 -
60 -
1 kg/yr TCDD Input
Photolysis Rate Constant (m/day):
0.20
2.5
0
40
80
120
160
200
Elapse Time (years)
58
-------
MONTE CARLO SIMULATION OF UNCERTAINTY
Introduction
The uncertainty associated with the modeling results for TCDD in Lake
Ontario can be associated with a number of causes. The analysis discussed
below addresses an important source of uncertainty - that associated with
uncertainties in the value of the model parameters. Other causes of
uncertainty such as the possibility that a flaw exists in the basic model
structure, or that a phenomena of first order importance has been left out of
the model cannot be addressed in a quantitative way. The reason is that they
relate to possibilities which would change the concentrations to an unknown
extent.
The consequences of parameter uncertainty can be analyzed directly since
the model can be used to compute the change in concentration that results
from a change in a parameter value. The procedure is as follows. Consider a
single parameter, e^, and let the concentration that results with this
value be c{0|<}. Assume that we can estimate the range of values:
ek,MIN> 0k,MAX that bound the true value:
ek,MIN - ek 1 ek,MAX (Equation 26)
Then for any value 0|<(i) in the range, we can compute the concentration
that results: cfejji)}. If we can generate a sequence of values:
ฉk(1)> 9k(2), 6k(n), that represent the distribution of possible
parameter values, then we can compute the sequence of concentrations:
c{9|cO)}> c{ejc(2)}, ..., cCo^n)} that represent the uncertainty of the
concentrations due to the uncertainty in 0^. This is the basic idea behind
59
-------
the Monte Carlo method for analyzing the consequences of parameter
uncertainty.
For a model with N uncertain parameters: 0j, 92, 0fl> the
concentration is a function of the N parameters: c{0j, 0 2ป ป Or).
The uncertainty in the concentration is computed by generating a set of
parameters: ej(i), 63(i)ป ป 6j^(i)ป i 0 1, H with each
parameter varying appropriately to reflect its uncertainty. The method
employed to choose the appropriate sets of parameters is discussed below.
Methodology
Choice of Parameter Uncertainty Probability Distribution:
As described above, the Monte Carlo method requires a sequence of
parameter values ฉj(i), 02(i), ป 6|y|(i), i = 1, ..., M, that are used
in the calculation of the concentrations, c{0j, ฉ3, .... 6^}. Consider
a single parameter: 0^. Suppose this parameter is experimentally measured
and a number of values are reported. Then the probability distribution of
the sample mean would represent the uncertainty of the parameter and it would
be the appropriate distribution from which the Monte Carlo samples: 0^0),
i = 1, ..., M, would be generated. The organic carbon partition coefficient,
Koc, is an example of such a parameter for which the uncertainty of the
sample mean is used in the Monte Carlo analysis.
The quantification of the uncertainty of most parameters, however, is
less rigorous. Usually a possible range of values: (ek,MIN* 0k,MAX) is
specified without any further details. If the parameter could be anywhere
within the range with equal probability, then the appropriate choice of
probability distribution for 0^ is a uniform distribution over the range:
60
-------
(0k,MIN' ek,MAX)- The photolysis rate constant, KPHOTO, was initially
treated as having such a uniform probability distribution.
However, if the range is interpreted as the confidence limits of the
parameter, and the actual values are most likely in the middle of the range,
then a probability distribution that is centered in the middle of the range
is preferred. For the calculation presented below, the log normal
probability density function is chosen for these centrally distributed
parameters. The reasons are: (1) for narrow ranges, the log normal density
is approximately the normal density function which is the natural choice; (2)
for parameters with a wide range of uncertainty, the distribution is spread
out logarithmically over the range which centers the parameter at the
geometric mean of the range; (3) the sample parameter values from this
distribution are always positive.
The conversion of the parameter ranges to the parameters of the lognormal
probability density function is made by assuming that the range of the
parameter values: k,MIN* 0k,MAX)> ฎre the 95th percentile confidence
limits of the parameter: e^. The parameter ranges and the resulting
coefficients of variation used in the analysis are listed in Table 5.
Generation Method for Monte Carlo Realizations:
The generation of the sets of parameter values for the Monte Carlo
analysis is accomplished using techniques for generating a set of pseudo
random variables with specified probability density functions. The preferred
method is called the Latin Hypercube Method (Mackay et al.., 1979). It is
equivalent to the inverse probability distribution function method
(Abramowitz and Stegun, 1954). Let F.(e) be the probability distribution
u
61
-------
TABLE 5. UNCERTAINTY
BOUNDS FOR THE
MODEL PARAMETERS
VARIABLE
LOWER BOUND
UPPER BOUND
LOG S.D.
CV
Ml Suspended Sol Ids Concentration (mg/L)
1.0
1.5
0.103
0.104
M2 Sedimented Solids Concentration (mg/L)
130
360
0.260
0.264
H2 Active Sediment Layer Thickness (cm)
1.0
5.0
0.411
0.428
WS Sediment Flux (gm/m2/year)
110
550
0.411
0.428
WRS Resuspension Flux (gm/m^/year)
90
440
0.405
0.422
KL Pore Water Exchange Coefficient (cm/day)
0.11
1.1
0.587
0.642
DOC2 Water Column NSOM Concentration (mg/L)
2.0
4.0
0.177
0.178
D0C2 Pore Water NSOM Concentration (mg/L)
5.0
20
0.354
0.365
FOCI Suspended Solids foc
0.10
0.15
0.103
0.104
F0C2 Sedimented Solids foc
0.026
0.035
0.076
0.076
LOGKOC (1) Log POC Partition Coefficient (L/kg)
6.3
6.8
0.019
0.019
KPHOTO (2) Photolysis Rate Constant (day"*)
0.025
2.5
1.17
1.725
KVOL Volatilization Rate Constant (iq/day)
0.075
0.47
0.468
0.495
(1) Normal Probability Distribution - Standard Error of the mean is used.
(2) Uniform Probability Distribution is used.
-------
function of 0. Then M samples from this distribution can be found by solving
the equation:
F0{0k(i)} = M I 1 (Equation 27)
for 0jc(i), i = 1, M. These samples correspond exactly to the i/M + 1
quantiles of the distribution.
The inverse probability method (Equation 27), is used to generate the M
samples for each of the N parameters. The initial ordering of the parameters
values is from low to high for each parameter. Thus, initially they are
perfectly cross correlated. The cross correlation structure for the
parameter uncertainties is easily specified if the parameter uncertainties
are independent of each other. In this case the ordering of the parameters
is completely random relative to each other. This is achieved by using a
random number generator to sample a uniform distribution (without
replacement) which represents the random ordering to be applied to each
parameter. For any other cross correlation, specification methods are
available that calculate the appropriate ordering (Iman and Conover, 1982).
It should be noted, the at the binding coefficient (K^) was not treated as
an independent parameter, but was calculated as Koc/I00.
Results
Parameter Distributions:
The uncertainty ranges of the parameters used in this analysis are listed
in Table 5. Figure 19 presents the histograms for the photolysis and
volatilization rate constants. Two hundred realizations are used for the
Monte Carlo analysis. The uniform distribution for photolysis indicates that
63
-------
Figure 19. Histograms of the log photolysis rate and the log volatilization rate.
PHOTOLYSIS
1.0-
M-
(Lป
OJ
oe-
&s-
OJt-
u
0.1-
VOLATILIZATION
-AJO -3.4 -2.8 -14
-LS -3.1 -JL7 -U -14 -U
LKPHOT
LKVOLA
LOG 10 REACTION RATE (1/DAY)
64
-------
any value within the range is equally likely. The normal distribution for
the logarithm of the volatilization rate constant is typical of the
distributions for all the other uncertain parameters. Log normal probability
plots for the photolysis, volatilization, sedimentation and sediment mixed
layer depths are shown in Figure 20. Note that the lognormal distributions
are exact - they are a perfect straight line on the lognormal probability
plot. This is the consequence of the inverse method, Equation 27, used to
compute the values. The uniform probability distribution for photolysis rate
produces an S-shaped curve as shown.
The cross correlations, Figure 21, are displayed in scatter plots
(Chambers et al., 1983) which plot the standardized variables: (e^ -
^{QlJVaCeiJ against each other. For any individual plot, the x axis
variable is named above the plot and the y axis variable is named to the
right of the plot. Note the lack of cross correlation as indicated by the
circular distribution of the points in each plot. The insert histograms give
the probability distribution for the log of each parameter.
Base Case Results:
The analysis of the base case uncertainty is shown in Figure 22 as
lognormal probability plots. The water column and sediment total TCDD
concentration (CT1 and CT2), the water column dissolved concentration
(CFREE1), and the sediment sorbed concentration (R2) are shown. These are
steady-state concentrations in response to a 1 kg/year TCDD input. A zscore
of two corresponds to a probability of exceedence of approximately 2.5%. The
horizontal lines on the figures are the concentrations of TCDD that would be
computed if TCDD were a conservative passive chemical (the largest
65
-------
Figure
20.
Log probability plots of the volatilization rate (K-VOLAT), the photolysis
rate (K-PHOTO), the sedimentation velocity (WS), and the depth of the
active sediment layer (H2).
K-VOLAT
K-PHOTO
PROBABILITY (ZSCORE)
-------
Figure 21. Scatter plots of the log volatilization rate (LKVOLA), the log photolysis
rate (LKPHOT), the log water column suspended solids concentration
(LM1), the log sedimentation velocity (LWS), and the log depth of the active
sediment layer (LH2).
LKVOLA
67
-------
Figure 22. Log probability plots of the total water column TCDD concentration (CT1),
the dissolved concentration (CFREE1), the total sediment TCDD
concentration (CT2), and the sediment sorbed TCDD concentration (R2).
WATER COLUMN
0
-1
-2
CT1 (pG/L)
-3 -2
CFREE1 (pG/L)
SEDIMENT
6J
CT2 (pG/L)
R2 (nG/KG)
-3-2-10 1 2 3
-3-2-10 1 ป 3
PROBABILITY (ZSCORE)
68
-------
concentration) and if all decay processes (photolysis and volatilization)
were set to zero. Note that the Monte Carlo analysis indicates that the
median concentration is well below the no reaction concentration (the lower
line) but that the upper end of the range of concentration approaches this
value. The S-shaped distributions for the water column concentrations
reflect the photolysis rate distribution. The straight line behavior of the
sediment concentrations indicates that the distributions are more closely log
normal.
Histograms of the water column dissolved concentration and the sediment
concentration are shown in Figure 23. The skewed distribution of the
concentrations are consistent with the log uniform and lognormal distribution
of the parameters.
A number of statistical methods can be used to analyze the results (Rose,
1982). The probability distribution and cross correlation of the logs of the
concentrations that result can be seen in Figure 24. Total and dissolved
water column concentrations of TCDD are quite highly cross correlated. The
strongest correlation, however, is between rj and the suspended
solids and sediment sorbed TCDD concentrations. The reason for this is well
understood (Di Toro et a].., 1982). The relationship ^ * rj occurs if
the intensity of particle mixing between sediments and overlying water is
larger than the rate of interstitial water mixing. This appears to be the
case for TCDD in Lake Ontario.
The importance of the various sources of variability in generating the
variability of the concentrations is examined in Figure 25 which presents the
cross correlations of the concentrations to the volatilization, photolysis,
69
-------
Figure 23. Histograms of the dissolved water column and sediment sorbed
TCDD concentration for the base case uncertainty.
g
I
g
S
S
l-o-l
0.9-
0.8-
0.7-
0.6-
CK*-
03
WATER COLUMN
r 30
1.0-
09-
0.8-
0.7-
0.8
0J5-
OJtr
OS'
A2
0.1
SEDIMENT
40
8
3
CFREE1
R2
CONCENTRATION (pG/L OR nG/KG)
70
-------
Figure 24. Scatter plots of the logs of the total water column concentration (LCT1), the
dissolved water column concentration (LCFREE1), the suspended solid
sorbed concentration (LR1), the total sediment concentration (LCT2), and
the sediment sorbed concentration (LR2).
LCT1
jyyih
,-0-
<>r-
LCFREE1
jซjL.
.iป %
ฆปV*
..n&r ฆ
1
*
m: "
LR1
r\
, c-1
\
H * '
1 v _
j%
-------
Figrure 25. Scatter plots of the concentrations and the parameters that affect removal
of TCDD. Log of the total water column concentration (LCT1), log
sediment sorbed concentration (LR2), log volatiliztion rate (LKVOLA),
log photolysis rate (LKPHOT) and log sedimentation velocity (LWS).
LCT1
uik
LR2
k sซ#
h1 ฆ
/
LKVOLA
- 'A"
. -V
1 ^ ,m
. .> . '
;
S-.t'N
ฆ . t/
; ป\
-------
and sedimentation parameter values. These three mechanisms are entirely
responsible for lowering the concentration of TCDD below that of a
conservative chemical. Note the inverse correlation between the
concentration in the water column and sediment to the photolysis reaction
rate. This relationship indicates that the uncertainty in this parameter is
strongly related to the uncertainty of the computed TCDD concentration.
By contrast the lack of cross correlation to the other uncertain
parameters, Figure 26, indicates that the uncertainty of the other uncertain
parameters is not significantly contributing to the uncertainty of the steady
state TCDD concentration. The exception is a positive correlation between
log Koc and the sediment concentration, This effect will be more
apparent subsequently.
Importance of Uniform Uncertainty for Photolysis:
The parameter uncertainty distribution assigned to photolysis is a
uniform distribution of log KPHOTO, spanning 0.025 to 2.5 m/day. This range
covers the maximum possible rate consistent with the liquid film transfer
coefficient, K] and a rate which is low enough to be essentially zero. The
uniform distribution is chosen to represent the fact that no particular
reason exists for choosing any value in this range over any other.
As a test of the severity of this assumption, a second case has been
examined for which the range in KPHOTO is treated as the 95% confidence
limits consistent with the other parameter. The results shown in Figure 27
are not dramatically different than the previous results indicating that the
analysis is not sensitive to this particular assumption.
73
-------
Figure 26. Scatter plots of the concentrations and the other uncertain parameters.
Log of the total water column concentration (LCT1), log sediment sorbed
concentration (LR2), log active layer depth (LH2), log of the diffusive
exchange coefficient (LKL), log of the sediment resuspension rate
(LWRS), and log KOC (LGKOC).
LCT1
ซ*r
\ ,
/
LR2
ป
: <
'< "
t *
ฆ "*. ' *
LH2
sji>A
,-^V
..
- v
'.-s -ฆ
ฆ<ฃk- *
' 'V.A.v
LKL
" *
" '.&ฃ*?.-r.
ฆ.. ฆ
'
' .' 1
v..',* .
-v.lt -1
*
# %
->k.
LWRS
'.s-X
I - * V- V *
%-V!ป.rV
% ;* *' I
' '
-
J-
.
ป JiWTi . *
. V
:.'*yv*.
* . .
/
LGKOC
74
-------
Figure 27. Histograms of the dissolved water column and sorbed sediment TCDD
concentration for the base case uncertainty, modified to include lognormal
distribution of photolysis rate uncertainty.
<
a
I.O-i
0.9-
0.8-
0.7-
0.6-
0.5-
0.4-
0.3-
0.2-
0.1-
WATER COLUMN
-40
30
20
10
0.0 04 OA 04
r
04
A
o
1.0-
0.9-
0.8-
0.7
0.6-
0.5-
0.4-
M-
0.2
0.1
SEDIMENT
r40
A
2
3
0 40 SO 120 100 200
CFREE1
R2
CONCENTRATION (pG/L OR nG/KG)
75
-------
Contributions to the Uncertainty:
The cross correlation plots demonstrate that the uncertainty in the
steady state concentrations are due mainly to the uncertainty in the
photolysis rate constant. The Monte Carlo method can be used to directly
examine the consequences of removing uncertainty from this and each of the
other sources sequentially. The method is simply to generate a set of
parameters with the uncertainty of the photolysis rate constant set to zero.
The histograms of the concentrations that result are shown in Figure 28. A
dramatic reduction in uncertainty of predicted steady state concentrations is
evident. The impact of the other important mechanisms can be seen in Figure
29. Volatilization and sedimentation are both inversely correlated to water
column TCDD concentration and Koc is positively correlated to the sediment
concentration. These remaining sources of uncertainty can be sequentially
eliminated. The three sets of histograms, Figure 30, correspond to setting
the volatilization rate, sediment velocity, and log Koc uncertainties to
zero successively. The bottom histograms, then, correspond to the
uncertainty generated by all the other parameters in Table 5.
A summary of these results is shown in Figure 31. The dissolved water
column (top) and sorbed sediment (bottom) TCDD concentrations are shown. The
solid bar corresponds to the median of the distribution. The unfilled bars
correspond to the 2.5 and 97.5 percentile concentrations of the
distribution. The values are listed in Table 6. Note that the largest
contributor to the uncertainty is the photolysis rate. Removing its
uncertainty (labeled No Photo) decreases the confidence limits. Reducing
volatilization and sedimentation velocity uncertainty further reduces the
76
-------
Figure 28. Histograms of the dissolved water column and sorbed sediment TCDD
concentration for the case of no uncertainty in photolysis rate.
s
as
2
1.0-
0.9-
(K8
0.7
0.ซ-
0.5-
Ou4-
as
0.2
0.1-
WATER COLUMN
A
BO
40
SO
20
10
1.0-
0.9-
OJB-
0.7'
0*
OJ-
0A-
0.3
O*
0.1'
SEDIMENT
0.0 04 0.4 0.0 (K8
O
8
80 120 ISO 200
CFREE1
R2
CONCENTRATION (pG/L OR nG/KG)
77
-------
Figure 29. Scatter plots of the concentrations and the parameters affecting TCDD
removal and partitioning, with zero photolysis uncertainty. Log of the total
water column concentration (LCT1), log sediment sorbed concentration
(LR2), log volatilization rate (LKVOLA), log sedimentation velocity (LWS),
and log KOC (LGKOC).
LCT1
,' v Cr .
LR2
ฆ v .. -
.jL
:
* '
' : -ii
ฆv%v-
' r-/':
-
LKVOLA
vV.
'.i w.*.
* .
" .
.
*
K 1.
LWS
. -jfN
k"v.
".V1 #
s
ฃ '
ปSฃjfc
. -
' -
ป
LGKOC
78
-------
Figure 30. Histograms of the dissolved water column and sediment sorbed TCDD
concentration for no volatilization uncertainty (top), and in addition no
sedimentation velocity uncertainty (middle), and in addition no KOC
uncertainty (bottom).
WATER COLUMN
S
2
80
-TO
60
SO
40
-30
20
-10
OA
0.6
r
as
o
s
1.0-
0>
04-
0.7
04-
0.5-
0>
04-
0.2-
0.1
SEDIMENT
ง
3
SO 120 160 200
CFREE1
R2
b
z
a
o
E
s
Z
S
1.0-
0.9-
OJ-
0.7-
0.6-
0.5-
0^4-
0.3-
0.2-
0.1-
WATER COLUMN
u
r60
ฆ70
-60
60
r*o
-30
20
10
0.0
OJ
OA
0.6
r
04
i-ซh
04-
0.7-
0.6-
04-
0*4-
04
0.2
0.1
SEDIMENT
C
c
-10
CFREE1
R2
t
Z
a
o
s
<
o
z
<
z
8
B
1.0-
0J9-
0.8-
0.7-
o.a-
0.5-
OA-
0.3-
0.2-
0.1-
WATER COLUMN
c
o
r 140
- 120
100
-80
60
40
20
0.0 04
I
0.4
I
0.6
1.0-
04-
04-
0.7-
04
04-
OA-
04-
04-
0.1
r
04
SEDIMENT
I
120
100
80
60
40
f-20
80
120 160 200
CFREE1 "2
CONCENTRATION (pG/L OR nG/KG)
79
-------
Figure 31. Median (filled), 2.5 and 97.5 percent confidence limits (unfilled) for
the dissolved water column and sediment sorbed TCDD concentrations.
The base case with uniform distribution of log photolysis rate uncertainty
(BASE CASE UN), the base case with lognormal distribution of
photolysis rate uncertainty (BASE CASE LN), and with the successive
removal of uncertainty in photolysis (NO PHOTO), volatilization (+ NO
VOLAT), and sedimentation velocity (+ NO WS).
00
o
\
0
a
2
D
-I
O
0
cc
ILI
1
o
\
(5
a
Ui
2
5
Ul
CO
150
100
50
<&
G*
&
&
se
~ CFREE1U
ฆ CFREE1M
~ CFREE1L
I
I
1 T" I I
J
J
I
I
I
rf
-
~ R2U
ฆ R2M
~ R2L
-------
TABLE 6. RESULTS OF THE UNCERTAINTY ANALYSIS
PERCENTILES 2.5 50 97.5
Dissolved Mater Column Concentration fpq/L)
BASE CASE 0.36 0.13 0.038
(Uniform log KPHOTO distribution)
BASE CASE 0.36 0.15 0.042
(Lognormal KPHOTO distribution)
NO PHOTOLYSIS UNCERTAINTY 0.27 0.17 0.10
+ NO VOLAT UNCERTAINTY 0.23 0.18 0.11
+ NO WS UNCERTAINTY 0.22 0.18 0.13
+ NO Koc UNCERTAINTY 0.21 0.18 0.15
Sorbed Sediment Concentration fpq/q)
BASE CASE 110 35 5.9
(Uniform log KPHOTO distribution)
BASE CASE 120 41 7.4
(Lognormal KPHOTO distribution)
NO PHOTOLYSIS UNCERTAINTY 92 47 18
+ NO VOLAT UNCERTAINTY 87 48 22
+ NO WS UNCERTAINTY 88 48 22
+ NO Koc UNCERTAINTY 59 51 42
81
-------
uncertainty of the resulting concentrations. It is interesting to note that
the sediment concentration is consistently more uncertain than the water
column concentration although the effect is slight. The reason is that Koc
uncertainty affects sediment concentration uncertainty more than the water
column. Once it is removed, the uncertainty is virtually the same in both
compartments.
Results of model uncertainty analysis are summarized in Table 6. The
2.5, 50, and 97.5 percentiles correspond to the upper 95% confidence limit,
mean, and lower 95% confidence limit of the predicted concentration
distributions.
COMPARISON TO DATA
Results of the steady-state model may be interpreted in two ways. The
input/response functions of Figures 5 and 6 may be considered predictions of
incremental lake response to Hyde Park input, or of total lake response to
all TCDD loads. The latter interpretation allows model predictions to be
compared to data.
The steady-state model results were compared to available data of TCDD
concentrations in Lake Ontario. Fifty-five sediment samples were collected
at 49 locations throughout the lake as a part of the Lake Ontario
Bioaccumulation Study (due to practical difficulties, water column samples
were not collected). Sediment sampling results are displayed in Figure 32.
An average TCDD sediment concentration of 110 ng/kg was determined from
analysis of 30 surficial sediment core samples from lake deposition basins.
This concentration corresponds to a steady-state TCDD loading from all
sources of 2.1 kg/year (Figure 5). No measurements of TCDD loading to Lake
82
-------
Figure 32. Lake Ontario TCDD Concentrations (ppt) in top 3 cm of Sediment Cores
NS
NS
47
ro
to
NS
110
oo
co
NS
140
44
180
^ 170
NS
140
190
87
52
140
190,
170
220
NS
170
ฆ : Deposition Zone Sample
ND: Not Detected (< 1 ppt)
NS: No Sample Retrieved
-------
Ontario are available. However, the model estimate of steady-state TCDD
loading to Lake Ontario may be compared with two estimates from the
literature. Neely (1985) estimated 2.4 kg/year total TCDD load to Lake
Ontario based upon a model incorporating fish bioconcentration. Hallet
(1985) estimated a TCDD load of .005 - .05 kg/year would result in observed
Lake Ontario salmonid concentrations, by assuming TCDD to be as persistent as
PCBs in the Great Lakes.
84
-------
REFERENCES
Allan et al.. (1983) J. Great Lakes Res., 9(2):111.
Abramowitz and Stegun (1954) Handbook of Mathematical Functions. U.S.
Government Printing Office, Washington, D.C.
Ambrose et al.. (1983) "User's Manual for Chemical Transport and Fate Model
(TOXIWASP)". U.S. EPA, Athens, Georgia. EPA-600/3-83-005.
Ambrose (1987) Journal Environ. Eng., 113(4):703.
Ambrose et al- (1988) "WASP4, A Hydrodynamic and Water Quality Model".
Athens, Georgia. EPA/600/3-87/039.
Astle et al. (1987) "Lake Sediments as Historic Records of Atmospheric
Contamination by Organic Chemicals" in "Sources and Fates of Aquatic
Pollutants". Adv. Chem. Ser. 216, American Chemical Society.
Atlas et al. (1981) Science, 211:163.
Aubert and Richards (1981) "IFYGL-The International Field Year For the Great
Lakes". NOAA, Ann Arbor, Michigan.
Baker et al. (1986) Environ. Sci. Technol., 20(11):1136-
Bidleman et al. (1981) J. Mar. Res., 39:443.
Bierman and Dolan (1981) J. Great Lakes Res., 7(4):409.
Bukata et al- (1985) "Application of Direct Measurements of Optical
Parameters to the Estimation of Lake Water Quality Indicators",
Scientific Series No. 140, Canada Centre for Inland Waters, Burlington,
Ontario.
Burkhard and Kuehl (1986) Chemosphere, 15:163.
Burns et al- (1982) "Exposure Analysis Modeling System (EXAMS): User Manual
and System Documentation". U.S. EPA, Athens, Georgia. EPA-600/3-82-023.
Calbins (1975) "Impacts of Climatic Change on the Environment, CIAP Monograph
5". Dept. of Transportation, Washington, D.C.
85
-------
Carter and Suffet (1982) Environ. Sci. Technol., 16:735.
Chambers et a].. (1983) Graphical Methods for Data Analysis. Duxbury Press,
Boston, Massachusetts.
Chiou et al. (1979) Science, 206:831.
Chiou et al. (1987) Environ. Sci. Technol., 21:12:1231.
Connolly and Winfield (1984) "A Users Guide for WASTOX, Part 1". U.S. EPA,
Gulf Breeze, Florida. EPA-600/3-84-077.
Connolly et al. (1987) "A Model of PCB in the Water, Bed, and Food Chain of
Lake Ontario". Prepared for International Joint Commission, Windsor,
Ontario.
Croley (1988) Personal Communication and Data Transfer.
Croley (1989) "Verifiable Evaporation Modeling on the Laurentian Great
Lakes". Great Lakes Environmental Research Laboratory, Ann Arbor,
Michigan. GLERL Contribution No. 631.
Czuczwa and Hites (1984) Environ. Sci. Technol., 18:6:444.
Czuczwa et al. (1985) Chemosphere, 14:623.
Di Toro and Connolly (1980) "Mathematical Models of Water Quality in Large
Lakes, Part 2: Lake Erie". EPA-600/3-80-065.
Di Toro et al. (1981) "Analysis of Fate of Chemicals in Receiving Waters".
Prepared for Chemical Manufacturing Association, Washington, DC.
Di Toro et al. (1982) "Simplified Model of the Fate of Partitioning Chemicals
in Lakes and Streams" in "Modeling the Fate of Chemicals in the
Environment", pp. 165-190. K.L. Dickson, A.W. Maki, Jr., and J. Cairns
(eds.). Ann Arbor Science, Ann Arbor, Michigan.
Di Toro et al. (1983) "Water Quality Analysis Simulation Program (WASP) and
Model Verification Program (MVP)". U.S. EPA, Duluth, Minnesota.
Di Toro et al. (1985) Chemosphere, 14:10:1503.
Dolan and Bierman (1982) J. Great Lakes Res., 8(4):676.
Dulin et al. (1986) Environ. Sci. Technol., 20:1:72.
Durham and Oliver (1983) J. Great Lakes Res., 9(2):160.
Eadie et al. (1982) Chemosphere, 11:847-858.
86
-------
Eadie et al. (1983) "The Cycling of Toxic Organics in the Great Lakes".
GLERL, Ann Arbor, Michigan, NOAA Technical Memorandum ERL GLERL-45.
Eadie et al. (1988) Manuscript Submitted to Environ. Sci. Technol.
Eadie (1988) Personal Communication.
Eisenreich et al. (1981) Environ. Sci. Technol., 15:30.
Esposito et al. (1980) "Dioxins". U.S. EPA, Cincinnati, Ohio. EPA-600/2-
80-197.
Gschwend and Wu (1985) Environ. Sci. Technol., 19(1):90.
Halfon (1983) Ecological Modeling, 22:213.
Halfon (1986) Ecological Modeling, 33:13.
Hallet (1985) Chemosphere, 14(6):745.
Hassett (1988) Personal Communication.
Iman and Conover (1982) A Distribution-Free Approach to Inducing Rank
Correlation Among Input Variables. Commun. Statist.-Simul. Comput.,
11(3):311-334.
International Joint Commission (1988) "Report on Modeling the Loading-
Concentration Relationship for Critical Pollutants in the Great Lakes".
Report to the Great Lakes Water Quality. Windsor, Ontario.
Jackson et al. (1985) "Leaching Potential of 2,3,7,8-TCDD in Contaminated
Soils". EPA-600/9-85-013.
Jaffe and Hites (1986) Environ. Sci. Technol., 20(3):267.
Jerome et al. (1983) J. Great Lakes Res., 9:60-68.
Johnson (1982) "Aquatic Fate Process Data for Organic Priority Pollutants".
EPA Final Report on Contract 68-01-3867, Washington, DC.
Kaiser (1983) Chemosphere, 12:1159.
Karickhoff et al. (1979) Water Research, 13:241.
Kemp and Harper (1976) J. Great Lakes Res., 2:324.
Kenaga (1980) Environ. Sci. Technol., 14:553.
Kenaga and Goring (1980) ASTM Third Aquatic Toxicology Symposium.
Philadelphia, Pennsylvania.
87
-------
Lee and Swartz (1980) "Biological Processes Affecting the Distribution of
Pollutants in Marine Sediments. Part II. Biodeposition and
Bioturbation" in "Contaminants and Sediments - Volume 2". Ann Arbor
Science, Ann Arbor, Michigan.
Liss and Slater (1974) Nature, 247:181.
Lodge and Cook (1988) Chemosphere (in press).
Lyman et al. (1982) "Chemical Property Estimation Methods". McGraw-Hill Book
Company, New York.
Mackay et a].. (1979) A Comparison of Three Methods for Selecting Values of
Input Variables in the Analysis of Output From a Computer Code.
Technometrics, 21(2):239-245.
Mackay and Yeun (1983) Environ. Sci. Technol., 17:211.
Marple et al. (1986a) Environ. Sci. Technol., 20:2:180.
Marple et al. (1986b) Environ. Sci. Technol., 20:4:397.
Marple et al. (1987) ACS Symposium Series No. 338, "Solving Hazardous Waste
Problems: Learning From Dioxins".
McLachlan and Mackay (1987) "A Model of the Fate of Organic Chemicals in the
Niagara River". Institute of Environmental Studies, University of
Toronto, Ontario, Canada.
Mill (1988) Personal Communication.
Miller et al. (1987) Environ. Sci. Technol., 21(12):1164.
Morehead et a]_. (1986) Chemosphere, 15:403.
Neely (1985) "A Model for Estimating the Loading Rate of TCDD to the Great
Lakes" in "Aquatic Toxicology and Hazard Assessment: Eighth Symposium".
ASTM STP 891, Bahner and Hansen (eds.), American Society for Testing and
Materials, Philadelphia, Pennsylvania.
O'Connor (1983) Journal of Environmental Engineering, 109:3:731.
O'Connor et al. (1983) Journal of Environmental Engineering, 109:2:396.
O'Connor (1985) "Modeling Frameworks, Toxic Substances Notes". Manhattan
College Summer Institute in Water Pollution Control.
Oliver and Charleton (1984) Environ. Sci. Technol., 18(12):903.
Perkow et al. (1980) "An Exposure and Risk Assessment for 2,3,7,8-TCDD".
Office of Water Regulations and Standards, USEPA, Washington, DC.
88
-------
Phillips and McCtilloch {1972} "The Climate of the Great Lakes Basin".
Environment Canada, Toronto, Ontario.
Podol1 el al. (1986) Environ, Sci, Technol., 20:5:490.
Quinn and Kelley {1983} "Great Lakes Monthly Hydro!ogic Data". NOAA Data
Report ERL 6LERL-26, Ann Arbor, Michigan.
Richardson {1982} "Dynamic Mass Balance of PCS and Suspended Solids in
Saginaw Bay" in "Physical Behavior of PCBs in the Great Lakes". Ann
Arbor Science, Ann Arbor, Michigan.
Robbins (1935) "Coupled Lakes Model for Estimating Long-Term Response of
Srest Lakes to Time-Dependent Loadings of Particle-Associated
Contaminants". NOAA Technical Memorandum ERL GLERL-S7, Ann Arbor,
Michigan.
Robbins (1988} Personal Communication.
Robertson and Jenkins (1978) Ambio., 7:106.
Rodgers gt $1. (1988} "LTI Toxics Model Application: PCBs in Lake Ontario".
Report for the International Joint CoBtmission, Windsor, Ontario,
Rosa (1985} J. Great Lakes Res., 11:13.
Rose (1982} "A Simulation Comparison and Evaluation of Parameter Sensitivity
Methods Applicable to Large Models" in "Analysis of Ecological Systems:
State of the Art in Ecological Modeling", pp. 129-140. W.K. Lauenroth,
G.V. Skogerboe and M. Flug (eds.). Elsevier Science Publishing Co.
Sarna ft &L. (1985) Chemosphere, 13:975.
Shiu et tl. (1988) Environ. Sci. Techno!., 22:6:651.
Simons et. al. (1985) J. Great Lakes Res., 11:423.
Smith et al. (1981) Chemosphere, 10:3:281.
Thoitiarm et a]_. (1975) "Mathematical Modeling of Phytoplankton in Lake
Ontario, Part IV Report for U.S. EPA, Corvallis, Oregon. EPA-600/3-75-
005.
Thomann gt 3I. (1976) "Mathematical Modeling of Phytoplankton in Lake
Ontario, Part 2". Report for U.S. EPA, Duluth, Minnesota. EPA-600/3-76-
065.
Thoitiann et ai. (1979) "Verification Analysis of lake Ontario and Rochester
Embayroent Three Dimensional Eutrophication Models". Report for U.S. EPA,
Grosse lie, Michigan. EPA-600/3-79-094.
89
-------
Thomann and Fitzpatrick (1982) "Calibration and Verification of a
Mathematical Model of the Eutrophication of the Potomac Estuary". Report
for Department of Environmental Services, Government of the District of
Columbia, Washington, D.C.
Thomann and Di Toro (1983) J. Great Lakes Res., 9(4):474-496.
Thomann and Connolly (1984) Environ. Sci. Technol., 18:2:65.
Thomas et al.. (1972) Journal of Sedimentary Petrology, 42:1:66.
U.S. Environmental Protection Agency (1987) "Lake Ontario TCDD
Bioaccumulation Study, Volume 1: Study Summary and Background".
U.S. Environmental Protection Agency (1988) "Report on Project Planning for
the Green Bay Physical-Chemical Mass Balance and Food Chain Models".
Large Lakes Research Station, Grosse lie, Michigan.
U.S. Environmental Protection Agency (1989) "Lake Ontario TCDD
Bioaccumulation Study Final Report"
Voice and Weber (1985) Environ. Sci. Technol., 19(9):789.
Walters and Guisseppi-Elie (1988) Environ. Sci. Technol., 22:7:819.
Webster et al. (1986) Chemosphere, 15:9.
Wetzel (1985) "Limnology". Saunders College Publishing, Philadelphia,
Pennsylvania.
Zepp and Cline (1977) Environ. Sci. Technol., 11:4:359.
Zepp and Wolfe (1987) in "Aquatic Surface Chemistry", W. Stumm (ed.), John
Wiley and Sons.
Zepp (1988) Personal Communication.
90
-------
APPENDIX A
WORKPLAN
5th Draft - May 10, 1988
MODELING TRANSPORT AND EXPOSURE OF TCDD IN LAKE ONTARIO
U.S. Environmental Protection Agency
Office of Research and Development
ERL-Duluth
Large Lakes Research Station
Prepared at the Request of Region II--Superfund
I. GOAL
Modeling will be used to set the APL Flux Action Level for the Hyde Park
landfill based upon maximum permissible levels of TCDD in Lake Ontario
sediment and fish.
A. Objectives:
1. Develop and apply mathematical models for Lake Ontario which will
simulate concentrations of 2,3,7,8-TCDD in sediment and water at
various space scales ranging from whole-lake to discrete Take
segments.
2. Given the allowable TCDD concentration in fish and the
bioaccumulation factor determined from laboratory and field BAF
studies, the APL Flux Action Level for Hyde Park will be
determined, based upon the modeled equilibrium lake sediment
concentration.*
*It is recognized that there are numerous sources of TCDD to Lake Ontario.
91
-------
3. The detailed goals of the modeling project are as follows:
a. Modeling will be used to predict sediment and water
concentrations of TCDD as a function of Hyde Park landfill
discharge.
b. Determine the uncertainty associated with model predictions.
This will include the investigation of both the uncertainty
arising from uncertain and variable model parameters and the
range of modeled system response, i.e., best-case,
worst-case, as well as expected behavior.
c. Assess the relative importance of the various processes of
TCDD transport, transformation and fate. Which are
controlling in Lake Ontario?
d. Determine the response time of Lake Ontario to a cessation
of TCDD inputs. How long before sediment concentration
declines to a level resulting in acceptable levels of TCDD
in fish? This determination will provide the time framework
of interest for modeling.
PROJECT PLAN
A. Background
Lake Ontario is a valuable resource for both the United States and
Canada and to the State of New York and Province of Ontario. It is
the last of the series of the Great Lakes. In the 1970's the major
water quality concern on the Great Lakes was eutrophication. Major
international research studies were conducted on Lake Ontario to
obtain data and develop models to understand cause and effect
relationships which eventually led to controls of phosphorus
discharges.
During this research it became evident that the lake was also
impacted by many toxic substances. With later concerns for toxics
emanating from abandoned waste sites along the Niagara River,
questions have arisen as to mitigative actions necessary to reduce
and eliminate the impacts of the discharges from these sites on the
ecosystem of Lake Ontario.
Particularly, in response to regulatory negotiations related to
toxic discharges from Hyde Park to the Niagara River and Lake
Ontario, the question of the impact of TCDD has arisen. Field and
laboratory studies have been initiated which will provide some
insight into TCDD transport in Lake Ontario and subsequent biotic
exposure in the lake.
92
-------
Mathematical modeling and simulation were selected by
Bioaccumulation Study participants to provide the relationship
between TCDO input and resulting concentrations in Lake Ontario.
Model simulations, based on available field and laboratory data,
will be used to estimate the TCDD concentrations in sediment and
water which result from an incremental load to the Niagara River.
6. Scope
1. This project will deal with modeling 2,3,7,8-TCDD in Lake
Ontario.
2. The project began in February 1988 and will extend to November
1988, the RRT Stipulation date. A follow-up period to March
1989 will be allowed for submission of a final report to Region
II.
3. The geographical scale of the model simulations and predictions
will be geared to the questions asked. At a minimum the lake
will be considered as a completely mixed flow (CMF) reactor, and
exposure concentrations will be calculated on a lake-wide basis.
The lake can also be divided into multiple segments, in order to
provide simulation of the spatial distribution of TCOD
concentrations.
C. Approach
1. Models will be used to simulate expected TCOD concentrations in
water and sediment.
2. Evaluations will be made on the expected concentrations in water
and sediment, first for average conditions in Lake Ontario and
later for various regions of the lake, depending on the needs of
the Lake Ontario TCDD Bioaccumulation Study.
3. This modeling effort will build upon research already completed
as a result of Lake Ontario eutrophication modeling and the
EPA/ORD Great Lakes toxic substance modeling research program at
ERL-Duluth/LLRS, the WASP models developed by Manhattan College,
and generic exposure assessment models developed at ERL-Athens.
The models are based upon a physicochemical mass balance
framework, with an emphasis upon predicting sediment
concentrations. Model predictions will be related to fish
concentrations using the BAF determined from laboratory and
field study.
4. A LEVEL I model will be developed in-house at the LLRS. This
model will use an existing framework developed for Lake Ontario
toxics modeling (Connolly and Oi Toro, "A Model of PCB in the
Water, Bed and Food Chain of Lake Ontario", Prepared for EPA
Region V, as a part of the IJC Task Force on Chemical Loadings).
93
-------
This model was developed by Manhattan College under an EPA
cooperative agreement, managed at the Large Lakes Research
Station and Great Lakes National Program Office. An updated
version of the WASP model, WASP-4, will be used for modeling of
TCDD in Lake Ontario. The Level I model will consider the lake
to be a completely mixed reactor and all simulations will
provide lake-wide average concentrations of sediment and water.
Model coefficients for solids transport, partitioning,
volatilization, photolysis, etc., will be based on most recent
experimental evidence. These will include actual data values,
such as sediment radionuclide profiles used to determine solids
settling and resuspension rates, and estimates based upon
physical-chemical parameters of TCDD. Literature searches will
be conducted to obtain information necessary to parameterize
TCDD in the model, and the assistance of environmental process
and modeling experts will be solicited. If a range of values
are suggested, then the problem will be "bracketed" by model
simulations for upper and lower parameter estimates as welT as
by the "most probable" values.
It will be necessary to make a number of modeling assumptions
for this work. We will model the Hyde Park landfill discharge
to Lake Ontario as a single source originating from the mouth of
the Niagara River. Other sources contributing TCDD to Lake
Ontario will be neglected in modeling, since accurate lake-wide
TCDD loading estimates are unavailable. Model simulations will
provide sediment and water concentrations resulting from an
incremental input of TCDD. In other words, the model will
predict concentrations due to a single load entering the lake.
No losses of TCDD in the Niagara River will be considered, again
due to a lack of data. However, loss mechanisms such as
volatilization and sediment burial are not expected to
significantly reduce the mass of TCDD carried by the Niagara
River into Lake Ontario.
5- A LEVEL II model will be developed by dividing the lake water
and underlying sediment layers into segments, to provide
simulation of the TCDD distribution in different regions of the
lake. The physical transport simulation will be based upon the
Lake Ontario segmentation used in the "LAKE 3" eutrophication
model (Thomann et a].., "Verification Analysis of Lake Ontario
and Rochester Embayment Three Dimensional Eutrophication
Models", ERL Duluth, ORD, USEPA, 1979). This work will be
carried out by the modeling team at the Large Lakes Research
Station, with technical assistance from Dominic Di Toro.
Model segmentation will be an iiqportant consideration at this
level of the project. The appropriate segmentation will depend
upon factors of the system, including:
94
-------
* Lake morphology and circulation pattern
* Expected concentration gradients
* Zones of solids deposition within the lake.
Model segmentation will be related to lake segments used in the
Bioaccumulation Study sampling design for sediment and fish.
Finally, the results of eutrophication modeling in Lake Ontario
may be valuable here, because phytoplankton and suspended solids
are strongly associated, and the partitioning of TCOD onto
solids is significant.
6. The need for a LEVEL HI model will be investigated as a tool to
estimate the "near field" concentrations in water and sediment
near at the mouth of the Niagara River. Whether this level of
modeling is conducted will depend upon the need for
highly-localized simulation of TCDO concentrations.
D. Project Management
The modeling project will be directed from the Large Lakes Research
Station, Grosse He, Michigan, under the guidance of the Project
Manager, Willi am Richardson, LLRS Station Chief. Project direction
and communications will be handled by the Project Manager at LLRS
with assistance from the in-house and on-site contractor staff. A
team of modelers and toxic substance exposure experts will be
assembled to provide additional technical direction and advice. All
modeling work, however, will be done or directed through the LLRS at
Grosse He, Michigan. A monthly status report will be submitted to
Region II.
1. Modeling
The development and application of models will be performed by
on-site personnel at LLRS. Model simulations will be run on the
LLRS Micro-VAX computer. Computer graphics will be developed
during this project, to improve the presentation of modeling
results.
2. Expert Panel
An expert panel will be assembled to assist in providing
scientific direction and credibility to the project. The panel
will suggest detailed processes, rates, coefficients, review
model code and output and review the final report. The panel
will be made up of experts in the following fields:
* General Water Quality Modeling
* Physicochemical Processes
* Sorption Dynamics
* Particle Transport (Sett!ing/Resuspension/Sedimentation)
* Biological-Chemical Interactions
95
-------
The "TEAM" will consist of the persons who will be doing the
"hands-on" modeling. The "PANEL" will include the team members
and the technical advisers who will be retained for their
national expertise in the fields of environmental modeling and
physicochemical processes.
The TEAM will present a detailed approach for initial review by
the PANEL. The PANEL will review the approach and provide a
critique with specific advice on improvements in model
framework, process equations, process rates, etc. The general
water quality modeling expert will review all model code as a QA
check.
3. Reports
In addition to the monthly reports a final project report will
be prepared. A Draft final report will be prepared for November
1988 to comply with the Requisite Remedial Technology
stipulation date. The Draft final report will include all
modeling results and necessary related information. An
unofficial period to March 1989 will be allowed for the LLRS to
produce the final, camera-ready report, complete model
documentation and quality control, and prepare any final
accounting reports.
4. Modeling Workshop
An intensive, three-day modeling workshop will be conducted at
the LLRS in July 1988. At this workshop, the expert panel will
review the work accomplished by the modeling team and, together
with the team, will discuss and investigate means of improving
the modeling effort. The LLRS computer facilities will be made
available to the participants, so that models can be accessed
and run during the workshop. This degree of interaction between
the modelers, experts and the models themselves should maximize
the productivity of the workshop.
96
-------
APPENDIX B
EXPERT PANEL CONTRIBUTION
A modeling expert panel was assembled for the Lake Ontario TCDD modeling
project. Members, and their areas of expertise, were:
Dominic Di Toro, Ph.D. - Water Quality Modeling and Statistical Analysis
Brian Eadie, Ph.D. - Great Lakes Solids and Organic Carbon Modeling
Samuel Karickhoff, Ph.D. - Hydrophobic Partitioning
James Martin, Ph.D. - Water Quality Modeling and WASP4
Theodore Mill, Ph.D. - TCDD Photolysis and Volatilization
The role of the expert panel was to provide peer-review of the TCDD
model development and application. This included reviewing model formulation
and parameterization, identifying/providing additional information and
suggesting improvements to the model, its application, and this report. The
expert panel was not involved in the design or conduct of other elements of
the Lake Ontario TCDD Bioaccumulation Study.
The expert panel was convened at the Large Lakes Research Station on two
occasions, July 18-10, 1988 and January 12, 1989. Three teleconferences were
also held with the expert panel.
97
-------
APPENDIX C
SAMPLE WASP4 INPUT AND OUTPUT
98
-------
TCDD IN 2-SEGMENT LAKE ONTARIO / WASP-4 MODEL
LOADING / DEPURATION SIMULATION
KSIM NSEG NSYS I CRD MFLG IDMP NSLN INTY ADFC DD HHMM
0
1
2
5.
73.
73.
0 0
2 0
0
1
1
1.95E10
2
4.7E-10
0 1
1 0
1.0
1
2
1 3
1
2
1.0
2
7100.
0
3
1
1.95E10
2
1.16E-5
1
1.95E10
2
2.349E-11
1
1.95E10
2
2.819E-11
0 0
2
1.0
02 0
73000.
1825.
56575.
1 0000
A:MODEL OPTIONS
12 1
365.
365.
3650.
62050.
+
1.000
1.000
1
2
2
0.0
2
0.0
2
1.0
1 2
22.09
2 2
270000.0
1
0.00274
1 4
1.0
0
0
5
R 5
0
0
,0
1
+
1.0
1.0
010
0.
30.
1.0
2
0
1
0.
4.7E-10
+
1
3
73000.
+
1.677E12
3.51E08
++++++++++++++++
.0
1.0
7100.
1 0
73000.
1.0
2
0. 1.16E-5
1
0. 2.349E-11
0
0. 2.819E-11
+ +
+
1.0
0.
+
1.0
0.
0.
1.0
+
0.
+
+
+
1.0
0.0
0.0
+
22.09
73000.
73000.
73000.
+ +
73000.
73000.
+ +
73000.
3650.
3650.
18250.
73000.
7300.
54750
+ + + +
(surface water)
(pore water)
B: EXCHANGES
+ C: VOLU+MES
1.0
1.0
0.0
0.0
++++++++++++
86.
0.018
D: FLOWS
0.0
0.0
.(pore water)
.. (settling)
.(resuspension)
..(deposition)
+ E:BOUNDARIES
(SOLIDS)
270000.0 73000.
+ + + + + +
(enter load in Kg/Yr.)
+ + + + + +
1.0 54749. 0.0
+ + + + + +
+ + + + + +
+ + + + + +
DOC 6 1.0 FOCI 7
Fl-ls POINT SOURCE LOADS
+ + + + +
54750. 0.0 73000
Fl-2: POINT SOURCE SOLIDS
F2: NON-POINT SOURCES
+ G:+ PAR+AMET+ERS
1.0 XKE2 12 1.
99
-------
TEMP 3
1
1.0
X
REAER 5
0.23
DOC 6
0.030
FOCI
7
0.12 XKE2 12 1.0
TEMP 3
'i
20.0
REAER 5
0.0
DOC 6
0.0887
FOCI
7
0.030 XKE2 12 0.0
TEMP 3
20.0
+ +==== + :
ป==+ซ
sss^asss^:
:===+=
:=== + =
===+=== +CONSTANTS
f>D
2
GENERAL
5
LKOC
101
6.5
XV
136
1
MOLWT
81
322
NUX
106
1.4
HENRY
137
1.0
PHOTO
5
XPHOTO
286
2
QUANTG(1)
551
1.0
QUANTG(2)
556
1.0
QUANTG(3)
561
1.0
KDPG
291
0.0023
SOLIDS
0
RATE CONST
0
0 +
+
+ +
+ +
+
+
+ I:TIME FUNCTIONS
TOX +
+
+ +
+
2
0.0
1.0E08 J:INIT. COND'S.
1: 0.0
1.0 2:
0.0
1.0
TOTAL SOLIDS
3
2.40
1.0E08
Is 1.20
0.0 2:
270000.
0.0
100
-------
Segment 1
Time (days) Tot Solid Tot Chem 1 Dis. Chem Doc Chem 1 Part Chem Part Chem
0.00
75.00
.150.00
ฆ220.00
295.00
365.00
440.00
515.00
585.00
660.00
730.00
805.00
880.00
950.00
1025.00
1095.00
1170.00
1245.00
1315.00
1390.00
1460.00
1535.00
1610.00
1680.00
1755.00
1825.00
2190.00
2555.00
2920.00
85.00
50.00
7300.00
10950.00
14600.00
18250.00
25550.00
32850.00
40150.00
47450.00
54750.00
54825.00
54900.00
54970.00
55045.00
55115.00
55190.00
55265.00
55335.00
55410.00
55480.00
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.000E+00
0.956E-07
0.150E-06
0.180E-06
0.199E-06
0.209E-06
0.216E-06
0.220E-06
0.223E-06
0.225E-06
0.226E-06
0.227E-06
0.228E-06
0.229E-06
0.230E-06
0.230E-06
0.231E-06
0.232E-06
0.232E-06
0.233E-06
0.233E-06
0.234E-06
0.235E-06
0.235E-06
0.236E-06
0.236E-06
0.239E-06
0.241E-06
0.244E-06
0.246E-06
0.248E-06
0.261E-06
0.267E-06
0.270E-06
0.271E-06
0.272E-06
0 -27 2E-06
0.272E-06
0.272E-06
0.272E-06
0.177E-06
0.122E-06
0.922E-07
0.734E-07
0.629E-07
0.561E-07
0.519E-07
0.493E-07
0.474E-07
0.462E-07
0.000E+00
0.664E-07
0.104E-06
0.125E-06
0.138E-06
0.146E-06
0.150E-06
0.153E-06
0.155E-06
0.156E-06
0.157E-06
0.158E-06
0.159E-06
0.159E-06
0.160E-06
0.160E-06
0.161E-06
0.161E-06
0.161E-06
0.162E-06
0.162E-06
0.163E-06
0.163E-06
0.163E-06
0.164E-06
0.164E-06
0.166E-06
0.168E-06
0.169E-06
0.171E-06
0.172E-06
0.181E-06
0.186E-06
0.188E-06
0.188E-06
0.189E-06
0.189E-06
0.189E-06
0.189E-06
0.189E-06
0.123E-06
0.848E-07
0.641E-07
0.510E-07
0.437E-07
0.390E-07
0.361E-07
0.343E-07
0.330E-07
0.321E-07
O.OOOE+OO
0.630E-08
0.991E-08
0.119E-07
0.131E-07
0.138E-07
0.143E-07
0.145E-07
0.147E-07
0.148E-07
0.149E-07
0.150E-07
0.150E-07
0.151E-07
0.151E-07
0.152E-07
0.152E-07
0.153E-07
0.153E-07
0.154E-07
0.154E-07
0.154E-07
0.155E-07
0.155E-07
0.155E-07
0.156E-07
0.158E-07
0.159E-07
0.161E-07
0.162E-07
0.163E-07
0.172E-07
0.176E-07
0.178E-07
0.179E-07
0.179E-07
0.179E-07
0.180E-07
0.180E-07
0.180E-07
0.117E-07
0.805E-08
0.608E-08
0.484E-08
0.415E-08
0.370E-08
0.342E-08
0.325E-08
0.313E-08
0.304E-08
0.000E+00
0.228E-07
0.359E-07
0.431E-07
0.476E-07
0.501E-07
0.517E-07
0.527E-07
0.533E-07
0.538E-07
0.541E-07
0.543E-07
0 -546E-07
0.547E-07
0.549E-07
0.551E-07
0.552E-07
0.554E-07
0.555E-07
0.557E-07
0.558E-07
0.560E-07
0.561E-07
0.562E-07
0.564E-07
0.565E-07
0.571E-07
0.577E-07
0.583E-07
0.588E-07
0.592E-07
0.624E-07
0.638E-07
0.645E-07
0.648E-07
0.650E-07
0.651E-07
0.651E-07
0.651E-07
0.651E-07
0.423E-07
0.292E-07
0.221E-07
0.175E-07
0.150E-07
0.134E-07
0.124E-07
0.118E-07
0.113E-07
0.110E-07
0.000E+00
0.190E-01
0.299E-01
0.358E-01
0.396E-01
0.416E-01
0.430E-01
0.438E-01
0.443E-01
0.447E-01
0.450E-01
0.452E-01
0.454E-01
0.455E-01
0.457E-01
0.458E-01
0.459E-01
0.461E-01
0.462E-01
0.463E-01
0.464E-01
0.466E-01
0.467E-01
0.468E-01
0.469E-01
0.470E-01
0.475E-01
0.480E-01
0.485E-01
0.489E-01
0.493E-01
0.519E-01
0.531E-01
0.537E-01
0.539E-01
0.541E-01
0.542E-01
0.542E-01
0.542E-01
0.542E-01
0.352E-01
0.243E-01
0.183E-01
0.146E-01
0.125E-01
0.112E-01
0.103E-01
0.981E-02
0.944E-02
0.918E-02
101
-------
segment x
Time (days) Tot Solid Tot Chem 1 Dis. Chem Doc Chem 1 Part Chem Part Chem
55555.00
55630.00
15700.00
|775.00
5845.00
55920.00
55995.00
56065.00
56140.00
56210.00
56285.00
56360.00
56430.00
56505.00
56575.00
56940.00
57305.00
57670.00
58035.00
58400.00
58765.00
59130.00
59495.00
59860.00
60225.00
60590.00
60955.00
61320.00
61685.00
f50.00
00.00
50.00
73000.00
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.120E+01
0.451E-07
0.441E-07
0.434E-07
0.426E-07
0.420E-07
0.413E-07
0.406E-07
0.400E-07
0.394E-07
0.388E-07
0.382E-07
0.376E-07
0.371E-07
0.365E-07
0.360E-07
0.333E-07
0.309E-07
0.286E-07
0.265E-07
0.245E-07
0.227E-07
0.210E-07
0.195E-07
0.180E-07
0.167E-07
0.155E-07
0.143E-07
0.133E-07
0.123E-07
0.114E-07
0.530E-08
0.246E-08
0.115E-08
0.313E-07
0.307E-07
0.301E-07
0.296E-07
0.292E-07
0.287E-07
0.282E-07
0.278E-07
0.274E-07
0.270E-07
0.266E-07
0.261E-07
0.258E-07
0.254E-07
0.250E-07
0.231E-07
0.214E-07
0.199E-07
0.184E-07
0.170E-07
0.158E-07
0.146E-07
0.135E-07
0.125E-07
0.116E-07
0.108E-07
0.997E-08
0.923E-08
0.855E-08
0.792E-08
0.368E-08
0.171E-08
0.796E-09
0.297E-08
0.291E-08
0.286E-08
0.281E-08
0.277E-08
0.272E-08
0.268E-08
0.264E-08
0.260E-08
0.256E-08
0.252E-08
0.248E-08
0.244E-08
0.241E-08
0.237E-08
0.220E-08
0.203E-08
0.188E-08
0.175E-08
0.162E-08
0.150E-08
0.139E-08
0.128E-08
0.119E-08
0.110E-08
0.102E-08
0.946E-09
0.876E-09
0.811E-09
0.751E-09
0.349E-09
0.162E-09
0.755E-10
0.108E-07
0.106E-07
0.104E-07
0.102E-07
0.100E-07
0.987E-08
0.972E-08
0.957E-08
0.942E-08
0.928E-08
0.914E-08
0.900E-08
0.886E-08
0.873E-08
0.860E-08
0.796E-08
0.738E-08
0.683E-08
0.633E-08
0.586E-08
0.543E-08
0.503E-08
0.466E-08
0.432E-08
0.400E-08
0.370E-08
0.343E-08
0.318E-08
0.294E-08
0.273E-08
0.127E-08
0.589E-09
0.274E-09
0.896E-02
0.878E-02
0.863E-02
0.848E-02
0.835E-02
0.821E-02
0.808E-02
0.796E-02
0.784E-02
0.772E-02
0.760E-02
0.748E-02
0.737E-02
0.726E-02
0.715E-02
0.663E-02
0.614E-02
0.568E-02
0.527E-02
0.488E-02
0.452E-02
0.418E-02
0.388E-02
0.359E-02
0.333E-02
0.308E-02
0.285E-02
0.264E-02
0.245E-02
0.227E-02
0.105E-02
0.490E-03
0.228E-03
102
-------
Segment 2
Time (days) Tot Solid Tot Chem 1 Dis. Chem Doc Chem 1 Part Chem Part Chem
0.00
75.00
150.00
220.00
295.00
365.00
440.00
515.00
585.00
660.00
730.00
805.00
880.00
950.00
1025.00
1095.00
1170.00
1245.00
1315.00
1390.00
1460.00
1535.00
1610.00
1680.00
1755.00
1825.00
2190.00
2555.00
2920.00
ฃ285.00
*50.00
7300.00
10950.00
14600.00
18250.00
25550.00
32850.00
40150.00
47450.00
54750.00
54825.00
54900.00
54970.00
55045.00
55115.00
55190.00
55265.00
55335.00
55410.00
55480.00
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0 .270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.210E+06
0 .270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0. 270E+06
0 .270E+06
0.270E+06
0.270E+06
0 .270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0 .270E+06
0.270E+06
0.270E+06
0 .270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.000E+00
0.492E-04
0.172E-03
0.324E-03
0.507E-03
0.689E-03
0.889E-03
0.109E-02
0.128E-02
0.148E-02
0.166E-02
0.186E-02
0.205E-02
0.223E-02
0.241E-02
0.258E-02
0.277E-02
0.294E-02
0.311E-02
0.328E-02
0.344E-02
0.361E-02
0.377E-02
0 .393E-02
0.409E-02
0.423E-02
0.497E-02
0.565E-02
0.628E-02
0.686E-02
0.740E-02
0.110E-01
0.127E-01
0.135E-01
0.139E-01
0.141E-01
0.142E-01
0.142E-01
0.142E-01
0.142E-01
0.141E-01
0.140E-01
0.139E-01
0.137E-01
0.135E-01
0.133E-01
0.131E-01
0.129E-01
0.127E-01
0.125E-Q1
0.000E+00
0.192E-08
0.672E-08
0.126E-07
0.198E-07
0.269E-07
0.347E-07
0.426E-07
0.499E-07
0.577E-07
0.649E-07
0.725E-07
0.800E-07
0.869E-07
0.942E-07
0.101E-06
0.108E-06
0.115E-06
0.121E-06
0.128E-06
0.134E-06
0.141E-06
0.147E-06
0.153E-06
0.160E-06
0.165E-06
0.194E-06
0.221E-06
0.245E-05
0.268E-06
O.289E-05
0.431E-06
0.497E-06
0.528E-06
0.542E-06
0.552E-06
0.554E-06
0.554E-06
0.554E-06
0.554E-06
0.552E-06
0.548E-06
0.542E-06
0.535E-06
0.527E-06
0.520E-06
0.512E-06
0.504E-06
0.497E-06
0.489E-06
0. OOOE+OO
0.607E-09
0.212E-08
0.399E-08
0.626E-08
0.850E-08
0.110E-07
0.135E-07
0.158E-07
0.182E-07
0.205E-07
0.229E-07
0.253E-07
0.275E-07
0.298E-07
0.319E-07
0.341E-07
0.363E-07
0.384E-07
0.405E-07
0.425E-07
0.445E-07
0.466E-07
0.484E-07
0.504E-07
0.522E-07
0.613E-07
0.697E-07
0.775E-07
0.847E-07
0.914E-07
0.136E-06
0.157E-06
0.167E-06
0.171E-06
0.17 4E-06
0.175E-06
0.17 5E-06
0.175E-06
0.175E-06
0.175E-06
0.173E-06
0.171E-06
0.169E-06
0.167E-06
0.164E-06
0.162E-06
0.159E-06
0.157E-06
0.155E-06
0.OOOE+OO
0.492E-04
0.172E-03
0.324E-03
0.507E-03
0.689E-03
0.889E-03
0.109E-02
0.128E-02
0.148E-02
0.166E-02
0.186E-02
0.205E-02
0.223E-02
0.241E-02
0.258E-02
0.277E-02
0.294E-02
0.311E-02
0.328E-02
0.344E-02
0.361E-02
0.377E-02
0.393E-02
0.409E-02
0.423E-02
0.497E-02
0.565E-02
0.628E-02
0.686E-02
0.740E-02
0.110E-01
0.127E-01
0.135E-01
0.139E-01
0.141E-01
0.142E-01
0.142E-01
0.142E-01
0.142E-01
0.141E-01
0.140E-01
0.139E-01
0.137E-01
0.135E-01
0.133E-01
0.131E-01
0.129E-01
0.127E-01
0.125E-01
0.OOOE+OO
0.182E-03
0.637E-03
0.120E-02
0.188E-02
0.255E-02
0.329E-02
0.404E-02
0.473E-02
0.547E-02
0.616E-02
0.688E-02
0.759E-02
0.825E-02
0.894E-02
0.957E-02
0.102E-01
0.109E-01
0.115E-01
0.122E-01
0.127E-01
0.134E-01
0.140E-01
0.145E-01
0 .151E-01
0.157E-01
0.184E-01
0 .209E-01
0.233E-01
0.254E-01
0.274E-01
0.409E-01
0.472E-01
0.5Q1E-01
0.514E-01
0.523E-01
0.525E-01
0.526E-01
0.526E-01
0 .526E-01
0.52 4E-01
0.520E-01
0.514E-01
0.507E-01
0.500E-01
0.493E-01
0.486E-01
0.479E-01
0.471E-01
0.464E-01
103
-------
Segment 2
Time (days) Tot Solid Tot Chem 1 Dis. Che,ฎ Doc Chem 1 Part Chem Part*Chem
55555.00
55630.00
55700.00
R 775.00
845.00
55920.00
55995.00
56065.00
56140.00
56210.00
56285.00
56360.00
56430.00
56505.00
56575.00
56940.00
57305.00
57670.00
58035.00
58400.00
58765.00
59130.00
59495.00
59860.00
60225.00
60590.00
60955.00
61320.00
61685.00
(2050.00
700.00
9350.00
73000.00
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.270E+06
0.123E-01
0.121E-01
0.120E-01
0.118E-01
0.116E-01
0.114E-01
0.112E-01
0.111E-01
0.109E-01
0.108E-01
0.106E-01
0.104E-01
0.103E-01
0.101E-01
0.996E-02
0.922E-02
0.854E-02
0.791E-02
0.733E-02
0.679E-02
0.629E-02
0.582E-02
0.540E-02
0.500E-02
0.463E-02
0.429E-02
0.397E-02
0.368E-02
0.341E-02
0.316E-02
0.147E-02
0.682E-03
0.317E-03
0.482E-06
0.474E-06
0.467E-06
0.460E-06
0.453E-06
0.446E-06
0.439E-06
0.433E-06
0.426E-06
0.420E-06
0.413E-06
0.407E-06
0.401E-06
0.395E-06
0.389E-06
0.360E-06
0.334E-06
0.309E-06
0.286E-06
0.265E-06
0.246E-06
0.228E-06
0.211E-06
0.195E-06
0.181E-06
0.168E-06
0.155E-06
0.144E-06
0.133E-06
0.123E-06
0.573E-07
0.266E-07
0.124E-07
0.152E-06
0.150E-06
0.148E-06
0.145E-06
0.143E-06
0.141E-06
0.139E-06
0.137E-06
0.135E-06
0.133E-06
0.131E-06
0.129E-06
0.127E-06
0.125E-06
0.123E-06
0.114E-06
0.105E-06
0.977E-07
0.905E-07
0.838E-07
0.776E-07
0.719E-07
0.666E-07
0.617E-07
0.571E-07
0.529E-07
0.490E-07
0.454E-07
0.421E-07
0.390E-07
0.181E-07
0.842E-08
0.391E-08
0.123E-01
0.121E-01
0.120E-01
0.118E-01
0.116E-01
0.114E-01
0.112E-01
0.111E-01
0.109E-01
0.108E-01
0.106E-01
0.104E-01
0.103E-01
0.101E-01
0.996E-02
0.922E-02
0.854E-02
0.791E-02
0.733E-02
0.679E-02
0.629E-02
0.582E-02
0.540E-02
0.500E-02
0.463E-02
0.429E-02
0.397E-02
0.368E-02
0.341E-02
0.316E-02
0.147E-02
0.682E-03
0.317E-03
0.457E-01
0.450E-01
0.443E-01
0.437E-01
0.430E-01
0.423E-01
0.417E-01
0.411E-01
0.404E-01
0.398E-01
0.392E-01
0.386E-01
0.380E-01
0.375E-01
0.369E-01
0.342E-01
0.317E-01
0.293E-01
0.272E-01
0.252E-01
0 .233E-01
0.216E-01
0.200E-01
0.185E-01
0.172E-01
0.159E-01
0.147E-01
0.136E-01
0.126E-01
0.117E-01
0.544E-02
0.253E-02
0.118E-02
104
------- |