United States ^ EPA Region 10 EPA 910-R-04-006
Environmental Protection AgencpEA-095 April 2004
&EPA
Water Quality Assessment of
American Falls Reservoir
April 2004
Office of Environmental Assessment
EPA Region 10
1200 Sixth Avenue
Seattle, Washington 98101
-------
EPA has developed this report as part of a multi-agency effort to Improve our
understanding of nutrient pollution in American Falls Reservoir. For more information about
this report, contact:
Ben Cope
Office of Environmental Assessment
EPA Region 10
1200 Sixth Ave, OEA-095
Seattle, Washington 98101
(206) 553-1442
cope.ben@epa.gov
-------
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
REGION 10
*"«<**• 1200 Sixth Avenue
Seattle, Washington 98101
February 1, 2008
Reply To
AttnOf: OEA-095
TECHNICAL MEMORANDUM
Subject: Errors in Loading Information in "Water Quality Assessment of American
Falls Reservoir" (EPA, 2004)
From: Ben Cope, Environmental Engineer
Office of Environmental Assessment
To: File
I recently re-visited the 2004 report on American Falls Reservoir and discovered an error
the figures and the text related to the contribution of phosphorus loading from the
Portneuf River to the American Falls Reservoir. The error originates in the flow
estimates for the Portneuf River. The error is evident in the phosphorus loading estimates
on graphs and in the text of the report.
The error was the use of flows from the station "Portneuf River at Pocatello" rather than
"Portneuf River at Tyhee". Tyhee is lower on the river and therefore closer to the inflow
to American Falls Reservoir. Due to groundwater inflow between Pocatello and Tyhee
locations, the flows are approximately 215 cfs higher at Tyhee than at Pocatello. This is
a significant increase relative to the base flow (and associated phosphorus loadings),
particularly in the summer.
See attached to the two corrected plots of estimated inflow loadings to American Falls
Reservoir (replace Figures 9 and 10) in the report.
The same flow error carries into the text on Page 17 of the report. The following is a
correction of the loading numbers in the text:
A comparison of the estimated daily phosphorus loadings from the Snake, Portneuf,
and ungauged inflows for 2001 is shown below. Cumulatively, these loads result in
an estimated annual loading of approximately 150,000 900,000 Ibs of total
-------
phosphorus, an average of 1,200 2,500 Ibs/day. The second figure is the estimated
loadings for 1968, with an estimated annual loading of approximately 2,000,000
3,500,000 Ibs of total phosphorus. For reference, note that the same ungauged
tributary/groundwater loads were assumed for both simulations and the y-scale is
markedly different. As a result of significantly higher flows and phosphorus
concentrations in 1968, the loads carried by the Snake and Portneuf Rivers were
much higher that year than in 2001.
Finally, the corrected loadings indicate that the Portneuf River carries a higher proportion
of the total load to the reservoir (estimated at approximately 70% of the total loading).
The text in the conclusions should be revised accordingly:
The Portneuf River and a number of ungauged tributaries carry relatively high
loadings of orthophosphate and total phosphorus to the reservoir, at times
exceeding the loading from the Snake River in a low water year (2001). Based
on 2001 data and load estimation, the Portneuf River contributes over two
thirds of the total loading to the reservoir.
Finally, the loading error also affects the water quality simulations. The corrected
simulation for 1968 showed small changes in the phytoplankton concentrations in the
surface layer because of the diminished influence of the Portneuf River in that year.
However, the simulation for 2001 showed more notable differences in the predictions,
compared to the original simulations, with an increase of approximately 11% in the peak
chlorophyll concentration. While re-calibration of the model could be considered, the
data and model limitations are unchanged, and a re-calibration is unlikely to significantly
reduce the uncertainty of the model.
Aside from the edited conclusion above, the remaining conclusions in the report remain
valid.
-------
2001 Loading Estimates - Total Phosphorus
•Snake
• Portneuf
• Ungauged/Groundw ater
0 31 62 93 124 155 186 217 248 279 310 341
1968 Loading Estimates -Total Phosphorus
30000
25000
20000
T3
03
15000
10000
Snake
Portneuf
Ungauged/Groundw ater
0 31 62 93 124 155 186 217 248 279 310 341
-------
Introduction 4
I. Management Objectives 4
Scope of problem 4
Technical analysis objectives 5
II. Conceptual Model 6
System boundaries 8
Important time and length scales 8
Important Processes 9
Phosphorus source characteristics 10
Available data sources (quality and quantity) 11
Data gaps 13
III. Choice of Technical Approach 13
Rationale for approach in context of management objectives and conceptual model . . 13
Important assumptions 14
IV. Parameter Estimation 14
Data used for parameter estimation 14
Reliability of parameter estimates 19
V. Uncertainty/Error 19
Error/uncertainty in boundary conditions 19
Structural assumptions in methodology 20
VI. Results 21
Parameter values used for analysis 21
Comparison of model simulations and measurements - 2001 and 1968 21
Discussion 22
2001 Spring Conditions 23
2001 Residual Orthophosphate Concentrations 23
Phosphorus Loading from Sediments 24
2001 Dissolved Oxygen 28
1968 Simulation 29
1968 Temperature 29
1968 Dissolved Oxygen 31
1968 Residual Orthophosphate Concentrations 31
Hydrology Sensitivity Test - 2001 and 1999 32
Conclusions of analysis in relationship to management objectives 34
VI. References 36
Appendix A : Water Quality Data
Introduction
-------
The Clean Water Act calls for states to identify waterbodies under their jurisdiction that do not
meet water quality standards and develop restoration plans called Total Maximum Daily Loads
(TMDLs) to reduce pollution of those waterbodies. States submit their TMDLs to EPA for
review and approval. In addition to conducting this oversight function, EPA provides funding
and technical assistance to states in support of their TMDL programs.
The state of Idaho is developing a TMDL for American Falls Reservoir on the Snake River near
Pocatello, Idaho. During the planning phase, the state determined that an analytical tool was
needed to assess the potential effects of a range of phosphorus loadings on phytoplankton growth
and dissolved oxygen levels in the reservoir. In the fall of 2003, the state requested technical
assistance from EPA to develop a water quality model for the reservoir. This assessment and
model report is a product of information sharing and discussions between the Idaho Department
of Environmental Qualitiy, EPA, and the Shoshone-Bannock tribe.
This report describes a mathematical model developed for the reservoir and the available water
quality data used to estimate boundary conditions and evaluate the model. This assessment
relies on available data that was collected prior to model development and testing. Based on this
assessment, recommendations for future monitoring are identified at the end of the report.
I. Management Objectives
Scope of problem
American Falls Reservoir is listed as an impaired waterbody by the state of Idaho (1998 303(d)
list). The listed parameters for the reservoir include dissolved oxygen, flow alteration, nutrients,
and sediment. This report contains an analysis of nutrient and dissolved oxygen levels in the
reservoir in support of a TMDL.
Elevated nutrients can cause excessive growth of phytoplankton, which can diminish the
recreational value of the waterbody and impact fish habitat by depleting dissolved oxygen in
deeper waters. Nitrogen or phosphorus can be the limiting nutrient controlling phytoplankton
growth. Phosphorus is most commonly the limiting nutrient in fresh water systems. Recent
studies by Idaho Power of the Brownlee Reservoir on the Snake River concluded that
phosphorus was the limiting nutrient for the reservoir (Harrison, et al, 1999). Generally, the
water quality data for American Falls Reservoir are consistent with a phosphorus-limited system.
In addition, the blue-green algae that are dominant in the summer months are capable of fixing
nitrogen. For these reasons, the model and assessment described in this report are focused on
phosphorus as the nutrient of concern.
-------
Figure 1: Study Area
Technical analysis objectives
The objective of this analysis is to develop a mathematical model of the chemical, physical, and
biological characteristics of the reservoir for use in development of a Total Maximum Daily
Load (TMDL). Specifically, the analysis focuses on those characteristics that pertain to
phosphorus uptake by phytoplankton and the effects of the phytoplankton activity on dissolved
oxygen levels in the water column. The model includes a water balance, boundary inputs
(energy, nutrients), reservoir heat budget, phosphorus transformations, and factors that affect
phytoplankton growth, mortality, and settling.
-------
II. Conceptual Model
A model framework similar to that used for the Winchester Lake nutrients TMDL (March 1999)
was adapted to American Falls reservoir. This framework is developed using the STELLA
software environment, which is specifically designed for mathematical modeling applications.
The model is a one-dimensional (two cells in the vertical) dynamic model framework that
includes modules for heat budgets, phosphorus cycling, phytoplankton kinetics, and dissolved
oxygen. Simulated variables include phosphorus (total, organic, inorganic, particulate),
dissolved oxygen (impacts from background detritus and primary productivity/respiration), water
temperature, and organic matter (phytoplankton, detritus).
The diagrams below depict the model system and processes included in the model for
phosphorus, phytoplankton, and dissolved oxygen.
SiiakefffNeeley
Sii;«ke <(Bl;
-------
Advection In Out)•*-*• Soluble Inorganic P-
Uptake
Mineral izati on
Decay
Detritus
i
Settling 1
/Vert
Vli,
Figure 3: Phosphorus and Phytoplankton Processes in the Model
(Atmospheric O2
Reaeratioii
Photosynthesis
DO top
DO bottom
Phytoplviiilvtoii
SOD
Veitical Diffii&'ion Decay
Detritus
Figure 4: Dissolved Oxygen Processes in the Model
-------
The following changes have been made to the Winchester Lake model to adapt it to American
Falls reservoir:
Variable Elevation and Volume - the water surface elevation and total volume varies
based on management of the reservoir.
Variable Layer Thickness - the thickness of the top layer (epilimnion) is assumed to be 5
meters unless the average depth of the reservoir falls below 5 meters, in which case the
top layer thickness is essentially equal to the reservoir depth (and the bottom layer
thickness less than one meter).
Uniform Inflows - a feature that placed the inflow into either the top or bottom layer
based on its density was removed based on the shallow depths of this reservoir. The
inflows and outflows occur in the top layer.
Zero Sediment Phosphorus Flux- the Winchester Lake model included a constant flux
from the sediments. Based on observed aerobic conditions in the bottom of American
Falls reservoir, which is generally shallow and weakly stratified, it was assumed that
there is zero influx of inorganic phosphorus from sediments. Further analysis (see
below) indicates that there are episodes of diminished oxygen in the hypolimnion and
orthophosphate flux from sediments.
Sediment Oxygen Demand - this process was added to the American Falls model to
achieve better agreement between observed and simulated dissolved oxygen.
System boundaries
The model domain is the American Falls Reservoir, bounded by the dam and the entry point of
the Snake River and other tributary inputs to the reservoir. Because it is a storage system, with
significant changes in surface water elevation, the reservoir boundaries change over the year.
These changes are incorporated into the model by dynamically adjusting the reservoir volume
based on the water budget and volume/elevation relationships for the reservoir.
Important time and length scales
Residence Time
The reservoir has a full pool volume of approximately 1.6 million acre-ft at an elevation of 4355
feet, but this volume is highly variable due to large agricultural withdrawals during periods of
low inflow in the summer, subsequent re-filling in the fall and winter, and flood control
management in the spring. At full pool and average June flow in the Snake River (11 kefs), the
residence time is approximately 76 days. During a typical October condition, with a pool
elevation of 4335 feet, the residence time assuming average October Snake River flows is
approximately 140 days.
Vertical, Longitudinal, and Lateral Mixing
-------
The model incorporates both advective and vertical mixing components of constituent transport.
The reservoir can be generally characterized as weakly stratified. Using the physical dimensions
of the system and velocity estimates, time scales for vertical, horizontal, and lateral mixing can
be estimated and compared. The characteristic time scale for vertical diffusion is estimated
using the value of the coefficient of eddy diffusivity used in the model (S.OxlO5 mVday) and the
length over which the diffusion is assumed to occur (assumed to be 6 m from top layer midpoint
to bottom layer midpoint). This results in time scale of approximately 8 days. Based on typical
flows in the Snake River, the time scale for longitudinal advection at a similar scale (6 m mixing
length) is on the order of 0.02 days. Based on a typical Portneuf River flow, the time scale for
lateral mixing at this scale is on the order of 1 day. These comparisons are indicative of a system
where the dominant mixing process is longitudinal advection.
Model Resolution
The reservoir is approximately 21 miles long with a maximum width of approximately 9 miles.
Because the reservoir is represented by a single cell in the model, lateral and longitudinal
variability seen in the water quality data are not resolved in the model. Based on the importance
of advective transport and the significant length and width of the reservoir, one would expect
longitudinal and lateral variation in the system. Longitudinal variation is evident in the water
quality data. For example, the chlorophyll-a data indicate that peak chlorophyll-a growth does
not occur simultaneously at the four monitoring sites. In the completely mixed model
representation, a single growth pattern is simulated, representative of an idealized average
condition for the reservoir as a whole.
Important Processes
Water quality and the resulting beneficial uses of American Falls Reservoir are impacted by
nuisance phytoplankton growth. There are numerous types of phytoplankton, and each has
specific ideal conditions for propagation. This model includes a single phytoplankton
community type and the main processes limiting growth, including light limitation, temperature
limitation, and nutrient limitation. Blue-green algae are the focus for this analysis based on the
high concentrations of the species Aphanizomenon observed in the reservoir in the summer.
-------
o^nnnn
!
0>
o 9nnoon
~m
D)
-2.
E
° 150000
-------
The model does not currently include specific pollutant sources that deliver phosphorus to the
American Falls Reservoir. Rather, phosphorus loads are introduced to the reservoir at the
mouths of the tributaries and through groundwater flows. Potential sources of phosphorus to the
reservoir include municipal treatment works such as the City of Pocatello and cities upstream of
the reservoir that discharge to the Snake River, industrial facilities such as FMC, agricultural
activities in the watershed, and natural background contributions.
Available data sources (quality and quantity)
The model relies on monitoring information for both estimation of boundary conditions and
comparison of simulations and observations within the reservoir. Past studies, particularly a
study by the Bureau of Reclamation (Bushnell, 1968), USGS (Kjelstrom, 1995), and Battelle
(Baca, et al, 1974), provide insights into historic conditions in the reservoir. Monitoring of
reservoir conditions by Idaho DEQ began in 2001 and continues to the present (Van Every,
2003). The locations of the DEQ sampling are depicted in Figure 6. This information, combined
with boundary condition estimates using data from USGS (flow, tributary quality) and the
National Weather Service (heat budget parameters), was used in the modeling analysis. Table 1
shows the types of monitoring information used in this analysis.
Sample Sites
Sampling transects
0.8 O 0.81.6 Miles
—i—
1:200000
re 6 : DEQ 2001-2003 Sampling Sites in American Falls Reservior
Fi
gu
11
-------
Table 1: Summary of Available Data
Parameters
total phosphorus
orthophosphate chlorophyll-a
dissolved oxygen temperature
nitrogen parameters
total phosphorus
orthophosphate dissolved
oxygen temperature
nitrogen parameters
flow
total phosphorus
orthophosphate
temperature
weather, including:
dry bulb air temp
dew point temp
wind speed
barometric pressure
cloud cover
weather, including:
dry bulb air temp
dew point temp
wind speed
barometric pressure
solar radiation
Frequency
approx. 2/month
between May and
early August
8 samples between
May and October
daily
118 samples over
period
27 samples
over period
46 samples
over period
approx. 20 samples
daily
daily
daily
Location
vertical profiles at 4 locations
in the reservoir
vertical sampling at 3-4
locations in the reservoir
Snake River near Blackfoot
(#13069500)
Snake River at
Neeley (#13077000)
Portneuf River at:
Pocatello ( #13075500)
Tyhee (#13075910)
Snake River near Blackfoot
Portneuf River
at Tyhee
Portneuf River
at Tyhee
American Falls drains and
tributaries
Snake River near Blackfoot
Surface Airways Station in
Pocatello
Agrimet station at Aberdeen
Period of
Record
2001-2003
1968
1910-2002
1907-2002
1897-2002
1985-1994
2001-2002
1972-2002
1970-1994
1999-2003
2001-2003
2000-2001
1968 to present
1983 to present
Agency
Idaho DEQ
BOR
uses
uses
uses
City of
Pocatello
BOR
uses
NWS
BOR
Data gaps
12
-------
Like many TMDL modeling analyses, this analysis relies on existing, available data. There are a
number of gaps in the data that require filling to accomplish a reasonable dynamic simulation.
The following table highlights the most significant data gaps and the manner in which they are
handled in the model.
Table 2: Data Gaps
Parameter(s)
water quality profiles in
reservoir
Snake inflows of
phosphorus
Portneuf inflows of
phosphorus
groundwater & ungauged
tributary phosphorus
ungauged flows
Portneuf flows at mouth
Problem
no information prior to
May or after early August
2001 sampling focused on
summer months
no sampling in 2001; grab
sampling over long term
very limited or no
sampling
no routine sampling
Tyhee gauge not operated
in 1997 and 1999
Model Assumptions or
Estimation
none
interpolation used in
winter/spring; constant
values assumed in fall
long term average used
assumed equal to Snake
River levels
constant value assumed
and water balance
checked for 1999 and
2001
constant value added to
Pocatello flows; checked
years when both gauges
operated
Comments
cannot evaluate
simulations of spring or
late summer conditions
simulated orthophosphate
in reservoir suggest that
inputs are reasonable
does not account for long
term changes in average
phosphorus
higher levels known to
exist in Portneuf - this is
addressed by data at
Tyhee gauge
constant value (2285 cfs)
resulted in good water
balance
constant value (215 cfs)
resulted in reasonable
agreement at Tyhee
III. Choice of Technical Approach
Rationale for approach in context of management objectives and conceptual model
The management objective for this work is to provide technical information for the establishment
of a TMDL. Because of resource constraints and lack of bathymetry data for the reservoir, the
option of developing a more complex, 2-D model using a framework such as CE-QUAL-W2 was
not pursued. At the same time, it was desirable to use a mathematical process model rather than a
more simple empirical model (e.g., Vollenweider loading plots) to enable some exploration of the
effects of the large seasonal variations in reservoir volume on water quality. In addition, the need
13
-------
for a dissolved oxygen analysis for the TMDL required a linkage between eutrophication and
dissolved oxygen conditions in the reservoir.
The selected model framework includes many of the processes in CE-QUAL-W2 (e.g., water
balance, reservoir heat budget, phosphorus transformations, and phytoplankton kinetics). It also
includes the effects of the phytoplankton activity on dissolved oxygen levels in the water column.
However, it does not provide the longitudinal spatial resolution that can be obtained using a
model like CE-QUAL-W2.
Important assumptions
The model developed for American Falls includes the following simplifying assumptions:
- The top and bottom layers are completely mixed.
- There is a single phytoplankton community (blue-green algae).
- Phosphorus is the limiting nutrient
- There is no wind mixing (general mixing is captured in the diffusion coefficient).
The temperature/density gradient occurs at 5 meter depth.
- There is no phosphorus loading from sediments.
The model rests primarily on mathematical formulations and literature values for parameters cited
in EPA guidance on surface water quality modeling (Bowie, et al, 1985).
IV. Parameter Estimation
Data used for parameter estimation
The model was developed using 2001 observations of the system. Model parameters related to
phytoplankton stoichiometry, growth, respiration, mortality, and settling were set in the range of
literature values (Bowie, et al, 1985) for blue-green forms such as genus Aphanizomenon.
Boundary Conditions
Hydrology, Inflow Temperatures, and Weather
Inflows to American Falls are estimated as the sum of the flows measured in the Snake River
below the Blackfoot River confluence, Portneuf River near Tyhee, and ungauged
tributaries/groundwater. This last inflow is a combination of numerous small creeks and
groundwater inflows to the reservoir, and the estimated magnitude of this inflow (2,300 cfs) is
significant with respect to the Snake and Portneuf River inflows in low flows years. The estimate
calculated for this analysis was found to be consistent with previous estimates for ungauged flows
to the reservoir by USGS (Kjelstrom, 1995). Also, review of long term flow records indicates
that this average net inflow to the reservoir does not vary significantly from year to year (Vaga,
personal communication). Based on BOR sampling of 10 creeks, the surface water contribution
to this inflow is estimated at approximately 750 cfs.
14
-------
Figure 7: Snake and Portneuf River Flows (1970-2002)
50000
45000
40000
35000 4
30000
25000
20000
15000
10000
5000
Snake@Blackfoot (13069500)
• Snake@Neeley (13077000)
-Portneuf River
Outflows are estimated as the flow measured in the Snake River at Neeley (see figure below for
upstream/downstream flow comparisons). The estimated annual water balances for the reservoir
were checked by calculating a predicted daily surface water elevation using volume/elevation
relationships for the reservoir (BOR, 1971) and measured/estimated inflows and outflows, and
comparing this predicted water surface elevation to the elevation observed at the dam. The
comparison for 2001 is shown below.
15
-------
Simulated and Observed Water Surface Elevations
(2001)
•observed
simulated
oo in CNI en
CNI LO oo o
CNI
OO LO CNI
en CNI uo
CNI CO CO
Day
Figure 8: Comparison of Simulated and Observed Reservoir Elevation (2001)
The USGS monitored the temperature of the Snake River continuously between April and
September of 2000 and 2001, and the daily average temperatures were used as input temperatures
for the model. For the period prior to April and after September, a trigonometric function was
used to fill the data gap. The same temperatures were assumed for the Portneuf River and
ungauged tributaries.
For weather data, the Pocatello station provides daily observations of air temperature, dew point,
wind speed, and pressure for the heat budget calculations. The Agrimet station at Aberdeen, in
the vicinity of the reservoir, records solar radiation. These data were used in the model and also
used in conjunction with clear sky radiation to estimate the cloud cover. Each component of the
heat budget has estimation uncertainty, but the evaporation estimates are typically the most
uncertain of the heat budget components. The evaporation rate coefficient was adjusted based on
observed temperatures.
Phosphorus Loadings
Orthophosphate inputs are estimated using observed total phosphorus samples from the Snake and
Portneuf rivers and an estimate of the ratio between total phosphorus and Orthophosphate. USGS
collected 12 samples for phosphorus and Orthophosphate in the Snake River below the confluence
of the Blackfoot River in the spring/summer of 2001. The total phosphorus concentration ranged
from 17-51 ug/1. The average ratio of total phosphorus and Orthophosphate was approximately
5:1 (the actual ratio may be higher, because many of the Orthophosphate samples were below
detection levels).
16
-------
For the Portneuf River, there are two datasets available. USGS collected approximately 20
samples from 1989 to 1994 at Tyhee. The total phosphorus concentrations ranged from 290-770
ug/1. The City of Pocatello has collected over 40 samples in the Portneuf River just upstream of
the Tyhee gauge from 1999-2003. The total phosphorus concentrations ranged from 200-1590
ug/1, with a mean of 960 ug/1. For both datasets, the average ratio of total phosphorus and
orthophosphate was approximately 1:1 (meaning virtually all of the phosphorus is in the
orthophosphate form).
As noted above, in addition to the Snake and Portneuf rivers, there is a significant flow from
smaller tributaries and groundwater into the reservoir. BOR has sampled a number of smaller
tributaries to the reservoir, such as Bannock Creek, Spring Creek, and McTucker Creek. For the
model, the flow-weighted average of all the samples collected from all tributaries (approximately
50 ug/1 total phosphorus) was used as a constant inflow concentration. The average ratio of total
phosphorus and orthophosphate for these tributaries was approximately 2:1
A comparison of the estimated daily phosphorus loadings from the Snake, Portneuf, and
ungauged inflows for 2001 is shown below. Cumulatively, these loads result in an estimated
annual loading of approximately 450,000 Ibs of total phosphorus, an average of 1,200 Ibs/day.
The second figure is the estimated loadings for 1968, with an estimated annual loading of
approximately 2,000,000 Ibs of total phosphorus. For reference, note that the same ungauged
tributary/groundwater loads were assumed for both simulations and the y-scale is markedly
different. As a result of significantly higher flows and phosphorus concentrations in 1968, the
loads carried by the Snake and Portneuf Rivers were much higher that year than in 2001.
17
-------
1000
Snake
Portneuf
Ungauged/Ground
31 61 91 121 151 181 211 241 271 301 331 361
Day
30000
25000 -
20000 -
J; 1 sooo
T3
(O
O
10000
5000
Snake
Portneuf
Ungauged/Ground
31 61 91 121 151 181 211 241 271 301 331 361
Figures 9 and 10: Estimated Total Phosphorus Loadings for 2001 and 1968
18
-------
Estimated Model Parameters
Model simulation results were compared to available DEQ observations of phosphorus,
chlorophyll-a, temperature, and dissolved oxygen in the reservoir. The following parameters,
aside from the phytoplankton parameters, were adjusted to the following values to achieve a
qualitative fit to the observations:
Table 3: Parameters adjusted during model development
Parameter
Evaporation Rate
Vertical Diffusion Coefficient
Sediment Oxygen Demand
Detrital Decay Rate
Value
l.OE-09
5.0E-5
0.10
0.15
Units
/mb
mVs
gm O2/m2-day
/day
It was noted that the vertical diffusion coefficient was comparable to the estimated value (l.OE-5
mVs) for American Falls Reservoir referenced in Bowie, et al (EPA, 1985). A table with all of the
parameter values used in the model is included in the Results section of this report.
Reliability of parameter estimates
It is important to note that it is not possible to determine a single set of parameter values that
uniquely and correctly represents the system. Parameters are estimated through a process of
literature review, sensitivity tests, and comparisons of simulated and observed conditions. The
reliability of parameter estimates can be examined by simulating a variety of waterbody
conditions with the same parameter sets and comparing model estimates and observations. The
recent monitoring of the reservoir by DEQ provides a set of observations for parameter
estimation. Unfortunately, this period of sampling has been characterized by a persistent drought.
Water quality information for the reservoir during wetter years is limited.
To evaluate reliability of the parameter estimates, the model parameters estimated for 2001 were
used in a simulation of 1968 conditions using reservoir and boundary observations from
Bushnell's report. The reservoir elevation, inflows, and water quality were significantly different
in 1968 than in 2001, so the observations from this year provide a useful test of the model.
V. Uncertainty/Error
Error/uncertainty in boundary conditions
There are a number of sources of uncertainty and error in the assumed boundary conditions,
including data gaps (described earlier), measurement error, and simplified representations of the
boundaries. The model includes three advective inflows (Snake River, Portneuf River, and
ungauged tributaries/groundwater). This last inflow is significant in magnitude with respect to
the other two inflows, but it is an aggregation of numerous small creeks and groundwater inflows
19
-------
to the reservoir. BOR has sampled 10 creeks with an average collective inflow to the reservoir of
approximately 750 cfs, out of an estimated aggregate inflow of 2300 cfs. Estimates of the
quantity and quality of aggregate groundwater inflows were not available for this analysis.
There are no measurements of sediment oxygen demand, and the model assumption is that this
demand is moderate and constant over the year. There may be significant variability in the
demand, particularly during and after spring and summer phytoplankton blooms, when organic
detritus settles to the bottom. The 2001 simulation, which does not include a spring diatom
bloom, predicts higher dissolved oxygen in the bottom than was observed in the June 2001
samples.
Structural assumptions in methodology (e.g., effects of aggregation or simplification)
Like any model, this model presents a simplified representation of the waterbody. The two-layer
approach assumes American Falls reservoir consists of two well-mixed layers with uniform total
depths across the reservoir, whereas the actual reservoir depths vary substantially (e.g., from 8 m
at County Boundary to 19 m depth at the dam in May 2001). The depth of the epilimion is
assumed to be a constant 5 m, whereas the available temperature data indicates substantial
variability in vertical gradients over the summer period.
Horizontal gradients also appear in some of the water quality observations that would not be
captured in the model. For example, the peaks in chlorophyll-a concentrations do not occur
simultaneously in the observed data. The peak at the County Line site is generally earlier than at
the other sites. One possible explanation for these observations would be earlier productivity due
to earlier peaks in temperature at this location.
The model accounts for a single phytoplankton community, and a blue-green algae community
was selected for analysis based on the prevalence of Aphanizomenon in the summer algal
samplings by DEQ. Low concentrations of orthophosphate and dissolved oxygen measured in
May 2001 suggest significant diatom productivity in the reservoir prior to the first chlorophyll-a
samples collected in June. Comments in the Bushnell report also suggest a significant diatom
bloom prior to the onset of the blue-green bloom, and Idaho Power observes spring diatom
blooms in Brownlee Reservoir. The model does not capture this spring productivity or its effect
on the initial conditions prior to the summer bloom.
20
-------
VI. Results
Parameter values used for analysis
In addition to the parameters adjusting in the parameter estimation process (see discussion and
table above), the following parameter values to characterize phytoplankton activity were selected
based on ranges reported in EPA guidance (Bowie, et al, 1985).
Table 4: Phytoplankton parameters in the model
Parameter
Detrital Sinking Velocity
Phytoplankton Mortality Rate
DO/Detrital Carbon ratio
DO/Phyto Carbon ratio
Mineralization rate - Org P to dissolved PO4-P
Max Phytoplankton Growth Rate
Max Phytoplankton Respiration Rate
Half-saturation constant - light limitation
Half-saturation constant - PO4-P
Phytoplankton sinking speed
Ratio - C to chlorophyll in phytoplankton
Ratio- phosphorus to C in phytoplankton
Value
0.20
0.20
2.0
2.0
0.10
2.0
0.10
.004
.02
0.10
.025
.025
Units
m/day
/day
gmDO/gmC
gmDO/gmC
/day
/day
/day
kcal/mete^-sec
gm P/m3
m/day
gm Chla/gm C
gm TP/gm C
Comparison of model simulations and measurements - 2001 and 1968
As discussed above, the model parameters were estimated based on comparisons of simulated and
observed water quality in the summer of 2001. Graphical comparisons of simulations and depth-
averaged observations were generated for temperature, chlorophyll-a, phosphorus, and dissolved
oxygen. In order to provide some perspective on the spatial/temporal variability in the
observations, observations from the individual sample collection sites are included on the graphs.
21
-------
Discussion
2001 Simulation
General
In general, the model predicts the observed patterns of water quality in the reservoir in the June
through early August time frame. The temperature simulations generally follow the warming and
cooling patterns and range of vertical stratification seen in the observations. Figure 11 depicts a
comparison of simulated and observed temperatures at the dam, as well as observations in the
surface at the county line station (at the shallower, upstream end of the reservoir). The plot shows
the vertical and longitudinal variability observed in reservoir temperatures. Waters at the
shallow, county line site are more responsive to weather changes, heating faster in the spring and
cooling faster in the fall.
00 _
91 -
1 Q
O
5? 1 7
¥ 15 ^
5
* HO
•~ 1 1
Y
X
>£-
n / »
"^o^ 2 •
/^ /^ ^^^=^ /7—
^ /*/ ^^
n/
^\*/ V-^^"
150 160 170 180 190
Day
° observed@dam - top
X observed@county- top
» observed@dam - bottom
simulated - bottom
200 210 220
230
Figure 11: Simulated and Observed 2001 Temperatures
For chlorophyll-a, a spike in the concentration was observed between mid-July and mid-August
2001 at the sampling sites, and the model generates a spike of similar magnitude in this general
time frame (see Figure 12). Note that the observed chlorophyll peak occurs at different times
across the reservoir, with later peaks at the Dam and Little Hole sites compared to the more
upstream County Line site).
22
-------
n o
1
|a1^
CL
2 01-
£ U. I
Q.
O
i
O rv rvc
fl
a
a observed- dam-top
A observed -dam-bottom
"\
\
* A
D
1 a
130 150 170
A
A
A
~^ V o
A ^^ 0 D
' \ ~
190 210 230 250
Day
Figure 12: Comparison of Simulated and Observed Chlorophyll for 2001
In both model simulations and observations, phosphorus concentrations are increasing in the late
spring and early summer. While temporal patterns in water quality vary between sites, top and
bottom layer dissolved oxygen levels are generally comparable to the observations. The
observations do not continue past early August, and the model predicts a recurrence of significant
biomass growth in the late summer and fall.
2001 Spring Conditions
As note earlier, the model includes a single phytoplankton community with growth requirements
of blue-green species. There is evidence that the reservoir experiences spring productivity in
addition to the blue-green productivity in July/August. The orthophosphate and dissolved oxygen
levels are significantly lower in the earliest samples available (May) than simulated levels (Figure
13). Unfortunately, there are no chlorophyll samples for May in the 2001-03 data.
2001 Residual Orthophosphate Concentrations
The model predicts that the July/August 2001 phytoplankton bloom is limited first by
temperature, then by light limitation, and finally by low phosphorus concentrations. This last
phase in the simulation is not consistent with the 2001 observations, which generally show fairly
high orthophosphate concentrations even after the bloom/crash in 2001. Tests with a variety of
parameter adjustments failed to produce both the observed chlorophyll-a pattern and a significant
post-bloom orthophosphate concentration.
23
-------
Figure 13: Simulated and Observed Orthophosphate for 2001
It should also be noted that data from different locations and different years indicates some
variability in the chlorophyll-a/orthophosphate pattern. For example, the last sample from 2003
at the Little Hole site indicates a combination of a chlorophyll-a spike and severe depletion of
orthophosphate concentrations similar to the pattern in the 2001 simulations (but not the 2001
observations).
Phosphorus Loading from Sediments
Because of the complexities of phytoplankton growth and reservoir dynamics, there are numerous
possible explanations for the observed relationship between phytoplankton biomass and
orthophosphate, and its departure from the mechanisms incorporated into the model. One
compelling hypothesis is that significant orthophosphate loads are released from the sediments in
the summer due to low dissolved oxygen at the bottom of the reservoir. The observations of
dissolved oxygen and orthophosphate near the dam for 2001-2003 show a consistent pattern that
supports this hypothesis. In July of each year, dissolved oxygen is depleted in the bottom during
periods of stratification, and there is a corresponding spike in orthophosphate concentrations at
depth (Figure 14). This spike occurs when dissolved oxygen levels reach about 2 mg/1. This
threshold concentration is identical to a threshold concentration for sediment phosphorus release
cited in the literature (Marsden, 1989).
24
-------
0.25
_£_
O
Figur
Dissolved Oxygen and Orthophosphate in Bottom Samples in 2001 and 2002
e!4
25
-------
45
40
35
30
25
20
15
10
5
0
0.25
•Stratification
-Chlorophyll
- PO4-P bottom
PO4-P top
^— co 10 T— co LO CN a> co
LO LO
Figure 15: Water Quality Observations for 2001. Note that "stratification" is the
difference between the surface and bottom temperatures
10
9
8
-S 4
' *
•*
> * .•• «*
i?5 L?> [fi
cBcBcS
5s
Figure 16: Wind Speed Recorded in 2001 at Pocatello, Idaho
26
-------
Once the episode of thermal stratification ends (see Figure 15 for measure of stratification),
dissolved oxygen concentrations in the bottom rebound and there is a corresponding drop in
orthophosphate concentrations. A relationship is apparent between the onset and breakdown of
stratification and wind speed recorded in the vicinity (Figure 16).
The reduction in orthophosphate concentrations after the breakdown in stratification is likely the
result of a combination of factors, including cessation of the sediment flux, vertical mixing with
less concentrated surface waters, and phytoplankton uptake (which occurs in July as well). As
noted above, Aphanizomenon has a buoyancy regulation mechanism and may extract
orthophosphate directly from deeper waters.
There are no data available for later dates in the summer to evaluate whether these patterns of
stratification, dissolved oxygen concentrations, and phosphorus concentrations were repeated
during calm periods over the remainder of the summer and early fall. The occurrence of this
pattern in 2001, when the reservoir was relatively shallow, also raises questions about water
quality conditions during years of average or above-average flow and reservoir depth. A model
sensitivity test comparing the effects of 2001 and 1999 hydrology on water quality indicates that
dissolved oxygen concentrations are lower when the reservoir is deeper (see discussion below).
Because of the amount of phosphorus commonly present in the eutrophic lake bed sediments,
sediment phosphorus release can hamper attempts to improve water quality by reducing external
phosphorus loads to the system. Marsden (1989) found that the persistence of sediment loading is
influenced by lake morphometry, flushing rate, sediment type, trophic state, and history of
enrichment. He offers examples of highly enriched, shallow lakes that were slow to respond to
significant reductions in external phosphorus loadings because of persistent sediment phosphorus
releases. In an extreme case, Lake Trummen in Sweden did not show water quality
improvements until enriched sediments were mechanically removed from the lake. Marsden
generally found that "in lakes with annual mean TP (total phosphorus) concentrations greater than
100 mg/m3, few improvements have been noted unless the reduction in loading were greater than
60%. Even where the reductions in loading have been greater a proportion of lakes still failed to
respond." For comparison, the mean total phosphorus concentration at the DEQ sampling site
near the American Falls dam from 2001 to 2003 was 90 mg/m3.
The efforts to identify and reduce external loadings to the reservoir are underway. The State of
Idaho developed a TMDL for nutrients in the Portneuf River in April 1999 (Idaho DEQ, 1999)
and EPA approved the plan in April 2001). This plan calls for a reduction of 81% in the major
sources of total phosphorus to the river upstream of the Tyhee gauge. The TMDL identified
springs and Pocatello municipal discharges as the largest sources of nutrients in this area.
27
-------
2001 Dissolved Oxygen
A comparison of simulated and observed dissolved oxygen is shown in Figure 17. The model
predicts a consistent difference between the top and bottom layers of approximately 1-3 mg/1 until
the period of maximum reservoir drawdown in the late summer. A significant spike in the top
layer oxygen (and a smaller increase in the lower layer) corresponds to the period of peak
phytoplankton activity.
In the observations, there is virtually no difference in the top and bottom in the spring, which
suggests that the model is under-predicting vertical mixing during this period. Beginning in July,
however, observed differences are similar to predicted differences. Consistent with the timing of
the phytoplankton activity at the dam and in the simulation, the spike in surface oxygen occurs
later at the dam than in the simulation.
14
2 --
A A
simulated - top layer
simulated - bottom layer
A observed - bottom
X observed - top
120
140
160
180
200
220
240
Day
Figure 17: Simulated and Observed Dissolved Oxygen - 2001
A comparison of simulations for 2001 (drought) and 1999 (average flow) under identical 2001
weather inputs indicates that predicted hypolimnetic dissolved oxygen is consistently lower in
1999. In theory, shallow conditions of 2001 would be conducive to more effective oxygen
transfer from the surface to the small hypolimnion volume. In the 1999 test, the effects on
dissolved oxygen of higher inflows are counteracted by the greater hypolimnetic volume and
depth of the reservoir. This larger, deeper pool of sequestered water is less efficiently
28
-------
oxygenated. While the model theory for these predictions is sound, we have no data from 1999 to
compare to the predictions.
1968 Simulation
General
Phosphorus concentrations in the Snake and Portneuf rivers were notably higher in 1968 than in
2001. The average total phosphorus in the Snake River samples was 0.260 mg/1 compared to
0.030 mg/1 in 2001, and the Portneuf averaged 0.46 mg/1 in the early 1990s versus 1.7 mg/1 in
1968 samples.
In addition to differences in phosphorus concentrations, the seasonal inflows were higher in 1968
than in 2001. The higher concentrations and flows in 1968 resulted in substantially higher
estimates of phosphorus loading to the reservoir compared to 2001 levels (see Figures 9 and 10
earlier). The higher flows also resulted in higher reservoir elevations in the summer.
Unfortunately, chlorophyll was not monitored in 1968. There are phytoplankton samples from
several locations on July 22 and August 4 of 1969 showing high Aphanizomenon counts; one
sample near the dam was higher than any individual sample in 2001.
The model simulation, with phosphorus limitation governing the ultimate peak concentration,
predicts a higher peak chlorophyll-a concentration in 1968 compared to 2001 (66 ug/1 compared
to 42 ug/1 for 2001). Unlike the 2001 simulation, where phytoplankton levels drop precipitously
after the initial peak, the 1968 simulations show sustained high phytoplankton concentrations
after the initial peak. This occurs even though simulated orthophosphate levels in the reservoir
remain low after the peak. This model prediction suggests that external phosphorus loadings in
1968 would have been sufficient to sustain phytoplankton activity in the absence of internal
loading.
1968 Temperature
The simulated and observed temperatures are quite similar except for two samples in the late
summer. The observations at different locations were highly variable during this sampling
period. The primary source of these discrepancies (e.g., local variation, sampling error, boundary
condition uncertainty) is unclear. It was noted that observed Snake River temperatures were
lower than the assumed temperatures used as boundary conditions in the model (2001 conditions);
however, simple tests with lower Snake River temperatures did not close the gap significantly.
29
-------
140
160
-top-sim
• top-obs
A bottom-obs
^—bottom-sim
120 140 160 180 200 220 240 260 280
day
Figures 18 and 19 : Simulated 1968 Chlorophyll and Temperature
30
-------
1968 Dissolved Oxygen
Given the paucity of chlorophyll data for 1968, the dissolved oxygen data provides an important
dataset for comparison to the simulations, because oxygen concentrations are influenced by
multiple processes included in the model, including phytoplankton growth/mortality, vertical
mixing, and heat budgets. While the model tends to over-predict the difference in dissolved
oxygen in the top and bottom layers, the model predicts concentrations (approximately 8 mg/1
and 4 mg/1 in the top and bottom, respectively) and differences (approximately 4 mg/1 between
top and bottom) that are similar to the observations in the critical August time period.
O)
£_
o>
O)
73
01
~
do1-sim
A do1-obs
— do2-sim
do2-obs
0
120 140
160
180
200
day
220
240
260 280
Figure 20 : Simulated and Observed Dissolved Oxygen - 1968
1968 Residual Orthophosphate Concentrations
Similar to the 2001 simulations, the Orthophosphate predictions reach peak concentrations similar
to the observations, but the overall simulated pattern is significantly different than the observed
pattern. First, the model predicts higher levels in the spring than observed, potentially due to a
diatom bloom. Second, the simulated Orthophosphate plunges to near-zero, while the observed
phosphorus level remains high. As noted earlier, this may be the result of internal loading, which
is not included in the model. Conditions in 1968 were more conducive to anoxia at the sediment
surface than in 2001, with average oxygen bottom layer concentrations about 2 mg/1 lower in
31
-------
1968 and individual samples near the bottom below 1 mg/1 on August 6, 1968 (Julian day 218).
0.25
ortho1-sim
ortho1-obs
140 160 180 200 220 240 260 280
day
Figure 21: Simulated and Observed Orthophosphate (top layer)
Hydrology Sensitivity Test - 2001 and 1999
As noted earlier, the recent sampling has occurred over a period of very low flows in the basin.
In order to provide insights into the effect of inflows and reservoir elevations on water quality, a
simulation was constructed using all of the 2001 boundary conditions except daily tributary
inflows, dam outflows, and reservoir elevations. Flows and elevations for 1999, a more typical
hydrologic year, were substituted for the 2001 inputs. The results indicate that despite the higher
inflows, predicted chlorophyll and lower layer dissolved oxygen concentrations are worse under
the 1999 conditions. This could be the result of greater reservoir depth and/or differences in
loading due to changes in relative flow of the Snake, Portneuf, and ungauged
tributaries/groundwater.
32
-------
140
190
90
340
Day
Figure 22: Comparisons of Predicted Chlorophyll - 1999 and 2001
140
190
240
Day
-1999 simulation
-2001 simulation
290
340
Figure 23: Comparisons of Predicted Lower Layer Oxygen - 1999 and 2001
Conclusions of analysis in relationship to management objectives
33
-------
As noted earlier, the state of Idaho determined that an analytical tool was needed for assessment
of the potential effects of a range of phosphorus loadings on phytoplankton growth and dissolved
oxygen levels in the reservoir. Based on the available data and model simulations, the following
conclusions are offered:
The American Falls water quality model provides useful information for the assessment of
water quality dynamics in the reservoir as a whole, despite the observed heterogeneity in
water quality across sampling locations. The model parameters estimated for 2001
resulted in reasonable estimates for chlorophyll, temperature and dissolved oxygen in
2001 and 1968 during the July/August period of interest.
Observations and simulations suggest that release of phosphorus from the sediments is a
significant source of phosphorus to the system during periods of stratification in July and
August.
Spring diatom activity and subsequent settling may be contributing to diminished oxygen
levels at depth during periods of stratification, thus contributing to release of
orthophosphate from sediments.
The Portneuf River and a number of ungauged tributaries carry relatively high loadings of
orthophosphate and total phosphorus to the reservoir, at times exceeding the loading from
the Snake River in a low water year (2001).
Simulations suggest that, with zero phosphorus release from sediments and consumption
of surplus orthophosphate in late July, phosphorus loadings from the tributaries would be
sufficient to drive measurable productivity for the remainder of the summer and fall.
Model simulations indicate that periods of low flow and reservoir elevation (e.g., 2001)
may not represent the worst-case conditions for water quality in the reservoir.
Recommendations for additional monitoring and analysis
This assessment has identified a number of information gaps that warrant additional sampling and
analysis. The following recommendations are offered:
Improvements to Sampling Plan
extend the time frame for lake and river sampling to April through October, with a
particular focus on potential spring diatom blooms
add routine sampling of the Snake River, Portneuf River near the mouth (at Tyhee), and
high load, ungauged tributaries
- conduct groundwater phosphorus sampling and reservoir sediment sampling
34
-------
conduct sampling closer to centerline of reservoir if possible
subject to agency resources, collect bathymetry information to support a two-dimensional,
laterally-averaged model
Additional Analysis
using existing and new data, refine the flow estimates and loading analysis of ungauged
tributaries/groundwater
subject to agency resources, examine potential improvements to the model, including
variation of the dispersion coefficient based on wind speed, addition of sediment
phosphorus flux tied to bottom water dissolved oxygen levels, and addition of a second
phytoplankton community (spring diatoms)
35
-------
VI. References
Baca, R.G., Lorenzen, M.W., Mudd, R.D., Kimmel, L.V. "A Generalized Water Quality Model
for Eutrophic Lakes and Reservoirs". Battelle Pacific Northwest Laboratories. 1974.
Bowie, G., Mills, W., Porcella, D., Campbell, C., Pagenkopf, J., Rupp, G., Johnson, K., Chan, P.,
and Gherini, S. "Rates, Constants, and Kinetics Formulations in Surface Water Quality
Modeling" (Second Edition). EPA. Environmental Research Laboratory. EPA/600/3-85/040.
June 1985.
Bureau of Reclamation. Letter from Regional Supervisor to John Yearsley (EPA) with
Attachments (area-elevation-volume relationships for American Falls Reservoir). Oct. 15, 1971.
Bushnell, V. "Eutrophication Investigation of American Falls Reservoir". Bureau of
Reclamation, Pacific Northwest Region. 1968-69.
California Dept. of Fish and Game. "Prediction of Thermal Energy Distribution in Streams and
Reservoirs". Water Resources Engineers, Inc. June 1967.
Harrison, J., Wells, S., Myers, R., Parkinson, S., and Kasch, M. "1999 Status Report on
Brownlee Reservoir Water Quality and Model Development". Draft Technical Report and
Appendices. Idaho Power. Boise, ID. November 1999.
Idaho Department of Environmental Quality. "Portneuf River TMDL: Waterbody Assessment
and Total Maximum Daily Load." Pocatello Regional Office. March 1999.
Kjelstrom, L.C. "Streamflow Gains and Losses in the Snake River and Ground-water Budgets for
the Snake River Plain, Idaho and Eastern Oregon" U.S. Geological Survey. Paper 1408-C. 1995.
Marsden, M. "Lake Restoration by Reducing External Phosphorus Loading: The Influence of
Sediment Phosphorus Release". Freshwater Biology. Vol. 21, 139-162. 1989.
Vaga, R. Environmental Protection Agency. Personal Communication. 2004.
Van Every, L. "Water Quality Monitoring Workplan: American Falls Reservoir". Idaho
Department of Environmental Quality. Pocatello Regional Office. April 2003.
Welch, E. "Ecological Effects of Wastewater: Applied Limnology and Pollution Effects". E and
FNSpon. 1992.
36
------- |