&EPA
United States Office of Water (4305) EPA-820-R-14-004
Environmental Protection April 2014
Agency
AQUATOX
MODELING ENVIRONMENTAL FATE
AND ECOLOGICAL EFFECTS IN
AQUATIC ECOSYSTEMS
Guidance in AQUATOX Setup and
Application
-------
Introduction 1
Model Installation 1
Very First Steps 1
Understanding Simulation Modeling 1
Common Problems with Simulation Setup 2
Working with Exam pie Sim ulations 5
Changing Site Characteristics to Match Your Site 6
Choosing Which State Variables Should Be Included 7
Working with Boundary Conditions 9
Model Calibration and Validation 10
Order of Model Calibration 10
Which parameters are likely to be most important? 10
Generality 10
Acceptance Criteria 11
Possible Model Applications 11
Screening-level Model 12
Site-specific Water Quality Criteria 12
Thermal Pollution and Climate Change 13
Invasive Species 13
Risk Assessment of New or Existing Chemical 13
Impacts of Contaminated Sites on Aquatic Ecosystems 14
Combined Sewer and Stormwater Discharge 14
Environmental Impact of Construction 14
Common Error Messages 15
Appendix: Guide to AQUATOX 3.1 Simulations 17
References 24
-------
Guidance in AQUATOX Setup and Application
AQUATOX is a simulation model for aquatic ecosystems. AQUATOX predicts the fate of
various pollutants, such as nutrients and organic chemicals, and their effects on the
ecosystem, including fish, invertebrates, and aquatic plants.
This document is intended to be a "quick-start" guide to introduce major model features
as well as being a type of "cookbook" to guide basic model setup, calibration and
validation. It is designed to supplement the AQUATOX Users' Manual and Technical
Documentation which are frequently referenced within this document. This guidance
document primarily pertains to Release 3.1 of the AQUATOX model, produced and
supported by the U.S. Environmental Protection Agency.
AQUATOX is fairly easy to install on a Windows system (Windows XP or subsequent
releases), however, administrator privileges are generally required for installation. To
get started, download the AQUATOX installer from the EPA or other website.
For more information on model installation, please see the AQUATOX Release 3.1
Installation guide.
Once the model has installed and is running, you may start to explore the features of
the model. AQUATOX is a multiple-document interface in that you can load many
different simulations and they will appear on the AQUATOX desktop under the main
window.
To get started, we recommend working with the "Simple Tutorial" that is available by
clicking on "Help, Tutorial" from the main menus (at the top of the AQUATOX screen
after startup), or can also be found in the AQUATOX Users' Manual. This tutorial will
guide you through adding and deleting a state variable, setting initial conditions, viewing
parameters, running a simulation, and viewing output. There is a more advanced
"stream tutorial" also available in these documents.
An AQUATOX model is primarily composed of the following components
"state variables"components of the modeling system, such as animal
biomasses, in which masses or concentrations are modeled and tracked;
"driving variables"time series inputs, such as temperature, that are not
changed by the model;
-------
Guidance in AQUATOX Setup and Application
"parameters"constant model inputs, such as "maximum photosynthetic rate,"
that are used to calculate the modeled state variables;
"boundary conditions"information about state variables from outside the model
domain, such as upstream loadings, or point-source loadings;
"physical characteristics"constant model parameters or time series such as
"mean depth" that describe the site being modeled.
As a further introduction to the model, we recommend reading the section 1.1 "overview
of the AQUATOX model" in the Technical Documentation, along with section 1.8
regarding the intended application of the model.
AQUATOX can be run as a point model, a stratified model (with seasonally varying
epilimnion and hypolimnion layers), and as a two- or three-dimensional model with
linked segments. The spatial resolution depends on the modeling goal. However, given
the trade-off between ecological realism and spatial resolution, the model domain is
usually spatially aggregated, in contrast to hydrodynamic models.
When embarking on a modeling project, it is important to understand the assumptions
regarding physical model setup and spatial and temporal resolution. Many of the basic
points regarding the AQUATOX model construction can be found in Chapter 2 of the
Technical Documentation. That could be a good section to read before running
AQUATOX.
Other resources include "What is AQUATOX," lecture materials that start on page 3 of
the AQUATOX one-day web training materials followed by the discussion of analytical
capabilities.
Many users decide to start a new simulation "from scratch" with all parameters zeroed
out. This is probably seen as a "safe" way to start because this method ensures that all
parameters will then reflect what is appropriate for their site, as opposed to some other
location. However, when we have worked with studies created in this manner, we have
found many parameters left as blank or "zero" as a default. This will not result in a
successful ecological simulation. For this reason, we generally recommend starting
with a surrogate simulation and modifying it to match your site's characteristics. This
also has the benefit of working with a calibrated parameter set as opposed to having to
perform all model calibration yourself. For more information about choosing a surrogate
site, see the section on Working with Example Simulations, which is later in this
document.
The first consideration when setting up an AQUATOX model is defining the physical
characteristics for your site properly. The majority of errant simulations that we have
-------
Guidance in AQUATOX Setup and Application
seen have problems with water volume, water depth, or water velocity.
Water Volume: There are many options to model water volume, so the user must be
careful in setting up this component of the model. The full set of water volume options
can be found in the section on "Water Volume Data" in the AQUATOX Users' Manual.
The step that users tend to neglect is graphing water volume after specifying their
choices and running the model. This is an important check that the parameters that
were imported into the model have the correct units and are being properly accounted
for by the model. To examine the water volume output, you should click on the "output"
button, then select "New" under Graph Library, then under "other" graphs, select "water
volume." You can then look at the water volume results for each segment as well as an
accounting of inflow and outflow of water to that segment. Many early implementations
of models have water volumes near zero or unreasonably large due to misspecification
of inputs. This error is such a common problem that a technical note has been written
on the subject and is available on the EPA website: Modeling Water Flows with
AQUATOX Release 3.1.
Specifying water volume in a linked-segment model is complicated and requires a full
external accounting of the water balance for the entire linked system. (AQUATOX is not
a hydrodynamic model so it does not account for the effect of river slope on water
volume, for example. The large cell sizes of AQUATOX preclude hydrodynamics from
being added to the model, but aggregation of data from smaller-cell-size hydrodynamic
models is often utilized.) If a linked-segment model takes too long to run and causes
problems debugging the water volume setup, a useful trick is to delete all biotic state
variables and then to iron out the water volume setup with the faster linked model. The
debugged water volume setup can then be imported into the slower full model once all
segments have appropriate water volumes calculated.
Water Depth: There are three options for modeling mean-water depth at your site. This
physical characteristic has an important effect on light climate, especially for periphyton
that reside at the bottom of the water column. As a first option, in the site "underlying
data" the "use bathymetry" option should be turned off, in which case mean depth is
calculated in a time-varying manner as volume over surface area. In this case, the
surface area does not change as volume increases, meaning this option has limited
utility for some sites. If mean water depth does not change much, it can be entered in
the underlying site data and kept as a constant. However, if mean depth is dynamic
then it is useful to determine a time-series of mean-depth data using a hydrodynamic
model or an external relationship of water volume to mean depths. This can then be
imported into the model by clicking the "Site" button and then "Show Mean Depth /
Evaporation" at the bottom of the window, and then importing a dynamic mean depth
time series.
-------
Guidance in AQUATOX Setup and Application
As with water volume, whenever model input is changed, it is useful to look at a graph
of the output to ensure that the expected model output has been produced. In this
case, select :"new" graph, select then "custom" graph, and then add "ZMean (Dynamic)
(m)" to one of the axes.
Water Velocity: Water velocity may also be imported or calculated by the model. By
clicking on the site button, a "Water Velocity" time series is visible along with the option
to have AQUATOX calculate water velocity or to import water velocity for the site.
Whichever option is selected, outputs for water velocity should also be graphed and
inspected. Water velocity has important implications for the scouring of periphyton,
breakage of macrophytes, oxygen reaeration, and scour of bottom sediments.
Light: Now that some critical variables in the physical setup of the site have been
entered and double-checked, a proper accounting of light climate is required. Primary
producers cannot function if light boundary conditions and water clarity are not within
the bounds of reason. Light boundary conditions are available by clicking "light" in the
AQUATOX state-variable list and are generally set using the annual mean and annual
range ("Average Light" and "Annual Light Range") parameters in the site's "underlying
data" record. Light penetration in the water column is a function of suspended inorganic
matter (usually TSS), organic matter, and algal growth. To double-check the light
climate in your model, graph the "secchi depth" output from your model. If this value is
small then light cannot penetrate deeply enough to enable algal growth. The "secchi
depth" time-series can also be plotted against observed data from your site, if available.
See "Importing Observed Data" in the AQUATOX Users' Manual for more information
about plotting observed data against model results.
Organic Matter: Organic matter in the water column cannot be ignored in the model for
many reasons. Some important examples are its role as a food source for
invertebrates, its effect on water-column light extinction, and its effect on biochemical
oxygen demand. To specify initial conditions and boundary conditions for organic
matter, click on "Susp. and dissolved detritus" in the state-variables list associated with
every simulation. Inputs in terms of "organic matter," "organic carbon," or "BOD" can
also be specified as percentage breakdowns into faster-reacting labile and slower-
reacting refractory components. More information on setting these parameters is
available in section 5.1 (especially Table 10) of the Technical Documentation.
Food Web: One more common mistake made when setting up the model is to forget to
specify linkages in the model's food web. A "food web" button is available on the main
simulation's screen along with a help button at the bottom to assist in setting up the
trophic interactions (feeding preferences and egestion coefficients) appropriate for your
ecosystem.
-------
Guidance in AQUATOX Setup and Application
AQUATOX 3.1 is delivered with 39 example studies (.aps or .als files) that include
single-segment models, multi-segment models, and also include rivers, streams, ponds,
lakes, reservoirs, and estuaries. A full list of these studies is provided in a color-coded
table in the appendix at the end of this document.
As mentioned above, the best way to apply AQUATOX to your particular site is usually
to select a "surrogate" site from the library of simulations that most closely matches your
simulation and then modify the physical setup, biota, nutrients, and organic matter
parameters to match your site.
A complete example of how to modify a surrogate site to match the characteristics of
your own may be found in Day 1 (especially Labs 2 and 3) of the web-training materials,
which were developed for a three day course on AQUATOX. These exercises (study
files and loadings files included) guide the user through modifications of a site's physical
setup, organic matter, inorganic matter, nutrient specifications in the water column, and
selection of biota.
AQUATOX is also has a "wizard" interface that allows you to step through many of the
most important parts of model setup. The parameters shown in the wizard are the
same as those accessed through the state variables lists, loadings screens, and
underlying data in AQUATOX's primary interface, but the organization is different.
There are 19 primary steps that the AQUATOX wizard uses sequentially or the user can
select a step from the progress window (Figure 1). The wizard doesn't include all of the
parameters and flexibility of the primary AQUATOX interface, but it provides a user-
friendly interface to work with the model, especially in adding and removing plant and
animal state variables.
-------
Guidance in AQUATOX Setup and Application
Step 1: Simulation Type
Step 2: Simulation Period
Step 3: Nutrients
Step 4: Detritus
Step 5: Plants
Step 6: Invertebrates
Step 7: Fish
Step 8: Site Characteristics
Step 9: Water Volume
Step 10: Water Temperature
Step 11: Wind Loading
Step 12: Light Loading
Step 13: Water pH
Step 14: Inorganic Solids
Step 15: Chemicals
Step 16: Inflow Loadings
Step 17: Direct Precipitation
Step 1S: Point-source Loadings
Step 19: Nonpoint-source Loads
(double click on any step to jump there)
Help
Wizard Sumr
Simulation Name: Galveston Bay TX
Simulation Type: Estuary
State Variables in Simulation:
Total Ammonia as N
Nitrate as H
Total Soluble P
Carbon dioxide
Oxygen
Tot. Susp. Solids
Salinity
Refrac. sed, detritus
Labile sed. detritus
Susp. and dissolved detritus
BuriedRefrDetr
Diatomsl: [Phyt Hi-Nut Diat Mar]
Diatoms2: [Peri Diatom, Marine]
Greensl: [Phyto, Green, P/larine]
Cyanobacterial: [Phyt, BI-Greens, Mar]
OtnerAlgl: [Cryptomonas]
QtherAlg2: [Dinoflagellate, Mar.]
SedFeedeM: [Polychaete]
SedFeeder2; [Amphipod]
SuspFeederl: [Copepod]
SuspFeeder2: [Rotiter, marine]
Claml: [Surf claml
Hide Summary
Figure 1. AQUATOX Wizard Progress and Wizard Summary screens
Changing Site Characteristics to Match Your Site
When modifying a surrogate site to match your site, you should generally consider the
following groups of variables:
Dates of simulation in the "setup" window. The period of your simulation
depends on what data sets are available for calibration and verification and what
your needs are for model projections.
Site characteristics in the site "underlying data" window (under the "Site"
button). Initial focus should be on length, surface area, depths, temperature
ranges, latitude, and average-light variables. Time-series of site characteristics
such as water velocities and mean depths can also be found in the site-type
window under the "Site" button.
Water volume setup can be found by double-clicking on "water volume" towards
the end of the state-variables list. See the discussion of modeling water volume
in the section above on common problems above.
-------
Guidance in AQUATOX Setup and Application
Water temperature setup is found by double-clicking on "temperature" towards
the end of the state-variables list. If you set the temperature ranges in the site
characteristics screen properly you can select to "use annual mean and range";
otherwise you can use a constant or a time series to model water temperature.
Nutrient and inorganic matter loadings and initial conditions may be found by
double-clicking on each of these state variables at the top of the list of state and
driving variables. Depending on your available data, inputs of phosphate and
nitrate can be loaded as TN or TP using the "inflows are Tot. N" (or "Tot. P")
checkboxes located on these screens. Point source, non-point source and direct
precipitation can also be separately specified. Total suspended solids may be
input as a driving variable and this can have a significant effect on light climate.
Organic matter loadings are available under "suspended and dissolved
detritus." See the discussion of modeling organic matter in the section on
common problems above.
All of the above variables may be viewed in the simulation output window after running
a preliminary simulation. Some examples are the "Nutrients," "Detritus," "Temperature,"
and "Water Volume" graphs that can be selected after opting to create a "New" graph
for the graph library. Custom graphs can also be produced for site mean depths and
water velocities, for example. (For more information on creating and editing graphs,
including a full list of the types in AQUATOX, see page 45 of the AQUATOX Users'
Manual.) Whenever a model input has been modified, it is important to double-check its
effects on relevant model outputs before assuming that a variable has been properly set
up. This can quickly solve potential problems with units or check-boxes not set
correctly.
Animal and plant state variables are discussed in the next section of this document.
,
AQUATOX is distributed with an abundance of state variables to facilitate application to
many different situations. The analyst should not feel compelled to use all the biotic
groups, and in fact, that is probably a bad idea! Depending on the goal, many groups
can be removed from a simulation, making it faster to run, simpler to calibrate, and
easier to defend.
Adding and removing state variables is often easiest through the AQUATOX wizard
interface which significantly facilitates the adding of size-class fish and selection of
appropriate animals depending on plant, fish, or invertebrate type (see Figure 2).
Some examples of simple vs. more complex AQUATOX simulations may be found on
pages 18-19 of the AQUATOX one-day web training materials. Before selecting the
food-web conceptual model for your site, it might also be worth reading the lectures on
-------
Guidance in AQUATOX Setup and Application
modeling plants (p52), modeling animals (p80), and modeling chemicals (p115) of those
same web training materials.
AQUATOX-- Simulation Setup Wizard
Step 6: Invertebrates to Simulate (Susp Feeders)
Within AQUATOX, invertebrates are classified as Shredders, Sediment Feeders,
Suspension Feeders, Clams, Grazers, Snails, and Predatory Invertebrates.
To add a Susp Feeder Compartment to the simulation, drag its name from the list of
available Susp Feeders to the simulation box on the right. To remove a Susp Feeder
Compartment from the simulation, select it and click the Remove button below.
Available Susp Feeders:
Bosmina Lonsirosrris Mayfly (Isonychia)
Susp Feeders in Simulation:
(Maximum of Two )
Caddisfiy
Caddisfly.Trichopter
Cladoceran
Copepod
Daphnia
R ii tiler. Brachiomis
Rotifer, Keratella
Rotifer, marine
Saltwater copepod
SuspFeederl: [Copepod]
SuspFeederl: [Rotifer, marine]
Show Progress
^^^^^^^^^^^^^m
Show Summary
Figure 2. Selection of suspension feeders from AQUATOX database
Parameters are often available for individual species, but in food-web modeling, a
species is usually representative of a larger group of species with similar life histories.
Circumstances under which you would choose to model specific organisms with explicit
attention to parameter values include:
Commercially important species (salmon, oysters)
Invasive species (Cylindrospermopsis cyanobacteria, Hydrilla, zebra mussels)
Keystone species (Pacific salmon, gizzard shad)
Species near the top of the food web (lake trout, largemouth bass).
Simpler food webs will be more subject to disruption from losses or gains in biomass of
one component of that food web. For example, a simple food-chain model would be
devastated by the loss of its single primary producer whereas a complex food web
might continue to function fairly well if that same primary producer were lost as a result
of prey switching and opportunistic feeding. Our comprehensive sensitivity analysis of
AQUATOX found that the model is sensitive to food-web construction:
8
-------
Guidance in AQUATOX Setup and Application
"Simpler food-web models are more sensitive to effects from food-web
interactions. For example a food web with five zoobenthos categories is less
sensitive to perturbations in a single zoobenthos parameter than a food web in
which all zoobenthos are represented by a single category."
. L.
"Boundary conditions" are state variables from outside of the spatial domain of a
modeled system, and the loss of state variables to a region beyond the spatial domain.
These are especially important to properly characterize in rivers or other systems with
low retention times. For example, concentrations of nutrients in a system with a water
retention time of less than one hour will largely reflect the loadings without modification.
There simply is not enough time for the nutrients to react within the system to noticeably
change their concentration in the water column. For this reason, the "retention time" of
the modeled system is another output variable that should be plotted and considered.
In standing-water systems with longer retention times, initial conditions become much
more important and boundary conditions have a smaller, more subtle effect on overall
model predictions.
Because phytoplankton and zooplankton wash out of stand-alone stream segments
rapidly, a special assumption has been put into place to handle plankton retention
times. In this case, AQUATOX takes into account the "Total Length" of the river being
simulated, as opposed to the length of the river segment or reach, so that phytoplankton
and zooplankton production upstream can be estimated. The "Total Length" parameter
can be directly entered at the bottom of the "Site Data" screen or estimated from the
watershed area in that same location. This parameter essentially slows down the
residence time for phytoplankton and zooplankton to account for up-stream production
and allows more reasonable in-stream predictions to be produced. For more
information about this feature, see the "Phytoplankton and Zooplankton Residence
Time" section in Chapter 4 of the Technical Documentation.
In a multi-segment system such as a river with multiple reaches or a complex reservoir,
tributaries are important sources of water, nutrients, organic matter, and toxicants.
There could be point-source inputs, such as effluent from wastewater treatment plants,
and nonpoint-source inputs, like direct runoff and groundwater input. All of these
sources may be separately characterized using a structure called "tributary input
segments." See in the Users' Manual file for more information
on how to set up and utilize these model structures.
-------
Guidance in AQUATOX Setup and Application
In general one should work with the portions of the model domain least affected by
other categories you will be calibrating in the future. Ordinarily the sequence would be:
Boundary conditions including nutrient loadings;
Plants;
Animals; and,
Toxicants.
The Sensitivity Analysis report provides many insights into the importance of various
parameters. As presented in the report's summary:
Biotic state variables are sensitive to temperature parameters.
Consumption and respiration parameters are sensitive, especially when
allometric formulations are used for fish.
Algae are sensitive to their maximum photosynthesis rate (Pmax).
Simpler food-web models are more sensitive to effects from food-web
interactions due to lack of alternative prey sources.
Periphyton biomass is quite sensitive to sloughing parameters such as "percent
lost in slough event."
Log octanol-water partition coefficient (Kow) is a highly sensitive parameter for
toxicant fate and effect.
Models cannot be realistic, general, and precise at the same time. However, for many
applications, models should be both realistic and general, and precision is not
necessary or even attainable (Park and Collins 1982). Generality is especially
important because models are usually used to predict responses under changing
conditions. Mechanistic constructs can extend the applicability beyond the immediate
domainunlike empirical models that are constrained by the limits of observed data
(DeAngelis and Mooij 2003). AQUATOX can be set up so that multiple sites, linked only
by a common parameter set, can be calibrated simultaneously. An example is the
Minnesota Rivers study in which rivers with low-, moderate-, and enriched-nutrients and
turbidity were calibrated together. In another study, the model was calibrated across
diverse reaches of the 60-mile long Lower Boise River in Idaho. Because of this
generality the periphyton parameter set can be easily applied to represent periphyton in
other wadeable rivers with little additional calibration.
10
-------
Guidance in AQUATOX Setup and Application
However, there are limits to generality. If the expectation of goodness of fit is too
demanding (see next section) then site-specific calibration may be required, keeping in
mind the tradeoff between precision and applicability to changing conditions. Perhaps
more important is the fact that not all processes are represented by mechanistic
constructs. For instance, calibration may be required to capture the responses of
phytoplankton to the fine-scale hydrodynamics that are not represented by AQUATOX.
A weight-of-evidence approach is usually recommended in terms of deciding whether
simulated results are acceptable and defensible. From simple to more complex lines of
reasoning, some of the following lines of evidence may be used:
Reasonable behavior based on general experience;
Visual inspection of data points and model plots;
Do model curves fall within error bands of data?
Do point observations fall within model bounds obtained through uncertainty
analysis?
Regression of paired data and model resultsis there concordance or bias?
Comparison of mean data and mean model results;
Comparison of frequency distributions;
o Relative bias;
o F test; and,
o Kolmogorov-Smirnov test.
For more information on measures of model performance please see section 2.6 of the
Technical Documentation
AQUATOX is a comprehensive but flexible ecosystem model, and has an almost
unlimited variety of potential applications. In this section we highlight some of the types
of model applications that we have worked with or are aware of. A good starting point
for investigating AQUATOX applications would be the AQUATOX bibliography and the
Annotated Bibliography for AQUATOX on the EPA website. An additional source of
information regarding model applications may be found in the "potential applications"
lecture starting on page 6 in the AQUATOX one-day web-training material. In the
discussion below, relevant example studies included with AQUATOX 3.1 are highlighted
in grey; more information about these studies can be found in the color-coded appendix
at the end of this document.
11
-------
Guidance in AQUATOX Setup and Application
Perhaps the simplest application of AQUATOX is as a screening-level model, applied
without calibration where relative, not absolute, differences are of interest. For
example, AQUATOX was highlighted in a recent Water Environment Research
Foundation report on using models to develop site-specific nutrient goals (Bowers and
Bell 2013). In this case study, AQUATOX was applied without calibration to four
streams in Virginia. Two streams had greater periphytic biomass and were dominated
by green algae, and two had lower biomass and were dominated by diatoms.
This application of AQUATOX was:
Capable of distinguishing between "higher biomass" and "lower biomass" sites.
Capable of predicting when filamentous greens would predominate
The default parameters were taken from (Park et al. 2009). Changing the critical force
(FCrit) values for scouring periphyton probably would have provided the biggest
adjustment.
At press time, AQUATOX was in the draft stages of being utilized as part of a TMDL
project on the Lower Boise River (LBR), Idaho. The model had previously been applied
to the LBR, so that study file was used as a starting point (see Lower Boise R. ID Seg_l-
s.ais). The goal of the model application was to estimate the necessary reductions in
nutrient loads to the river in order to meet state water quality standards. (The TMDL,
when completed, is subject to review and approval by EPA.) The Idaho water quality
standards include the following applicable requirement:
"Excess Nutrients. Surface waters of the state shall be free from excess
nutrients that can cause visible slime growths or other nuisance aquatic growths
impairing designated beneficial uses."
Because of the emphasis on nuisance algal endpoints, but not higher organisms, while
also trying to minimize overparameterization of the model relative to the available data,
the modeling team felt that the food web included in the model could be simplified. All
invertebrate and fish state variables were removed, and the only algal groups used
were those necessary to obtain an acceptable fit to the chlorophyll a observations.
Using a process-based model such as AQUATOX can help to provide a mechanistic
link between nutrients and algal responses. This can be used in conjunction with other
efforts and approaches to develop nutrient targets. For example, AQUATOX could be
used to evaluate which factor or factors are controlling algae levels or to evaluate the
12
-------
Guidance in AQUATOX Setup and Application
effects of agricultural practices or land-use changes on chlorophyll a concentrations.
Some examples of recent use follow:
Short course Day 2 demonstration: "Modeling Nutrients for Criteria Support in
Tenkiller Lake, OK" (Tenkiller Ferry Lake OK.als). For more information, see page 41
of the second-day lectures in the AQUATOX three-day web-training materials.
Please note, when accessing these lecture materials you can look at the slides
only or you can also view slides with detailed notes.
AQUATOX is being used at the present time to develop nutrient standards for
Nevada (Smith and Fritsen 2011, Smith et al., In press). This model application
remains under development and is subject to review and approval by all
applicable agencies.
'"'. ;"" /' :"!. '-.,"'. : : . ' :. ' ';' :. i- ''"'. \ " '.'"'. { ' ;_ . ' . -. _. . '". '. '= :. ,"''-. ~: !J _
AQUATOX has realistic temperature responses that include limited adaptation.
Therefore, boundary conditions for temperature can be varied to represent a variety of
climate scenarios. In particular, the impacts of climate change on Lake Onondaga, New
York, were forecast using AQUATOX (Taner et al. 2011). The AQUATOX sensitivity
analysis report suggests that AQUATOX biotic state variables are sensitive to
temperature parameters both due to direct effects (on metabolism rates, for example)
and indirect effects (food-web interactions).
AQUATOX can be used to evaluate potential responses to invasive species and to
evaluate various mitigation measures. For example, direct and indirect impacts on
native species may be simulated. As a teaching example of this type of model
application, control of Hydrilla in Clear Lake, California, was simulated. For more
information, please see Lab 6 in the second-day of the three-day web-training materials.
To model invasive species, an AQUATOX user needs a good idea of why an invasive
species is expected to be more successful than native species (faster growth rates or
better feeding efficiency, for example) and include those characteristics within the
parameter set for that species.
AQUATOX can be used in risk assessment of bioaccumulation and ecotoxicity of
pesticides and industrial chemicals. Examples include:
Microcosms
o Replication of aquarium with HCB and macrophytes (Gobas et al.
1991) (HCB Tank.aps)
Mesocosms
13
-------
Guidance in AQUATOX Setup and Application
o Probably one of the best examples is an application in France
(Sourisseau et al. 2008) in which the model was successfully
calibrated and validated to biomass dynamics of various biological
compartments in artificial streams designed for measuring pollutant
effects on aquatic communities.
Estuarine environment
o Galveston Bay, Texas in which the bioaccumulation of PFOS
throughout the estuarine food web was simulated. For more
information, please see "Modeling Estuarine Conditions" at the start of
the third-day lectures within the three-day web-training materials.
The model has been used to analyze the potential bioaccumulation of, ecotoxicity of,
and recovery from chemicals in a variety of ecosystems, including:
A small stream in Denmark polluted with TCE from a leaking tank (Funder 2009,
McKnight et al. 201 Oa, McKnight et al. 201 Ob) (Skensved Denmark TCE.aps). The
model predicted limited ecological changes in the aquatic life in the stream as a
result of the TCE contamination.
A small creek in Oregon with chlorpyrifos and legacy dieldrin, (Zollner Creek OR w
chlorpyr dieldrin-pulse.aps).
"Lake Onondaga is arguably the most polluted lake in the United States." This excerpt
comes from the preface of a Effler's book (1996) which served as the source of data for
this study. The lake has significant nutrient inputs from wastewater treatment plant
("Metro") and combined sewers, resulting in successive algal blooms, hypoxic
hypolimnion, and build-up of organic sediments (Onondaga Lake NYSed Diagenesis.aps). An
earlier version of AQUATOX was verified using data from Lake Onondaga (see page 1-
1 of the Release-1 Model Validation Reports). The non-calibrated model successfully
predicted oxygen and chlorophyll a dynamics in this vertically stratified systemfurther
calibration subsequently improved model fit.
Nearby construction can increase the flux of nutrients and sediment into receiving
waters. In a hypothetical example, total suspended sediment (TSS) loadings to the
Cahaba River were doubled, increasing the embeddedness and affecting both
periphyton and zoobenthos (Cahaba R AL X2 TSS.aps).
14
-------
Guidance in AQUATOX Setup and Application
For the most part, AQUATOX error messages are comprehensive explanations that
should help you to resolve whatever problem that you are having, but several common
error messages have the potential to cause confusion.
AQUATOX is not set up in a correct directory structure! AQUATOX will not
function. AQUATOX must be set up in a directory structure with the PROGRAM,
OUTPUT, DATABASE, and STUDIES directories in their original locations after the
installer has been executed. If a new "exe" file is to be used, it must be copied into the
PROGRAM directory of an existing installation.
Because LipidFrac differs from that in the Chem Tox records, AQUATOX will
update the LipidFrac fields in the chemical toxicity records for all toxicants in the
study. To allow chemical toxicity records to be edited as a unique database (when not
associated with a simulation) chemical toxicity records have their own set of "lipid
fraction" data. However, this option can cause a discontinuity in a simulation if an
animal's "underlying data" has a lipid fraction different from that in the chemical toxicity
record. This dialog informs the user that it will modify the lipid fraction in the chemical
toxicity record to remove this discontinuity. Press cancel to leave all data unchanged.
Because LipidFrac was changed, AQUATOX will update the LipidFrac fields for all
toxicants in the study. AQUATOX is attempting to reconcile the lipid data in the
chemical toxicity record with the lipid in the Animal's "underlying data," in this case
because the underlying data has changed. Press cancel to leave all data unchanged.
Because Lipid Fraction data may have been changed in the Toxicity Screen, each
Lipid Fraction from this chemical's toxicity record will be copied over to the other
toxicants in this study. Lipid Frac will change in each relevant (linked) organism's
underlying data. Again, AQUATOX attempts to equalize the lipid data in the chemical
toxicity record with the lipid in the Animal's "underlying data," in this case because the
chemical toxicity record data has changed. Press cancel to leave all data unchanged.
File version unreadable: File predates Version 1.03, or other error. This error is
usually shown when a user tries to load a non-AQUATOX file (non .aps or .als file) into
the model or drags and drops a non-AQUATOX file onto the AQUATOX desktop.
Key violation. When importing a time-series. This error message is produced by the
AQUATOX database manager when an imported time series has the same date
repeated twice, which is not allowed in the AQUATOX data structures.
No "Partial" runs are permitted when exporting or viewing Linked Results. Linked-
result output and export relies on all segments being complete and having the same
number of data points. When a linked-mode run is stopped by the user part of the way
through a simulation, these functions are unavailable. A user can still view partial output
for an individual segment by first clicking on that individual segment and then clicking on
the "output" button associated with that segment.
Warning, in the control run, variable {variable name} becomes zero or so tiny as to
result in infinite differences being calculated. AQUATOX will plot these differences
as zero. When plotting "difference" data in the output window. The difference graph
15
-------
Guidance in AQUATOX Setup and Application
shows percent differences in perturbed data in comparison to the control run. If a
variable in the control run has a value of zero, this quantity is not calculable (due to
division by zero). This dialog warns the user that the differences are sometimes being
plotted as "zero" because of this problem.
Warning: Periphyton {plant name} is not linked to a phytoplankton compartment.
Chlorophyll may be undercounted in a scour event. Observed chlorophyll a data at
the simulated river could include the effects of scoured periphyton biomass. To allow for
better comparison between observed and simulated time series, periphyton may be
linked to a phytoplankton compartment so that simulated chlorophyll a will include the
effects of periphyton sloughing. See the section on "periphyton-phytoplankton link" in
chapter 4 of the Technical Documentation.
Warning: Your study has inputs of organic matter in BOD units. AQUATOX 3.1
uses a different method than Release 3.0 for converting CBOD to organic matter,
based on percent refractory rather than BOD5_CBODu. (Please see equation 148c
in the latest Tech. Doc.). The quantity of OM loaded into your system may be
different than in previous model results. Note, the default BOD5_CBODu ratio was
2.47 which corresponds to a 60% refractory loading. This wordy dialog was
designed to inform users of older versions of AQUATOX that our assumptions regarding
the conversion of BOD to organic matter have been refined in the latest version of the
model. The error message alerts the user to this fact as well as the relevant equation
(148c) in the Technical Documentation.
You have entered a value outside of the recommended range for Relative Error:
(0.0005 to 0.01). The "relative error" defines how much error is allowed by the
AQUATOX Runge-Kutta differential equations solver before it moves on to the next step.
Setting this to be too small could cause the program to execute too slowly; setting it too
large could create large errors in model accuracy. See section 2.1 of the Technical
Documentation and especially Figures 4 and 5 for more information.
16
-------
Guidance in AQUATOX Setup and Application
Appendix: Guide to AQUATOX 3.1 Simulations
AQUATOX is distributed with a variety of self-contained studies (Table 1) that can be
used as tutorial examples, templates, or starting points for developing new applications.
They are color-coded here to give the user a rough idea of their applicability. There are
four general classes of studies:
Nutrient studies that are designed to examine the effects of organic matter,
nitrogen, and phosphorus levels on primary productivity and the consequent
effects on the food web.
Microcosm and mesocosm studies in which the model is applied to
experimental facilities or sites that are in themselves physical models with
controlled boundary conditions; these range from simple aquaria to experimental
streams to pond enclosures.
Chemical fate and effects studies that examine bioaccumulation and the direct
and indirect effects of organic chemicals on the food web as well as the
persistence of those chemicals.
Studies intended for teaching purposes that are not closely based on
observed data, but that are included to illustrate particular AQUATOX features or
site types.
The table below is organized by study type in the following order: nutrient studies,
micro- and mesocosm studies, chemical fate and effects studies, and teaching studies.
Well-calibrated studies1 for each type are presented first.
Well-calibrated nutrient study
Well-calibrated micro- or mesocosm study
Well-calibrated chemical fate/effects study
Roughly-calibrated nutrient study
Roughly-calibrated mesocosm study
Roughly-calibrated chemical fate/effects study
Study intended for teaching purposes
In this case, the term "well calibrated" is a function of the available data to calibrate against and the
goals of the study. The term does not necessarily mean that all state variables in the study have been
calibrated against an extensive data set.
17
-------
Guidance in AQUATOX Setup and Application
Table 1. Description of Example Study Files for AQUATOX 3.1
Study Name
Blue Earth
R.MN.aps
(well-calibrated
nutrient)
Blue Earth R.MN
BMP Criteria, aps
Cahaba R AL.aps
(well-calibrated
nutrient)
Cahaba R AL X2
TSS.aps
Crow Wing R.
MN.aps
(well-calibrated
nutrient)
DeGray Res AR.aps
(well-calibrated
nutrient)
Lake George
NY.aps
(Well-calibrated
nutrient)
Lake George NY
smelt, aps
Site
Type
River
River
River
Reser-
voir
Lake
Location
Southern
MN
Near
Birming-
ham AL
North
central
MN
Near Hot
Springs
AD
Mr\
Upstate
NY
Run time
(h:mm;
2.66 GHz
Quad CPU)
0:14 for
2yr
0:1 4 for
2yr
0:47 for
2 yr
0:29 for
2yr
0-13 for
\J . 1 \J l\Jl
2\/r
yr
0:1 6 for
2yr
0:01 for
Syr
0:03 for
A O *
13 yr
Notes
The Blue Earth River drains a watershed in
the Western Corn Belt Plains ecoregion that
is 95% agricultural, planted in corn and
soybeans. Suspended sediments are
important most of the time; otherwise, algal
blooms predominate.
Study set up to evaluate nutrient reduction
due to best management practices (BMPs).
A shallow stream incised in the southern
Appalachians, located in a rapidly urbanizing
area and receiving effluent from wastewater
treatment plants. Good calibration data on
periphyton, invertebrates, and fish.
TSS is doubled to demonstrate
embeddedness and impact on zoobenthos; it
also decreases periphyton growth and speeds
up simulation.
Shallow, relatively low-nutrient river that
drains a predominantly forested watershed in
the Northern Lakes and Forests ecoregion.
Mile 72 is in the headwaters and drains
numerous small lakes.
A mesotrophic-eutrophic impoundment of the
Caddo River in the Ouachita Mountains
ecoregion. Most of the watershed is forested.
Study shows transient response to drowned
forest shortly after dam construction. Uses
sediment diagenesis model.
Mesotrophic end of large, deep lake in
Adirondacks.
Introduction of smelt changes food web and
favors diatom blooms.
18
-------
Guidance in AQUATOX Setup and Application
Study Name
Lower Boise R. ID
Seg_1-3.als
(Well-calibrated
nutrient)
Lower Boise R. ID
Seg_1-3Diel.als
MN Rivers, a Is
(Well-calibrated
nutrient)
Onondaga Lake NY
Sed Diagenesis.aps
(Well-calibrated
nutrient)
Rum R MN.aps
(Well-calibrated
nutrient)
Tenkiller Ferry Lake
OK.als
(Well-calibrated
nutrient)
Site
Type
River
River
Rivers
Lake
River
Reser-
\if\\f
VUll
Location
Boise ID
Boise ID
North,
central,
and
southern
MN
North of
Syracuse
NY
north of
St. Paul
MN
Eastern
i~ii^
\Jr\
Run time
(h:mm;
2.66 GHz
Quad CPU)
2:49 for
Syr
2:35 for
1 yr
0:40 for
2yr
0:01 for
2yr
(steady-
state
aerobic
layer)
0:13 for
2yr
0:51 for
2\/r
yr
Notes
Three upstream linked segments of the lower
Boise River, a shallow river with abundant
periphyton. Flow is controlled by upstream
releases and irrigation diversions. Two
segments are low-nutrient and the third
receives WTP effluent. Also has hourly
simulation to predict diel oxygen, which is
dominated by throughflow except during low
flow.
Crow Wing, Rum, and Blue Earth Rivers as
linked segments sharing the same parameter
set (Park et al. 2005).
"Lake Onondaga is arguably the most
polluted lake in the United States" from the
preface of a book (Effler 1996), which served
as the source of data for this study. The lake
has significant nutrient inputs from
wastewater treatment plant ("Metro") and
combined sewers, successive algal blooms,
hypoxia in hypolimnion, build-up of organic
sediments in bottom, and high mercury levels
and high salinity (the latter two are not
modeled at present). Run with sediment
diagenesis submodel (Di Toro 2001), with
steady-state aerobic layers.
Rum River is a shallow river, with moderate
nutrients and low suspended solids that
drains forests and dairy farms in the North
Central Hardwoods Forest ecoregion.
Linked segments representing a eutrophic
reservoir impaired by nutrients and organics,
especially from upstream poultry and swine
farms; there are excessive algae, and the
hypolimnion is anoxic during the summer.
However, it is one of the most important
recreational lakes in the state. The sediment
diagenesis submodel is necessary to simulate
the anoxic hypolimnion.
19
-------
Guidance in AQUATOX Setup and Application
Study Name
Cheney Res KS.aps
(roughly-calibrated
nutrient)
Lake Jesup FL.aps
(roughly-calibrated
nutrient)
Lake Pyhajarvi
Finland, aps
(roughly-calibrated
nutrient)
Farm Pond MO. aps
Farm Pond MO
Esfenval.aps
(Well-calibrated
mesocosm)
HCB Tank, aps
(Well-calibrated
microcosm)
Ponds MN
Chlorpyrifos.als
(Well-calibrated
mesocosm)
Expr Stream
Esfenval.aps
(Roughly-calibrated
mesocosm)
Site
Type
Reser-
voir
Lake
Lake
Pond
Aquari-
um
Enclos-
ures
Stream
Location
Near
Wichita
KS
North of
Orlando
SW
Finland
Central
MO
IVIW
Experime
ntal lab
Duluth
MN
Idaho
Run time
(h:mm;
2.66 GHz
Quad CPU)
0:01 for
15 mn
0:01 for
7yr
0:04 for
10yr
0:01 for
1 yr
0:01 for
1 yr
n-nn-ni
U.UU.U 1
for ? mn
I\JI £- 1 1 1 1 1
0:00:15
(perturbed
& control)
for 3 mn
0:15 for
10 mn
(perturbed)
Notes
City of Wichita acquires about 70 percent of
its daily water supply from Cheney Reservoir.
It is believed that objectionable tastes and
odors in Cheney Reservoir result from
cyanobacteria (blue-green algae), and there
is concern with proliferation of algal growth.
Both nutrients and suspended solids affect
algal growth and could be a concern fortaste-
and-odor issues (USGS 2008).
Lake Jesup is a large, shallow lake. Urban
storm water and agricultural runoff impact the
lake, as well as historic wastewater
discharge. Blooms of the invasive
cyanobacteria Cylindrospermopsis have been
increasing.
Mesotrophic boreal lake simulated by Anne
Makynen, Jyvaskyla University. The
difference between observed and simulated
phosphorus concentration corresponds
perfectly with the mass removed by fishing.
Generic pond built to USDA specifications.
Esfenvalerate loadings are the worst-case
scenario using runoff from an adjacent corn
field predicted by the PRZM model.
Represents an experiment in which an
aquarium tank containing macrophytes was
dosed with hexachlorobenzene (Gobas et al.
1991).
Pond enclosures dosed with 0.5, 6, and 32
ug/L chlorpyrifos at an EPA lab.
Based on Lower Boise River, this is a reach
with a volume of 400 m3 and a retention time
of 0.1 day. Set up for constant dosing for a
period of time. Study uses fixed time step so
it can be used for detecting lowest effect
levels.
20
-------
Guidance in AQUATOX Setup and Application
Study Name
Ohio stream
Chlorpyrifos
constant, aps
(Roughly-calibrated
mesocosm)
Ohio stream
Chlorpyrifos
pulsed. aps
Coralville Res IA
Dieldrin.aps
(Well-calibrated
chemical
fate/effects)
Coral Res IA
Sens, aps
Evers Res FL.aps
(Well-calibrated
chemical
fate/effects)
Site
Type
Stream
Reserv-
oir
Reserv-
oir
Location
North
1 NUI LI 1
central
OH
Vu/l 1
Near
Iowa City
IA
Bradento
nFL
Run time
(h:mm;
2.66 GHz
Quad CPU)
0:07 for
2yr
ry-n
\J . i \J
(perturbed)
0:10
(control)
for 9 yr
2:41 for
1 yr
rynr; fnr
\J ,\J\J 1 \Ji
5 yr
(perturbed)
Notes
A small creek draining agricultural area, used
as a generic study for various pesticides. One
study has constant exposure and other has
pesticide runoff during summer storms.
Coralville Reservoir is a large, shallow,
eutrophic reservoir. The drainage area is over
90% agricultural, especially corn. Runoff
carries large amounts of fertilizer, animal
wastes, silt, and pesticides into the reservoir.
By the early 1970's, the population of
largemouth bass and fish other than
buffalofish began to decline and residues of
the pesticides aldrin and dieldrin greatly
increased in tissue samples (Mauriello and
Park 2002).
Study set up for sensitivity analyses, 54
parameters.
A reservoir with increasing algal blooms,
treated with copper sulfate and hydrogen
peroxide. Simulated by Dr. Don Blancher,
Sustainable Ecosystem Restoration, LLC
21
-------
Guidance in AQUATOX Setup and Application
Study Name
Lake Ontario
PCBs.aps
(well-calibrated
chemical
fate/effects)
Skensved Denmark
TCE.aps
(well-calibrated
chemical
fate/effects)
Skensved Denmark
Atrazine.aps
Clear Lake CA
Fluridone.aps
(Roughly-calibrated
chemical
fate/effects)
East Fork Poplar
Creek TN PCBs.aps
(Roughly-calibrated
chemical
fate/effects)
Galveston Bay
TX.aps
(Roughly-calibrated
estuary)
Site
Type
Lake
Stream
Lake
Of rp o m
Oil Cdl 1 1
Estuary
Location
US-
Canada
Denmark
Central
CA
Oak
Ridge TN
Near
Houston
TX
Run time
(h:mm;
2.66 GHz
Quad CPU)
1 :55 for
4 yr
0:1 5 for
1 yr
(perturbed)
0:1 4 (both
perturbed
& control)
for 3 yr
1 :09 for
Syr
0-11 for
U.I 1 I\JI
3\ir
yr
Notes
Demonstration of bioaccumulation simulation
for numerous PCB congeners compared to
data of (Oliver and Niimi 1988)see also
(Burkhard 1998); this implementation uses
Barber (2003) k2 estimation.
Groundwater with trichloroethene from a
leaking tank is polluting a small stream.
Simon Funderand Dr. Ursula McKnight of the
Technical Univ. of Denmark, used AQUATOX
to show the impacts are probably
negligible. The same setup with atrazine
does show some direct and indirect
ecotoxicological effects. Concentrations are
near the no effects level so the option for a
fixed time step was chosen.
Roughly based on Clear Lake CA, a large,
shallow, eutrophic lake with cyanobacteria
blooms. Sonar (fluridone) has been used
successfully in Clear Lake to eradicate
Hydrilla. Although Hydrilla did not appear
until 1 994, the study is set up with 1 970-1 971
ecosystem data. Note that the fluridone
loadings are for 1971 but without bracketing
the simulation period with 0 loadings. The
fluridone loadings are repeated in each of the
three years. Also note that the entire lake
was modeled for convenience; in reality,
Hydrilla spread slowly, so only selected areas
needed to be treated. Our simulation is
therefore a worst-case scenario.
A small stream that drains the Y-12 plant at
Oak Ridge National Lab with PCB
contamination. The simulation runs for eight
years to illustrate gradual recovery.
A shallow, productive bay that receives runoff
from Central TX, including the Houston Ship
Channel.
22
-------
Guidance in AQUATOX Setup and Application
Study Name
Zollner Creek OR w
chlorpyr dieldrin-
pulse.aps
(roughly-calibrated
chemical
fate/effects)
Impact of
anadromous fish.aps
(Study intended for
teaching purposes)
Nockamixon Res
PA.aps
(Study intended for
teaching purposes)
Site
Type
Stream
1 flkp
Reserv-
oir
Location
Willamette
Valley
OR
Based on
Lake
George
MV
IN Y
eastern
PA
Run time
(h:mm;
2.66 GHz
Quad CPU)
0:01 for
Syr
0:00:30
for 2 yr
Notes
The watershed is >90% agricultural, with row
crops, orchards and vineyards, grain and
grass fields, and large poultry farms. It is a
USGS National Water Quality Assessment
Program (NAWQA) site, and also a principal
TMDL site. State criteria for chlorpyrifos and
legacy dieldrin were exceeded (Williams and
Bloom 2008).
Mesotrophic lake based on Lake George NY,
with Chinook salmon representing
anadromous fish. Nutrients are imported into
lake.
Heavily impacted reservoir downstream of the
Quakertown wastewater treatment plant
outlet.
23
-------
Guidance in AQUATOX Setup and Application
Bowers, L, and C. Bell. 2013. AQUATOX Modeling of Attached Algae: WERF Case
Study and Lessons Learned.
Burkhard, L. P. 1998. Comparison of Two Models for Predicting Bioaccumulation of
Hydrophobic Organic Chemicals in a Great Lakes Food Web. Environmental
Toxicology and Chemistry 17:383-393.
DeAngelis, D., and W. Mooij. 2003. In praise of mechanistically rich models. Pages 63-
82 in C. D. Canham, J. J. Cole, and W. K. Lauenroth, editors. Models in
Ecosystem Science. Princeton University Press, Princeton, NJ.
Effler, S. W., editor. 1996. Limnological and Engineering Analysis of a Polluted Urban
Lake. Springer, New York.
Funder, S. G. 2009. Risk Assessment of the Skensved A Field Site: Review and
Application of Surface Water Models, Bachelor's Thesis Technical University of
Denmark Lyngby Denmark.
Gobas, F. A. P. C., E. J. McNeil, L. Lovett-Doust, and G. D. Haffner. 1991.
Bioconcentration of Chlorinated Aromatic Hydrocarbons in Aquatic Macrophytes
(Myriophyllum spicatum). Environmental Science & Technology 25:924-929.
Mauriello, D. A., and R. A. Park. 2002. An Adaptive Framework for Ecological
Assessment and Management. Pages 509-514 in Integrated Assessment and
Decision Support. International Environmental Modeling and Software Society,
Manno Switzerland.
McKnight, U. S., S. G. Funder, J. J. Rasmussen, M. Finkel, P. J. Binning, and P. L.
Bjerg. 201 Oa. An integrated model for assessing the risk of TCE groundwater
contamination to human receptors and surface water ecosystems. Ecological
Engineering 36:1126-1137.
McKnight, U. S., J. J. Rasmussen, S. G. Funder, M. Finkel, P. L. Bjerg, and P. J.
Binning. 201 Ob. Integrated modelling for assessing the risk of groundwater
contaminants to human health and surface water ecosystems in 7th International
Groundwater Quality Conference, Zurich, Switzerland.
Oliver, B. G., and A. J. Niimi. 1988. Trophodynamic Analysis of Polychlorinated
Biphenyl Congeners and Other Chlorinated Hydrocarbons in the Lake Ontario
Ecosystem. Environ. Sci. Technol. 22:388-397.
Park, R. A., J. N. Carleton, J. S. Clough, and M. C. Wellman. 2009. AQUATOX
Technical Note 1: A Calibrated Parameter Set for Simulation of Algae in Shallow
Rivers. EPA-823-R-09-003, U.S. Environmental Protection Agency, Washington
D.C.
Park, R. A., J. S. Clough, M. C. Wellman, and A. S. Donigian. 2005. Nutrient Criteria
Development with a Linked Modeling System: Calibration of AQUATOX Across a
Nutrient Gradient. Pages 885-902 in TMDL 2005. Water Environment Federation,
Philadelphia, Penn.
24
-------
Guidance in AQUATOX Setup and Application
Park, R. A., and C. D. Collins. 1982. Realism in Ecosystem Models. Perspectives in
Computing 2:18-27.
Smith, D., and C. Fritsen. 2011. Modeling Nutrient Dynamics and Benthic Algal
Relationships on the South Fork Humboldt River, NV. ASCE.
Smith, D., J. Warwick, and C. Fritsen. In press. Modeling Nutrient Dynamics and
Benthic Algal Relationships on the South Fork Humboldt River, NV. Pages 1147-
1150 World Environmental and Water Resources Congress 2011.
Sourisseau, S., A. Basseres, F. Perie, and T. Caquet. 2008. Calibration, validation and
sensitivity analysis of an ecosystem model applied to artificial streams. Water
Research 42:1167-1181.
Taner, M. U., J. N. Carleton, and M. Wellman. 2011. Integrated model projections of
climate change impacts on a North American lake. Ecological Modelling
222:3380- 3393.
USGS. 2008. The Cheney Reservoir and Watershed Study.
25
------- |