Unitad States
Environmental Protaction
Agency
Permits Division
(EN--336)
Washington DC 2046C
OWEP 83-03
August 1983
Water
&EPA
Development of improved
Ammonia Fate IVSodeSs
for the State of Iowa
-------
DEVELOPMENT OF IMPROVED AMMONIA FATE
MODELS FOR THE STATE OF IOWA
Prepared by
Dr. Elizabeth Southerland
JRB Associates
A Company of Science Applications, Inc.
8400 Westpark Drive
McLean, Virginia 22102
Prepared for
U.S. Environmental Protection Agency
401 M Street SW
Washington, D.C. 20460
Bruce Newton, Project Officer
Permits Division (EH-33 6)
August 1983
-------
DISCLAIMER
Mention of trade names or commercial products does not constitute endorse-
ment or recommendation for use by the Environmental Protection Agency.
Similarly, publication of studies reporting better results from one model
vis-a-vis others does not constitute endorsement.
-------
FOREWARD
A major function of the Office of Water Enforcement and Permits is to
provide technical assistance to the Regional Offices and States in the
area of permit issuance. Water quality is an important aspect of permit
issuance particularly as requests for waivers under section 301(g) of the
Clean Water Act become more numerous. The project reported in this docu-
ment was undertaken at the request of Region VII and the State of Iowa.
It is being distributed so that other Regions and States may benefit from
its findings.
J. William Jordan
Chief
NPDES Technical Support Branch
-------
NOTE ON MODEL AVAILABILITY
The official EPA version of QOAL II is currently being revised, in part as
a result of this study. In the interim,.copies of the two models described
in this report are available from the EPA Modeling Center with the under-
standing that no user's assistance can be provided for the models. Users
should rely on "User's Manual for Modified Iowa DEQ Model" and "User's
Manual for Vermont QUAL-II" also available from the Modeling Center.
Inquiries should be addressed to Thomas Barnwell, USEPA Modeling Center,
College Station Road, Athens, Georgia 30613.
-------
ABSTRACT
The purpose of this study was to provide technical assistance to the
State of Iowa to improve that State's ability to develop water quality-based
NPDES permit limits through improved modeling techniques. A secondary purpose
of this case study was to encourage other States to evaluate the effectiveness
of their water quality models.
The project evaluated three models for assessing the impacts of BOD and
ammonia discharges on receiving water quality. The models included the cur-
rent Iowa model which is a simple Streeter-Phelps DO sag equation; a version
of the current Iowa model which was developed as part of this project to in-
clude algal photosynthesis-respiration effects on DO and preferential uptake
of ammonia; and a version of Qual-II developed by the State of Vermont. The
models were evaluated on the basis of predictive capabilities, data require-
ments, and costs. Predictive capability could not be fully examined because
of the lack of data available for the study site. Descriptions of the modi-
fied Iowa and Vermont Qual-II models are included in this report, and User's
Manuals are available under separate cover.
The improved ammonia predictions of the modified Iowa and Qual-II models
led to significantly different water quality-based permit limits than the
existing Iowa model. In contrast to the existing model, the modified Iowa and
Qual-II models predicted that under both summer and winter ice cover condi-
tions, discharges of ammonia at the study site POTW are limited by DO consi-
derations and not toxicity concerns. Toxicity determined permit limits on
total ammonia only during winter no ice conditions. While the permit limits
derived with the existing model required year around operation of nitrifica-
tion facilities, the modified Iowa and Qual-II models indicated that operation
was only needed during winter conditions.
The study concluded that computer costs are comparable for all three
models. Monitoring costs are comparable for both the existing and modified
Iowa models but considerably higher for the Qual-II model. The report re-
commended that the modified Iowa model be used for preliminary water quality
assessments to identify stream segements where the Qual-II model should be
used to develop final permit limits.
-------
TABLE OF CONTENTS
Page
1. INTRODUCTION 1-1
1.1 BACKGROUND 1-1
1.2 ORGANIZATION OF THE FINAL REPORT 1-2
2. DESCRIPTION OF MODIFIED IOWA DEQ MODEL 2-1
2.1 BACKGROUND 2-1
2.2 MODEL THEORY 2-2
2.2.1 Predictive Equations 2-2
2.2.2 Rate Constants 2-6
2.2.3 Temperature Effects 2-7
3. DESCRIPTION OF VERMONT QOAL-II MODEL 3-1
3.1 BACKGROUND . 3-1
3.2 MODEL THEORY 3-2
4. COMPARISON OF MODELS 4-1
4.1 LIMITATIONS IN AVAILABLE DATA 4-1
4.1.1 Modified Iowa Model — Assumptions for Missing
Data '. 4-6.
4.1.2 Vermont QUAL-II Model — Assumptions for Missing
Data 4-9
4.2 PREDICTIVE ACCURACY COMPARISONS 4-12
4.2.1 Existing Iowa Model Calibration/Verification 4-14
4.2.2 Modified Iowa Model Calibration/Verification 4-22
4.2.3 Vermont QUAL-II Model Calibration/Verification 4-26
4.3 COST COMPARISONS 4-37
4.3.1 Sampling Costs 4-41
4.3.2 Personnel Time 4-41
4.3.3 Computer Costs 4-42
4.4 RECOMMENDATIONS FOR MODEL USE ".....< 4-43
5. WASTE LOAD ALLOCATION ANALYSES 5-1
5.1 STATE NH. STANDARDS 5-2
5.2 PROPOSED EPA AMMONIA TOXICITY CRITERIA 5-3
5.3 CRITICAL FLOW CONDITIONS 5-3
5.4 RESULTS OF MODIFIED IOWA WLA 5-4
-------
5.4.1 Summer Waste Load Allocation 5-7
5.4.2 Winter Waste Load Allocation 5-?
5.4.3 Proposed EPA NH^ Toxicity Criteria 5-15
5.5 RESULTS OF VERMONT QUAL-II WLA 5-15
5.5.1 Summer Waste Load Allocation 5-18
5.5.2 Winter Waste Load Allocation 5-18
5.5.3 Proposed EPA NH^ Toxicity Criteria 5-25
5.6 POTENTIAL COST SAVINGS 5-25
6. STREAM MONITORING REQUIREMENTS." 6-1
6.1 SAMPLING LOCATIONS AND TIMING 6-1
6.2 CONSTITUENTS MONITORED 6-2
7. REFERENCES 7-1
-------
LIST OF TABLES AND FIGURES
TABLE
4.1 Comparison of Data Needs For Model Calibration 4-4
4.2 Input Data - Existing and Modified Iowa Models 4-10
4.3 Input Data - Vermont QUAL-II Model 4-13
4.4 Instream Process Parameter Values - Existing Iowa Model 4-15
4.5 Comparison of Relative Error Statistics For Each Model 4-18
4.6 Comparison of Standard Error Statistics For Each Model 4-18
4.7 Instream Process Parameter Values - Modified Iowa Model 4-23
4.8 Instream Process Parameter Values - Vermont QUAL-II Model 4-29
4.9 Calibrated Diurnal DO Fluctuations - Vermont QUAL-II Model 4-38
4.10 Required Input Data Cards 4-41
4.11 Job Costs For Model Calibration 4-43
5.1 Low Flows For Waste Load Allocation Analyses 5-4
5.2 Results of Waste Load Allocation Analyses -
State NHj Standards - Modified Iowa Model 5-5
5.3 Results of Waste Load Allocation Analyses -
Proposed EPA Toxicity Criteria - Modified Iowa Model 5-6
5.4 Results of Waste Load Allocation Analyses -
State NHj Standards - Vermont QUAL-II 5-16
5.5 Results of Waste Load Allocation Analyses -
Proposed EPA NH^ Toxicity Criteria - Vermont QUAL-II Model 5-17
5.6 Operation and Maintenance Costs For Nitrification Facilities 5-26
FIGURE
4.1 Map of Study Area 4-2
4.2 Outlet Creek - Calibration - Existing Iowa Model 4-16
4.3 North Raccoon River - Calibration - Existing Iowa Model 4-17
4.4 Outlet Creek - Verification - Existing Iowa Model 4-20
-------
4.5 North Raccoon River - Verification - Existing Iowa Model 4-21
4.6 Outlet Creek - Calibration - Modified Iowa Model 4-24
4.7 North Raccoon River - Calibration - Modified Iowa Model 4-25
i
4.8 Outlet Creek - Verification Modified Iowa Model 4-27
4.9 North Raccoon River - Verification - Modified Iowa Model 4-28
4.10 Outlet Creek - Calibration - QUAL-II Model - B0Du, DO and NH^ 4-30
4.11 Outlet Creek - Calibration - QUAL-II Model - NO-j, NH^, Org-N 4-31
4.12 Outlet Creek - Calibration - QUAL-II Model - Diss-P and Org-P 4-32
4.13 North Raccoon River - Calibration - QUAL-II Model - BOD ,
DO and NH^ U 4-33
4.14 North Raccoon River - Calibration - QUAL-II Model - NO,,
NH3 and Org-N 4-34
4.15 North Raccoon River - Calibration - QUAL-II Model - Diss-P
and Org-P 4-35
4.16 Outlet Creek - Verification - QUAL-II Model 4-39
4.17 North Raccoon River - Verification - QUAL-II Model 4-40
5.1 North Raccoon River - Waste Load Allocation -
Summer Annual 7Q10 - Modified Iowa Model 5-8
5.2 North Raccoon River - Waste Load Allocation -
Summer Seasonal 7Q10 - Modified Iowa Model 5-9
5.3 North Raccoon River - Waste Load Allocation -
Winter Annual 7Q10 - Modified Iowa Model - Ice Cover 5-11
5.4 North Raccoon River - Waste Load Allocation -
Winter Seasonal 7Q10 - Modified Iowa Model - Ice Cover 5-12
5.5 North Raccoon River - Waste Load Allocation -
Winter Annual 7Q10 - Modified Iowa Model - No Ice Cover 5-13
5.6 North Raccoon River - Waste Load Allocation -
Winter Seasonal 7Q10 - Modified Iowa Model - No Ice Cover 5-14
5.7 North Raccoon River - Waste Load Allocation -
Summer Annual 7Q10 - QUAL-II Model 5-19
5.8 North Raccoon River - Waste Load Allocation -
Summer Seasonal 7Q10 - QUAL-II Model 5-20
-------
5.9 North Raccoon River - Waste Load Allocation -
Winter Annual 7Q10 - QUAL-II Model - Ice Cover 5-21
5.10 North Raccoon River - Waste Load Allocation -
Winter Seasonal 7Q10 - QUAL-II Model - Ice Cover 5-22
5.11 North Raccoon River - Waste Load Allocation -
Winter Annual 7Q10 - QUAL-II Model - No Ice Cover 5-23
5.12 North Raccoon River - Waste Load Allocation -
Winter Seasonal 7Q10 - QUAL-II Model - No Ice Cover 5-24
-------
1. INTRODUCTION
1.1 BACKGROUND
The Iowa Department of Environmental Duality (DEQ) has evidence of
potentially toxic levels of NH^ occurring in some State streams. This has led
to the imposition of stringent NPDES permit limits for NH^ discharges and the
establishment of compliance schedules for the upgrading of numerous wastewater
treatment plants to include nitrification facilities. The U.S. Environmental
Protection Agency (EPA) will approve funding of these advanced treatment pro-
jects only if it believes that the need for such facilities exists and that
the environmental benefits justify the costs.
The Iowa DEQ uses a mathematical computer model developed by Stanley
Consultants to establish waste load allocations (1). Model calibration and
verification performed by TenEch Environmental Consultants in 1978 raised
questions about,the model's ability to predict accurately the assimilative
capacity of Iowa's receiving waters (2). As a result, JRB Associates, in
conjunction with a separate project being conducted in the State, evaluated
the performance of the Iowa DEQ model and outlined an approach to improve its
capabilities.
The ensuing technical assessment revealed several weaknesses in the
model's ability to predict ammonia (NH^) concentrations during winter and
summer conditions and to predict dissolved oxygen (DO) concentrations during
summer conditions. JRB concluded that the following modifications would sig-
nificantly improve the model's performance:
• Use of a more widely accepted temperature correction function for nitri-
fication in order to improve NH^ simulation in winter
• Development of an expression to account for the uptake of NH- by aquatic
plants in order to improve NH^ simulation in summer
• Development of a "photosynthesis minus respiration" (P-R) term in order
to improve DO simulation in summer
1-1
-------
Subsequently, JRB Associates was contracted by the EPA, Permits Division,
Washington, DC to evaluate alternative models, conduct case studies, and
provide recommendations to the State. The existing Iowa model was modified to
include the recommended improvements, and the more sophisticated Vermont
Qual-II model was investigated for use by the State. The project was intended
not only to assist the State of Iowa but also to serve as a case study to
encourage other States to evaluate the effectiveness of their models.
1.2 ORGANIZATION OF THE FINAL REPORT
Chapter 2 describes the modified Iowa DEQ model and comments on the limi-
tations of its simplified approach. The capabilities of the Vermont Qual-II
model are described in Chapter 3. In Chapter 4, the existing Iowa, modified
Iowa, and Qual-II models are compared in terms of simulation accuracy and
cost, and recommendations for their use are made. Chapter 5 summarizes the
waste load allocation analyses performed for the case study area. Stream
monitoring requirements for model calibration and verification are included in
Chapter 6. Chapter 7 contains the list of references for the report.
1-2
-------
2. DESCRIPTION OF MODIFIED IOWA DEQ MODEL
2.1 BACKGROUND
JRB Associates used two approaches In investigating improved waste load
allocation modeling techniques for the State of Iowa. The first approach was
to modify the existing model to include better algorithms for NH^ and DO pre-
dictions. The second approach was to evaluate the most comprehensive waste
load allocation model currently available. After evaluating several versions
of QUAL II, it was decided that the Vermont version was the most comprehensive
model available. The advantage of the first approach is that the model
remains simple and familiar — only minimal time would be required to train
State personnel in its use. The disadvantage of this approach is that the
revised model is greatly improved but still too simple to simulate accurately
all the water quality processes which determine receiving water concentrations
of NHj and DO.
The revised Iowa model was constrained to follow the structure of Che
existing Iowa model In simulating only DO, BOD, and NH^. In order to simulate
the algal growth, which has such an influence on NH^ and DO in Iowa's
nutrient-enriched streams, a model must Include, as a minimum', the reactions
and transport of nitrate (NO^) and inorganic phosphorus (PO^). The revisions
to the existing model attempt to circumvent the absence of NO^ and PO^ simu-
lation by having the modeler calculate algal growth rates outside the model,
using observed instream concentrations of NO^, NH^ and PO^. This calculated
value can then be entered in the model and used in subsequent algorithms that
compute photosynthetic production of oxygen and preferential uptake of
ammonia. The revised model cannot, however, overcome the need to simulate NO^
in order to ensure that calibrated nitrification rates do not result in exces-
sive conversion of NH^ to N0^> This deficiency in the model may lead to the
overestimation of nitrification rates and the underestimation of preferential
uptake of ammonia.
More detailed discussion of the limitations and appropriate use of the
modified Iowa model will be provided in Chapter 4. In summary, JRB recommends
that Iowa DEQ use this revised model as a screening tool to select those
2-1
-------
streams that appear to require advanced treatment facilities. The Vermont
QUAL II model should then be used to develop waste load allocations for these
potential AWT streams.
2.2 MODEL THEORY
The existing Iowa DEQ model is a steady-state model which predicts re-
ceiving water dissolved oxygen (DO), carbonaceous biochemical oxygen demand
(BOD), and ammonia (NH^) concentrations, assuming completely mixed conditions
in each stream reach. DO deficit is calculated using a modified Streeter-
Phelps equation which includes only instream microbial nitrification and bio-
logical oxidation. No benthic oxygen demands or photosynthesis-respiration
effects on DO are included. The rates of oxygen utilization due to both car-•
bonaceous and nitrogenous biochemical oxygen demand are expressed as first
order equations. Data supplied to the program as input include the first
order decay rate constants for BOD (K.) and ammonia (K ). The travel time and
1 n
reaeration rate constant (1^) are calculated within the model.
JRB modified the existing DEQ model by improving algorithms for NH^ and
DO and adding new ones to simulate previously neglected instream processes.
New algorithms were added to revise the first order nitrification rate con-
stant as a function of temperature and stream DO concentration, and to
simulate the uptake of NH^ by algae. Unlike nitrification, algal uptake of
involves ammonia removal without oxygen utilization; therefore, the simu-
lation of algal NH^ uptake will also affect the DO simulation. The relation-
ship between NH^ and nitrogenous BOD was changed to reflect a more reasonable
oxygen demand for NH^, and a photosynthesis-respiration term was added to
improve DO simulation in enriched streams. The following subsections describe
the revised modeling expressions in detail. JRB has also prepared a detailed
User's Manual for the modified model which is available under separate cover.
2.2.1 Predictive Equations
The equations which predict stream concentrations of carbonaceous BOD,
nitrogenous BOD, and DO deficit are listed below:
-K t
L(t) =» L e (1)
o
2-2
-------
where
L(t) 8 ultimate carbonaceous BOD at time t (mg/1)
Lq = initial ultimate carbonaceous BOD concentration (mg/1)
Kj =» carbonaceous deoxygenation rate constant (day *)
t = time of travel through reach (day)
"V
N(t) =• N e (2)
o
where
N(t) ° nitrogenous BOD concentration at time t (mg/1)
N ° Initial nitrogenous BOD concentration (mg/1)
0 -1
=» nitrogenous deoxygenation rate constant (day )
-K t -K_t -K t -K,t -K,t
D(t) =» K1Lq (e -e z (e -e L ) + DQe z + (3)
V*N
(R-P) (1-e 2 )
*2
where
D(t) « DO deficit at time t (mg/1)
= reaeration rate constant (day ^)
Dq = initial DO deficit (mg/1)
R 3 algal respiration oxygen utilization (mg/l/day)
P = photosynthetic oxygen production (mg/l/day)
Only the DO deficit equation differs from those in the existing model.
The equations used to calculate P and R are taken from the MS-ECOL fresh
(3)
water stream model. They are:
(OP)(GP - DPKCHLA) (4)
AP
2-3
-------
where
OP = mg oxygen produced by algae/mg algae
AP ' ug chlorophyll-a/mg algae
GP » algal growth rate (day *)
DP 3 algal death rate (day *)
CHLA ¦ chlorophyll a concentration (ug/1)
and
R = 0.025 CHLA (5)
The values of OP, AP, and DP must be selected from the literature by the
modeler. It Is essential that chlorophyll-a measurements be available from
the sampling data. If not, chlorophyll-a values must be estimated by field
observation and calibration, which detracts from the credibility of the
calibration. Since nitrate and inorganic phosphorus are not Included in the
model, the growth rate (GP) must be calculated outside the model using the
equation:
GP - u ( N )( P )( LI ) (6)
(N + V(P * Vai * *LI>
where
GP = local algal growth rate at 20°C (day *)
u =» maximum specific algal growth rate at 20°C (day-1)
N = sum of observed instream concentrations of NH^-N and NO^-N (mg/1)
= Michaelis-Menton half-saturation constant for total inorganic
™ N (mg/1)
P = observed instream concentration of inorganic phosphorus (mg/1)
= Michaelis-Menton half-saturation constant for inorganic P (mg/1)
LI = average incident light intensity (kcal/m^-sec)
2
=¦ Michaelis-Menton half-saturation constant for light (kcal/m -sec)
The values of OP and AP are input as constants for the entire stream, while
GP, DP, and CHLA are specified for each reach. The Michaelis-Menton constants
are used to adjust the maximum potential algal growth rate by the amounts of
2-4
-------
light, nitrogen, and phosphorus that can limit algal growth. Each constant is
the concentration at which that particular constituent limits algal growth to
half the maximal or "saturated" value.
The relationship between NH^-N and nitrogenous BOD (NOD) has also been
revised. In the existing model, the value of NOD is calculated by multiplying
the simulated instream NH^-N concentration by 4.57. This constant, which
represents the grams of oxygen required to convert one gram of ammonia to
nitrate, was calculated using the oxidation-reduction equations for nitrifi-
cation. As long as nitrification is not limited by insufficient inorganic
carbon, some oxygen will be obtained from inorganic carbon sources. When this
is taken into account, the synthesis-oxidation equations show that only A.33
grams of oxygen are required to convert one gram of ammonia to nitrate. Thus,
the revised model uses the constant 4.33 instead of 4.57.
Another new feature in the revised model is the simulation of algal up-
take of NH^-N. The instream concentrations of inorganic nutrients are reduced
by phytoplankton consumption. Phytoplankton requirements for inorganic N may
involve both ammonia nitrogen (NH^-N) and nitrate nitrogen (NO.J-N). The frac-
tion of consumed nitrogen which is NH^-N must be known if instream concentra-
tions of NH^-N are to be properly simulated. This fraction is the preferen-
tial NH^ uptake factor.
The amount of NH^-N removed by algae in a reach is calculated by the
following equation taken from the MS-ECOL model (3):
UP - (GP)(ANP)(NF)(CHLA)(e(GP"DP)(t)-e"(KN)(t)) (7)
(GP - DP +
where
UP a amount of NH^-N removed in a reach (mg/1)
ANP 3 mg N / ug chlorophyll-a
NF = fraction of NH^ preferred for algal uptake (0 - .9)
t = time of travel through reach (day)
The model calculates t internally, the value of ANP must be input by the
/
modeler, and the value of NF is calibrated. The model assumes that algal
2-5
-------
uptake of ammonia occurs uncil Che instream concentration of NH^-N is equal to
the inorganic N half-saturation constant If the instream concentration
of NH^ is below the half-saturation constant, the technical literature indi-
cates that algae will switch to NO^ as the sole source of nitrogen.
2.2.2 Rate Constants
The predictions of equations (l)-(3) depend upon the values of rate con-
stants Kj, K^, and , which represent the carbonaceous deoxygenation rate,
nitrogenous deoxygenation rate, and reaeration rate, respectively. The model-
er sets initial values of and K^; final values of these rate constants are
established through calibration and verification of the model. Unlike and
K^, the reaeration rate constant is calculated internally in the model by
the equation:
K, = (ICE)(C)(Ah) (8)
L t
where
ICE 3 factor reflecting effect of ice cover on reaeration rate
C = escape coefficient (ft *)
Ah = difference in water surface elevation between upstream and
downstream ends of reach (ft)
t = time of travel through reach (day)
The value of C is specified by the modeler, and the appropriate value is re-
fined through calibration and verification. TenEch recommends ICE factors
ranging from 0.05 for complete ice cover to 1.0 for zero cover (2).
The modified model alters the value of within each reach as a function
of the stream DO concentration. Because nitrifying bacteria are very sensi-
tive to DO levels, is reduced when low DO conditions exist. The following
equation, which accounts for the effect of DO concentrations on nitrification
rates, is taken from the Wisconsin Qual III Model (4):
PN = l-e"(,52)(DO) (9)
2-6
-------
where
PN » nitrification reduction factor
DO = dissolved oxygen concentration (mg/1)
The value input to the model is multiplied by the reduction factor PN. The
product is the value of which is used in equations (2) and (3).
2.2.3 Temperature Effects
Each rate constant is affected by changes in stream temperature. The
equations within the model which simulate the effect of temperature are:
' T-20
x 1.047 (10)
T-20
x 1.0159 (11)
K1(T)
s
o
CM
y:
D
K2(T)
° K2(20)
Vt)
o
-------
3. DESCRIPTION OF VERMONT QUAL-II MODEL
3.1 BACKGROUND
The QUAL-II model is an extension of the stream model, QUAL-I, developed
by F. D. Masch and Associates, and the Texas Water Development Board in 1971.
QUAL-I was originally designed to simulate the dynamic behavior of conserva-
tive materials, temperature, BOD and DO in streams.
Water Resources Engineers, Inc. (WRE) revised the QUAL-I model to include
the steady-state simulation of ammonia, nitrite, nitrate, dissolved phospho-
rus, algae and coliforms, as well as BOD and DO. This WRE QUAL-II model has
since undergone numerous revisions to Incorporate additional parameters and
changes in constituent interactions. The version of QUAL-II which JRB used on
the North Raccoon River is the Vermont version of QUAL-II.
The Vermont QUAL-II is basically a version developed by Meta Systems,
Inc. (June 1979), with later modifications by Walker (1980, 1981) and the
Vermont Department of Water Resources and Environmental Engineering. The
changes Meta Systems introduced in 1979 to USEPA's version of QUAL-II Includes
the following:
• Incorporation of the simulation of organic nitrogen
• Provision for algal uptake of ammonia as a nitrogen.source
• Steady state calculation of diurnal oxygen variations due to algal
photosynthesis and respiration based on diel curve analysis
• Changes in the model to delete the dynamic simulation of DO, thus
allowing dynamic simulation of temperature only
• Inclusion of dam reaeration
• Changes in the methods used to calculate the reaeration coefficient,
3-1
-------
To this Meta Systems version of QUAL-II, Vermont has added the simulation
of organic phosphorus and has modified the expressions for algal kinetics.
In order to develop a useful manual which incorporates all of the recent
changes made to the Vermont model, JRB had to consolidate the material from
four separate reports into one User's Manual. The new manual is provided
under separate cover as part of this project for the State of Iowa.
3.2 MODEL THEORY
The Vermont Qual-II is a steady state water quality model that can
simulate any completely mixed branching stream or river system. It can handle
multiple waste discharges, water withdrawals, tributary flows, and incremental
inflows, all of which are considered constant. Advection and dispersion are
the major pollutant transport mechanisms. Stream channel geometry and
velocity are held constant in each reach. The following constituents are
simulated in the model:
•
Dissolved oxygen (DO)
•
Carbonaceous biochemical oxygen demand (CBOD)
•
Nitrogenous biochemical oxygen demand (NBOD)
•
Temperature
•
Algae
•
Organic Nitrogen
•
Ammonia (NH^-N)
•
Nitrite (N02-N)
•
Nitrate (N03-N)
•
Organic phosphorus
•
Dissolved phosphorus
•
Conforms
•
Up to three conservative substances.
3-2
-------
The Vermont Qual-II model assumes first order decay for CBOD and also
allows for CBOD settling. The effects on DO of benthlc oxygen demand, algal
production, and nitrification are included in the model. The model- utilizes a
simplified nutrient-algal model with MONOD kinetics and includes the cycling
of nitrogen and phosphorus forms between inorganic and organic forms. Local
algal growth rates are computed as being limited by light and either nitrogen
or phosphorus, but not both. An ammonia preference factor is input to the
model so that algae can use ammonia and/or.nitrate as a source of nitrogen.
When oxygen is simulated, and photosynthesis is considered either by simu-
lating algae or by inputing P and R values, a diel curve analysis is performed
after the steady state solution is reached. This analysis estimates for each
computational element in the system the daily minimum and maximum DO concen-
trations resulting from photosynthesis and respiration.
3-3
-------
4. COMPARISON OF MODELS
Sampling data from a summer and a winter period were used Co compare the
predictive capabilities and costs of the existing Iowa, modified Iowa, and
Vermont QUAL-II models. Figure 4.1 Is a map of the study area, which extends
from the City of Storm Lake on Outlet Creek to Sac City on the North Raccoon
River. Only two point source discharges are Included In the modeling
analysis: the City of Storm Lake's wastewater treatment plant, a two-stage
trickling filter system, and the IBP Meat Packing Company's aerobic lagoon
system. Both of these treatment facilities discharge to Outlet Creek. This
study area was selected by the State of Iowa because additional nitrification
Is currently scheduled for the Storm Lake municipal plant based on waste load
allocation analyses performed using the existing Iowa DEQ model.
4.1 LIMITATIONS IN AVAILABLE DATA
The data from both the September 1981 and the January 1978 sampling
events were obtained from reports prepared by the Hygienic Laboratory at the
University of Iowa (6,7). Both the trickling filter plant .and the lagoon were
discharging during the September sampling, but only the Storm Lake facility
was active during the January period.
The September sampling involved grab samples at the two dischargers and
in the stream every six hours, beginning at noon on September 14 and ending at
6 a.m. September IS. Two more samples were then taken at 6 a.m. September 16
and 6 a.m. September 17. JRB calibrated the models using the mean values for
the September 14 to 15 period.
During the January sampling period, single grab samples were taken at the
three instream stations on January 16 and at the Storm Lake effluent on
January 13, 16, 17 and 18. The January 16 Instream values were used for model
verification purposes. However, since BOD was not measured in the effluent on
January 16, JRB was forced to use data from January 18 as the best estimate
for effluent quality preceding the stream sampling.
4-1
-------
p.
Out\ex
CfeeVt
North
Raccoon
River
(T) Municipal Plant Discharge — Storm Lake
(^Industrial Lagoon Discharge
^I^Municipal Plant Discharge — Sac City
• Sampling Stations
Miles
Figure 4.1
Map of Study Area
4-2
-------
The sampling methods employed for both September and January are not the
most appropriate for model calibration and verification. The Ideal monitoring
program consists of plug flow sampling, In which the sampling of each station
Is staggered In accordance with the travel time between stations. Any
tributaries or point sources contributing to the slug of water being sampled
should be monitored as the slug passes these points. In the case of the
Outlet Creek dischargers, situated at the headwaters of the modeled stream,
24-hour composite sampling of the effluent should have been performed one day
prior to the lnstream sampling In order to allow for the monitored discharge
to affect receiving water concentrations.
The other major deficiency In the calibration sampling program involved
the inclusion of a period in which the industrial lagoon was discharging. The
lagoon is not permitted to discharge during flows at or below the 7Q10 level.
Consequently, waste load allocation analyses are not supposed to include this
discharge. The lagoon discharge has such an overwhelming effect on receiving
water quality in the summer time that all the algal growth parameters and
related reaction rate coefficients established during model calibration will
reflect the impact of the lagoon and not be appropriate for use in the
wasteload allocation modeling.
Table 4.1 summarizes the gaps that existed in the sampling data available
compared to the data required for each model's calibration. The effluent data
was deficient from the very beginning because the sampling was performed on
the same day as lnstream sampling. The measured discharge had no time to
influence the receiving water. Monitoring data Indicate that effluent
concentrations at the two treatment plants varied significantly from day to
day, thus making it even more Important that the effluent be monitored one day
before lnstream sampling. Another significant limitation in the effluent data
was that it only included DO, BOD and NH^. No measurements were made of
chlorophyll-a concentrations or of any nutrients other than NH^.
Table 4.1 Indicates that appropriate lnstream data were also lacking.
The most serious deficiencies include the lack of chlorophyll-a data during
the summer, sampling period and the absence of flow, velocity and ice cover
4-3
-------
TABLE 4.1
COMPARISON OF DATA NEEDS FOR MODEL
CALIBRATION/VERIFICATION
REQUIRED SAMPLING DATA
EXISTING IOWA
AVAILABLE SAMPLING DATA
9/81 1/78
QUAL-II
MODIFIED IOWA
PARAMETER
POINT
SOURCE
DISCHARGE
STREAM
POINT
SOURCE
DISCHARGE
STREAM
POINT
SOURCE
DISCHARGE
STREAM
POINT
SOURCE
DISCHARGE
STREAM
POINT
SOURCE
DISCHARGE
STREAM
DO X
BODu X
ORGANIC N X
NH -N X
NO^-N X
NO^-N X
DISSOLVED X
P
ORGANIC P X
TEMPERATURE X
CHLOROPHYLL-A X
FLOW X
VELOCITY
CHANNEL GEO-
METRY
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
-------
measurements during Che winter period. In addition, there were no measure-
ments made of pollutant concentrations in the Outlet Creek waters above the
dischargers, or in the agricultural tile drainage that contributes much of the
incremental inflow to both Outlet Creek and North Raccoon River.
The lack of chlorophyll-a data was a critical deficiency in calibrating
both the modified Iowa and the Vermont QUAL-II models for the September sam-
pling period. The instream concentration of chlorophyll-a is used in both of
these models to calculate photosynthetic oxygen production, algal respiration,
and algal uptake of ammonia. During the September 1981 monitoring, the field
crew observed a dark green coloration extending from the lagoon discharge to
the confluence with the North Raccoon River. The lagoon was apparently seed-
ing Outlet Creek with algae. The wide range of diurnal DO values monitored
during this period also confirmed the presence of high Instream concentrations
of algae.
The absence of flow, velocity and ice cover data during the winter period
greatly hindered verification of all the models. The flow could be estimated
by proportioning USGS flows at Sac City by the appropriate drainage area, but
the missing velocity and ice cover data could not be easily replaced. Since
velocity determines travel time, poor estimation of this parameter has a
detrimental effect on the prediction of all water quality parameters. Without
documented estimates of the extent of ice cover, the reaeration capacity of
the receiving waters during the January sampling period is completely open to
question.
The lack of data on the background concentrations for Outlet Creek head-
waters and the Incremental inflow is not as critical as the other deficiencies
because these components make up such a small percentage of the total flow
during the sampling events. However, studies by EPA's OR&D which Involved
calibrating the HSPF model to Iowa watersheds indicate that nitrate concentra-
tions in tile drainage may be quite high. In the future BOD, DO and nutri-
ents should be measured in both headwaters and tile drainage during waste load
allocation monitoring studies.
4-5
-------
JRB attempted-to work around all these data limitations by supplementing
the sampling information with reasonable assumptions and appropriate values
from the technical literature. It is important to note, however, that the
lack of essential data is so significant that none of the models can be
considered calibrated or verified. JRB thus recommends that this project be
viewed only as a modeling exercise to investigate the relative advantages and
disadvantages of the existing Iowa, modified Iowa, and Vermont QUAL-II models.
The statistical tests used to compare the predictive accuracy of the models
are only included to demonstrate how proper goodness-of-fit measurements
should be conducted in addition to graphical comparisons of observed and
simulated data. Consequently, the waste load allocation analyses Included in
Chapter 5 are presented only as a theoretical demonstration of how improved
waste load allocation models can lead to cost savings in the design and
operation of treatment facilities.
The following two subsections of this chapter describe in detail the
various assumptions JRB made to fill in the data gaps and proceed with
the model comparisons.
4.1.1 Modified Iowa Model—Assumptions for Missing Data
As explained in Chapter 2, the modified Iowa model only simulates DO, BOD
and NH^. The modified Iowa model is not a deterministic model like QUAL-II,
which routes all forms of N and P and calculates the algal growth rate in each
reach, using simulated inorganic N and P concentrations. Instead the growth
rate calculation is performed by hand using observed instream inorganic N and
P data. The computed algal growth rate is then input to the model, where it
is used to calculate photosynthetic oxygen production and uptake of ammonia.
As a result, the calibration of the modified Iowa model is not affected by the
lack of effluent nutrient data.
Model calibration for the September period is, however, significantly
harmed by the absence of Instream chlorophyll-a measurements. These chloro-
phyll-a data are required for model calculations of photosynthetic oxygen pro-
duction, algal respiration oxygen demand, and preferential uptake of ammonia.
4-6
-------
Since the field crew observed that the discharge from the industrial lagoon
was seeding Outlet Creek and the North Raccoon River with algae, JRB referred
to the technical literature on aerobic ponds to estimate chlorophyll-a concen-
trations in the lagoon effluent. The literature suggests that algae in an
aerobic lagoon must be maintained at a level from 40 to 100 mg/1 for effective
operation (8). Assuming a typical value of SO ug chlorophyll-a/mg algae, the
IBP lagoon on Outlet Creek can be expected to contain chlorophyll-a concentra-
tions of at least 2,000 ug/1. In the absence of corroborating sampling data,
JRB assumed a discharge of less than half this amount (800 ug/1). The use of
this chlorophyll-a discharge resulted in an instream concentration after dilu-
tion of about 400 ug/1. This assumed instream concentration is reasonable,
based on the situation, and may be quite conservative.
Because chlorophyll-a is not routed through the system like NH^-N, BOD,
and DO, the modeler must specify the chlorophyll-a concentrations for each
reach. The value of 400 ug/1 was specified for all reaches of Outlet Creek
downstream from the lagoon discharge. The concentrations in the North Raccoon
reaches downstream from the river's confluence with Outlet Creek were based on
simple dilution calculations, considering the flow and chlorophyll-a concen-
trations contributed by both the Creek and the River. Headwater chlorophyll-a
concentrations for Station it 1 in the North Raccoon River above the confluence
wich Outlet Creek, and in Outlet Creek above the lagoon discharge, were esti-
mated at 25 ug/1, a reasonable value for enriched streams during summer condi-
tions. Measured concentrations of DO, BOD and the nutrients from Station 91,
above the confluence with Outlet Creek, were used to represent the headwaters
of the North Raccoon River. Since there were no samples taken in the Outlet
Creek headwater, JRB made^use of a combination of estimates from the State of
Iowa, and the measured values from the North Raccoon headwaters. The DO and
BOD values input to the model were those recommended by the Iowa DEQ, while
the nutrient concentrations came from Station #1. Groundwater inflow values
were obtained from TenEch estimates for the North Raccoon River (2). Neither
the headwaters or groundwater inflow values exerted any significant effect
on simulated instream concentrations since their contribution to the total
September flow is so small.
4-7
-------
The absence of flow and velocity data in the January 1978 sampling
prevented verification of the modified Iowa model. The flow values were esti-
mated using data from the Sac City USGS gaging station, located on the North
Raccoon River 30 miles below the confluence with Outlet Creek. The sum of the
point source discharges above the gage was subtracted from the measured flow
to estimate the natural inflow to the stream system. This inflow was then
proportioned by drainage area to Outlet Creek, North Raccoon River, and Big
Cedar Creek. The flows estimated for Outlet Creek and the North Raccoon River
were converted to velocities in the modified Iowa model with the Leopold-
Maddock equation. The empirical coefficients, "a" and "bH, used in the equa-
tion were those determined from the September dataset.
The extent of ice cover during the January 1978 sampling was not
documented in the monitoring report. Estimates from the field crew indicate
that the North Raccoon had complete ice cover and that Outlet Creek was at
least partially covered with ice during the sampling event. The modified Iowa
model retained the ICE variable from the existing model. The reaeration rate
constant (K^) is multiplied by the ICE factor to calculate a reduced reaera-
tion capacity due to ice cover conditions. In a modeling study for the Iowa
DEQ, TenEch Environmental Consultants found that calibrated ICE values for
complete ice cover ranged from 0.01 to 0.40, with 0.05 resulting in good DO
calibrations for the majority of streams (2). Due to the uncertainty of the
extent of ice cover during the January monitoring period, the value of ICE was
calibrated for Outlet Creek. ICE was assumed to be 0.05 for the North Raccoon
River, because the field crew recalled that complete ice cover existed at that
-time.
The estimation of headwater and groundwater quality was another consider-
ation in the January verification. The only existing headwater data was from
a grab sample taken approximately 8 miles upstream of the confluence of Outlet
Creek and the North Raccoon River. Those conditions were used as headwater
conditions for both the Creek and the River. Groundwater inflow values recom-
mended by TenEch for winter conditions for both the Creek and the River were
used for the January verification (2).
4-8
-------
Table 4.2 lists all the Input data developed for both the existing and
the modified Iowa model from the assumptions discussed in this section of the
report.
A.1.2 Vermont QUAL-II Model—Assumptions for Missing Data
The Vermont QUAL-II model must route all inorganic and organic forms of
i nitrogen and phosphorus in order to calculate the effects of algal growth on
DO and NH^. Therefore, it could not be calibrated for the September sampling
event without data on the effluent concentrations of organic nitrogen,
nitrate, dissolved phosphorus and organic phosphorus. In order to-estimate
these concentrations, JRB consulted the following sources:
• Literature values of typical nutrient concentrations in the effluent
of secondary treatment plants
• Effluent data from similar trickling filter plants and lagoons in Iowa
• Historical data from the Storm Lake POTW and IBP Meat Packing Lagoon.
After reviewing all the available data on trickling filter plants, JRB
decided to utilize the Storm Lake POTW data from April 1972 to estimate organ-
ic nitrogen, nitrate and phosphorus values for the September 1981 sampling
period. Secondary treatment plant data from the literature were used to dis-
tribute measured total P concentrations between organic and dissolved P. The
April 1972 data was selected because its concentrations of NH^ and BOD were
similar to those monitored In September 1981. Both sets of data were provided
by the State of Iowa and are listed below:
POTW
Effluent Concentrations September April
(mg/1) 1981 1972
BOD 30.0 35.0
DO u 2.23
NH -N 14.0 13.0
Organic N — 3.8
Nitrate — 2.9
Total Phosphorus — 11.0
4-9
-------
TABLE 4.2
INPUT DATA
EXISTING AND MODIFIED IOWA MODELS
Calibration
Verification
Input Data
Storm Lake POTW
Flow (cfs)
BOD (mg/1)
DO Fmg/1)
NH3-N (mg/1)
IBP Lagoon
Flow (cfs)
BOD (mg/1)
DO lfmg/1), .
NH3-N (mg/1)
Headwater
Flow (cfs)
BOD (mg/1)
DO If mg/1)
NH -N (mg/1)
Chlorophyll-a (ug/1)*
Inflow
Flow (cfs)
BOD (mg/1)
DO Ng/1)
NH -N (mg/1)
Chlorophyll-a (ug/1)*
Receiving Water
Chlorophyll-a (ug/1)*
Outlet Creek
2.74
30.0
2.2
14.0
2.46
19.1
14.9
1.04
0.10
6.00
8.2
0.02
25.00
0.0
6.0
2.0
0.02
0.0
400.00
North
Raccoon River
4.20
6.0
10.2
0.02
10.0
0.28
6.0
2.0
0.02
0.0
259.00
Outlet Creek
1.33
27.0
7.3
21.0
North
Raccoon River
0.0
0.10
1.5
6.5
0.29
5.0
0.06
2.0
2.0
0.05
0.0
5.0
4.20
1.5
6.5
0.29
5.0
0.10
2.0
2.0
0.05
0.0
5.0
*Chlorophyll-a values are not included in existing model.
4-10
-------
No historical data were found for the IBP lagoon or similar lagoon
systems in Iowa. To estimate lagoon effluent quality, JRB was forced to use
literature values for the mean concentration of organic N, NO^, and phosphorus
discharged from "moderately well-operated" secondary treatment systems. These
values compare to the measured effluent concentrations as follows:
Lagoon
Effluent
Concentrations IBP Literature
(mg/1) data Values
BOD 19.1 30-45
DO U 14.9
NH -N 1.04 1.0
Organic N — 1.0
NO -N — 16.9
Total Phosphorus — 7-9
The same assumptions regarding the lagoon discharge of chlorophyll-a, the
headwater concentrations, and groundwater inflow concentrations were made for
QUAL-II as were made for the modified Iowa model (See Section 4.1.1). Unlike
the modified Iowa model, the QUAL-II model does route chlorophyll-a through
the reach system. The dilution calculations performed externally for input to
the modified Iowa model are done internally by the QUAL-II 'model.
The January 1978 sampling program also failed to measure the concentra-
tions of organic N, N03, dissolved P and organic P in the POTW effluent. This
was not as critical a deficiency for model verification as it was for model
/~
calibration since algal growth does not affect receiving water quality during
the winter. The lack of NO^ effluent data did, however, prevent an evaluation
of the nitrification rate. NO^ concentrations should be simulated along with
NH^ in order to ensure that calibrated nitrification rates do not lead to an
excess of NO^. Because of the lack of effluent data and the insignificance of
winter algal growth, JRB decided not to simulate algae and phosphorus in the
verification model run. The Vermont QUAL-II program requires that all nitro-
gen components be simulated when any individual parameter is to be modeled.
All nitrogen constituents except NH, were set to zero, and QOAL-II was opera-
J 1
ted like the modified Iowa model to simulate only BOD, DO and NHy.
4-11
-------
The same assumptions regarding flow and velocity, ice cover factors, and
headwater/inflow concentrations were made for QUAL-II as were made for the
modified Iowa model (see Section 4.1.1).
Table 4.3 lists all the input data developed for the Vermont QUAL-II
model from the assumptions discussed in this section of the report.
4.2 PREDICTIVE ACCURACY COMPARISONS
The ultimate goal of the calibration process is to obtain an assessment
of stream system behavior that can be used to support decision making. Such
an assessment is realized by defining reaction rates which, when combined with
the measured inputs, yield constituent values that are consistent with the
observed data (9). As discussed in the preceding section, the lack of appro-
priate data prevented any real calibration of the models. JRB thus recommends
that the following discussion regarding calibration/verification results be
viewed only as a comparative exercise to illustrate the relative capabilities
of each model.
In order to evaluate the predictive accuracy of each model, JRB used
graphical comparisons of observed versus simulated data and statistical mea-
surements of goodness-of-fit. The use of statistical tests is not really
appropriate for this study because of the inadequacy of the input data. Des-
pite these problems, JRB decided to include the analyses in order to demon-
strate how visual comparisons should be supplemented with statistics during
the calibration/verification process. A number of statistical measurements
are recommended for use as goodness-of-fit indicators, including the student's
t-test comparison of means, regression analysis, and relative error calcula-
tions. Relative error and standard error measurements are used in this study.
Relative error is calculated for each parameter at each sampling station
using the following equation:
4-12
-------
Input Data
Storm Lake POTW
Flow (cfs)
BOD (mg/1)
DO ?mg/l)
NH.-N (og/1)
Chlorophyll-a (ug/1)
ORG-N (mg/1)
NO.-N (mg/1)
DISS-P (mg/1)
ORG-P (mg/1)
IBP Lagoon
Flow (cfs)
BOD (mg/1)
DO Fmg/D
NH -N (mg/1)
Chlorophyll-a (ug/1)
ORG-N (mg/1)
NO -N (mg/1)
DISS-P (mg/1)
ORG-P (mg/1)
Headwater
Flow (cfs)
BOD (mg/1)
DO IVg/1)
NH -N (mg/1)
Chlorophyll-a (ug/1)
ORG-N (mg/1)
NO -N (mg/1)
DISS-P (mg/1)
ORG-P (mg/1)
Inflow
Flow (cfs)
BOD (mg/1)
DO Fmg/1)
NH.-N (mg/1)
Chlorophyll-a (ug/1)
ORG-N (mg/1)
NO.-N (mg/1)
DISS-P (mg/1)
ORG-P (mg/1)
TABLE 4.3
INPUT DATA
VERMONT QUAL II MODEL
Verification
North
Outlet Creek Raccoon River
Calibration
North
Outlet Creek Raccoon River
2.74
30.0
2.2
14.0
0.0
3.8
2.9
8.9
2.1
1.33
27.00
7.3
21.00
0
0
0
0
0
19.1
14.9
1.04
800.00
1.01
16.9
7.86
0.14
0.10
6.0
8.2
0.02
25.0
0.93
0.05
0.06
0.06
4.20
6.0
10.2
0.02
10.0
0.93
0.05
0.06
0.06
0.10
1.5
6.5
0.29
5.00
0.34
7.8
0.12
0.0
4.20
1.5
6.5
0.29
5.00
0.34
7.8
0.12
0.0
0.0
6.0
2.0
0.02
0.0
0.93
0.05
0.06
0.06
0.28
6.0
2.0
0.02
0.0
0.93
0.05
0.06
0.06
0.06
2.00
2.00
0.05
0.00
0.34
7.8
0.12
0.0
0.10
2.00
2.00
0.05
0.00
0.34
7.8
0.12
0.0
4-13
-------
where
e^ - Relative error at the sampling station
= Measured value at the sampling station
° Simulated value at the sampling station
Standard error combines relative errors at all the stations for each
parameter using the following equation:
Relative error calculations are very sensitive when the concentrations of a
parameter are low. Small differences between observed and simulated values
can lead to relatively large relative error calculations. As long as the
modeler is aware of this problem, it should not lead to erroneous interpre-
tation of results.
4.2.1 Existing Iowa Model Calibration/Verification
Table 4.4 shows the final calibration dataset of instream process para-
meters for the existing Iowa model. Figures 4.2 and 4.3 contain the plots of
observed versus predicted concentrations of BODu> an<* DO for Outlet
Creek and the North Raccoon River during the September 1981 sampling event.
Tables 4.5 and 4.6 present the statistical measurements of goodness-of-flt. A
good calibration of BODu was achieved using reasonable deoxygenation rate
coefficients of 0.60 on Outlet Creek (where stream flow is almost 100 percent
effluent), and 0.05 on the North Raccoon River (which is over 10 miles down-
stream of the two discharges). The good results for BOD^ were expected since
JRB's technical evaluation of the existing model did not indicate any problems
with the BOD algorithms. In Contrast, observed NH^ concentrations were ap-
proximated only by using unusually high nitrification rates of 4.0 on Outlet
S
e
where
S ° Standard error of the estimate
n 3 Number of paired values
4-14
-------
TABLE 4.4
INSTREAM PROCESS PARAMETER VALUES
EXISTING IOWA MODEL
Calibration
Outlet
Creek
North
Raccoon
Verification
Outlet North
Creek Raccoon
Carbonaceous 0.60
Oxygenation Rate
Nitrogenous 4.00
Deoxygenation
Rate
Tsivoglou 0.32
Coefficient
ICE Factor 1.00
0- uptake by 4.57
Nitrification
(mg O^/mg NH3)
0.05
2.50
2.00
1.00
4.57
0.60
4.00
0.32
0.70
4.57
0.05
2.50
2.00
0.05
4.57
4-15
-------
30
28
26
24
22
20
18
16
14
12
10
8
6
4
2
0
Figure 4.2.
Outlet Creek
Calibration
September 1981 Data
Existing Iowa Model
0 Actual BODu
0 Actual NH3-N
A Actual DO
©
3 4 5 6
Stream Mile Below Trickling Filter Plant Discharge
4-16
-------
30
28
26
24
22
20
18
16
14
12
10
8
6
4
2
0
Figure 4.3.
North Raccoon River
Calibration
September 1981 Data
Existing Iowa Model
© Actual BODu
~ Actual NH3-N
A. Actual DO
Predicted" NH3-N
l " P l =t=—.
4 6 8 10 12 14 16 18 20 22 24
Stream Mile Below Confluence with Outlet Creek
4-17
-------
TABLE 4.5
COMPARISON OF RELATIVE ERROR
STATISTICS FOR EACH MODEL
CALIBRATION VERIFICATION
SAMPLING STATION SAMPLING STATION
PARAMETER MODEL 1 2 4 5 6 7 3 5 8
DO
BOD
Existing Iowa
Modified Iowa
QUAL II
Existing Iowa
Modified Iowa
QUAL II
0
.013
.161
.185
.259
.240
.365
1.11
1.55
0
.134
.098
.056
.008
.075
.022
.165
.011
001
.146
.164
.094
.006
.084
.050
.165
.138
143
.040
.070
.040
.033
—
.074
.697
2.09
143
.053
.057
.020
.016
—
.089
.697
2.09
123
.033
.032
0.0
.047
—
.004
.658
1.96
NH--N Existing Iowa 0.0 .742
Modified Iowa 0.0 1.10
QUAL II .176 1.03
.582 1.00
.766 .400
.239 2.85
.231
.025
.974
.655
.310
.172
.350
.285
.430
1.17
0.0
.126
1.03
.255
.441
TABLE 4.6
COMPARISON OF STANDARD ERROR
STATISTICS FOR EACH MODEL
CALIBRATION VERIFICATION
PARAMETER EXISTING IOWA MODIFIED IOWA QUAL II EXISTING IOWA MODIFIED IOWA QUAL II
DO 2.49 0.89 1.24 12.1 1.12 1.44
BOD 0.94 0.95 0.67 3.93 3.95 3.54
NH-^N 0.70 1.03 0.94 5.01 3.17 4.82
4-18
-------
Creek and 2.5 on the North Raccoon River. This force-fitting of NH^ values
resulted in a significant undersimulation of DO, even when extremely high
Tsivoglou C values were used to increase the reaeration rate beyond expected
levels. Previous modeling studies in Iowa determined that the Tsivoglou C
coefficient should be set at 0.11 when flows are below 15 cfs, as in this case
(2).
The unrealistic rate coefficients established during the calibration of
the existing model resulted in extremely poor model verification results. In
accordance with proper procedures, all rate coefficients were held constant
for both the calibration and verification model runs (see Table 4.4). The
only instream process parameters that had to change were ice cover factors
and, of course, flow, velocity and water temperature data. During the January
1978 verification period, the existing Iowa model significantly overpredicted
BOD^, NH^ and DO in both Outlet Creek and the North Raccoon River (see Figures
4.4 and 4.5). The poor B0Du simulation can be attributed more to the absence
of effluent BOD data for the day preceding instream sampling than to errors in
the calibrated deoxygenation rate coefficient or its temperature correction
function. Daily effluent data collected from the POTW on January 16, 17, and
18 showed wider variations in BOD^ concentrations (27 to 48 mg/1) than in NH^
(21 to 24 mg/1) or DO (7.3 to 9.1 mg/1). Consequently, the BOD simulation
should be the most affected by the lack of appropriate model input data. The
oversimulation of NH^, even though the nitrification rate was probably cali-
brated too high, can be explained by the inadequate temperature correction
function in the existing model. The 0°C water temperatures observed during
the January sampling event resulted in zero nitrification being simulated by
the model. The observed data Indicate, however, that nitrification was occur-
ring since concentrations of NH^ declined more than could be explained by
dilution processes. The existing model's calibration/ verification results
confirm JRB's conclusions that the existing model needs to be improved by: 1)
including algal uptake of NH^ in order to simulate the rapid decline in NH^
without exerting an unrealistic oxygen demand; 2) adding the effects of algal
photosynthesis and respiration on DO concentrations; and 2) using a more up-
to-date temperature correction function for nitrification rates.
4-19
-------
30
28
26
24
22
20
18
16
14
12
10
8
6
4
2
0
Figure 4.4.
Outlet Creek
Verification
January 1978 Data
Existing Iowa Model
Ice Cover
© Actual BODu
~ Actual NH3-N
A Actual DO
r&y
Predicted DO
®
~
~
T 1 1 1 1 1 1 1 1 1 1
1 23456789 10 11
Stream Mile Below Trickling Filter Plant Discharge
4-20
-------
30
28
26
24
22
20
18
16
14
12
10
8
6
4
2
0
Figure 4-5-
North Raccoon River
Verification
January 1978 Data
Existing Iowa Model
Ice Cover
© Actual BODu
0 Actual NH3-N
A Actual DO
Predicted DO
JL
Predicted BODu
©
~
~
©
Predicted NH3-N
i
4
I
8
12
14
16
10 12 14 16 18
.Stream Mile Below Confluence with Outlet Creek
20
22
24
4-21
-------
4.2.2 Modified Iowa Model Calibratlon/Verlfication
Table 4.7 shows Che final calibration dataset of instream process para-
meters for the modified Iowa model. Figures 4.6 and 4.7 contain the plots of
observed versus predicted concentrations of BODu, NH^-N and DO for Outlet
Creek and the North Raccoon River during the September 1981 sampling event.
Tables 4.5 and 4.6 present the statistical measurements of goodness-of-fit.
The same deoxygenation rates obtained through calibration with the existing
model were also used in the modified Iowa model but slightly lower BOD concen-
trations were predicted because JRB added a term to the modified model which
reduces nitrification rates when DO concentrations fall to low levels. In
both models, values of BOD, NOD and DO deficit are calculated for each stream
section. After these values are computed, the models check to see if the DO
deficit exceeds the saturated DO level. If it does, they internally reduce
the predicted removal of BOD and NOD to account for the insufficient oxygen
levels, and set the DO deficit equal to the DO saturation value. In the ex-
isting model, calculations for both sections of the reach between the munici-
pal plant and industrial lagoon discharges predicted an insufficient oxygen
supply, and ROD and NOD removals were reduced proportionately in both sec-
tions. In the modified model, an insufficient oxygen supply was predicted
only in the first section of the reach. Because the predioted oxygen concen-
tration entering the second section was zero, the value was greatly reduced
by the newly added nitrification rate equation. The reduction in the nitrifi-
cation rate caused so little oxygen demand that it did not result in a DO
deficit less than the DO saturation value. Consequently, the deoxygenation
rate for that section did not have to be reduced and the predicted BOD removal
was greater than the removal in the existing model.
In comparison with the existing model results, the NH^ calibration with
the modified model was almost as good on Outlet Creek and much better on the
North Raccoon River. This Improved NH^ calibration was achieved using more
reasonable nitrification rates because of the preferential uptake of ammonia
included in the modified model. The DO simulation on both Outlet Creek and
the North Raccoon River was so much better than the existing model results
that the reaeration coefficient did not have to be adjusted beyond the value
4-22
-------
TABLE 4.7
INSTREAM PROCESS
PARAMETER VALUES
MODIFIED IOWA MODEL
CALIBRATION
VERIFICATION
MODEL
NORTH
NORTH
PARAMETER
OUTLET RACCOON
OUTLET RACCOON
CREEK RIVER
CREEK RIVER
Algal Preference for NH-
0.85
0.85
0.0
0.0
mg N/ug Chlorophyll-a
0.007
0.007
0.007
0.007
Nitrogen Half-saturation
0.20
0.20
—
—
Constant (og/1)
Phosphorus Half-saturation
0.05
0.05
—
—
Constant (mg/1)
Light Half-saturation
0.0035
0.0035
—
—
Constant (kcal/m - s)
ug Chlorophyll-a/mg Algae
50.00
50.00
50.00
50.00
0- Production by Algae
Carbonaceous Deoxygenation Rate
1.63
1.63
1.63
1.63
0.60
0.05
0.60
0.05
Nitrogenous Deoxygenation Rate
2.50
1.00
2.50
1.00
Tsivoglou Coefficient
0.11
0.11
0.11
0.11
Maximum Algal Growth Rate
3.0
3.0
—
—
Algal Death Rate
0.24
0.24
—
—
Ice Factor
1.0
1.0
0.70
0.05
0. Uptake by Nitrification
4.33
4.33
4.33
4.33
(mg O./mg NH-)
Chloropnyll-a Concentration
400.0
259.0
5.0
5.00
(ug/1)
4-23
-------
30
28
26
24
22
20
18
16
14
12
10
8
6
4
2
0
©
Figure 4.6.
Outlet Creek
Calibration
September 1981 Data
Modified Iowa Model
0 Actual BODu
~ Actual NH3-N
A Actual DO
!Q,y,
©
A
0
Predicted NH3-N
T
7
~
~r
9
2
i
10
11
Stream Mile Below Trickling Filter Plant Discharge
4-24
-------
30
28
26
24
22
20
18
16
.14
12
10
8
6
4
2
0
Figure 4.7.
North Raccoon River
Calibration
September 1981 Data
Modified Iowa Model
© Actual BODu
0 Actual NH3-N
A Actual DO
~
®
Predicted DO
predicted Rnr>..
Predicted NH3-N
E |D |
6 8 10 12 14
Stream Mile Below Confluence with Outlet Creek
1
16
i
18
20
i
22
24
4-25
-------
recommended for these low flow conditions. The improved DO calibration is
attributable to the calibration of more realistic nitrification rates and the
inclusion of algal effects on DO.
The January 1978 model results shown in Figures 4.8 and 4.9 and analyzed
in Tables 4.5 and 4.6 confirm the calibration evidence that the modified
model's rate coefficients are more realistic and, therefore, more verifiable
than the existing model's values. The B0Du simulation is, of course, equally
as poor as the existing model predictions, but this can be explained in large
part by the lack of appropriate BOD effluent data. The DO and NH^ predictions
are significantly improved over the existing model results for the winter
period. The improvement in NH^ is due to the new temperature correction
function for nitrification, and the improvement in DO is attributable to the
more reasonable reaeratlon values (Tsivoglou's Coefficient).
4.2.3 Vermont QUAL-II Model Calibration/Verification
Table 4.8 shows the final calibration dataset of instream process parar-
meters for the Vermont QUAL-II model. Figures 4.10 through 4.15 contain the
plots of observed versus predicted concentrations of B0Du> NH^-N, DO, NO^,
Organic N, Dissolved P, and Organic P for Outlet Creek and the North Raccoon
River during the September 1981 sampling event. Tables 4.5 and 4.6 present
the statistical measurements of goodness-of-fit for BOD^, DO and NH^. It is
difficult to evaluate the calibration results of the QUAL-II model since the
effluent data needed for proper operation of the model was completely missing
and had to be estimated from literature values and historical data (see Sec-
tion 4.1.2). The modified Iowa model is designed to simulate the impact of
algae on DO and NH^ without having inorganic N and P effluent data since algal
growth is calculated outside the model using observed instream concentrations
of these constituents. The Vermont QUAL-II model, however, calculates local
algal growth rates within the program by routing point source discharges of N
and P through the system. The lack of accurate effluent data for the two
treatment facilities means that the QUAL-II calculations of photosynthetic DO
production and algal uptake of ammonia will be hindered. The lack of effluent
data also prevents any calibration of organic N, NO^, dissolved P or organic
P. Despite these problems, JRB attempted calibration of the Vermont QUAL-II
model in order to demonstrate its capabilities.
4-26
-------
28-
Figure 4.8.
Outlet Creek
Verification
January 1978 Data
Modified Iowa Model
Ice Cover
©
0
A
Actual BODu
Actual NH3-N
Actual DO
&-
>
4^
-r
2
I
4
5 6 7 8
Stream Mile Below Trickling Filter Plant Discharge
i
10
I
II
4-27
-------
30 -i
28 -
26 -
Figure 4.9.
North Raccoon River
Verification
January 1978 Data
Modified Iowa Model
Ice Cover
© Actual BODu
0 Actual NH3-N
A Actual DO
24 -
22
20 -
18
1
01
- 16 H
14 -
12 -
o
O
Q 10
4 -
2
Predicted BODu
Predicted DO
©
Predicted NH3-N
-1
24
10
1
14
16
i 8 10 12
Stream Mile Below Confluence with Outlet Creek
1
18
1
20
22
4-28
-------
TABLE 4.8
INSTREAM PROCESS
PARAMETER VALUES
VERMONT'QUAL II MODEL
CALIBRATION VERIFICATION
MODEL
NORTH
NORTH
PARAMETER
OUTLET
RACCOON
OUTLET
RACC0(
CREEK
RIVER
CREEK
RIVER
Algal Preference for NH^
0.89
0.89
0.89
0.89
mgN/mg Algae
0.09
0.09
0.09
0.09
Nitrogen Half-saturation
0.01
0.01
0.01
0.01
Constant (mg/1)
Phosphorus Half-saturation
0.01
0.01
0.01
0.01
Constant (mg/1)
Light Half Saturation
0.03
0.03
0.03
0.03
Constant (Langleys/min)
ug Chlorophyll-a/mg Algae
50.0
50.0
50.0
50.0
02Production by Algae
1.63
1.63
1.63
1.63
Carbonaceous Deoxygenation Rate
0.553
0.061
0.553
0.061
0.869
0.191
0.869
0.191
Nitrogenous Deoxygenation Rate
4.00
2.50
4.00
2.50
Tsivoglou Coefficient
0.11
0.11
0.11
0.11
Maximum Algal Growth
4.00
4.00
2.00
2.00
Algal Death Rate
0.24
0.24
0.24
0.24
Ice Factor
—
—
0.70
0.05
Manning's n
0.035
0.035
0.035
0.035
07 Uptake by NO,
3.33
3.33
3.33
3.33
(Oxidation toJNO,)
0„ Uptake by N02
1.00
1.00
1.00
1.00
(Oxidation to NO-)
0. Uptake by Algae
Algal Respiration Rate
2.00
2.00
2.00
2.00
0.275
0.275
0.275
0.275
Organic P Decay Rate
0.35
0.35
0.35
0.35
Organic P Settling Rate
0.00
0.00
0.00
0.00
Dissolved P Decay Rate
0.35
0.35
0.35
0.35
Organic N Decay Rate
0.01
0.01
0.01
0.01
BOD Settling Rate
0.00
0.00
0.00
0.00
Algal Settling Rate
0.50
0.50
0.50
0.50
Light Extinction Coefficient
0.01
0.01
0.01
0.01
4-29
-------
30
28
26
24
22
20
18
16
14
12
10
8
6
4
2
0
Figure 4.10.
Outlet Creek
Calibration
September 1981 Data
Qual-ll Model 0 Actual BODu
Stream Mile Below Trickling Filter Plant Discharge
4-30
-------
30
28
26
24
22
20
18
16
14
12
10
8
6
4
2
0
Figure 4.11.
Outlet Creek
Calibration
September 1981 Data
Qual-ll Model ® Actual NOyN
(3 Actual NH3-N
A Actual ORG-N
©
0
PredictedNOyN.
Predicted ORG-N
i
5
6
8
10
11
Stream Mile Below Trickling Filter Plant Discharge
4-31
-------
30
28
26
24
22
20
18
16
14
12
10
8
6
4
2
0
Figure
Outlet Creek
Calibration
September 1981 Data
Qual-ll Model @ Actua) diss-p
A Actual ORG-P
0
¦£"
Predicted Organic-P
A
T
2
~r
4
T"
6
"I
11
10
Stream Miles Below Trickling Filter Plant Discharge
.4-32
-------
Figure 4.13.
North Raccoon River
Calibration
September 1981 Data
Qual-ll Model
O Actual BODu
0 Actual NH3-N
A Actual DO
Predicted DO
Predicted NH3-N
ffl I lU tp 1 I I I I I I I 1
2 4 6 8 10 12 14 16 18 20 22 24
Stream Mile Below Confluence with Outlet Creek
4-33
-------
Figure 4.14.
North Raccoon River
Calibration
September 1981 Data
Quai-ll Model
20 -i
18-
16 -
0 Actual N03-N
0 Actual NH3-N
A Actual ORG-N
©
?ndicted NOn-to
~
A A
Predicted ORG-N
-SB:
~r
2
P H
Predicted NH3-N
6 8 10 12 14 16 18
Stream Miles Below Confluence with Outlet Creek
1
10
14
I
20
22
I
24
4-34
-------
Figure 4.15.
North Raccoon River
Calibration
September 1981 Data
Qual-ll Model
Stream Miles Below Confluence with Outlet Creek
4-35
-------
Deoxygenation rates which are slightly higher than those used in the
existing and modified Iowa models were calibrated for QUAL-II. This resulted
in slightly better B0Du simulations.
The same high nitrification rates that were used in the existing model
also had to be Included in the QUAL II model in order to simulate the rapid
decline in NH^ concentrations which was observed during the summer sampling
event. Unlike the modified Iowa model, the simulated uptake of NH^ by algae
was not adequate in QUAL-II to account for this decline. Since algal uptake
of NH^ is a function of algal growth, the differences in uptake simulation can
be explained by the growth rates. The Vermont QUAL-II model uses a more
sophisticated algal growth rate equation than the modified Iowa model. The
primary difference in the computed values lies in the light term used in the
two models. The modified Iowa model uses a simple Michaelis-Menton relation-
ship that divides measured light intensity by the sum of the half-saturation
constant and the light intensity. The QUAL-II model modifies the Michaelis-
Menton light expression by che fraction of daylight hours in the day, the
light extinction coefficient in the stream, and the mean stream depth. These
values reduce the Michaelis-Menton expression by approximately one-half. This
difference in algal growth calculations also explains the difference between
the Michaelis-Menton half-saturation constants calibrated for the modified
Iowa model and the QUAL-II model. Each constant represents the concentration
at which that particular factor limits algal growth to half the maximal or
"saturated" value. The nitrogen and phosphorus half-saturation constants in
QUAL-II had to be set as low as possible in order to counterbalance the re-
duced light term and simulate as much algal growth as possible. The QUAL-II
growth rate equation is more representative of actual conditions and could be
substituted for the simple modified Iowa term, but it would require the addi-
tional calculation of a light extinction coefficient from empirical estimates
of algal and non-algal extinction effects. It was JRB's opinion that the
extra effort was not worthwhile, given the limitations of the modified model
which does not even route nitrate, phosphorus, or chlorophyll-a concentrations.
The Vermont QUAL-II DO calibration was significantly better than the
existing model but not quite as good as the modified Iowa model results. The
improvement over the existing model is attributable to the inclusion of
4-36
-------
photosynthetic production of DO* Even Chough QUAL-II and the existing model
used the sane high nitrification rates, photosynthesis maintained observed DO
levels without requiring the unreasonably high reaeration rates used in the
existing model. The modified Iowa model had a slightly better DO simulation
than QUAL-II primarily because of the lower nitrification rates calibrated for
that model.
QUAL-II has a distinct advantage over the other two models in that it
predicts dally minimum and maximum values of DO, as well as mean DO concen-
trations. This is valuable Information because minimum DO concentrations have
greater implications for receiving water quality than calculated mean values.
Table 4.9 compares the simulated and observed diurnal DO fluctuations for the
September 1981 sampling event. QUAL-II reproduced the range of daily values
reasonably well, even though model input data were so faulty.
The January 1978 model results shown in Figures 4.16 and 4.17 and ana-
lyzed in Tables 4.5 and 4.6 confirm the calibration evidence that the QUAL-II
model performed much better than the existing model but not quite as well as
the modified Iowa model because of the lack of appropriate input data for the
calibration of rate coefficients. The BOD^ simulation during, the verification
period was slightly better than the other models because of the higher rate
coefficients calibrated for QUAL-II. The QUAL-II NH^ verification was much
better than Che existing model even Chough Che same nicrlflcaCion races were
used. This resulc can be aCCrlbuCed to the use of the Improved temperature
correction funcCion. The QUAL-II DO simulaCion in January was also much
beccer Chan Che exlsCing model because of Che more reallscic reaeration races
calibrated for this model.
4.3 COST COMPARISONS
The selection of a model cannot be based solely upon the model's predic-
tive accuracy but must also consider the cost of using the model. In this
study, the additional simulation capabilities of the QUAL II model must be
weighed against the additional costs of using the model. Incremental costs
will be required for the laboratory analysis of more parameters and the in-
creased personnel time needed to set up the run stream and calibrate the more
complex model.
4-37
-------
TABLE 4.9
CALIBRATED DIURNAL DO FLUCTUATIONS
VERMONT QUAL II MODEL
SAMPLING STATION DO DO DO
MIN MAX AVG
1 Observed 7.0 13.0 10.2
Simulated 9.6 10.7 10.2
2 Observed 3.5 10.8 7.0
Simulated 4.9 8.1 6.0
4 Observed 6.9 13.7 10.6
Simulated 7.6 11.4 8.9
5 Observed 6.9 15.3 11.0
Simulated 5.5 15.0 9.9
6 Observed 8.8 15.7 12.0
Simulated 7.1 17.2 12.0
7 Observed 7.6 16.4 11.8
Simulated 8.9 17.0 12.8
4-38
-------
Figure 4.16.
Outlet Creek
Verification
January 1978 Data
Qual-il Model
Ice Cover
© Actual BOOu
0 Actual NH3-N
A Actual DO
,0^
(**
0°
2
-r
3 4 5 6 7 8
Stream Mile Below Trickling Filter Plant Discharge
10
I
II
4-39
-------
Figure 4.17.
North Raccoon River
Verification
January 1978 Data
Qual-il Model
Ice Cover
0 Actual BODu
~ Actual NH3-N
A Actual DO
Predicted NH3-N
T
T
I
6 8 10 12 14 16 18
Stream Mile Below Confluence with Outlet Creek
i
14
20
22
24
4-40
-------
4.3.1 Sampling Cost3
It is evident from Table 4.1 that the QUAL-II model requires more samp-
ling data than the modified Iowa DEQ model. To properly calibrate the QUAL-II
model, organic and inorganic nitrogen and phosphorus data are required for
both the point source discharges and the stream samples. By contrast, the
modified Iowa model does not require such data for point source discharges and
uses only inorganic nutrient concentrations in the stream to calculate the
local algal growth rates. Both models require chlorophyll-a measurements, but
the modified Iowa model only needs stream samples.
The Hygienic Laboratory at the University of Iowa estimates laboratory
costs at $24, $16, and $10 per sample for the analysis of organic N, NO^, and
NH^; dissolved P and organic P; and chlorophyll-a, respectively. The existing
Iowa DEQ model requires only the NH^-N measurements which cost approximately
$8 per effluent or stream sample. The modified Iowa model requires NO^, NH^,
dissolved P, and chlorophyll-a instream measurements at a total cost of about
$34 per stream sample, and only $8 per effluent sample for NH^ analysis. The
QUAL II model requires all analyses for both stream and point source samples
at a total cost of $50 per effluent or stream sample.
4.3.2 Personnel Time
Because of the additional data cards and input formats used in the QUAL
II model, more personnel time is needed to establish the input data set.
Table 4.10 lists the number of data cards used by both models.
TABLE 4.10
REQUIRED INPUT DATA CARDS
QUAL II MODIFIED IOWA EXISTING IOWA
Cards per Reach
Cards per Headwater
Cards per Junction
Cards per Point Source
Other Required Cards
51
10
2
1
2
2
0
1
1
1
0
2
1
1
1
4-41
-------
The "Other Required Cards" in the QUAL II model include cards which identify
the parameters to be simulated, define the overall reach system, separate one
type of data from another, and establish factors relating to algae, nitrogen
and phosphorus. Clearly, the QUAL II model requires many more input state-
ments than the other models. The existing and modified Iowa models also main-
tain a consistent card format while the QUAL II model employs card formats
which vary throughout the data set. This increases the time required to code
the input data and increases the chance of making a mistake while entering the
data.
If the QUAL-II model is used to simulate algae and all forms of N and P,
calibration of this model will also require more personnel time than the
calibration of the modified Iowa model which predicts only DO, BODu, and NH^.
More rate constants must be calibrated for the more complex model including
settling rates, bed activity, and benthic inputs which are not considered in
the modified Iowa model.
4.3.3 Computer Costs
The difference in computer costs will be small in comparison to the
difference in personnel time costs, but the QUAL II model will be the more
expensive model. Computer costs can be broken down into connect costs, job
costs, and storage costs. Connect costs are the charges for accessing the
computer and are directly proportional to the time spent interacting with the
computer. Connect time is highly variable and difficult to estimate, depend-
ing on the editing required for each job. Consequently, these costs will not
be estimated here. It can be assumed, however, that connect charges will be
higher for QUAL-II since it has a longer, more complicated run stream and
requires more calibration runs. Storage costs include the expense of storing
the model, input data files, and output data files on a disk or tape. These
costs are virtually the same for each model. Job costs include the expense of
CPU time and EXCP's used by the model (one EXCP a one block of data trans-
ferred into or out of the system). Table 4.11 shows job costs for a calibra-
tion run of each model at rates charged by the USEPA National Computer Center.
4-42
-------
TABLE 4.11
JOB COSTS FOR MODEL CALIBRATION
CPU
CPU
EXCPS
TIME
COST
COST
MODEL
(SEC)
EXCPS
(0 $425/hr)
(@ $.74/1000)
JOB COST
Existing
Iowa
0.40
3
$0.05
$0.00
$0.05
Modified
Iowa
0.53
16
$0.06
$0.01
$0.07
QUAL II
1.86
63
$0.22
$0.05
$0.27
Despite the differences in cost, the total cost is so low that the QUAL II
model will not be prohibitively expensive when compared to the modified Iowa
model.
4.4 RECOMMENDATIONS FOR MODEL USE
After reviewing the above comparisons of the predictive accuracy and
costs of each model, JRB recommends that use of the existing Iowa model be
discontinued. This model has serious deficiencies which prevent the repre-
sentation of instream processes and the prediction of receiving water quality.
In its place, DEQ should consider using the modified Iowa model to identify
those stream segments which appear to require advanced treatment facilities in
order to meet State water quality standards. JRB recommends that the Vermont
QUAL-II model be used for the final determination of NPDES permit limits for
point source dischargers in these potential AWT streams. In comparing the
expenses of the modified Iowa model to the Vermont QUAL-II, the differences in
monitoring costs of $34 versus $50/sample and in computer costs of $0.07 ver-
sus $0.27/model run are not significant enough to merit any use of the modi-
fied Iowa model. The differences in personnel time are, however, quite sub-
stantial. Until DEQ staff become more familiar with QUAL-II, it will probably
be more efficient to use the modified Iowa model for most waste load alloca-
tions and save QUAL-II for AWT determinations.
4-43
-------
5. WASTE LOAD ALLOCATION ANALYSES
Waste load allocations are used to establish a quantitative relationship
between a particular discharge and its Impact on water quality. Knowledge of
such a relationship makes it possible to compare incremental changes in the
concentration of specific constituents in the receiving water. This capabil-
ity allows identification of the maximum effluent concentration that can be
discharged without violating a water quality standard. A determination can
then be made of a cost effective level of treatment in order to meet this
water quality standard (9).
The Iowa DEQ has developed future permit limits for the Storm Lake POTW
based on waste load allocations developed with the existing Iowa model. These
future permit limits assume a flow increase at the Storm Lake POTW from the
current level of 2.74 cfs to 6.67 cfs. In order to meet Iowa water quality
standards for DO and NH^, the current NH^ concentration must be reduced from
14 rag/1 to 3 mg/1 in the summer, and from 21 mg/1 to 5 mg/1 in the winter.
Year around operation of nitrification facilities will be needed to meet these
new NH^ limits.
Using both the modified Iowa and Qual-II models, JRB conducted additional
waste load allocation analyses for the Storm Lake POTW. Because inadequate
monitoring data prevented calibration and verification of these models, new
permit limits cannot be derived from these analyses. JRB conducted the study
only to illustrate the various permit limits that can be developed based on
different assumptions regarding the waste load allocation model, the critical
flow conditions, ice cover, and water quality standards. The study was done
with both seasonal and annual low flows to determine if appreciable cost
savings could be achieved through the use of seasonal flows with the State of
Iowa's seasonal NH^ standards. The winter allocations were modeled with and
without ice cover to Illustrate the impact of the reduced reaeration capacity
caused by ice cover. As a matter of interest, the summer and winter waste
load allocations were also performed with proposed new EPA ammonia toxicity
criteria.
5-1
-------
With a few exceptions, the waste load allocations were generated using
the sane rate coefficients which were derived for each model through the
calibration process. The exceptions refer to the algal growth and death
rates, and the instream concentration of chlorophyll-a. These parameters were
reduced for the waste load allocations because the industrial lagoon is not
permitted to discharge during low flow (7Q10) conditions, and thus was not
simulated in the waste load allocation.
Both models were calibrated for a summer situation when the lagoon was
discharging, seeding the stream with algae and influencing all rate constants.
With this significant discharge missing for waste load allocation purposes, it
was necessary to reduce the algal growth and death rates, as well as the
estimated instream concentration of chlorophyll-a.
Inflow values- for the waste load allocations were changed to reflect the
assumed seasonal or annual natural inflow, and point source water quality
parameters were iteratively adjusted to determine the maximum allowable
concentrations.
It should be noted that at the direction of the Iowa DEQ, the waste load
allocations were based only on the water quality of the North Raccoon River
and not on Outlet Creek, which is comprised of almost 100% effluent flow.
5.1 STATE NH3 STANDARDS
The current seasonal Total NH^ standards for the State of Iowa are 2 mg/1
in summer (April 1 - November 1) and 5 mg/1 in winter. These figures were
derived from the USEPA "Red Book" criterion of 0.020 mg/1 un-ionized ammonia
(10). The fraction of instream total ammonia which is un-ionized and,
therefore, toxic to fish and aquatic invertebrates, is related to the stream
pH and temperature. As pH and temperature increase, the fraction of total
ammonia which is un-ionized increases; hence, total ammonia is more toxic in
summer than in winter. This is reflected in the Iowa standards, which were
calculated by assuming typical seasonal values of pH and temperature.
5-2
-------
5.2 PROPOSED EPA AMMONIA TOXICITY CRITERIA
The USEPA has recently issued a draft document providing ammonia criteria
based on the toxicity of NH^ to fish and invertebrates (11). This document
proposes the use of an equation to calculate allowable ammonia concentrations
as a function of pH and temperature. Assuming typical Iowa conditions of 1°C
stream temperatures in winter and 26°C in summer with pH estimated at 7.5 for
both seasons, the new EPA function results in a winter ammonia criterion of
12.8 mg/1 total NH^ and a summer criterion of 3.91 mg/1 total NH^. Both
values are about double the numerical limits adopted by the State of Iowa.
The EPA draft criteria must go through a complete agency review and public
comment in the Federal Register before it is approved.
5.3 CRITICAL FLOW CONDITIONS
The streamflow used in a waste load allocation study is usually a low
flow because it provides protection during "worst case" conditions. The Iowa
DEQ uses the 7Q10 (one-in-ten-year, seven-consecutive-day) low flow as the
basis for its modeling. Calculation of low flows is based on daily streamflow
records collected by the U.S. Geological Survey at gaging stations throughout
the State.
Both the annual and seasonal low flows at the Sac City gaging station on
the North Raccoon River were calculated using the STORET computer package and
flow data for the years 1959 through 1976. The STORET program uses the log
Pearson type III distribution to calculate the 7Q10 values. The annual low
flow was determined using every month's gaging data; the summer low flow wad
determined using data from April through October; and the winter low flow was
determined using data from November through March.
Table 5.1 lists the annual and seasonal low flows generated by the STORET
program and shows the calculated proportion of point source discharge and
natural inflow.
5-3
-------
TABLE 5.1 LOW FLOWS FOR WASTE LOAD ALLOCATION ANALYSES
LOW FLOW
STREAM FLOW (cfs)
POINT SOURCE
DISCHARGE (cfs)
NATURAL
INFLOW (cfs)
Annual 7Q10
4.45
3.40
1.05
Summer 7Q10
7.77
3.40
4.37
Winter 7Q10
2.94
3.40
0
The point source discharge value is the sum of the average flows from the
treatment plants above the Sac City gaging station which discharge to the
North Raccoon River and its tributaries. The discharge of the lagoon system
on Outlet Creek was omitted because its operation is not permitted when stream
flows are at or below the 7Q10 level. The difference between the total stream
flow and the point source discharge was attributed to natural inflow, and was
proportioned to the river and its tributaries by drainage area.
The results of the seasonal low flow calculations were unexpected, with
the summer low flow exceeding the winter low flow. In most stream systems,
the lowest flows occur during dry summer conditions. In the case of the North
Raccoon River, however, the lowest flows are in winter, apparently due to
extensive freezing of the shallow river. By using seasonal low flows, the
predicted dilution capability of the river will be greater in summer and
smaller in winter than that predicted using the annual low flow.
5.4 RESULTS OF MODIFIED IOWA WASTE LOAD ALLOCATION
The results of the waste load allocations using the modified Iowa DEQ
model are provided in Tables 5.2 and 5.3, along with the future permit limits
based on the results of the existing Iowa DEQ waste load allocation model.
As can be seen in Tables 5.2 and 5.3, the waste load allocations based on
seasonal flows are virtually identical to those based on annual flows. This
is due to the low natUral inflows relative to the assumed discharge flow from
the Storm Lake POTW. Larger stream systems with much higher flows usually
5-4
-------
TABLE 5.2
RESULTS OF WASTE LOAD ALLOCATION ANALYSES
STATE NH STANDARDS
MODIFIED IOWA MODEL
EFFLUENT CONC. LIMITS
MODIFIED
IOWA MODEL
EXISTING
IOWA MODEL
SEASON
7Q10
ICE BODs NH.-N BOD,. NH,-N
COVER (mg/l) (mg/1) (mg/l) (mg/l)
Summer
Summer
Annual
Seasonal
No
No
30
30
13
13
30
30
3
3
Winter
Winter
Winter
Winter
Annual
Annual
Seasonal
Seasonal
Yes
No
Yes
No
25
25
25
25
5
8
5
8
25
25
25
25
5
5
5
5
5-5
-------
TABLE 5.3
RESULTS OF WASTE LOAD ALLOCATION ANALYSES
PROPOSED EPA TOXICITY CRITERIA
MODIFIED IOWA MODEL
EFFLUENT CONC. LIMITS
MODIFIED
IOWA MODEL
EXISTING
IOWA MODEL
SEASON
7010
ICE BOD NH -N BOD,. NH -N
COVER (mg/1) (mg/1) (mg/l) (mg/1)
Summer
Summer
Annual
Seasonal
No
No
30
30
13
13
30
30
Winter
Winter
Winter
Winter
Annual
Annual
Seasonal
Seasonal
Yes
No
Yes
No
25
25
25
25
5
21
5
20
25
25
25
25
5-6
-------
exhibit a wider variation in seasonal flows. Because of this wider variation,
significant differences in waste load allocations can result from using
seasonal flows.
5.4.1 Summer Waste Load Allocation
The summer waste load allocation results based on the State of Iowa's
water quality standards are shown in Table 5.2 and illustrated by Figures 5.1
and 5.2. These results indicate that the proposed NH^ discharge limits may be
too conservative and could be raised considerably without violating stream
water quality standards. Future permit limits developed by DEQ using the
existing model require that the Storm Lake POTW only discharge 3 mg/1 NH^ in
the summer. In contrast to this result, the modified model shows that the
discharge could be as high as 13 mg/1 before the nitrogenous oxygen demand
would result in a violation of the State Water Quality Standards for DO of 5
mg/1. The modified Iowa Model predicts that the allowable NH^ effluent
concentration in the summer is limited by the State of Iowa's water quality
standard for DO rather than the NH^ standard for toxicity.
The Storm Lake POTW is currently discharging an average NH^ concentration
of 14 mg/1 in the summer and an average BOD concentration of 30 mg/1. These
data are based on studies conducted by the University of Iowa's Hygienic Lab-
oratory for the Iowa DEQ. Using the results of the modified Iowa waste load
allocation model, the allowable effluent concentration of 13 mg/1 is almost as
high as the current level.
5.4.2 Winter Waste Load Allocation
In winter, the waste load allocations are dominated by the ice cover, as
is shown in Table 5.2. Ice cover on.streams during winter low flow conditions
reduces the surface area of the air-water interface through which reaeration
occurs. In order to represent this effect in the models, reaeration rates
must be multiplied by an ice cover factor. The waste load allocation analyses
presented here illustrate the Importance of correctly estimating the extent of
5-7
-------
30
28
26
24
22
20
18
16
14
12
10
8
6
4
2
0
Figure 5.1.
North Raccoon River
Waste Load Allocation
Annual 7Q10 - Summer
Modified Iowa Model
Predicted DO
DO Water Quality Standard
NH^N Water Quality Standard
Predicted NH3-N
6 8 10 12 14 16 18
Stream Mile Below Confluence with Outlet Creek
5-8
-------
30
28
26
24
22
20
18
16
14
12
10
8
6
4
2
0
Figure 5.2.
North Raccoon River
Waste Load Allocation
Seasonal 7Q10 - Summer
Modified Iowa Model
Predicted DO
DO Water Quality Standard
NH3-N Water Quality Standard
Predicted NH3-N
i
4
10
12
16
8 10 12 14
Stream Mile Below Confluence with Outlet Creek
i
18
i
20
22
24
5-9
-------
ice cover. The most reliable estimates are based on extensive field observa-
tions. On Iowa streams, TenEch calibrated a range of values from 0.01 to 0.40
for the ice cover factor representative of complete ice cover but recommended
the use of 0.05 (2). In order to test the effect of this recommended value on
waste load allocations, JRB performed the analyses once with an ICE factor of
0.05 (assuming complete ice cover) and again for 1.0 (assuming zero ice
cover).
The difference ice cover can make in the winter waste load allocations
can be seen by comparing winter conditions with 100 percent ice cover,
illustrated by Figures 5.3 and 5.4, with winter conditions and no ice cover,
illustrated by Figures 5.5 and 5.6. Under 100 percent ice cover conditions,
the DO in the North Raccoon River exhibited a continual decrease as the stream
flowed toward Sac City. JRS had to lower the NH^ discharge concentration
until the DO concentrations in the stream remained above the Iowa water
quality standard of 5 mg/1. The instream concentration of DO limited the
POTW' s effluent NH^ concentration when ice was present on the stream.
The modified Iowa model's waste load allocation with complete ice cover
predicts that, for both seasonal and annual low flows, the discharge levels of
NH^ would not differ from the current future permit limits determined using
the existing Iowa model. Both models estimated that a discharge of 5 mg/1
would be required to meet Iowa water quality standards.
Without the reduced reaeration imposed by ice cover, the modified model
Indicates that NH^ discharge concentrations could be increased somewhat, as
Illustrated in Figures 5.5 and 5.6. Under no ice conditions, Table 5.2 shows
that the Storm Lake POTW could discharge up to 8 mg/1 NH^. JRB's study
suggests that the State NH^ standard for toxicity limits the effluent NH^
concentrations in winter when no ice cover is present, in contrast to the ice
cover situation when the State DO standard limits the concentrations.
According to Iowa DEQ-staff, the Storm Lake POTW currently discharges an
average of 21 mg/1 of NH^ during the winter. Waste load allocation results
from both the existing and modified Iowa models, with and without ice cover
5-10
-------
30
28
26
24
22
20
18
16
14
12
10
8
6
4
2
0
Figure 5.3.
North Raccoon River
Waste Load Allocation
Annual 7Q10 - Winter
Modified Iowa Model
Ice Cover
NH3-N and DO Water Quality Standards
Predicted NHi-N
T
8
"T"
12
16
6 8 10 12 14 16 18
Stream Mile Below Confluence with Outlet Creek
20
1
22
i
24
5-11
-------
Figure 5.4.
North Raccoon River
Waste Load Allocation
Seasonal 7Q10 - Winter
Stream Mile Below Confluence with Outlet Creek
5-12
-------
30
28
26
24
22
20
18
16
14
12
10
8
6
4
2
0
Figure 5.5.
North Raccoon River
Waste Load Allocation
Annual 7Q10 - Winter
Modified Iowa Model
No Ice Cover
Predicted DQ
NH3-N and DO Water Quality Standards
Predicted NH3-N
1l 1 1 1 1 1 1 l l l l
2 4 6 8 10 12 14 16 18 20 22 24
Stream Mile Below Confluence with Outlet Creek
5-13
-------
30
28
26
24
22
20
18
16
14
12
10
8
6
4
2
0
Figure 5.6.
North Raccoon River
Waste Load Allocation
Seasonal 7Q10 * Winter
Modified Iowa Model
No Ice Cover
Predicted DO
NH3-N and DO Water Quality Standards
Predicted NHvm
1 1 1 1 1 1 1 1 1 1 1 1
2 4 6 8 10 12 14 16 18 20 22 24
Stream Mile Below Confluence with Outlet Creek
5-14
-------
assumptions, require increased nitrification at the plant in winter to produce
NHj concentrations considerably lover than existing levels.
5.4.3 Proposed EPA NH^ Toxicity Criteria
As shown on Table 5.3, the use of the calculated EPA summer criterion of
3.91 mg/1 total NH^ in place of the State standard of 2 mg/1 did not change
the permit limits developed from the modified Iowa model's waste load
allocation. This is because summer NH^ discharges are limited by the DO
standard, not by NH^ toxicity concerns.
The use of the calculated EPA winter criterion of 12.8 mg/1 total NH^ in
place of the State standard of 5 mg/1 also had no effect on the modified
model's winter waste load allocations when 100Z ice cover was assumed. Again
this is due to the fact that the reduced reaeration caused by ice cover makes
the DO standard the limiting parameter for effluent limits. This is not the
case, however, when zero ice cover is assumed. .Under the no ice cover
condition, the NH^ standard limits the permissible level of effluent NH^
concentration. Consequently, the higher EPA criterion results in a permit
limit of 21 mg/1 Total NH^ instead of the 8 mg/1 predicted with the modified
model using the State NH^ standard. The current winter discharge level of NH^
measured at the Storm Lake treatment plant is 21 mg/1, so that adoption of
EPA1s proposed ammonia toxicity standards could result in little change in the
level of treatment under \ri.nter conditions.
5.5 RESULTS OF VERMONT QUAL-II WASTE LOAD ALLOCATION
Results of the waste load allocations for the North Raccoon River using
the Vermont QUAL-II Model are provided in Tables 5.4 and 5.5. These permit
limits are more stringent than those generated using the modified Iowa model.
This is due to the differing rate constants each derived from the calibration
I
process, including different carbonaceous deoxygenation rates and
nitrification rates.
The waste load allocation results from the QUAL-II model, as from the
modified Iowa DEQ model, Indicate that allocations based on annual and season-
al low flows are almost identical. The one difference results from using
5-15
-------
TABLE 5.4
RESULTS OF WASTE LOAD ALLOCATION ANALYSES
STATE NH, STANDARDS
VERMONT QUAL II
EFFLUENT CONC. LIMITS
QUAL II
MODEL
EXISTING
IOWA MODEL
SEASON
7Q10
ICE
COVER
BOD,. NH,-N
(mg/1) (mg/1)
BOD,.
(mg/1)
NH.-N
(mg/1)
Slimmer
Summer
Annual
Seasonal
No
No
30 8.5
30 9.5
30
30
3
3
Winter
Winter
Winter
Winter
Annual
Annual
Seasonal
Seasonal
Yes
No
Yes
No
25 3.0
25 8.0
25 3.0
25 8.0
25
25
25
25
5
5
5
5
5-16
-------
TABLE 5.5
RESULTS OF WASTE LOAD ALLOCATION ANALYSES
PROPOSED EPA NH, TOXICITY CRITERIA
VERMONT QUAL II MODEL
EFFLUENT CONC. LIMITS
QUAL II
MODEL
EXISTING
IOWA MODEL
SEASON
7Q10
ICE
COVER
BOD,.
(og/1)
NH,-N
ui/D
BOD,.
(mg/1)
NH.-N
(mi/1)
Summer
Summer
Annual
Seasonal
No
No
30
30
8.5
9.5
30
30
3
3
Winter
Winter
Winter
Winter
Annual
Annual
Seasonal
Seasonal
Yes
No
Yes
No
25
25
25
25
3.0
16.0
3.0
15.0
25
25
25
25
5
5
5
5
5-17
-------
Che simmer seasonal low flow (Figure 5.8) which Is slightly higher Chan the
summer annual low flow (Figure 5*7) and, thus, allows Che POTW Co discharge 1
mg/1 more and still meet Iowa water quality standards.
5.5.1 Summer Waste Load. Allocaclon
Summer annual and seasonal NH^ effluenC concentrations of 8.5 and 9.5
mg/1 respectively, derived from the QUAL-II waste load allocaClons (Figures
5.7 and 5.8), and based on Iowa waCer quality standards, are higher than those
proposed by the existing Iowa DEQ model of 3 mg/1. They are, however, lower
than the modified Iowa model's predicted NH^ concentrations of 13 mg/1. This
is because the calibrated QUAL-II model had higher deoxygenatlon and nitrifi-
cation rate constants than the modified model. As a result, NH^ discharges
exert a greater oxygen demand In this model and permit limits must be more
stringent to meet Che Scace DO standard.
5.5.2 Winter Waste Load Allocation
The QUAL-II winter waste load allocations which assume 100 percent ice
cover require even lower permit limits than the existing or modified Iowa
models. Again this is because the higher nitrification and deoxygenatlon rate
constants in QUAL-II exert a greater oxygen demand. A comparison of Figures
5.9 and 5.10, with ice cover, to Figures 5.11 and 5.12, without ice cover,
graphically illustrates the profound effect ice cover exerts on the stream's
capacity to assimilate wastes. The simulated DO concentrations in Figures 5.9
and 5.10 exhibit a steady decline throughout the North Raccoon River to Sac
City. In contrast, Figures 5.11 and 5.12 Indicate that DO concentrations
remain high and that the NH^ standard is the limiting factor for point source
discharges. For winter conditions with no ice cover, the Vermont QUAL-II
results parallel those of the modified Iowa model. Both models suggest that,
with no ice cover, a higher concentration of NH^ (8 mg/1) can be discharged
from the POTW than with ice cover (3 mg/1). QUAL-II indicates, as do the
other two models, that increased nitrification is required at the Storm Lake
facility in order Co reduce effluenc concentrations from the current level of
21 mg/1 Total NH^.
5-18
-------
30
28
26
24
22
20
18
16
14
12
10
8
6
4
2
0
Rgure 5.7.
North Raccoon River
Waste Load Allocation
Annual 7Q10 - Summer
Qual-ll Model
NH3-N Water Quality Standard
2 4 6 8 10 12 14 16 18 20 22 24
Stream Miles Below Confluence with Outlet Creek
5-19
-------
30
28
26
24
22
20
18
16
14
12
10
8
6
4
2
0
Figure 5.8.
North Raccoon River
Waste Load Allocation
Seasonal 7Q10 - Summer
Qual-ll Model
NH3-N Water Quality Standard
Predicted NH3-N
1 I I I I I I i i i i 1
2 4 6 8 10 12 14 16 18 20 22 24
Stream Mile Below Confluence with Outlet Creek
5-20
-------
30
28
26
24
22
20
18
16
14
12
10
8
6
4
2
0
Figure 5.9.
North Raccoon River
Waste Load Allocation
Annual 7Q10 - Winter
Qual-ll Model
Ice Cover
5-21
-------
30
28
26
24
22
20
18
16
14
12
10
8
6
4
2
0
Figure 5.10.
North Raccoon River
, Waste Load Allocation
Seasonal 7Q10 - Winter
Qual-ll Model
Ice Cover
5-22
-------
Figure 5.11.
North Raccoon River
Waste Load Allocation
Annual 7Q10 - Winter
Qual-ll Model
30-i No Ice Cover
28 -
26 -
24 -
22 -
20 -
5-23
-------
30
28
26
24
22
20
18
16
14
12
10
8
6
4
2
0
Figure 5.12.
North Raccoon River
Waste Load Allocation
Seasonal 7Q10 - Winter
Qual-ll Model
No Ice Cover
NH3-N and 00 Water Quality Standards
"1 I I I I l I I I 1 1 l
2 4 6 8 10 12 14 16 18 20 22 24
Stream Mile Below Confluence with Outlet Creek
5-24
-------
5.5.3 Proposed EPA NH^ Toxicity Criteria
The Vermont QUAL-II waste load allocations were also conducted using the
proposed EPA NH^ toxicity criteria. Results of the waste load allocation
using this assumption are provided in Table 5.5. The outcome of this exercise
is comparable to that using the modified Iowa model.
Wasteload allocations for the summer low flows (annual and seasonal) did
not differ from those using the Iowa water quality standards. This is due to
the fact that, as mentioned above, the effluent concentration is controlled by
the Iowa DO standard and not the NH^ toxicity standard.
Winter wasteload allocation results for complete ice cover conditions are
also controlled by the DO level, and raising the NH^ standard does not affect
these results. Thus, using the Vermont QUAL-II Model with winter ice cover and
the proposed EPA NH^ criteria does not alter the allowable effluent concentra-
tions at the Storm Lake Plant from what they were using the Iowa NH^ standards.
When there is no ice cover, however, the winter waste load allocations
show that the NH^ effluent levels can increase to 16 and 15 mg/1 for annual
and seasonal low flows respectively, and still not exceed the proposed EPA NH^
toxicity criterion and the Iowa standard for DO. The limiting factor, in this
case, is the instream NH^ criterion. As explained earlier, the QUAL-II model
predicts lower permissible effluent concentrations of ammonia than the
modified model because the nitrification and deoxygenation rates calibrated
for this model are higher. While the modified model predicts that no decrease
in existing NH^ discharges are required for winter no ice conditions using the
EPA NH^ criterion, the QUAL II model indicates that current ?OTW effluent must
be decreased from 21 mg/1 to 16 mg/1 Total NH^.
5.6 POTENTIAL COST SAVINGS
The existing Iowa model for the North Raccoon River and its tributaries
Indicates the need for advanced wastewater treatment at the Storm Lake Treat-
ment Plant. Both the QUAL-II and modified Iowa model results suggest that
advanced treatment may not be required during the summer months. If this is
5-25
-------
confirmed through improved calibration/verification of the models, then
operation and maintenance cost savings could be realized.
The magnitude of the cost savings depends upon the nitrification method.
Table 5.6 lists operation and maintenance cost estimates taken from Culp,
based on an influent flow of 6.98 cfs and an influent NH^ concentration of 20
mg/1. These flow and NH^ assumptions approximate the future Storm Lake POTW
characteristics.
TABLE 5.6
OPERATION AND MAINTENANCE COSTS
FOR NITRIFICATION FACILITIES
Nitrification Method Estimated O&M Cost ($/yr)
Breakpoint Chlorination $190,000
Selective Ion Exchange $130,000
Ammonia Stripping $ 60,000
These figures are Fourth Quarter 1975 costs, and should be increased to
reflect 1983 costs. In addition, the estimates do not include the costs of pH
adjustment, which is necessary for ammonia stripping, and may be used in
breakpoint chlorination. Obviously, significant savings would occur If the
nitrification facility could be shut down for part of the year.
Instead of updating the existing municipal plant, the City of Storm Lake
plans to abandon the trickling filter plant and build an oxidation ditch to
achieve the future permit limits established by the existing waste load
allocation model. Because winter conditions control the capacity of the
oxidation ditch, the less stringent summer NH^ limits predicted in this study
would not result in capital cost savings. Significant cost savings under less
stringent summer NRj- limits could, however, result from reduced operation of
aeration equipment during the summer months.
5-26
-------
6. STREAM MONITORING REQUIREMENTS
This case study of Outlet Creek and the North Raccoon River illustrates
the importance of proper stream monitoring. For both calibration and veri-
fication periods, essential data regarding the quality of point source dis-
charges and the receiving water were missing and had to be estimated based on
engineering judgement. Consequently, the models were not convincingly cali-
brated or verified, and the resulting waste load allocation is open to ques-
tion. Model calibration and verification cannot be proven without sufficient
and accurate data.
The key questions in sampling are when to sample, where to sample, and
what data to acquire through sampling. Sampling stations must be located so
that the instream water quality processes are adequately represented. The
scheduling of sampling times at each station is also important. Finally, the
sampling must be complete enough to provide all required data. Most of the
water quality data can be analyzed in the laboratory, but onsite flow and
velocity measurements must be made at the time of sampling.
6.1 SAMPLING LOCATIONS AND TIMING
Sampling stations should bracket point source discharges and major
tributaries to the receiving stream. Samples should be taken upstream and
downstream of the discharge, and the discharge must also be sampled. Enough
samples must be taken between discharges so that the reaction rates of water
quality parameters can be determined. For example, In the likely event that a
DO sag occurs downstream of a discharge, enough sampling stations should be
established so that the location of the DO sag can be identified.
The most efficient sampling for calibration and verification data is that
performed during flows similar to the critical design flow for the waste load
allocation analysis. If flow.variable permitting is being considered, sam-
pling should be done when actual flows approximate the various seasons or
permitting periods. Before sampling is initiated, these flows should be
computed using the low flow statistics package and USGS streamflow database
6-1
-------
which are available on STORET. A plug flow sampling method should be used In
which samples at each station are taken in accordance with the travel time
between stations, so that the same slug of water is sampled as it flows
downstream. Any tributaries or point sources contributing to the slug of
water being sampled should be monitored as the slug passes these points.
Point source discharges in the headwaters of a stream should be monitored one
travel time before the receiving water is sampled. The sampling should be
arranged so that the calibration and verification periods have differing point
source loads, water temperature, and/or flows.
6.2 CONSTITUENTS MONITORED
Since travel time is a factor in many of the predictive equations in each
model, its accurate simulation is essential. The determination of travel time
is best accomplished through dye tracer studies, particularly under the low
flow conditions on many Iowa streams. At low velocities and shallow depths,
parts of the receiving stream can exhibit highly forked and meandering flow
regimes, with velocities much slower than the corresponding mean main-stem
velocity. Under low flow conditions, many Iowa streams are less than or equal
to one foot deep, with velocities less than 1 foot per second.
If dye tracer studies are not used, flow and velocity measurements should
be made at enough locations to represent the hydraulics properly. Using the
U.S.G.S. method, the stream should be divided into sections and the average
velocity in each section determined with a current meter. Ideally, these
measurements would be made over the entire range of flows considered for flow
variable permitting. Waste load allocations using differing flows could then
be developed with the appropriate velocities.
Table 4.1 lists the water quality parameters which must be monitored In
both the stream and point source effluent in order to calibrate and verify the
modified Iowa and Vermont QUAL-II models. The data limitations encountered in
the calibration and verification of the QUAL-II model Illustrate the impor-
tance of analyzing all parameters in the effluent as well as the receiving
stream. The water quality parameters that should be monitored for QUAL-II
6-2
-------
include organic N, NH^, NO^, dissolved P, organic P, BOD, DO, temperature and
chlorophyll-a. Organic N and P are the only instream parameters that do not
have to be monitored for the modified Iowa model. In the receiving stream,
sampling for DO should be conducted every 4 to 6 hours of the day in order to
monitor diurnal fluctuations. In both stream and point source analyses, the
BOD samples should be inhibited for nitrification so appropriate values of
carbonaceous BOD are available for the models. Sediment oxygen demand (SOD)
should also be measured if benthic deposits exert a significant influence on
stream vater quality. The modified Iowa model does not include benthic
effects, but the QUAL-II model Includes sediment oxygen demand as well as
benthic releases of NH^ and dissolved P.
6-3
-------
7. REFERENCES
1. Stanley Consultants, Inc. January 1975. Computer Modeling Techniques
for Waste Load Allocations. Prepared for Iowa Department of
Environmental Quality.
2. TenEch Environmental Consultants, Inc. July 1978. Waste Load Allocation
Verification Study; Final Report. Prepared for Iowa Department of
Environmental Quality.
3. Shindala, A., Corey, M. W., Hill,D. 0. May 1981. MS ECOL: An Opdated
Water Quality Model for Fresh Water Streams.
4. Wisconsin Department of Natural Resources. November 1979. QOAL-III
Water Quality Model Documentation.
5. Walker, W. W. December 1981. Q0AL2 Enhancements and Calibration to the
Lower Wlnooskl. Prepared for the Vermont Agency of Environmental
Conservation.
6. University of Iowa Hygienic Laboratory. September 1981. Outlet Creek -
North Raccoon River Water Quality Study. Prepared for the Iowa
Department of Environmental Quality.
7. University of Iowa Hygienic Laboratory. January 1978. Winter Water
Quality Survey of the North Raccoon River. Prepared for the Iowa
Department of Environmental Quality.
8. U.S. Environmental Protection Agency. March 1983. Draft, Technical
Guidance Manual for the Regulations Promulgated Pursuant to Section
301(g) of the Clean Water Act~
9. Driscoll, E. D., Mancini, J. L., and Mangarella, P. A. January 1981.
Technical Guidance Manual for Performing Waste Load Allocations, Book II
Streams and Rivers, Chapter 1 Biochemical Oxygen Demand/Dissolved Oxygen
and Ammonia Toxicity.
10. U.S. Environmental Protection Agency. September 1976. Quality Criteria
For Water. Water Quality Management Guidance No. EPA-440/9-76-023,
USEPA, Washington, D.C.
11. U.S. Environmental Protection Agency. January 1983. Ammonia: Water
Quality Criteria for the Protection of Aquatic Life and Its Uses.
7-1
------- |